【布莱克威尔哲学指南】分析哲学

【布莱克威尔哲学指南】分析哲学

ACompaniontoAnalyticPhilosophy\nBlackwellCompanionstoPhilosophyThisoutstandingstudentreferenceseriesoffersacomprehensiveandauthoritativesurveyofphilosophyasawhole.Writtenbytoday’sleadingphilosophers,eachvolumeprovideslucidandengagingcoverageofthekeyfigures,terms,topics,andproblemsofthefield.Takentogether,thevolumesprovidetheidealbasisforcourseuse,represent-inganunparalleledworkofreferenceforstudentsandspecialistsalike.Alreadypublishedintheseries15.ACompaniontoBioethicsEditedbyHelgaKuhseandPeterSinger1.TheBlackwellCompaniontoPhilosophyEditedbyNicholasBunninandEric16.ACompaniontothePhilosophersTsui-JamesEditedbyRobertL.Arrington2.ACompaniontoEthicsEditedbyPeterSinger17.ACompaniontoBusinessEthicsEditedbyRobertE.Frederick3.ACompaniontoAestheticsEditedbyDavidCooper18.ACompaniontothePhilosophyofScience4.ACompaniontoEpistemologyEditedbyW.H.Newton-SmithEditedbyJonathanDancyandErnestSosa5.ACompaniontoContemporaryPolitical19.ACompaniontoEnvironmentalPhilosophyPhilosophyEditedbyRobertE.GoodinandPhilipPettitEditedbyDaleJamieson6.ACompaniontoPhilosophyofMind20.ACompaniontoAnalyticPhilosophyEditedbySamuelGuttenplanEditedbyA.P.MartinichandDavidSosa7.ACompaniontoMetaphysicsEditedbyJaegwonKimandErnestSosaForthcoming8.ACompaniontoPhilosophyofLawandACompaniontoGenethicsLegalTheoryEditedbyJohnHarrisandJustineBurleyEditedbyDennisPatterson9.ACompaniontoPhilosophyofReligionACompaniontoAfrican-AmericanEditedbyPhilipL.QuinnandCharlesPhilosophyTaliaferroEditedbyTommyLottandJohnPittman10.ACompaniontothePhilosophyofACompaniontoAfricanPhilosophyLanguageEditedbyKwasiWireduEditedbyBobHaleandCrispinWright11.ACompaniontoWorldPhilosophiesACompaniontoAncientPhilosophyEditedbyEliotDeutschandRonBontekoeEditedbyMaryLouiseGill12.ACompaniontoContinentalPhilosophyACompaniontoEarlyModernEditedbySimonCritchleyandWilliamPhilosophySchroederEditedbyStevenNadler13.ACompaniontoFeministPhilosophyEditedbyAlisonM.JaggarandIrisACompaniontoPhilosophicalLogicMarionYoungEditedbyDaleJacquette14.ACompaniontoCognitiveScienceACompaniontoMedievalPhilosophyEditedbyWilliamBechtelandGeorgeEditedbyJorgeJ.E.Gracia,GregGrahamReichberg,andTimothyNoone\nBlackwellCompanionstoPhilosophyACompaniontoAnalyticPhilosophyEditedbyA.P.MARTINICHandDAVIDSOSA\nCopyright©BlackwellPublishersLtd2001Firstpublished200124681097531BlackwellPublishersInc.350MainStreetMalden,Massachusetts02148USABlackwellPublishersLtd108CowleyRoadOxfordOX41JFUKAllrightsreserved.Exceptforthequotationofshortpassagesforthepurposesofcriticismandreview,nopartofthispublicationmaybereproduced,storedinaretrievalsystem,ortransmitted,inanyformorbyanymeans,electronic,mechanical,photocopying,recording,orotherwise,withoutthepriorpermissionofthepublisher.ExceptintheUnitedStatesofAmerica,thisbookissoldsubjecttotheconditionthatitshallnot,bywayoftradeorotherwise,belent,resold,hiredout,orotherwisecirculatedwithoutthepublisher’spriorconsentinanyformofbindingorcoverotherthanthatinwhichitispublishedandwithoutasimilarconditionincludingthisconditionbeingimposedonthesubsequentpurchaser.LibraryofCongressCataloging-in-PublicationDataAcompaniontoanalyticphilosophy/editedbyA.P.MartinichandDavidSosa.p.cm.–(Blackwellcompanionstophilosophy)Includesbibliographicalreferencesandindex.ISBN0-631-21415-1(hb:alk.paper)1.Analysis(Philosophy)2.Philosophy,Modern–19thcentury.3.Philosophy,Modern–20thcentury.I.Martinich,Aloysius.II.Sosa,David,1966–III.Series.B808.5.C5552001146¢.4–dc2100-050770BritishLibraryCataloguinginPublicationDataACIPcataloguerecordforthisbookisavailablefromtheBritishLibrary.Typesetin10on12.5ptPhotinabyBest-setTypesetterLtd.,HongKongPrintedinGreatBritainbyMPGBooksLtd,Bodmin,CornwallThisbookisprintedonacid-freepaper.\nContentsListofContributorsviiiIntroduction1A.P.MARTINICH1GottlobFrege(1848–1925)6MICHAELDUMMETT2BertrandRussell(1872–1970)21THOMASBALDWIN3G.E.Moore(1873–1958)45ERNESTSOSA4C.D.Broad(1887–1971)57JAMESVANCLEVE5LudwigWittgenstein(1889–1951)68P.M.S.HACKER6RudolfCarnap(1891–1970)94SAHOTRASARKAR7KarlPopper(1892–1994)110W.H.NEWTON-SMITH8GilbertRyle(1900–1976)117AVRUMSTROLL9AlfredTarski(1902–1983),AlonzoChurch(1903–1995),andKurtGödel(1906–1978)124C.ANTHONYANDERSON10FrankP.Ramsey(1903–1930)139BRADARMENDT11CarlG.Hempel(1905–1997)148PHILIPKITCHERv\nCONTENTS12NelsonGoodman(1906–1998)160ISRAELSCHEFFLER13H.L.A.Hart(1907–1992)169SCOTTSHAPIRO14CharlesStevenson(1908–1979)175JAMESDREIER15W.V.Quine(1908–2000)181PETERHYLTON16A.J.Ayer(1910–1989)205T.L.S.SPRIGGE17J.L.Austin(1911–1960)218JOHNR.SEARLE18NormanMalcolm(1911–1990)231CARLGINET19WilfridSellars(1912–1989)239JAYF.ROSENBERG20H.P.Grice(1913–1988)254STEPHENNEALE21G.H.vonWright(1916–)274FREDERICKSTOUTLAND22RoderickChisholm(1916–1999)281RICHARDFOLEYANDDEANZIMMERMAN23DonaldDavidson(1917–)296ERNESTLEPORE24G.E.M.Anscombe(1919–2001)315ANSELMMÜLLER25R.M.Hare(1919–)326WALTERSINNOTT-ARMSTRONG26P.F.Strawson(1919–)334P.F.SNOWDON27PhilippaFoot(1920–)350GAVINLAWRENCE28RuthBarcanMarcus(1921–)357MAXCRESSWELL29JohnRawls(1921–)361NORMANDANIELSvi\nCONTENTS30ThomasS.Kuhn(1922–1996)371RICHARDGRANDY31MichaelDummett(1925–)378ALEXANDERMILLER32HilaryPutnam(1926–)393JOHNHEIL33DavidM.Armstong(1926–)413FRANKJACKSON34NoamChomsky(1928–)419PETERLUDLOW35RichardRorty(1931–)428MICHAELWILLIAMS36JohnR.Searle(1932–)434A.P.MARTINICH37JerryFodor(1935–)451GEORGESREY38SaulKripke(1940–)466DAVIDSOSA39DavidLewis(1941–)478ROBERTSTALNAKERIndex489vii\nContributorsC.AnthonyAndersonProfessorofPhilosophy,UniversityofCalifornia,SantaBarbaraBradArmendtAssociateProfessorofPhilosophy,ArizonaStateUniversityThomasBaldwinProfessorofPhilosophy,UniversityofYork,EnglandMaxCresswellProfessorofPhilosophy,VictoriaUniversityofWellington,NewZealandNormanDanielsProfessorofPhilosophyandGoldthwaiteProfessorofRhetoric,TuftsUniversityJamesDreierAssociateProfessorofPhilosophy,BrownUniversitySirMichaelDummettWykehamProfessorofLogicEmeritus,UniversityofOxfordRichardFoleyProfessorofPhilosophyandDeanoftheFacultyofArtsandSciences,NewYorkUniversityCarlGinetProfessorEmeritusofPhilosophy,CornellUniversityRichardGrandyCarolynandFredMcManisProfessorofPhilosophy,RiceUniversityP.M.S.HackerFellowofSt.John’sCollege,UniversityofOxfordJohnHeilProfessorofPhilosophy,DavidsonCollegePeterHyltonProfessorofPhilosophy,UniversityofIllinois,Chicagoviii\nCONTRIBUTORSFrankJacksonProfessorofPhilosophyinthePhilosophyProgram,ResearchSchoolofSocialSciences,andDirectoroftheInstituteofAdvancedStudies,AustralianNationalUniversityPhilipKitcherProfessorofPhilosophy,ColumbiaUniversityGavinLawrenceAssociateProfessorofPhilosophy,UniversityofCalifornia,LosAngelesErnestLePoreProfessorofPhilosophyandDirectoroftheCenterforCognitiveScience,RutgersUniversityPeterLudlowProfessorofPhilosophy,StateUniversityofNewYork,StonyBrookA.P.MartinichRoyAllisonVaughanCentennialProfessorofPhilosophy,ProfessorofHistory,UniversityofTexas,AustinAlexanderMillerSeniorResearchFellow,CardiffUniversityAnselmWinfriedMüllerProfessorofPhilosophy,UniversityofTrier,GermanyStephenNealeProfessorofPhilosophy,RutgersUniversityW.H.Newton-SmithFairfaxFellow,JowettFellow,JowettLecturerandTutorinPhilosophy,BalliolCollege,UniversityofOxfordGeorgesReyProfessorofPhilosophy,UniversityofMaryland,CollegeParkJayF.RosenbergTaylorGrandyProfessorofPhilosophy,UniversityofNorthCarolina,ChapelHillSahotraSarkarAssociateProfessorofPhilosophy,UniversityofTexas,AustinIsraelSchefflerProfessorEmeritusofPhilosophyandProfessorEmeritusandDirector,PhilosophyofEducationResearchCenter,GraduateSchoolofEducation,HarvardUniversityJohnR.SearleMillsProfessorofPhilosophy,UniversityofCalifornia,BerkeleyScottShapiroAssistantProfessorofLaw,BenjaminCardozoSchoolofLaw,YeshivaUniversityWalterSinnott-ArmstrongProfessorofPhilosophy,DartmouthCollegeix\nCONTRIBUTORSP.F.SnowdonLecturerinPhilosophyandFellow,ExeterCollege,UniversityofOxfordDavidSosaAssociateProfessorofPhilosophy,UniversityofTexas,AustinErnestSosaRomeoEltonProfessorofNaturalTheology,BrownUniversity,andProfessorofPhilosophy,RutgersUniversityT.L.S.SpriggeHonoraryFellow,UniversityofEdinburghRobertStalnakerProfessorofPhilosophy,MassachusettsInstituteofTechnologyFrederickStoutlandProfessorEmeritusofPhilosophy,St.OlafCollege,andPermanentVisitingProfessor,UniversityofUppsalaAvrumStrollResearchProfessor,UniversityofCalifornia,SanDiegoJamesvanCleveProfessorofPhilosophy,BrownUniversityMichaelWilliamsProfessorofPhilosophy,JohnsHopkinsUniversityDeanZimmermanAssociateProfessorofPhilosophy,SyracuseUniversityx\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd2001IntroductionA.P.MARTINICHThoughanalyticphilosophywaspracticedbyPlatoandreinvigoratedinthemodernerabyRenéDescartesandThomasHobbesamongothers,weareconcernedwithitonlyinitstwentieth-centuryforms.Assuch,itwasrevivedintwocenters,GermanyandEngland.InGermany,GottlobFregewasexploringthefoundationsofmath-ematicsandlogic.Hiseffortsintroducednewstandardsofrigorthatmadetheirwayintoanalyticphilosophygenerally,throughtheworkofBertrandRussellandLudwigWittgenstein.Hisdiscussionsofthenatureoflanguageandreasoninghavealsobecomepowerfultoolsinthehandsoflaterphilosophers.AmongFrege’smanybooksandarticles,theGrundgesetze,Begriffsschrift,“OnSenseandReference”(“ÜberSinnundBedeutung,”1892)and“Thoughts”(“Gedanken,”1918)standoutasespeciallysignificant.DuringaboutthesameperiodinEngland,G.E.Mooreledthewayinopposingthethen-dominantphilosophyofBritishidealism.While“TheNatureofJudgment”isanearlycriticismofapointinF.H.Bradley’sLogic,thelocusclassicusofBritishanalyticphilosophyislikely“TheRefutationofIdealism”(1903),acriticismoftheformulaesseestpercipi(“tobeistobeperceived”).AcrucialpartofthatargumentisMoore’sclaimthattheconceptofthesensationofyellowcontainstwoparts:thesensationthatisuniquetoeachpersonandtheyellownessthatcanbeperceivedbymanypeople.Evenwhenidealistsconcededthattherewassomekindofdualityhere,theyinsistedonakindofinseparability.Touseageneralnameforthekindofanalyticphilosophypracticedduringthefirsthalfofthetwentiethcentury,initiallyinGreatBritainandGerman-speakingcountries,andlaterinNorthAmerica,Australia,andNewZealand,“conceptualanalysis”aimsatbreakingdowncomplexconceptsintotheirsimplercomponents.Successiveanaly-sesperformedoncomplexconceptswouldyieldsimplerconcepts.AccordingtoMoore,theprocessmightleadultimatelytosimpleconcepts,ofwhichnofurtheranalysiscouldbegiven.Thedesignation“conceptual”wassupposedtodistinguishthephilosophicalactivityfromvariousanalysesappliedtononconceptualobjects.Physicswasfamousinthetwentiethcenturyforbreakingdownatomsintoprotons,neutrons,andelectrons,andthesesubatomicparticlesintoanarrayofmoreexoticcomponents.Andanalyticchemistryaimsatdeterminingchemicalcompositions.Theanalogybetweenphiloso-phyandscienceinspiredthename“logicalatomism,”atheorythatflourishedbetween1\nA.P.MARTINICH1920and1930.BothWittgensteinandBertrandRussellmaintainedthattheremustbesimple,unanalyzableobjectsatthefundamentallevelofreality.Wittgensteinthoughtthatthesimplesexistedindependentlyofhumanexperience,Russellthattheyexistedonlyforaslongasone’sattentionwasfixedonthem.Notwithstandingtheanalogybetweenscientificandphilosophicalanalysis,mostphilosophersinthefirsthalfofthetwentiethcenturymaintainedthatphilosophywasverydifferentfromscience.InhisTractatusLogico-Philosophicus(1921),Wittgensteinwrote:“Philosophyisnotoneofthenaturalsciences.(Theword‘philosophy’mustmeansomethingwhoseplaceisaboveorbelowthenaturalsciences,notbesidethem.)”(4.111).Thisconvenientlyleftopenwhichwassuperior.Butifthereisanythingconstantinanalyticphilosophy,itischange,andtheoppo-siteviewoftherelationbetweenscienceandphilosophyhasdominatedthesecondhalfofthecentury.LargelyowingtotheinfluenceofW.V.Quine,manyphilosophershavecometobelievethatphilosophyiscontinuouswithscience.Yesterday’sheresyistoday’sorthodoxy.Whicheverviewiscorrect,thedivisionbetweenthephilosophicalanalysisofconceptsandthenonphilosophicalscientificanalysisofnonconceptualobjectsshouldperhapsnotbetakentoostrictly.Conceptsandhencephilosophywouldbeofnouseiftheydidnotmakecontactwiththenonconceptualworld.Inaddition,scienceusesconcepts,manyofwhichmaybeamongthemostfundamentalofreality.ToparaphraseKant,perceptionswithoutconceptsareblind;conceptswithoutpercep-tionsareempty.Overlappingwiththelatterperiodoflogicalatomismislogicalpositivism,whichmaybedatedfromMoritzSchlick’sfoundingoftheViennaCirclein1924.Oneofitsprincipaldoctrineswasthatscienceisaunity;andoneofitsprincipalprojectswastoshowhowtotranslateallmeaningfullanguageintoscientificlanguage,inotherwords,toreducemeaningfulnonscientificlanguagetoscientificlanguage.Thisprojectcannotbesuccessfulunlesssomethingdistinguishesmeaningfulfromnonmeaningfulexpres-sions.A.J.Ayerprobablydevotedmoreenergyanddisplayedmoreingenuityintryingtoformulateacriterionofmeaningfulnessthananyoneelse.HisfirsteffortwaspresentedinLanguage,TruthandLogic(1936),thebookthatbecamethemostwidelyknownstatementoflogicalpositivismandwhichintroducedthatphilosophytotheanglophonepublic.Thebasicideaisthatasentenceismeaningfulifandonlyifitiseitheranalytic(orcontradictory)orempiricallyverifiable.Variousobjectionswereraisedtothis,andtoeveryrevisionofthiscriterion.Partoftheproblemwasthestatusofthecriterionitself.Eitheritwouldbeanalyticandhencevacuous,oritwouldbeempiricalbutthennotcompletelyconfirmed.LogicalpositivismhadbeendeadforsometimewhenitwasburiedbyCarlG.Hempel’s“ProblemsandChangesintheEmpiricistCriterionofMeaning”(1950)andW.V.Quine’s“TwoDogmasofEmpiricism”(1951).Nevertheless,Ayerandothersneverabandonedthespiritofverifiability.Whathadalreadybeguntotaketheplaceoflogicalpositivisminthe1940swasordinary-languagephilosophy,onestrandofwhichemanatedfromCambridgeinthelaterphilosophyofWittgenstein,theotherfromOxford.OneofWittgenstein’smoti-vatingbeliefswasthatphilosophycreatesitsownproblems,andthatmeansthattheyarenotgenuineproblemsatall.Theconfusionarisesfromphilosophers’misuseofordi-narywords.Theytakewordsoutoftheirordinarycontext,theonlycontextinwhichtheyhavemeaning,usethemphilosophically,andtherebydiscoveranomalieswiththe2\nINTRODUCTIONdisplacedconceptsexpressedbythesewords:“Forphilosophicalproblemsarisewhenlanguagegoesonholiday.”Wittgensteinquestionedmanyoftheassumptionsofana-lyticphilosophy–fromthenatureandnecessityofanalysistothenatureoflanguage–inadiscursiveanddialecticalstylesoinimitablethatitwasasifLudwigweretalkingtoWittgenstein.Hisoracularaphorisms,suchas“Don’taskforthemeaning,askfortheuse”and“Tounderstandasentenceistounderstandalanguage”stimulatedavarietyofreactions,fromtheFregeaninterpretationsofPeterGeachandMichaelDummett,totheholismofQuineandDonaldDavidson,tothedeconstructivistapproachesofO.K.BouwsmaandD.Z.Phillips.Theotherstrandofordinary-languagephilosophycamefromOxford,undertheleadershipofGilbertRyleandJ.L.Austin.Thesephilosophers,morenumerousthantheCambridgegroup(AntonyFlew,J.O.Urmson,andG.J.Warnock,deservetobemen-tioned),didnotsomuchthinkthattherewerenophilosophicalproblemsassaythatphilosophicalproblemscouldbesolvedthroughthecarefulanalysisofthedistinctionsinherentinordinarylanguage.ThepurposeofAustin’s“IfsandCans”and“APleaforExcuses”wastoelucidatetheproblemoffreedomanddeterminism,whicharosefromhisunderstandingofAristotle(seehisPhilosophicalPapers,2ndedition,p.180).Hesaidthatwhileordinarylanguagewasnotthelastwordinphilosophy,itwasthefirst.Hecertainlywasnotopposedtophilosophersdevelopingtheories.Austin,whohadbeenaclosetlogicalpositivistaccordingtoA.J.Ayer,coinedtheterm“performativeutterance”aspartofhisrefutationofthecentralthesisoflogicalpositivism,namely,thatallsentencesthatwerecognitivelymeaningfulwereeithertrueorfalse.Austinpointedoutthatsomestraightforwardlymeaningfulsentences,sen-tencesthatdidnotcontainsuspiciouswordslike“beautiful,”“good,”or“God,”werenotthekindofsentencesthatcouldhaveatruth-value:“Ibequeathmywatchtomybrother,”“IchristenthisshiptheQueenElizabethII,”and“IbettendollarsthatClevelandwinsthepennant.”Althoughtheconceptofperformativesdidtheworkitwasdesignedtodo,thedistinctionbetweenperformativesand“constatives”(roughly,statements)couldnotbesustained;andAustinreplacedthatdistinctionwithanother,betweenlocutionary,illocutionary,andperlocutionaryacts.Inthe1960s,JohnSearle,whowastrainedatOxfordbyordinary-languagephilosophers,showedthatAustin’slattertheorywasitselfinadequateandreplaceditwithhisownfully-developedtheoryinSpeechActs(1969)andExpressionandMeaning(1979).Bythelate1960sordinarylanguagehadlostitsdominance.SomeoftheOxfordphilosopherswereinstrumentalinitsdemise.Searle,asmentioned,developedafull-fledgedtheoryofspeechacts,andthenuseditasinspirationforfoundationalworkonthenatureofintentionalityandthesocialworld.OneofhisteachersandacolleagueofAustin’s,H.P.Grice,developedhisowntheoryoflanguageuse,atheorycomple-mentaryinmanywaystoSearle’s.Amoredramaticcauseofthedemiseofordinarylanguagephilosophyisattribut-abletooneofitschiefpractitioners,P.F.Strawson.InIndividuals(1959),heresurrectedmetaphysics,anareaofphilosophythatwasconsideredunacceptablebylogicalposi-tivism.Strawsondistinguishedbetween“stipulative”(bad)metaphysicsand“descrip-tive”(good)metaphysics.Hisdescriptiveproject,tolay“barethemostgeneralfeaturesofourconceptualstructure,”wassupposedtodifferfromlogicalorconceptualanaly-sisonly“inscopeandgenerality.”Atalmostthesametime,theAmericanW.V.Quine3\nA.P.MARTINICHpublishedWordandObject(1960).HisapproachdifferedfromStrawson’sprimarilyinemphasizingthegenesisofthemostgeneralconceptsandinaccommodatingitselfexplicitlytoempiricalpsychologyandphysics.Oncemetaphysicshadbeenmaderespectableagain,philosophersfeltmorecom-fortablepursuingalargevarietyofproblemsinavarietyofways.MetaphysicalsystemsbecamemoreelaboratewhenSaulKripkeusedpossibleworldstoprovetheoremsaboutmodallogic.Somesubsequentpositionscanevenbethoughtoutlandish,suchasDavidLewis’sviewthateverypossibleworldexists,andexistsinthesamesenseourownworlddoes–outlandishbutnotdisreputable.Somedisciplinesthathadbeenrelativelyneglectedbetween1930and1960werereinvigorated,forexample,ethicsandpoliti-calphilosophybyJohnRawls,mostnotablyinATheoryofJustice(1971);andsomequestions,suchasthemeaningoflife,weremulledoverby,forexample,ThomasNagelinananalyticallyrespectableway.Perhapstwoofthemostsalientcharacteristicsoftheperiodfrom1970onwardswerefirst,theinterestofanalyticphilosophersinthefoundationsofempiricalsciences,fromphysicsthroughbiologytopsychology,andsecond,theiruseofandcontributiontoartificialintelligenceandcognitivescience.Analysiswaslargelyabandonedandreplacedbyadesireforphilosophicaldoctrinesthatwerevariouslymoreintelligibleorintellectuallyrespectabletophysicists,logicians,orpsychologists.Thiswouldexplainthelargepresenceofphilosophersincognitivescience,linguistics,logic,andthephilosophyofscience;buthasperhapsalsoledtowhatSearlehascalled“therediscoveryofthemind”inabookbythatname.Therewereotherconsequencesoftherevivalofmetaphysics.Somephilosophers,respectedfortheirworkasearlyasthe1950s,forexampleRoderickChisholmandWilfridSellars,butnotcloselyassociatedwithanyoftheschoolswehavementioned,grewinsignificance.Somephilosophersturnedtothehistoryofmodernphilosophy,notably,StrawsonandJonathanBennettonKant,BennettonLocke,Berkeley,andHume,andBernardWilliamsandMargaretWilsononDescartes.Somephilosopherswhobecameimportantinthelastquarterofthetwentiethcentury,notablyRichardRorty,declaredanalyticphilosophymisconceived,bankrupt,orsimilarlydeficient.Inmakingtheirpositionclearandinaimingatcogency,theyareanalyticphilosophersinspiteofthemselves.Itislikelylesshelpfultotalkaboutoneoranothermovementinphilosophyafter1965.Noonemethodordoctrinedominated.Sometimesaphilosopherchampioningaviewbecameitsmostsignificantcriticoratleastmovedontosomethingquitedif-ferent,paradigmaticallyHilaryPutnam.Whatcanbesaidaboutthelastquarterofthetwentiethcenturyisthattheoriginalconceptionofanalysisandmostofitspresuppo-sitionswereabandonedbyalmostallanalyticphilosophers.Goneistheassumptionthatconceptsofphilosophicalimportanceareoftencomposedofsimplersharply-definedconcepts.Quine’sargumentsthatthereisnoprincipleddistinctionbetweenanalyticandsyntheticstatementsisjustaspecialcaseofthebroaderthesisthatlanguageandhencethoughtareessentiallyindeterminate.Wehavebeenexplainingandillustratinganalyticphilosophyinthelastcenturywithoutdefiningit.Itprobablydefiesdefinitionsinceitisnotasetofdoctrinesandnotrestrictedinitssubjectmatter.Itismorelikeamethod,awayofdealingwithaproblem,butinfactnotonemethodbutmanythatbearafamilyresemblancetoeachother.4\nINTRODUCTIONOncewhenGilbertHarmanwasasked,“Whatisanalyticphilosophy?,”hesaid(tonguefirmlyincheek),“Analyticphilosophyiswhoyouhavelunchwith.”Ingeneral,analyticphilosophyhasbecomehighlypluralisticandinmanywayshardlyresembleswhatwasdoneinthefirsthalfofthecentury.Therefectoryofanalyticphilosophyisnotasclubbyasitoncewas.Manymorepeoplesitatthetable,andmanymorediffer-entkindsoffood,preparedinmoreways,areserved.Perhapswhatmakescurrentana-lyticphilosophersanalyticphilosophersisacounterfactual:theywouldhavedonephilosophythewayMoore,Russell,andWittgensteindiditiftheyhadbeendoingphi-losophywhenMoore,Russell,andWittgensteinwere.Themultiplicityofanalyticalstylesisonereasonfororganizingthevolumebyindividualphilosopherandnotbytheme.Overfortyofthegreatestanalyticphilosophersofthelastcenturyarediscussedinthisvolume.Atleastthirtyofthem,webelieve,wouldbeonvirtuallyanysensiblelistoffortyoutstandinganalyticphilosophers.Manyotherphilosophershavealmostasgoodaclaimtobeincludedinthisvolume.Tonameonlysomeofthosewhoarenotalive,thefollowingwereconsideredandfinally,reluctantly,notincluded:MaxBlack,GustavBergmann,HerbertFeigl,PaulFeyerabend,GarethEvans,C.I.Lewis,J.L.Mackie,ErnestNagel,H.H.Price,H.A.Prichard,A.N.Prior,HansReichenbach,MoritzSchlick,GregoryVlastos,FriedrichWaismann,andJohnWisdom.Somephilosopherswereexcludedbecausetheydonotfitsquarelywithinthetradi-tionofanalyticphilosophyasordinarilyunderstood:JohnDewey,WilliamJames,CharlesSandersPierce,JohnCookWilson,and,ironically,AlfredNorthWhitehead,co-authorwithRussellofoneofthecentury’sgreatestworksoflogic,PrincipiaMathematica.Whilethereputationsofsomeofthephilosophersincludedareashighastheyeverwere,e.g.FregeandRussell,thoseofothershavedeclined,notalwaysjustifiably,forexample,thoseofC.D.BroadandRudolfCarnap.Inmakingourdecisionswehavetriednottobeprejudicedeitherfororagainstanyschool,method,ortimeperiod,buttoreflecttherelativeimportanceofvariousphilosophersovertheentiretwentiethcentury.Weknowthatourselectionwillbecontroversial,eventhoughitwasinfluencedbythejudgmentsofmanycolleagues.Arefereeofourproposalwrotethattheeditorsseemto“aimatenragingthereader.”Mostanalyticphilosopherswillbelievethatsomeotherlistofpeoplewouldhavebeenbetter.Wearesympathetic.Neitherofuscom-pletelyagreeswiththefinalselection.Eachbelievesthatatleastthreeotherphiloso-phershaveabetterclaimtobeincludedthansomethatwere.Inordertopreserve“plausibledeniability,”wehaveagreednottocommentfurtheronthelistsinanywrittenform,andnottoappeartogetheratanypublicgatheringofphilosophersforfiveyears.5\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd20011GottlobFrege(1848–1925)MICHAELDUMMETTFriedrichLudwigGottlobFregewasborninWismarin1848,anddiedinBadKleinenin1925.Hiswholeadultcareerwasspent,from1874to1918,intheMathematicsDepartmentofJenaUniversity.Hedevotedalmostallhislifetoworkontheborderlinebetweenmathematicsandphilosophy.Inhislifetime,thatworkwaslittleregarded,savebyBertrandRussellandLudwigWittgenstein;hisdeathwasmarkedbyveryfew.Yettodayheiscelebratedasthefounderofmodernmathematicallogicandasthegrandfatherofanalyticalphilosophy.Frege’sintellectualcareerwasunusual.Mostphilosophersandmathematiciansmakecontributionstodiversetopicswithintheirfields;butFregesethimselftoachieveoneparticular,thoughextensive,task.From1879to1906hepursuedthissingleambi-tiousaim:tosetarithmeticuponsecurefoundations.Virtuallyallthathewroteduringthoseyearswasdevotedtothisproject,ortotheelaborationofideasthathedevelopedinthecourseoftryingtocarryitthroughandtookasintegraltoit.Theterm“arith-metic,”asheusedit,istobeunderstoodinabroadsense:forhim,itcomprised,notonlynumbertheory(i.e.thetheoryofthenaturalnumbers),butalsoanalysis,thatis,thetheoryofrealandcomplexnumbers.Hedidnotattempttoconstructthefounda-tionsofgeometry.Heviewedmathematicsinthetraditionalway,asdividedintothetheoryofquantity,andthusofcardinalnumbersandofnumbersmeasuringthemag-nitudesofquantities(realnumbers),andthetheoryofspace(points,lines,andplanes).Hebelievedthesetwopartsofmathematicstorestondifferentfoundations.Thefoundationsofarithmetic–ofnumbertheoryandanalysis–arepurelylogical.Butalthoughthetruthsofgeometryareapriori,theyrestuponspatialintuition:theyaresyntheticapriori,intheKantiantrichotomyFregeaccepted.Kantwasthereforerightaboutgeometry;buthewaswrongaboutarithmetic.Allappealtospatialortemporalintuitionmustbeexpelledfromarithmetic:itsconceptsmustbeformulatedanditsbasicprinciplesestablishedwithoutrecoursetointuitionofanykind.ItwastothetaskofestablishingthepurelylogicalfoundationsofarithmeticthatFregedevotedhiswholeintellectualendeavor.Incarryingoutthistask,hewasledintosomepurelyphilosophicalinvestigations;itisforthisreasonthat,althoughamathe-matician,heisnowheldinsuchhighregardbyphilosophersoftheanalyticschool.Hisattempttoprovidearithmeticwithsecurelogicalfoundationswasembodiedinthreebooks,allofhighimportance,althoughhealsopublishedanumberofarticles,6\nGOTTLOBFREGEspin-offsfromhiscentralendeavor.ThefirstofthesethreebookswastheshortBegriffsschrift(ConceptualNotation)of1879:thisexpoundedanewformalizationoflogic.Itwasthefirstworkofmodernmathematicallogic:itcontainedanaxiomaticsystemofpredicatelogicofpreciselythetypethatwastobecomestandard.Thenota-tionwasutterlydifferentfromthatwhichwouldbecomestandard,andwasessentiallytwo-dimensional,thetwoclausesofaconditionalbeingwrittenondifferentlines;butthenotationwasessentiallyisomorphictothatwhichPeano,Hilbert,andRusselllatermadestandard.AnEnglishtranslationofthebookandofsomerelatedarticleswaspublishedbyT.W.Bynumin1972.TheBegriffsschriftreceivedsixreviews,includingalengthyonebyErnstSchröder;butnoneofthereviewersunderstoodFrege’sintentionorhisachievement.Thesecondbookwasalsoshort,thoughnotquitesoshortasBegriffsschrift.ItwascalledDieGrundlagenderArithmetik(TheFoundationsofArithmetic)andappearedin1884.Inorderthatitshouldbeaccessibletoasmanyaspossible,thebookwaswrittenwithouttheuseoflogicalsymbols.Fregefirstsurveysarangeofrivaltheoriesonthestatusofarithmeticalpropositionsandthenatureof(cardinal)number,anddemolishesthemwithtrenchantarguments;thispartofthebookislargelyphilo-sophicalincharacter.Then,havingtohissatisfactionleftnospaceforanyothertheorybuthisown,heproceedstosketchapurelylogicalderivationofthefundamentallawsofarithmetic.Itistobedoubtedwhetheranyotherphilosophicaltreatiseofcomp-arablelengthhas,sincePlato,evermanifestedsuchbrillianceasFrege’sGrundlagen.AnEnglishtranslationbyJ.L.Austinwaspublishedin1950,andacriticaleditionbyC.Thielin1986.ThistimeFrege’sbookreceivedfivereviews,againnoneofthemadequatetotheirsubjectmatter.OnewasbyGeorgCantor,whounhappilydoesnotseemtohavetriedtounderstandthework,withwhichhemighthavebeenexpectedtobeinlargesympathy.Fregehadthoughtthathewasonthevergeofsuccessinconstructingdefinitivefoundationsforarithmetic.HehadthoughtthathisGrundlagenwouldmakethisplaintotheworldofphilosophersandmathematicians.Hebecameintenselydepressedbyhisfailuretohaveconveyedtothatworldthemagnitudeofhisachievement.Atthesametime,hebecameawareofdeficienciesinthephilosophicalbasisonwhich,inGrundlagen,hehadrestedhisargumentsandwhichunderpinnedhisformallogic.Therefollowedfiveyearsduringwhichhepublishednothing,butengagedinathoroughgo-ingrevisionofhisphilosophyoflogicandofhisformallogic.Theoutcomeofthisrevi-sionheexpoundedinalecture,FunctionundBegriff(FunctionandConcept),givenin1891.Hethensetaboutacompleteformalexpositionofhisfoundationsofarithmeticinhisthirdgreatbook,GrundgesetzederArithmetik(BasicLawsofArithmetic).ThiswasutterlyunlikeGrundlagen.Thefirstvolumecameoutin1893,thesecondvolumenotuntiltenyearslater,in1903.Thebookisincomplete;athirdvolumemusthavebeenplanned,butitneverappeared.InthefirstvolumeofGrundgesetze,thereisnoargu-ment,onlyexposition.Fregebeganbyexplaininghisformallogicalsystem,andexpounding,withoutgivinganyargumentfororjustificationofthem,thephilosophi-cal,or,moreexactly,semanticnotionsthatunderpinnedit.Heprovideswhatisineffectaprecisesemantictheoryfortheformallanguageusedinthebook.ThismakesupPartI,ofwhichanEnglishtranslationbyM.Furthwaspublishedin1964.TherefollowsinPartIIastringofformalderivations,carriedoutinFrege’sfarfromeasilyreadsymbolism,whichexecuteindetailtheprogramsketchedinGrundlagenfor7\nMICHAELDUMMETTconstructingalogicalfoundationforthetheoryofcardinalnumbers,includingthenaturalnumbers,understoodasfinitecardinals.PartIIwasconcludedinthesecondvolumeof1903,andisfollowedbyaPartIIIdevotedtotherealnumbers,atopicFregehadnevertreatedinanydetailbefore.PartIIIisnotcompletedinthesecondvolume;aclearintentiontocompleteitprovesthatathirdvolumewascontemplated.ThefirsthalfofPartIIIisinprose,notsymbols;Fregehaschangedhisapproach.HeattemptsinthishalftodoforrealnumberswhathehaddoneinGrundlagenforcardinalnumbers:toreviewandcriticizerivaltheoriessothattherewouldappearnoalterna-tivetohisownconstructionofafoundationforanalysis,carriedoutbyproofinhisformalsysteminthesecondhalfofPartIII.Unhappily,hehadlosthistouch.WhereasinGrundlagennothingismentionedsavewhatwillcarrytheargumentforward,thefirsthalfofPartIIIofGrundgesetzereadsasifhewasmerelydeterminedtogethisownbackontheothertheoristswhohadneglectedhim,CantorandDedekindincluded.Criticismsaremadeoffeaturesoftheirworkquiteirrelevanttothemainstrandoftheargument;errorsarepointedoutwhichcouldeasilyberectified,withoutanyindicationofhowtorectifythem.Apowerfulcritiqueofformalismisincluded.ItispossibletoextractFrege’sreasonsforthestrategyheadoptsforconstructingfoundationsofanalysis;butthewholelacksthebrillianceandtheexquisiteplanningofGrundlagen.WhenhehadfinishedcomposingthesecondvolumeofGrundgesetze,Fregemusthavefeltadeepcontentment.Stillembitteredbytheneglectofhiswork,hesurelybelievedthathehadattainedhislife’sgoal:hehadconstructedwhathethoughttobedefinitivefoundationsforthetheoriesbothofnaturalnumbersandofrealnumbers.But,whilethebookwasinpress,hereceivedtheheaviestblowofall,deliveredbyoneofhisfewadmirers,theyoungBertrandRussell.RussellwroteinJune1902toexplainthecelebratedcontradictionhehaddiscoveredinthe(naive)theoryofclasses,andtopointoutthatitcouldbederivedinFrege’slogicalsystem(seeRUSSELL).AtfirstFregewasshattered.Then,ashereflectedonthematter,hedevisedaweakeningofhisBasicLawV,whichgovernedtheabstractionoperatorusedforformingsymbolsforclassesandwasresponsibleforthecontradiction;hewasconfidentthatthismodificationwouldrestoreconsistencytohissystem.ThemodificationwasexplainedinanAppendixaddedtoVolumeIIofGrundgesetze.Thefactwas,however,thatthemodifiedBasicLawVstillallowedthederivationofacontradiction.WhetherFregeeverdiscoveredthisisuncertain;butwhathemusthavediscoveredwasthat,initspresence,noneoftheproofshehadgivenofcrucialtheoremswouldgothrough.Hiswifediedin1905.IttookhimuntilAugust1906toconvincehimselfthathislogicalsystemcouldnotbepatchedup;atthatpointhehadtofacethefactthattheprojecttowhichhehaddevotedhislifehadfailed.GrundgesetzeremainstheonlypartofFrege’spublishedworkofwhichnofullEnglishtranslationhasyetappeared(saveforthetranslationbyFurthofPartI,whichalsocontainstheAppendixtoVolumeII,andexcerptsinthevolumeoftransla-tionsbyGeachandBlack).AbrieffragmentamongFrege’sliteraryremains,datedAugust1906,askstheques-tion,“WhatcanIregardastheoutcomeofmywork?”Inotherwords,“Whatremainsnowthatthecontradictionhasdestroyedmylogicalfoundationsforarithmetic?”Frege’sanswerwasthatwhatsurvivedofenduringvaluewashislogicalsystem,strippedoftheabstractionoperator,andthewholestructureofphilosophicallogic8\nGOTTLOBFREGEwhich,from1891onwards,hadunderpinnedit.Hecontinuedtothinkaboutthesetopics,althoughitwasnotforyearsthathepublishedanythingmore.Eventually,duringthewaryears,hepublishedthefirstthreechaptersofwhatwasplannedasacomprehensivetreatiseonlogic,althoughitwasneverfinished.Intheverylastyearsofhislife,hefinallyturnedagaintothefoundationsofmathematics;reversinghislifelongview,hebeganaderivationofarithmeticfromgeometry,butdidnotcarryitveryfar.Frege’sjudgmentof1906aboutwherethevalueofhisworklaywasatfirstthejudgmentofthosewhoparticipatedintherevivedstudyofFrege’swritings.InthePrefacetohisTractatusLogico-Philosophicusof1922,LudwigWittgensteinhadwrittenof“thegreatworksofFregeandtheworksofmyfriendBertrandRussell,”andhadreferredfrequentlytoFregeinthebook.DespitethecelebrityoftheTractatusanditswideinfluence,thisfailedtostimulatemorethanaveryfewphilosopherstofindoutanythingaboutFrege.RudolfCarnap,whohadactuallyheardFrege’slectures,AlonzoChurch,andPeterGeachcontinuedtoholdhiminhighregard(seeCARNAP);butthemajorityofphilosophers,British,German,andAmerican,wentonignoringhim.Therevivalofinterestinhimbeganinaslowwayinthe1960s,andgatheredmomentuminthe1970s;bytheendofthatdecade,hehadbecomerequiredreadingforanystudentofanalyticphilosophy.Buttheinterestinhisphilosophyofarithmeticwasmeager;itwastakenforgrantedthathissystem’shavingrunfoulofRussell’scontradictiondestroyedallitspretensionstoseriousconsideration.InterestinFregethereforecon-centratedonwhatmadehimthegrandfatherofanalyticphilosophy:hisphilosophicalanalysisoflanguageandofthoughtwhichunderlayhisformallogic.Thekeytoamodernsystemofpredicatelogicisofcoursethequantifier-variablenotationforgenerality,whichFregeintroducedforthefirsttimeinBegriffsschrift.Heemployedonlynegation,theconditional,andtheuniversalquantifier;hedidnotusesymbolsforconjunction,disjunction,ortheexistentialquantifier,butexpressedthesebymeansofthethreelogicalconstantsforwhichhedidhavesymbols.Fregeinsistedthathissymbolism,unlikethatintheBooleantraditionsuchasSchröder’s,couldincor-porateaformallanguage:itneededonlytheadditionofsuitablenonlogicalconstants,predicates,etc.,tobecapableofframingsentencesonanysubjectmatterwhatever,andofcarryingoutdeductivereasoningconcerningit.FregedidnotconceiveofformulaeinhissymbolisminthewaythatTarskiwastodo,namelyasbuiltupfromatomicfor-mulaecontainingfreevariableswaitingtobecomeboundintheprocessofformingcomplexformulae.Rather,hethoughtofthemasbuiltupoutofatomicsentences.Thisrequiredthat,beforeattachingaquantifier,theremustfirstbeformed,fromasuitablesentence,whatweshouldcallapredicate,but,forFrege,wasafunctionalexpressionorexpressionforaconcept.(Heeschewedtheterm“predicate”astoocloselyassociatedwiththetraditionalsubject–predicatelogic.)Suchanexpressionwas“incomplete”or“unsaturated”:itcouldnotstandonitsown,buthadgapsinit,beingformedfromasentencebyomittingoneormoreoccurrencesofasingularterm.Whenaquantifierwasattachedtoit,theboundvariablesgovernedbyitweretobeinsertedintothesegaps,therebyshowingtowhatexpressionforaconceptthequantifierhadbeenattached.Whenhewantedtospeakofaparticularexpressionforaconcept,Fregeusedthelower-caseGreekletterxtoindicatethegapsinit;butanexpressioncontainingxwas9\nMICHAELDUMMETTnopartoftheformallanguage,butonlysomethingtobeusedmetalinguisticallytospeakabouttheformallanguage.Thusfromthesentence“PittrespectsPitt’sfather”(whichrepresentsthewaywearemeanttounderstandthecolloquial“Pittrespectshisfather”)threedifferentexpressionsforconceptscouldbeformed:“xrespectsPitt’sfather,”“Pittrespectsx’sfather,”and“xrespectsx’sfather.”Thisexemplifiesthefactthatasentencecanbeanalyzedindifferentways.FregeinsistsinBegriffsschriftthatthesedifferentanalyseshavenothingtodowiththecontentofthesentence,butonlywithourwayoflookingatit;inotherwords,ourgraspofthecontentofthesentencedoesnotdependuponournoticingthatitispossibletoanalyzeitinonewayoranother,forexamplethatthepropername“Pitt”occurstwicewithinit.Inonesense,eachofthethreeexpressionsforconceptsoccurswithinthesentence;butnoneofthedifferentconceptsispartofthecontentofthesentence.Byattachingtheuniversalquantifiertothesethreeexpressionsforconcepts,weobtainrespectively“Foralla,arespectsPitt’sfather,”“Foralla,Pittrespectsa’sfather,”and“Foralla,arespectsa’sfather,”or,colloquially,“EveryonerespectsPitt’sfather,”“Pittrespectseveryone’sfather,”and“Everyonerespectshisownfather.”Andnow,inthesequanti-fiedsentences,Fregesays,theexpressionfortherelevantconceptispartofthecontentofthesentence.Theprocessofformingexpressionsforconceptsmaylikewisebeusedtoformexpressionsforfunctions,asweordinarilyconceivethem,namelybystartingwithacomplexsingulartermwithinwhichsomesimplersingulartermoccurs.Anditmayalsobeusedtoformexpressionsforrelationsbetweentwoobjects,namelybyremovingfromasentenceoneormoreoccurrencesofeachoftwodifferentsingularterms.Thiswasofimportanceforsecond-orderlogic,admittingquantificationoverfunctionsandrelations.ThusFrege’sinventionofthequantifier-variablenotationyieldedhimseveralfun-damentalinsights.First,theconceptionofconcept-expressionsasincompletesolvedtheproblemoftheunityofsentencesandthethoughtstheyexpress.Noglueisneededtomakethepartsofthesentenceadheretooneanother.Theconcept-expressionorrelation-expressionisofitsnatureincapableofstandingalone,butcanbepresentonlywhenitsargument-placesarefilledbysingulartermstoformasentence.Orelseitisitselftheargumentofaquantifier,formingadifferentkindofsentence:itismadetoadheretosuchterms,ortohaveaquantifierattachedtoit,andcannotexistotherwise.Secondly,concept-formationdoesnotconsistsolelyofthepsychologicalabstractionofsomecommonfeaturefromindividualobjectsoroftheprocessofapplyingBooleanoperationstogivenconcepts(conjoiningordisjoiningthem).Bytheprocessofomit-tingsingulartermsfromcompletesentences,or,equivalently,ofthinkingofthemasreplaceablebyothersingularterms,wecanarriveatexpressionsforconceptswithnewboundaries,andsoattheconceptsthusexpressed.Moreover,suchexpressionswerenot,ingeneral,actualpartsofthesentencesfromwhichtheywereformed;theywere,rather,patternsexemplifiedbydifferentsentences.Theexpression“xrespectsx’sfather”occursbothin“PittrespectsPitt’sfather”andin“FoxrespectsFox’sfather”;thetwosentenceshave,notjustcommonwords,butacommonpattern.Thusweshouldnotthinkofasentence,orthethoughtitexpresses,asformedoutofitscomponentparts,butofthecomponentsasattainablebyanalyzingthesentence;stilllessshouldwethinkofaconcept-expressionasformedoutofitscomponents,butasaresultofanalyzing10\nGOTTLOBFREGEasentence.Becauseapprehendingthepossibilityofanalyzingasentenceinaparticu-larwayrequiresustoseeitasmanifestingacertainpattern,whichisnotrequiredforasimplegraspofthesentence’scontent,andbecauseapprehendingthispossibilitymaybeessentialtorecognizingthevalidityofsomedeductiveinference,thereisacreativeingredientindeductiveinference.Suchinferencedoesnotdependonlyuponagraspofthecontentsofthesentencesthatfigureinit;andthisexplainshowdeductiveinfer-encecanleadustonewknowledge,whichconsiderationofitsroleinmathematicsmakesevidentthatitdoes.TheseideaswereexpressedinBegriffsschriftandinGrundlagen.PartIofBegriffsschriftwasdevotedtosententiallogic,andPartIItofirst-orderpredicatelogic.AlthoughFregedidnothavetheconceptofthecompletenessofalogicalsystem,hehadinfactframedacompleteformalizationoffirst-orderlogic.PartIIIofBegriffsschriftisdevotedtosecond-orderpredicatelogic,involvingquantificationoverconcepts,relations,andfunctions;Fregeneversawanyreasonforregardingthefirst-orderfragmentasespe-ciallysignificant.Toexplainsecond-orderquantificationinthesamewayasfirst-orderquantification,Fregehastoadmitthenotionofanexpressionforaconceptofsecondlevel,formedbyremovingfromasentenceoneormoreoccurrencesofanexpressionforafirst-levelconcept,orofarelationalorfunctionalexpression;thesesecond-levelconcept-expressionsallhavedifferenttypesofincompleteness.InPartIIIFregegavehispurelylogicaldefinitionofasequence;sincepreviouslythenotionofaninfinitesequencehadusuallybeenexplainedintemporalterms,asinvolvingitssuccessivecon-structionstepbystep,orelseasuccessivediversionofattentionfromonetermtothenext,Fregeregardedthisasanessentialcontributiontotheprogramofexpellingintui-tionfromarithmeticinfavorofpurelylogicalnotions.Itwasespeciallyimportantfornumbertheory,sincethenaturalnumbersthemselvescouldbedefinedasthetermsofafinitesequencebeginningwith0andproceedingfromeachtermtoitssuccessor.Frege’sdefinitionofasequencewassoframedthat,whenthenaturalnumbersaresodefined,theprincipleoffiniteinduction,sometimesclaimedasamethodofreasoningpeculiartoarithmetic,becomesadirectconsequenceofthedefinition.Frege’sdefinitionof“sequence”isnowgenerallyknownasthedefinitionoftheancestralofarelation,namelytherelationwhichthefirsttermofafinitesequencehastothelasttermwheneachtermbutthelaststandsintheoriginalrelationtothenextterm.(Itisnamedthe“ancestralrelation”becausetherelation“ancestorof”istheancestraloftherelation“parentof.”)TherearethreefeaturesofGrundlagenofespecialinteresttophilosophyingeneral.Thefirstisthedistinctionbetweentheactual(wirklich)andtheobjective.Fregeused“actual”tomean“concrete”inthesenseinwhichconcreteobjectsaredistinguishedfromabstractones;anobjectisactualifitiscapableofaffectingthesenses,directlyorindirectly.Butsomethingmaybeobjectiveeventhoughitisnotactual;anexamplehegivesistheEquator.YoucannotseeortripovertheEquator,butitisnotfictitiousorsubjective;statementsaboutitmaybeobjectivelytrueorfalse.Wecanmakereferencetoobjectswhich,thoughobjective,arenotactual,andmakeobjectivelytruestatementsaboutthem.Fregethusrejectedwhatisnowcalled“nominalism”asbasedonafun-damentalerror.Thiswascrucialforhisphilosophyofarithmetic.Hetooknumberstobeobjects,objectivebutnotactual:wecanrefertothemandmakeobjectivelytruestatementsaboutthem.11\nMICHAELDUMMETTAprinciplegreatlystressedinGrundlageniswhathascometobeknownasthe“contextprinciple”:thatitisonlyinthecontextofasentencethatawordhasameaning.Itisnoteworthythattheprincipleisformulatedlinguistically,asconcerningthemeaningsofwordsinsentences,ratherthan,say,as“Wecanthinkofanythingonlyinthecourseofthinkingthatsomethingholdsgoodofit.”Theinterpretationofthedictumiscontentious.Attheveryleast,itisanassertionoftheprimacyofsen-tencesintheorderofexplanationofmeaning.Wemustfirstexplainwhat,ingeneral,constitutesthemeaningofasentence,andthenexplainthemeaningsofallsmallexpressionsastheircontributionstothemeaningsofsentencesinwhichtheyoccur.WhenwelookathowFregeappliestheprincipleinthebook,itappearstohaveamuchstrongersignificance:namelythat,tosecureameaningforanexpressionortypeofexpression,itsufficestodeterminethesensesofallsentencesinwhichitoccurs.Fregeneverreiteratedthecontextprincipleinanysubsequentwriting,althoughthereisastrongechoofitinPartIofGrundgesetze.Theothersalientfeatureisthefirstclearexampleofthelinguisticturn,givingFregeastrongclaimtobethegrandfatherofanalyticphilosophy.Atacriticalpointofthebook,Frege,havingalreadyarguedthatournotionofnumberisnotderivablefromsense-perceptionorintuition,asks,“How,then,arenumbersgiventous?”Theques-tionisbothepistemologicalandontological:howareweawareofnumbers,andwhatguaranteeistherethatsuchobjectsasnumbersexist?Inansweringit,Fregesimplyassumesthatitcanbeequatedto“Howaremeaningsconferredonnumericalterms?”Heappealsimmediatelytothecontextprinciple;invirtueofthis,thequestionreducesto,“Whatsenseattachestostatementscontainingnumericalterms?”Aquestionaboutwhatobjectsexistandhowweknowofthemisthustransformedintoaquestionaboutthemeaningsofcertainsentences.However,thoseofFrege’sideasthatmostinterestedanalyticphilosopherswheninterestinhisworkrevivedweretheonesheexpoundedinhismiddleperiod(1891–1906).Fregehadnogeneraltermformeaning,inthesenseinwhichthemeaningofawordorexpressioncompriseseverythingthataspeakermustimplicitlyknowaboutitinordertounderstandit.Hedistinguishedthreefeatureswhich,inthissense,maycontributetothemeaningofawordorsentence:force,tone,andsense.Theforceofanutteranceiswhatdistinguishesanassertionfromaquestion,andFregerec-ognizedonlythesetwotypesofforce:assertoricandinterrogative.InEnglishinterroga-tiveforceisusuallyindicatedbytheinversionoftheverbandsubject;Fregeinsistedthatthesenseofaquestioninvitingtheanswer“Yes”or“No”willcoincidewiththatofthecorrespondingassertoricstatement.Whatdifferentiatesthemisthesignificanceoftheutterance:inonecaseweaskwhetherthethoughtexpressedistrue,intheotherwecommitourselvestoitstruth.ItwasimportantforFregethatonlyacompleteutterancecancarryforce;adeclarativesentenceservingas,say,oneclauseofadisjunctivestate-mentorastheantecedentofaconditionalonedoesnothaveassertoricforce,whichisattachedonlytothestatementasawhole.Itwasessential,Fregethought,nottocon-struetheverborpredicateofasentenceasintrinsicallycontainingtheassertoricforcewithinit.Naturallanguagesusuallylackanyexpressmeansofindicatingthatasser-toricforceistobeattachedtoasentence,butFregeconsideredthisadefectofthem.Inhisformallanguageheusedasymbolforjustthispurpose,the“judgment-stroke”(oftencalledbyothersthe“assertionsign”).Itisaphilosophicalmistaketospeakof“judg-12\nGOTTLOBFREGEments”whenallthatweareconcernedwithistheircontents;unlessweareactuallyconcernedwiththeactofrecognizingthemastrue,weshouldspeakinthisconnectionofthoughts,ratherthanofjudgments.Fregedidnotrecognizeimperativalorotherkindsofforce,thoughitmayplausiblybearguedthatanimperativesentenceexpressesthethoughtthatistrueifthecommandisobeyedorthedemandcompliedwith.Hesimplydeclaredthatsuchasentenceexpressesacommand,notathought.WhatIhavecalled“tone,”andFregecalledFärbung(coloring)isdistinguishedfromsenseinthatitcannotaffectthetruthorfalsityofwhatissaid.TheEnglishsentences“Hehasdied,”“Heisdeceased,”and“Hehaspassedaway”donotdifferinsense,butonlyintone.Likewise,whereAandBaresentences,thecomplexsentences“AandB”and“NotonlyAbutB”donotdifferinsense,butintone:ifeitheristrue,theotheristrue,evenifitconveysaninappropriatesuggestion.Thesenseofawholesentenceisthethoughtthatitexpresses;thesenseofapartofacomplexexpression,includingasentence,ispartofthesenseofthewhole.Inhismiddleperiod,Fregedrewadistinctionbetweenthesignificanceofanexpres-sionandwhatitsignifies,whichhehadnotdoneinhisearlyperiod.Forthethingsignified,heconfusinglychosetheword“Bedeutung,”theordinaryGermanwordfor“meaning”:buttheBedeutungofanexpressionisnotpartofitsmeaning,where“meaning”isunderstoodasspecifiedabove.Itisnotnecessary,inordertounderstandawordorphrase,toknowitsBedeutung,onlyitssense.Frege’stermisconventionallyrenderedinEnglisheither“meaning”or“reference”;neitherishappy.TheBedeutungofasingulartermistheobjectweusethetermtotalkabout.ItisimpossiblejusttoknowtheBedeutungofasingularterm,evenifthattermislogicallysimple,i.e.itisapropernameintherestrictedsense(Fregemisleadinglycalledallsingularterms“Eigennamen”–propernames).FregefollowedKantinholdingthateveryobjectofwhichweareawareisgiventousinaparticularway;thesenseofasingulartermembodiestheparticularwayinwhichitsBedeutungisgiventousinvirtueofourunderstandingoftheterm.ButitwasnotonlysingulartermswhichFregetookashavingBedeutungen:heascribedthemtoeveryexpressionthatcouldbeagenuineconstituentofasentence,includingincompleteonessuchasconcept-expressionsandsentencesthemselves.HedoesnotarguethatanysuchexpressionmusthaveaBedeutung;hetakesitforgranted.TheonlyquestionhecanvassesiswhatkindofthingtheBedeutungofanexpressionofanygiventypeshouldbetakentobe.ThiscausesmuchperplexitytothosereadingFregeforthefirsttime:surelythereisnothingtowhichaconcept-expressionorasentencestandsasanamestandstotheobjectnamed.TheonlywaytoarriveatanunderstandingofFrege’snotionofBedeutungistolookattheusetowhichheputsit.Thatuseisgovernedbyfourfundamentaltheses:1TheBedeutungofapartofacomplexexpressionisnotpartoftheBedeutungofthewhole.2ButtheBedeutungofthewholedependsuniquelyupontheBedeutungenofitsparts.3IfapartlacksBedeutung,thewholelacksBedeutung.4TheBedeutungofasentenceisitstruth-value–itsbeingtrueoritsbeingfalse.Thesis(1)followsfromthefactthatSwedenisnotpartofStockholm,thecapitalofSweden;andthesis(3)derivesfromtheconsiderationthat,ifthereisnosuchcountry13\nMICHAELDUMMETTasRuritania,thenthereisnosuchcityasthecapitalofRuritania.Asfortheidentifi-cationoftruth-valuesastheBedeutungenofsentencesinaccordancewiththesis(4),thatfollowsfromFrege’sextensionalistlogic,despiteitsfailuretofitnaturallanguage.Accordingtoit,asubsentenceofacomplexsentencecontributestothetruth-valueofthewholesolelybyitsowntruth-value.Counterexamplesfromnaturallanguagethenhavetobeexplainedaway.Instancesaresentencesinindirectspeechfollowingverbslike“saidthat”and“believesthat”;asiswellknown,Fregehandledthesebydeemingthesentencesfollowing“that”tohaveaspecial,indirectsense,wherebytheirBedeutungbecamethesensesthattheywouldexpresswhenindirectspeech.ItisplainfromthesefourthesesthattheBedeutungofanexpressionconstitutesitscontributiontothedeterminationofthetruth-valueofanysentenceinwhichitoccurs.ThisexplainswhyFregetakesitforgrantedthatanyexpressioncapableofoccurringinasentencewithoutdenyingitatruth-valuemusthaveaBedeutung.ItalsomakesaFregeantheoryofBedeutungforalanguageequivalenttowhatweunderstandasasemantictheoryforthatlanguage,whichisatheoryexplaininghowsentencesofthelanguagearedeterminedastrueorasfalseinaccordancewiththeircomposition.Thesemanticvalueofanexpression,insuchatheory,ispreciselythatwhichcontributestothedeterminationofthetruth-valueofasentenceinwhichthatexpressionoccurs.WemaythereforeequatethenotionoftheBedeutungofanexpression,asFregecon-ceivedit,withthatofitssemanticvalue.Inaconventionalsemantictheoryforaformalizedlanguage,thesemanticvalueofanindividualconstantorothersingulartermisanelementofthedomaindenotedbytheterm.Thesemanticvalueofaone-placepredicateisaclassofelementsofthedomain;thesentenceresultingfromputtingthetermintheargument-placeofthepredicateistrueiftheelementdenotedbythetermisamemberoftheclassconstitut-ingthesemanticvalueofthepredicate,falseotherwise.Fregedidnotspeakofthedomainofquantification;sofarascanbedetermined,hetooktheindividualvariablestorangeoverallobjectswhatever,theBedeutungofanytermbeingsuchanobject.FregedidnottaketheBedeutungofaconcept-expressiontobeaclass.HecalledtheBedeutungofaconcept-expressiona“concept.”Thismustnotbeunderstoodinthesenseinwhichwemayspeakofacquiringaconceptorgraspingaconcept,whichhastodowiththesensesexpressedbywords.InGrundlagen,theword“concept”(Begriff)hadbeenusedbothinthiswayandinconformitywithwhatwastobecomeFrege’susageinhismiddleperiod;but,inthatperiod,hetookaconcepttostandtoaconcept-expressionasanobjectstandstoasingularterm,andthusnotatallasthesenseofthatexpression.ForFrege,aconceptmustbedistinguishedfromaclass,whichwasforhimapar-ticularkindofobject.Aclassistheextensionofaconcept,comprisingthoseobjectsthatfallundertheconcept;buttheextensionofaconceptisaderivativenotion,onlytobesoexplained.Therelationofbeingamemberofaclasscanbeexplainedonlyasthatoffallingunderaconceptofwhichtheclassistheextension;anyattempttoexplainitinanyotherwayturnstherelationintothatofparttowhole,whichisquitedifferent.Sowecanattaintheconceptofaclassonlyviathatofaconcept;andwecancharacterizeanyparticularclassonlybycitingaconceptofwhichitistheextension.Sinceaconcept-expressionwasforFregeincomplete,soitsBedeutungcannotbeanobjectofanykind,butmustbelikewiseincomplete,anentityneedinganobjectto14\nGOTTLOBFREGEsaturateit.Thisisadifficultconception,butistobethoughtofbyanalogywithhowwethinkoffunctions.Wethinkofanarithmeticalfunctionasaprincipleaccordingtowhichonenumberisarrivedat,givenanother:notamethodofarrivingatthevalue,giventheargument,butsimplytheassociationofthevaluetotheargument.Itsin-completenessconsistsinthefactthatthereisnothingtoitsavethisassociation:itsexistenceconsistssolelyinitslinkingargumentstovalues.Likewise,theexistenceofaconceptconsistssolelyinitshavingcertainobjectsfallingunderit,andothersnotfallingunderit.Infacts,accordingtothedoctrinesofFrege’smiddleperiod,conceptssimplyareaparticulartypeoffunction.Forheregardedtruth-values–truthandfalsity–asthemselvesbeingobjects.Soaconceptisafunctionwhichtakesonlytruth-valuesasvalues,mappinganobjectfallingunderitontothevaluetrue,andonenotfallingunderitontothevaluefalse.Inthesameway,arelationisafunctionwithtwoargu-ments,allofwhosevaluesaretruth-values.Inhisearlyperiod(1874–85),Fregehadnotdistinguishedbetweensignificanceandthingsignified;hehadusedtheoneterm“content”forbothwithoutdifferentiation.ItwasagreatadvancethatinhismiddleperiodhesharplydistinguishedthemassenseandBedeutung.Whatwetacitlyknowinunderstandingawordorexpressionisitssense;itssenseisthewayitsBedeutungisgiventous.Itisnotonlyofanobjectthatitholdsgoodthatitmustbegiventousinaparticularway:thesameholdsgoodofconcepts,relations,andfunctions.Forinstance,anarithmeticalfunctionmaybegiventousbymeansofaparticularprocedureforcomputingitsvalue,givenitsargumentorargu-ments;otherproceduresmightservetodeterminethevaluesofjustthesamefunction.Forthisreason,thecontentofanypieceofknowledgethatwemayhaveconcerningagivenexpressioncanneversimplyconsistinourknowingitsBedeutung,butmustbeourknowingitsBedeutungasgiveninaparticularway.TheBedeutungofanexpressionisthereforenopartofitsmeaning,wherethisiswhatwegraspinunderstandingtheexpression:whatwegraspisitssense.Wemayindeedgraspmorethanitssense,namelywhatwascalledaboveitstone.Senseisthatpartofthemeaningoftheexpressionthatisrelevanttothedeterminationofasentencecontainingitastrueorasfalse.ButthenotionsofsenseandBedeutungarecloselyconnected:again,thesenseofanexpressionisthewayinwhichitsBedeutungisgiventous.(ThesenseisdieArtdesGegebenseins,usuallyclumsilytranslated“themodeofpresentation.”)ItisnotonlythateachindividualspeakermustthinkoftheBedeutungofawordasgiveninsomeparticularway,leavingitpossiblefordifferentspeakerstothinkofitasgivenindifferentways.Forsuccessfulcommunication,thespeakersmustknowthattheBedeutungofaword,aseachisusingit,isthesame.Toensurethis,itmustbeaconventionofthelanguagethateachassociateswiththewordthesamesense,thatis,thesamewayofthinkingofsomethingasitsBedeutung.AnimaginaryexamplegivenbyFregeinalettertoJourdainisthattheBedeutungofthename“Afla”mightbegivenasthemountainvisibleonthenorthernhorizonfromsuch-and-suchaplace,andthatofthename“Ateb”asthemountainvisibleonthesouthernhorizonfromacertainotherplace.ItmayprovethatthetwonameshaveasBedeutungtheverysamemoun-tain,whichwasnotatfirstevident;theidentity-statement“AflaandAtebarethesame”isinformativeandreportsanempiricaldiscovery.Famously,inhiscelebratedessay“ÜberSinnundBedeutung”of1892,Fregeusedtheexampleofthenames“theMorningStar”and“theEveningStar,”whichbothdenotetheplanetVenus,toillustratethe15\nMICHAELDUMMETTdistinctionbetweensenseandBedeutungandsoexplainhowatruestatementofiden-titycouldbeinformative(hehadusedthesameexampleearlierinhislecture“FunctionandConcept”of1891).Toknowthesenseofanexpressionis,therefore,toknowhowitsBedeutungisdeter-mined:notnecessarilyhowwecandetermineit,sincewemaylackaneffectivemeansofdoingso,buthow,asitwere,realitydeterminesitinaccordancewiththesensewehavegivenit.Thesenseofapartispartofthesenseofthewhole;thesenseofanygivenexpressionispartofthesenseofanymorecomplexexpressionofwhichthegivenexpressionispart.Soagraspofthesenseofanexpressioninvolvesknowinghowitmaybeputtogetherwithotherexpressionstoformacomplexexpression–ultimately,asentence–andhowthesenseofthecomplexisdeterminedfromthesensesofitsparts.Tograspthesenseofaconcept-expressionistoapprehendaparticularwayofthinkingofsomethingincompleteasitsBedeutung,somethingthatassociateseacharbitraryobjectwithatruth-value:aconceptthatcarrieseachobjectintothevaluetrueorthevaluefalseaccordingasitfallsundertheconceptornot.Ingeneral,wegraspthesenseofawholesentencebygraspingthesenseofeachexpressioncomposingit,whichisitscontributiontothesenseofthesentenceasawhole;andtodothisistohaveaparticularconceptionoftheBedeutungofeachconstituent,togetherwithagraspofhowtheseBedeutungencombinetoyieldtheBedeutungofeachphraseandulti-matelyofthesentenceitself.ButtheBedeutungofasentenceisatruth-value;itssenseFregetermsathought.Inthecaseofasentence,thedistinctionbetweensenseandBedeutungisthatbetweenathoughtanditstruth-value.Thustograspathoughtistoapprehendhowitisdetermined–byreality,thoughnotnecessarilybyus–astrueorasfalse.Andtograspasensethatgoestocomposeathoughtbybeingthesenseofaconstituentofasentencethatexpressesthatthoughtistounderstandhowthecontri-butiontodeterminingthetruth-valueofthethoughtthatismadebythatconstituentisitselfdetermined.InthewordsofGrundgesetze,PartI,thethoughtexpressedbyasentenceisthethoughtthattheconditionforitstruthisfulfilled.Thiswasanexpres-sionofwhathasbecomethemostpopularformofatheoryofmeaning,atruth-conditionaltheory:truthisthecentralnotionofsuchatheory,andmeaningistobeexplainedintermsofit.Fregeheldthatanyonewhomakesajudgmentknowsimplicitlywhattruthandfalsityare.Wecanexpressathoughtwithoutassertingorjudgingittobetrue,whichwedowheneverweutterasentencewhosesenseitisbuttowhichassertoricforceisnotattached(e.g.whenweaskwhetheritistrue).Whenwejudgethethoughttobetrue,we“advancefromthethoughttothetruth-value.”Butthisadvanceisnotafurtherthought,totheeffectthattheoriginalthoughtistrue;byprefacingthesen-tenceexpressingthethoughtwiththewords“Itistruethat”,wedonotconferasser-toricforceonit,butmerelyexpresstheverysamethoughtasbefore.ThatiswhyFregesays,inoneofhisposthumouslypublishedwritings,thattheword“true”seemstomaketheimpossiblepossible.Fregeheldthenotionoftruthtobeindefinable:herejectedthecorrespondencetheoryoftruth,andanyothersuchtheorythatprofessestosaywhattruthis.Fregewasvehementlyopposedtopsychologisticexplanationsofconcepts,thatis,ofthesensesoflinguisticexpressions.Heopposedexplanationsintermsoftheinnermentaloperationsbywhichweacquiresuchconcepts.Thesenseofanyexpressionhad16\nGOTTLOBFREGEtobeexplainedobjectively,notsubjectively,intermsoftheconditionsforthetruthofsentencescontainingtheexpression.Athought,forFrege,isnotoneofthecontentsofthemind,asisasense-impressionoramentalimage.Thesearesubjectiveandincommunicable;butitisoftheessenceofthoughtstobecommunicable.Differentpeoplecangrasptheverysamethought;itcannotthereforebeacontentofanyoftheirminds.Thisrejectionofpsychologismwasofthegreatestimportance:itrescuedthephilosophyofthoughtandoflanguagefromexplanationsgivenintermsofprivatepsychologicalprocesses.Frege’salternativeexplanationwasneithersopopularnorsosuccessful.Herecognizednointermediatecategorybetweenthesubjectiveandthewhollyobjective.Hetookthoughtsandtheircomponentsensestoconstitutea“thirdrealm”:likethephysicaluniverse,itsinhabitantsareobjective,but,unlikeit,theyarenotintimeorspaceorperceivedbythesenses.Butitisonlythroughourgraspoftheinhabitantsofthethirdrealmthatmeresense-impressionsareconvertedintoper-ceptions,andsowebecomeawareoftheexternalworld.Wecangraspthoughtsandexpressthem:butwehumanbeingscangraspthemonlyasexpressedinlanguageorinsymbolism.Frege’sattitudetolanguagewasambivalent.Heviewednaturallanguageasfullofdefects:onlywhenitwasconductedbymeansofapurifiedlanguage,suchashislogicalsymbolism,coulddeductivereasoningbeconfidentlyreliedon.Sosomeofthetimeheinveighsagainstlanguage,declaringthatphilosophymuststruggleagainstitandthathisrealconcerniswiththoughtsandnotwiththemeansoftheirexpression.Yetagreatdealofhisdiscussionsareconcernedpreciselywithlanguageanditsworkings.Hisphilosophicallogicisnotatheoryofthought,independentoflanguage:itisasystematictheoryofmeaning,applicabledirectlytoalanguagepurifiedofthedefectsofoureverydayspeech,butindirectlytonaturallanguage.Thepowerofhistheoryofmeaningrestsuponthecapacityofpredicatelogic–thelogichefirstinvented–toanalyzethestructureofagreatrangeofsentencesandofthethoughtstheyexpress.Althoughmanyofhisideaswerenotfoundacceptablebylateranalyticphilosophers,histheorieswereseenasabettermodelofwhatphilosophyshouldaimat,inframingitsbasictheoriesofmeaningandofthought,thananythingsuppliedbyanyotherphilosopher;andhisdiscussionsofproblemswithinthatrealmabetterplacetostartfromthananyother.InrecentyearstherehasbeenagreatrevivalofinterestinFrege’sphilosophyofmathematics,thelateGeorgeBoolosbeingoneofthosetohavecontributedgreatlytoit.ThecomparisonbetweenFrege’sDieGrundlagenderArithmetikandRichardDedekind’sWassindundwassollendieZahlen?,twobookswhichapproachthesamesubjectmatterverydifferently,isextremelyfruitful.Dedekindisconcernedtocharac-terizetheabstractstructureofthesequenceofnaturalnumbers;havingdoneso,hearrivesatthatspecificsequencebyanoperationofpsychologicalabstraction,aquiteillegitimatedevicemuchfavoredbymathematiciansandphilosophersofthetime.Heacknowledgestheuseofthenaturalnumberstogivethecardinalityoffiniteclasses,butonlyasaminorcorollary.ForFrege,bycontrast,thatuseiscentral.Itwasforhimtheprimaryapplicationofthenaturalnumbers,andmustthereforefigureintheirdefinition.“Itisapplicabilityalone,”hewroteinPartIIIofGrundgesetze,“thatraisesarithmeticfromtherankofagametothatofascience.”Hestronglyopposedappeal,suchasthatmadebyJ.S.Mill,toempiricalnotionshavingtodowithoneorotherpar-17\nMICHAELDUMMETTticulartypeofapplication,indefiningthenaturalnumbersortherealnumbers.Buthethoughtitessentialthat,indefiningthem,thegeneralprincipleunderlyingalltheirapplicationsshouldbemadecentraltotheirdefinition.Hencenaturalnumbersweretobepresentedasfinitecardinals:theoperatorintermsofwhichallnumericaltermsweretobeframedwas“thenumberofx’ssuchthat...x...,”whereofcoursethegapwastobefilledbyanexpressionforaconceptoffirstlevel.Frege’saimwastoshowthatarithmeticalwerederivablefrompurelylogicalprin-ciples.Adescriptionofphysicalspaceasnon-Euclideanisintelligible;soEuclideangeometryisnotanalyticallytrue.Bycontrast,anyattempttodescribeaworldinwhichthetruthsofarithmeticfailisincoherent.SinceFregecharacterizeslogicalnotionsasthosewhicharetopic-neutral,applyingtothingsofeverykind,arithmeticalnotionsarealreadylogicalones.But,likeRussell’s“axiomofinfinity,”apropositionmaybeexpressedinlogicaltermswithoutitstruthbeingguaranteedbylogic.Itthereforeremainstobeshownthatwhatwetaketobethefundamentaltruthsofnumbertheoryarederivablefrompurelylogicalprinciples.Fregeendorsedthedefinitionofequicardinalitythatwasbecominggenerallyacceptedbymathematicians,inparticularCantor:TherearejustasmanyFsasGsiffthereisarelationwhichmapstheFsone-to-oneontotheGs.Ifthereisacuponeverysauceronthetable,andeverycuponthetableisonasaucer,weshallknowthattherearejustasmanycupsassaucersonthetablewithoutneces-sarilyknowinghowmanyofeachthereare.InGrundlagenFregeenunciatesabasicprinciplegoverninghiscardinalityoperator:(*)ThenumberofFs=thenumberofGsifftherearejustasmanyFsasGs,“justasmanyas”beinginterpretedinaccordancewiththeforegoingdefinition.Hedecidesthatthecardinalityoperatorcannotbedefinedcontextually,butrequiresanexplicitdefinition:theonethathechoosesis:ThenumberofFs=theclassofconceptsGsuchthattherearejustasmanyFsasGs.HereFregeappealstothenotionofaclassforthefirsttime,althoughheneveragainconsidersclassesofconceptsratherthanofobjects.Buttheappealissolelyforthepurposeofframinganexplicitdefinitionofthecardinalityoperator;Fregeusesitfornothingelsethanprovingtheprinciple(*)fromit:allthetheoremshegoesontoproveaboutthenaturalnumbersarederivedfrom(*)alone,withoutfurtherrecoursetothedefinitionof“thenumberof.”ThetheorysketchedinGrundlageniselaboratedandfullyformalizedinGrundgesetze,PartII.Grundgesetzemakesextensiveuseofthenotionofclasses,or,rather,ofFrege’sgeneralizationofit,thatofvalue-ranges:aclassistheextensionofafirst-levelconcept,whileavalue-rangeistheextensionofafirst-levelfunctionofoneargument.ThelatternotionisforFregethemorefundamentalone,sinceconceptsareforhimaspecialkindoffunction.Fregehadconvincedhimselfthatthenotionofavalue-rangewasalogicalone.TheBasicLawgoverningtheoperatorformingtermsforvalue-rangesisLawV:18\nGOTTLOBFREGEthevalue-rangeoff=thevalue-rangeofgifff(x)=g(x)foreveryx.Itwasofcoursethislawwhichgaverisetothecontradiction.Becauseofthis,interesthascentereduponapossiblemodificationofFrege’sconstructionofnumbertheory,inwhichtherearenovalue-rangesorclasses,butthecardinalityoperator,governedby(*),istreatedasprimitive.Attentionhasfocusedonwhatisnowcalled“Frege’sTheorem,”namelythepropositionthat,usingFrege’sdefinitionsof“0,”“successor,”and“naturalnumber,”allofPeano’saxioms,andhencethewholeofsecond-orderPeanoarithmetic,canbederivedinasecond-ordersystemfrom(*)alone.OpinionsvaryabouthowclosethisresultbringsustoFrege’sgoalofprovingthetruthsofnumbertheorytobeanalytic.WhilemostattentionhasbeenpaidtoFrege’sfoundationsfornumbertheory,somehasbeengiventohisfoundationforthetheoryofrealnumbers,expoundedintheincompletePartIIIofGrundgesetze.UnlikebothCantorandDedekind,Fregedoesnotfirstconstructtherationalnumbersandthendefinerealnumbersintermsofthem:truetohisprinciplethattypesofnumberaredistinguishedbytheirapplications,andholdingthatbothrationalsandirrationalsservetogivethemagnitudeofaquantity,hesimplytreatsrationalnumbersasakindofrealnumber,definingthelatterdirectly.Whilecardinalnumbersanswerquestionsoftheform“Howmany...?,”realnumbersanswerthoseoftheform“Howmuch...?”Anysuchquestionthatcanbeansweredbyarationalnumbercanalsobeansweredbyanirrationalnumber.Therearevariousquantitativedomains–lengths,durations,masses,electriccharge,etc.;withineach,themagnitudeofaquantityisgivenastheratioofthegivenquantitytosomechosenunitquantity;theseratiosarethesamefromdomaintodomain.Thusrealnumbersaretobedefinedasratiosofquantitiesbelongingtothesamedomain;suchadefini-tionaccordswithFrege’sgeneraltenet,thatthedefinitionofatypeofnumbershouldincorporatethegeneralprincipleunderlyingallitsapplications.InthesectionsofPartIIIincludedinVolumeIIofGrundgesetze,Fregeisconcernedtocharacterizequantitativedomains,andheidentifiesthemasgroupsofpermutationsofanunderlyingsetsatisfyingcertainconditions.UnknowntoFrege,thisworkhadbeenpartiallyanticipatedbyOttoHölderinanarticleof1901.NeitherFregenorHölderusesexplicitgroup-theoreticalterminology.Bothofthemwereconcernedwithgroupswithanorderinguponthem.HölderisgenerallycreditedwithhavingprovedtheArchimedeanlawfromthecompletenessoftheordering,whichFregealsoproved;butFrege’sassumptionsaremuchweakerthanHölder’s.Fregeassumesonlythattheorder-ingisright-invariantandthatitisuppersemi-linear(theorderingislinearupontheelementsgreaterthananygivenelement);Höldermakesthefurtherassumptionsthatitisalsoleft-invariant,fullylinear,anddense.ThispreliminarypartofFrege’scon-structionofthefoundationsofanalysiscontainssubstantialcontributionstogrouptheory,andPartIIIasawholepresentspregnantideasabouthowrealnumbersshouldbeexplained.Frege’sworkonthephilosophyofmathematicsofferedanexplanationofhowdeductivereasoningcanextendourknowledge,andaconceptionofthesignificanceoftheapplicationsofatheorytoitsfoundations.Italsochallengesustosayonwhatourrecognitionofmathematicaltruthrests,ifnotonpurelogicor,moregenerally,onpurelyconceptualtruths.Butitoffersanotherchallengenotsooftenrecognized.19\nMICHAELDUMMETTFrege’sattemptinPartIofGrundgesetzetojustifyhisintroductionofvalue-rangeswasundoubtedlyafailure:hewasattemptingsimultaneouslytospecifythedomainofhisindividualvariablesandtointerprethisprimitivesymbolsoverthatdomain.Buthewasfacingaproblemthatisusuallyleftuntackled:howcanwewithoutcircularityjustifytheexistenceofdomainssufficientlylargetocontaintheobjectsofourfundamentalmathematicaltheoriessuchasnumbertheoryandanalysis?Untilaconvincinganswerisgiventothisquestion,weshallnothaveasatisfyingphilosophyofmathematics.20\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd20012BertrandRussell(1872–1970)THOMASBALDWINRussellwasthemostimportantBritishphilosopherofthetwentiethcentury.Atthestartofthecenturyhehelpedtodevelopthenewtheoriesthattransformedthestudyoflogicatthistime,buthisgreatestcontributionwasnottologicitself.Insteaditlayindevelopinganddemonstratingthephilosophicalimportanceofthisnewlogicandtherebycreatinghis“logical-analyticmethod,”whichisthebasisoftheanalyticalstyleofphilosophyasweknowittoday.TheresultisthatwecanstillreadRussell’swritingsascontributionstocontemporarydebates.Heisnotyetsomeonewhoseworksbelongonlytothehistoryofphilosophy.Helivedtobenearly100andthereiseveryreasontoexpectthatsomeofhiswritingswillhaveanactivelifeandageatleastasgreatashis.Aclassicinstanceisprovidedbyhisintroductiontophilosophy,TheProblemsofPhilosophy(his“shillingshocker”ashelikedtocallit),which,thoughpublishedin1912,remainsoneofthebestpopularintroductionstothesubject.EarlylifeDespitethefactthatasaphilosopherRussellremainsalmostacontemporary,inotherrespectshislifenowseemsverydistantfromus.Hisfamily,theRussells,wasoneofthegreatLiberalfamiliesofBritishpolitics:hispaternalgrandfather,LordJohnRussell,hadbeenPrimeMinistertwiceduringthefirsthalfofthenineteenthcentury,andRusselldescribesmeetingMr.Gladstoneseveraltimes.Hisparents,ViscountAmberleyandhiswifeKate,werefriendswithJohnStuartMill,whoagreedtoactasanhonorarygod-fathertotheiryoungsonBertrand.Millinfactdiedduringthefollowingyear,toosoonforBertrandtomakehisacquaintance.Muchmoretraumaticfortheyoungchild,however,wasthedeathofbothhisparentssoonafterwards,sothatin1876hewasleftattheageof4inthecareofhisgrandparents,indeedofjusthisgrandmotherafterhisgrandfather’sdeathin1878.RusselldescribedhislonelychildhoodinhisAutobiography.HisrebelliouselderbrotherFrankwassentawaytoschool,butBertrand(“asolemnlittleboyinabluevelvetsuit,”1967:30)waseducatedathome,broughtupinaconstrictingatmospherewhosenarrowlimitswerefixedbyhisgrandmother’sstrictPresbyterianbeliefs.TheonlyrefugethattheyoungBertrandfoundwasintheprivacyofhisownthoughts;hekeptasecretdiaryincodeinwhichhesetdownhisgrowingdoubtsaboutreligiousorthodoxy.TherecanbelittledoubtthatRussell’s21\nTHOMASBALDWINtroubledlateremotionallife(hehadfourmarriages)wasaffectedbythischildhood.Inhiswritingstherearemanypassagesinwhichherefersindirectlytoit;forexample,whenwritingin1916aboutthedifficultiesofmarriage,heremarksthat“Thefunda-mentallonelinessintowhichwearebornremainsuntouched,andthehungerforinnercompanionshipremainsunappeased”(1916:191).Fromanearlyagehisprecocioustalentinmathematicshadbeenrecognized,andin1890hewenttoTrinityCollege,Cambridgetostudymathematics.Despitehisdelightinescapingfromhisgrandmother,however,hesoonfoundhimselfdissatisfiedwiththeantiquatedteachingofmathematicsatCambridge.Soin1893heswitchedtothestudyofphilosophy,asubjectintowhichhehadbeeninitiatedthroughmembershipoftheCambridge“Apostles”(aprivatesocietylargelydedicatedtothediscussionofphiloso-phy)andfriendshipwiththephilosopherJ.M.E.McTaggart,thenayoungFellowofTrinity.In1894heobtainedafirstclassresultinhisfinalexaminationsandalmostimmediatelystartedworkonadissertationinthehopeofwinningaprizefellowshipatTrinityCollege.Atthesametimehemarried,andthentravelledwithhisfirstwife,Alys,toGermany.InhisAutobiographyherecountsamomentofclear-mindedfuturereso-lutionduringthishoneymoon:Duringthistimemyintellectualambitionsweretakingshape.Iresolvednottoadoptapro-fession,buttodevotemyselftowriting.Irememberacold,brightdayinearlyspringwhenIwalkedbymyselfintheTiergarten,andmadeprojectsoffuturework.IthoughtthatIwouldwriteoneseriesofbooksonthephilosophyofthesciencesfrompuremathematicstophysiology,andanotherseriesofbooksonsocialquestions.Ihopedthatthetwoseriesmightultimatelymeetinasynthesisatoncescientificandpractical.MyschemewaslargelyinspiredbyHegelianideas.Nevertheless,Ihavetosomeextentfolloweditinlateryears,asmuchatanyrateascouldhavebeenexpected.Themomentwasanimportantandformativeoneasregardsmypurposes.(1967:125)Thispassage(thoughwrittenwiththebenefitofhindsight)isremarkablyprophetic;Russellhadnosettledprofession(heheldteachingpositionsforonlyabouttenyears)andformostofhislifehemadehislivingbywriting,inwhichhehadremarkablepro-ficiency.Hewroteaboutseventybooks,whichdoindeedformtwoseries:thereareabouttwentybooksofphilosophy,mostlyon“thephilosophyofthesciences”;andmanyoftherestconcern“socialquestions,”thoughitcannotbesaidthatthetwoseriesmeetinasynthesis.HismostpopularbookwasAHistoryofWesternPhilosophy,despitethefactthatthisisanunreliableanddistinctlyidiosyncraticbookinwhichRusselldevotesmostspacetoancientandmedievalphilosophy.BreakingwithidealismDuringthe1890sthedominantschoolofphilosophyinCambridge,asinBritaingen-erally,wasidealist.UnderMcTaggart’sinfluenceRussellchosetoworkwithinabroadlyidealistframework(astheTiergartentestamentquotedaboveshows),andtheprojectheselectedforhisfellowshipdissertationwasthatofprovidingarevisedapriorifoun-dationforgeometry,onethatwouldtakeaccountofthepossibilityofnon-EuclideangeometriesinawaythatKant’sfamousaccountdoesnot.RussellarguedthatKant’s22\nBERTRANDRUSSELLconceptionoftheconditionsofthepossibilityofexperiencehadbeentoolimited.Whatisimportant,andthusapriori,isthatspacebeofaconstantcurvature,butitisnotanapriorimatterjustwhatitscurvatureis–forexamplewhetheritiszero,asEuclidmain-tained,orpositive,asRiemannproposed.Russellwasdulyelectedtoasix-yearprizefel-lowshipatTrinityCollegein1895andin1897hepublishedarevisedversionofthedissertation,AnEssayontheFoundationsofGeometry.AswellasofferingaKantianfoundationforgeometryRussellarguedinaHegelianfashionthatwithintheabstractconceptionofpointsinspacecharacteristicofgeometrythereare“contradictions,”whichcanonlyberesolvedbyincorporatinggeometryintoanaccountofthephysicalstructureofspace.Thisledhimintoastudyofthefoundationsofphysicsandinparticulartoastudyoftheproblemsassociatedwiththecontinuityofspaceandtime.Heinitiallyapproachedthesematterswiththepresumptionthattheseproblemsarisefromaconflictbetween,ontheonehand,thefactthatindividualpointsorinstantsdifferonlyinrespectoftheirrelationsand,ontheother,therequirementthat“allrelationsareinternal,”whichhetooktoimplythatdifferencesintherelationshipsbetweenthingsaredependentuponotherdifferencesbetweenthesethings.Russellsummeduptheconflicthereas“thecontradictionofrelativity”:“thecontradictionofadifferencebetweentwoterms,withoutadifferenceintheconceptionsapplicabletothem”(“AnalysisofMathematicalReasoning,”Papers,2:166).Thepresumptionthatthereisacontradictionherewasacommonplaceamongtheidealistlogiciansoftheperiod,suchasF.H.Bradley,andwascentraltotheirdenialthatthereareanyrelationaltruthsandthustotheirmetaphysicalmonism.ItwasthereforebycallingthispresumptionintoquestionthatRussellmadehisbreakwithidealism.Thekeytothiswashisaffirmationoftheindependentrealityofrelations,whichheproposedinhis1899paper“TheClassificationofRelations”(seePapers,2).Oncethismovewasmade,thealleged“contradictionofrelativity”isdissipatedandRussellwasfreetoapproachtheissuesraisedbythecontinuityofspaceandtimeafresh.AlthoughRussell’spapersfromthisperiodshowhimfindinghisownwaytothisanti-idealistthesis,healwaysacknowledgedthedecisiveimportanceofG.E.Moore’swritingsatthistime(“Itwastowardstheendof1898thatMooreandIrebelledagainstbothKantandHegel.Mooreledtheway,butIfollowedcloselyinhisfootsteps,”1995a:42).G.E.MoorewastwoyearsyoungerthanRussell.HavingbeendrawnfromthestudyofclassicstothatofphilosophyatTrinityCollegepartlythroughRussell’sinfluencehegraduatedin1896andcompletedhisown,successful,disserta-tionforaprizefellowshipin1898.ItisinthisdissertationthatMooreworksouthisownbreakwithidealism.Hisbasicclaimisthatoftheunqualifiedrealityoftheobjectsofthought,propositions,asthingsisinnowaydependentuponbeingthoughtabout.Moorefurthermaintainedthatthereisnoreasontoduplicateontologicalstruc-turesbyhypothesizingtheexistenceoffactsfortruepropositionstocorrespondto.Insteadtherearejustpropositionsandtheirconstituents:theworldjustcomprisesthetotalityoftruepropositions,andanaccountofthestructureofpropositionsisanaccountofthestructureofrealityitself.Oneimplicationofthisisthatthestructureofspaceisindependentofourexperienceofitand,therefore,oftheconditionsunderwhichexperienceofitispossible.SoMoorewasverycriticalofRussell’sneo-Kantianaccountofgeometry,andRussellquicklycametoagreewithMooreonthismatter(seeMOORE).23\nTHOMASBALDWINRussellreadMoore’sdissertationatanearlystage,andimmediatelyacceptedmanyofMoore’scentralpoints,includinginparticularhisconceptionofaproposition.Asweshallsee,manyofhislaterdifficultiescanbetracedbacktothis.ButatthetimeRussellwasexhilaratedbythepossibilitiesthatthisnewrealistphilosophyopenedoutbeforehim:Butitwasnotonlytheseratherdry,logicaldoctrines[concerningtherealityofrelations]thatmademerejoiceinthenewphilosophy.Ifeltit,infact,asagreatliberation,asifIhadescapedfromahot-houseontoawind-sweptheadland.(1995a:48)TheprinciplesofmathematicsThistremendoussenseofintellectualliberationquicklybecamefocusedonanewproject,whichwastodominateRussell’sthoughtandlifeforthenexttenyears:the“logicist”projectofdemonstratingthat“allmathematicsisSymbolicLogic.”Theocca-sionwhichfiredRussell’senthusiasmforundertakingthisprojectwashisvisitinJuly1900totheInternationalCongressofPhilosophyinParis,whereheheardPeanodiscusshisformalizationofarithmeticusingnewlogicaltechniques.Peanodidnothimselfseektoprovideapurelylogicalfoundationformathematics;hedidnotofferlogicaldefinitionsoftheconcepts“0,”“successor,”and“number”whichoccurinhispostulates.But,onhearinghim,Russelljumpedtothehypothesisthatdefinitionsofthiskindshouldbepossible,andthusthatmathematicsis,intheend,justlogic.InmakingthisjumpRussellwasdrawingonhisrecentclosestudyofthephilosophyofLeibniz,which,fortuitously,hehadundertakentheyearbefore(simplybecausehestoodinforMcTaggartwhoshouldhavebeenteachingit).RussellrecognizedthatLeibnizhadalsoconceivedthishypothesisbuthadbeenpreventedfromdemonstratingit,largelybecauseoftheinadequaciesofthetraditionallogictowhichheadhered.ButwiththericherresourcesofthelogicemployedbyPeano(whichRussellimmediatelyusedtodevelopanewlogicofrelations),RussellsupposedthatLeibniz’slogicisthypoth-esiscouldnowbevindicated.AnimportantaspectofRussell’snewprojectwastheopportunityitprovidedhimtocontinuehiscriticismsofidealistphilosophy:Thequestionsofchiefimportancetous,asregardstheKantiantheory,aretwo,namely,(1)arethereasoningsinmathematicsinanywaydifferentfromthoseofFormalLogic?(2)arethereanycontradictionsinthenotionsoftimeandspace?IfthesetwopillarsoftheKantianedificecanbepulleddown,weshallhavesuccessfullyplayedthepartofSamsontowardshisdisciples.(1903:457)InprovidinganegativeanswertothesecondofthesequestionsRusselldrewontheworkofthegreatGermanmathematiciansofthenineteenthcentury,Dedekind,Weierstrass,andCantor,whoseworkhehaddiscoveredafewyearsearlierbuthadnotthenappreciatedfullybecauseofhisattachmenttoidealistdoctrines.Henowdevotedacentralsectionofhisnewbook,ThePrinciplesofMathematics,toacarefulexpositionoftheirphilosophyoftheinfinite,fromwhichheconcluded“thatalltheusualarguments,bothastoinfinityandastocontinuity,arefallacious,andthatnodefinite24\nBERTRANDRUSSELLcontradictioncanbeprovedconcerningeither”(1903:368).Sincetheidealistlogicianshadadvancedtheoppositeview,andthenusedtheinfinityofspaceandtimetoarguefortheirunreality,Russellfeltthathewashereprovidingadefinitiverefutationoftheirposition.InlayingoutthisnewphilosophyRussellworkedatextraordinaryspeed.HestartedwritingThePrinciplesofMathematicsinOctober1900andbytheendoftheyearhehadcompletedafirstdraft:thebookaswenowhaveitismorethan500pageslongand300ofthesecomeunchangedfromthatfirstdraft.Thisperiodwas,hewrote,anintellectualhoneymoonsuchasIhaveneverexperiencedbeforeorsince.EverydayIfoundmyselfunderstandingsomethingthatIhadnotunderstoodonthepreviousday.Ithoughtalldifficultiesweresolved,allproblemswereatanend.(1995a:56)But,hecontinues,Thehoneymooncouldnotlast,andearlyinthefollowingyearintellectualsorrowdescendeduponmeinfullmeasure.(1995a:56)Themainreasonfortheonsetofthissorrowwashisdiscovery,earlyin1901,of“thecontradiction,”nowusuallyknownas“Russell’sparadox.”Thisisacontradictionthatcanbeeasilydemonstratedjustatthepointatwhichoneseekstodevelopelementarylogicintosettheoryinordertoshowhowarithmeticcanbeestablishedonthebasisoflogicalone.RusselldiscoveredthecontradictionwhenreflectinguponCantor’s“paradox”thatthereisnogreatestcardinalnumber.Cantor’sparadoxrestsonthetheoremthatthenumberofsubsetsofagivensetSisalwaysgreaterthanthenumberofmembersofSitself.Cantorprovesthistheorembydeducingacontradictionfromthehypothesisthatthereisaone-to-onecorrelationbetweenthesubsetsofSandthemembersofS,whichwouldimply,onthecontrary,thattheirnumbersarethesame.Thecontradictionarisesasfollows:considerthatsubsetofSwhosemembersarejustthosemembersofSwhichdonotbelongtothesubsetofSwithwhichtheyarecorre-latedunderthehypothesizedcorrelation.Sincethisso-called“diagonal”setDisasubsetofSittoomustbecorrelatedwithamemberofS,sayd.CantornowaskswhetherdbelongstoD:giventhewaydandDhavebeendefined,itturnsoutthatdbelongstoDifandonlyifddoesnotbelongtoD,fromwhichitiseasytoderivetheexplicitcontra-dictionthatdbothbelongstoDanddoesnotbelongtoD.ThestepfromCantor’stheoremtoRussell’sparadoxisverysimple.InsteadofCantor’shypotheticalone-to-onecorrelationbetweenthemembersofasetanditssubsets,considerinsteadthenon-hypotheticalidentityrelationbetweenanythinganditself.ThentheanalogueofCantor’s“diagonal”setDofthingsthatarenotmembersofthesetwithwhichtheyarehypotheticallycorrelatedbecomessimplythesetRofthingsthatarenotmembersofthemselves.Underthe“identity”correla-tionRisofcourse“correlated”withitself;henceinaskingwhetherRbelongstothatwithwhichitiscorrelatedwearesimplyaskingwhetherRbelongstoR.ButsinceRjustisthesetofthingswhichdonotbelongtothemselves,itfollowsthatRbelongstoRifandonlyifRdoesnotbelongtoR.Thistooimmediatelygivesrisetoanexplicitcontradiction,butinthiscasethederivationdoesnotdependonahypothetical25\nTHOMASBALDWINcorrelationwhichistherebyprovednottoexist,butonlyonthenon-hypotheticalidentityofathingwithitselfwhichcannotberejected.SointhiscasethereisnoobviouspositiveconclusiontobedrawncomparabletoCantor’stheorem.Thereisinsteadtheutterlydismayingimplicationthatthereissomethingseriouslyamissinthefoundationsoflogic.AssoonashehaddiscoveredthiscontradictionRussellcommunicatedittoFrege,whoseworkshehadjustrecentlyreadproperlyforthefirsttimeandrecognizedforwhattheywere,namelymuchthemostsophisticatedattempttodevelopalogicistprogramofthekindhewasalsoengagedupon.FregereceivedRussell’sletterjustasthesecondvolumeofhisGrundgesetzewasinpress,andaddedthefamousAppendixII,whichbegins,“Hardlyanythingmoreunfortunatecanbefallascientificwriterthantohaveoneofthefoundationsofhisedificeshakenaftertheworkisfinished.”FregedidinfactsuggestawayofcircumventingRussell’sparadox,buthecouldnotshowthatitworked,anditisnowknownnotto.RussellalsoattemptedtofindawayaroundtheparadoxandendedThePrinciplesofMathematicswithatentativeproposalthatthedefinitionofthesetRisill-formedbecauseaset,beingofadifferent“type”fromthatofitsmembers,cannotbeamemberofitself.Asweshallseebelow,thisisapro-posaltowhichhereturnedlater;butinthecontextoftheThePrinciplesofMathematicsitwasnoteasyforhimtoadvanceit,sinceitconflictswiththeconceptionoflogicadvancedthere,namelythatthetruthsoflogicaretruthswhichareabsolutelyuniversal,andarethereforenottoberestrictedbyconsiderationsarisingfromthetypeofthingunderdiscussion.ThecontradictionwasnottheonlyproblemtodelaypublicationofThePrinciplesofMathematicsuntil1903andtodominatehisresearchforthenextfewyears.Healsoranintoatangleofdifficultiesconcerningthestructureofjudgment,withwhichhecontinuedtograpplethereafter.Thesedifficultiesare,broadly,oftwokinds:concern-ing(1)theunityofjudgmentand(2)thestructureofgeneraljudgments.InThePrinciplesofMathematicsRussell’streatmentofthesemattersisexpressedthroughadiscussionofthestructureofpropositions,which,followingMoore,hetakestocompriseboththeobjectsofjudgmentandtheobjectivestructureoftheworld.Soconceived,propositionsarenotrepresentationswhich,whentrue,correspondtoafact.Insteadtruepropositionsjustarefacts–thereisnodifferencebetweenthedeathofCaesarandthe(true)propositionthatCaesarisdead.SincehisdeathissomethingthatbefellCaesarhimself,Caesarishimselfa“constituent”ofthepropositionthatCaesarisdead.Indeedthepropositionjustisa“complex”whoseconstituentsareCaesaranddeath(whichisa“predicate”).Thedifficultythatnowarisesconcerningthe“unity”ofjudgmentisthatofexplaininghowitisthatacompletepropositionisconstituted.RusselldiscussesthisinconnectionwiththepropositionthatAdiffersfromB.TheconstituentshereareA,B,anddifference;butspecifyingthemdoesnotyetspecifythepropositioninquestion,fortheyarealsotheconstituentsofthedifferentpropositionthatBdiffersfromA(becausedifferenceisasymmetricrelationthisisactuallyapoorcasetohavetaken.Thepointismuchclearerwithanasymmetricrelation,suchasoccursinthepropositionthatAislargerthanB,whichismanifestlydifferentfromthepropo-sitionthatBislargerthanAalthoughithasthesameconstituents).Russellsumsuphisdiscussion:26\nBERTRANDRUSSELLThedifferencewhichoccursinthepropositionactuallyrelatesAandB,whereasthedifferenceafteranalysisisanotionwhichhasnoconnectionwithAandB.(1903:49)Furthermore,Russellnotes,itdoesnothelpifoneaddsthatthedifferenceinthecasewewantisadifferenceofAfromB,forallthattheseadditionsdoistoaddfurtherrela-tionstothesupposedconstituentsofthepropositionwithout“actuallyrelating”therelationofdifferencetotherightterms.TheproblemisthatAproposition,infact,isessentiallyaunity,andwhenanalysishasdestroyedtheunity,noenumerationofconstituentswillrestoretheproposition.(1903:50)Thepointraisedhereisonethatwastocomebacktoplaguehim.InThePrinciplesofMathematicsRussell,havingidentifiedit,simplysetsitasideforfurthertreatment.Thisisdisappointing,forherethereisastraightforwardchallengetotheMoore–Russellconceptionofapropositionasa“complexwhole”comprisedofitselementarycon-stituents.Indeedthepointwasnotnew:thisdifficultyconcerningrelationaljudgmentshadbeenfamouslysetoutbyF.H.BradleyinchapterIIIofAppearanceandReality(1893).ThepersonwhofirstsawclearlythewaytodefusetheissueherewasFrege:forthroughhisfamous“context”principle(presentedinhisGrundlagen,1884)thatitisonlyinthecontextofasentencethatawordhasameaning,heacknowledgestheirre-ducibleprimacyofjudgmentsinsteadofregardingthemas“complexes”tobecon-structedoutofelementaryconstituents.WeshallseebelowthatRussell’stheoryofdescriptionsincludesapartialacknowledgmentofthecontextprinciple;buthehimselfnevergeneralizesittosolvethisproblemoftheunityofjudgment.TheothergeneraltypeofdifficultythatRussellencounteredinThePrinciplesofMathematicsconcernsthestructureofgeneralpropositionssuchasthepropositionthatImetaman.Thedifficultyhereissupposedtocomefromthefactthat,ontheonehand,theconceptamanisaconstituentofthisproposition;but,ontheotherhand,suchaconceptdoesnotwalkthestreets,butlivesintheshadowylimboofthelogic-books.WhatImetwasathing,notaconcept,anactualmanwithatailorandabank-accountorapublichouseandadrunkenwife.(1903:53)Russell’sargumenthereisintuitiveandquestionable.Butitisclearthatthetensionarisesfromthedualroleofpropositionsasbothobjectsofthought(andthusconsti-tutedofconceptssuchasaman)andsituationswithintheworld(andthereforecon-stitutednotfromgeneralconcepts,butfromparticularmen).Russell’swayofresolvingthistensionistosaythatconceptssuchasaman“denote”thethingsthatapropositioninwhichtheyoccuris“about”;anditisthethingsthataredenotedinthiswaythat“walkthestreets”etc.Itisdifficultatfirstnottointerpretthistalkofthatwhichapropositionis“about”asawayofimplicitlyspecifyinga“truth-maker”forthepropositiondistinctfromthepropositionitself;butthisofcoursewouldbeentirelyinimicaltoRussell’sconceptionofaproposition.Furthermore,Russelldevelopshisaccountofdenotinginsuchawayastomakethisinterpretationinappropriate.Forhegoesontoarguethat27\nTHOMASBALDWINsomemanmustnotberegardedasactuallydenotingSmithandactuallydenotingBrownandsoon:thewholeprocessionofhumanbeingsthroughouttheagesisalwaysrelevanttoeverypropositioninwhichsomemanoccurs,andwhatisdenotedisessentiallynoteachseparateman,butakindofcombinationofallmen.(1903:62)AsRussellacknowledges,sucha“combinationofallmen”isaveryparadoxicalobject;inthecaseofaman,whatisdenotedissupposedtobeSmithorBrownor...(forthewholehumanrace).ItisinfactclearenoughwhatRussellisseekingtodohere:namely,toeffectareduc-tionofgeneralpropositionstopropositionsthatinvolveonlydisjunctionsorconjunc-tionsofsingularpropositions,forwhichthetensionbetweenpropositionsasobjectsofthoughtandastruthmakersisnotsoacute.Hewritesthat“thenotionofdenotingmaybeobtainedbyakindoflogicalgenesisfromsubject-predicatepropositions”(1903:54),andasGeachhasobserved,Russell’saccountisinthisrespectcomparabletomedievaltheoriesofsuppositio.ButRussellmuddlesthingsbysupposingthatheneedstoholdthattherearedisjunctiveandconjunctivecombinationsofthingsdenotedbythedenot-ingconceptsthatoccurinthegeneralpropositions.Thesearethe“paradoxicalobjects”whichthepropositionsinquestionaretobe“about.”Notonlyaresuchobjectsintrin-sicallyobjectionable(asRussellhimselfacknowledges,1903:55n.),thiswayofcomingatthemattermakesitimpossibletoprovideacoherenttreatmentofpropositionsinvolvingmultiplegenerality,sincethereisnowayofrepresentingscopedistinctions,suchasthedistinctionbetweenthetwowaysofinterpreting“Everyonelovessomeone.”Russellineffectacknowledgesthispointhimselfwhendiscussingvariables,bothfreeandbound.Hewouldliketohandlefreevariableswithinhistheoryofdenotingascasesofthedenotingconceptanyterm;buthecanseethatthisapproachdoesnotdealprop-erlywiththeroleofrepeatedvariables.HisdiscussionofboundvariablesoccursaspartofhisexpositionofPeano’sconceptionof“formalimplication”asauniversallyquan-tifiedconditional,asin“forallx,ifxisamanthenxismortal.”Russellwantstobeabletoapplyhistheoryofdenotingconceptstothequantifier“forallx,”asaconceptdenotingsomecombinationofthingswhichthewholepropositionis“about.”Buthecanalsoseethatwheretherearemultiplequantifiersbindingdifferentvariables(asinbothinterpretationsof“Everyonelovessomeone”)thevariablesaretiedtothequan-tifiersinawaythatblocksoffthisconceptionofthedenotationofaquantifier.Soalthoughhecannotbringhimselftosaysoexplicitly,histheoryofdenotingconceptsisinadequatetothenewlogicofquantifiersandvariablesuponwhichthelogicistprojectofThePrinciplesofMathematicsisfounded.ThetheoryofdescriptionsTwoyearslater,in1905,Russellpublishedhismostfamouspaper,“OnDenoting.”Hebeginsby,ineffect,developinghisearlierdiscussionofformalimplicationintoasys-tematicaccountofthepropositionsexpressedbysentencesinvolvingwhathenowcalls“denotingphrases”suchas“allmen,”“aman,”and“noman.”Thereisnownotalkofdenotingconcepts;insteadheusestheuniversalquantifierandboundvariablestospecifythepropositionsexpressedbythesesentencesbyreferencetothetruthofsimplerpropositions.Thushenowsaysthatthepropositionexpressedby“Imetaman”isthe28\nBERTRANDRUSSELLpropositionthatpropositionsofthetype“Imetx,andxishuman”arenotalwaysfalse(Papers,4:416).Clearly,muchishereassumed,forexampletheinterdefinabilityoftheexistentialanduniversalquantifiers.Thereisalsoadegreeofoversimplification,since,asherecog-nizeswhendealingwithmultiplequantifiers,heactuallyneedstospecifythevariableinthequantifier;inthecaseaboveheshouldhavesaid“arenotalwaysfalseofx.”Settingthesepointsaside,whatisworthconsideringiswhetherRusselloffersanygeneralaccountofquantifiersandvariablestoreplacethetheoryofdenotingconceptsthathasbeentacitlydiscarded.In“OnDenoting”itselfRussellisunhelpful:hejustsays“Herethenotion‘C(x)isalwaystrue’istakenasultimateandindefinable”(Papers,4:416).If,however,onelooksaheadtothediscussionoftheuniversalquantifierintheintroductionthatRussellwrotetoPrincipiaMathematica(1910),onefindshimusingthelanguageof“ambiguousdenotation”tosketchwhatisnowrecognizableasasub-stitutionalaccountofthequantifier.Forhenowsaysthatweassertauniversalpropo-sitioninordertocondensetheassertionofallthesubstitutioninstances“ambiguouslydenoted”(1910:40)bythepropositionalfunctionthatoccursinouruniversalpropo-sition;andthetruthoftheuniversalpropositiondependsonthe“elementarytruth”ofallthesesubstitutioninstances(p.42).Sincethemotivationbehindtheoriginaltheoryofdenotingconceptswasthatpropositionsinvolvingallmen,aman,etc.areinsomeway“about”theirinstances,itisunsurprisingthatheendsupwithasubstitutionaltreatmentofquantification,dealing,ofcourse,withsubstitutionsinpropositions,notsentences.Russell’smaintopicin“OnDenoting”isthestructureofpropositionswhoseexpres-sioninvolvesdefinitedescriptions,phrasessuchas“ThepresentPresidentoftheUSA.”InThePrinciplesofMathematicshehadappliedhistheoryofdenotingconceptstosuchpropositionsinordertoprovideanaccountofwhyitisthattrueidentities,propositionsexpressedbysentencessuchas“BillClintonisthepresentPresidentoftheUSA,”areofinteresttous(1903:62–4).Theproblemhereisfamiliar:ifwejusttakeitthatwehavetwonamesforthesamething,andthattheproposition’sconstituentsarejustthethingthusnamedtwiceandtherelationofidentity,itseemsthattheverysamepropo-sitionisalsoexpressedby“BillClintonisBillClinton,”whichisofnointeresttous.InThePrinciplesofMathematicsRusselltookitthatthisproblemissolvedbythehypoth-esisthatthedescription“thepresentPresidentoftheUSA”introducesacorrespondingdenotingconceptintothepropositionexpressedthroughitsuse,whichofcoursedoesnotoccurasaconstituentofthepropositionexpressedby“BillClintonisBillClinton.”Russelldoesnotexplainhowthisdenotingconceptoccursasaconstituentofthepropositionexpressed,andhesaysotherthingsaboutitonthebasisofwhichitiseasytoreopentheoldproblem.SincethepropositioncannotcompriseBillClinton’sidentitywiththedenotingconceptinquestion,itseemsthatitmustcompriseBillClinton’siden-titywiththethingdenotedbythedenotingconcept,thatwhichthepropositionis“about”inRussell’sintuitivesense.ButthisthingisofcoursejustBillClintonhimself,andwehavenowcomebacktothedifficultyofshowingwhythispropositionisofanyinteresttoussinceitisequallyexpressedby“BillClintonisBillClinton.”IfonelookstoRussell’sextensiveunpublishedwritingsonthismatterfromtheperiod1903–5(seePapers,4),onecan,Ithink,seehimidentifyingthisdifficultyforhisoldposition.Healsobeginstothinkaboutadifferentissue,alsoproblematicforhisoldposition,which29\nTHOMASBALDWINarisesfromhiscriticalreactiontosomeworksbythecontemporaryAustrianphiloso-pherAlexiusMeinong,whichhestudiedcloselyatthistime.Inthiscasetheissueconcernsthepropertreatmentof“empty”descriptions,descrip-tionssuchas“thepresentKingofFrance”which,thoughmeaningful,describenothingthatactuallyexists.RussellinterpretedMeinongasadvancingatheoryof“objects”accordingtowhichemptydescriptionsofallkindsdescribeanobject,evendescriptionsofimpossibleobjectssuchas“theroundsquare.”In“OnDenoting”andthereafterRussellridiculedthispositionasconflictingwith“therobustsenseofreality”which“oughttobepreservedeveninthemostabstractstudies”(1919:169–70);butintruthMeinong’spositionwasagooddealmoresubtlethanRussellappreciated.Whatisimportanthere,however,isthatthroughthinkingaboutMeinong’sworkRussellcametoappreciatetheimportanceofconstructingatheorythatwouldallowforthepossi-bilityofmeaningfulsentencesthatincludeemptydescriptions–sentencessuchas“ThepresentKingofFranceisbald.”SinceRusselltookitthatthemeaningofanamewasjustthename’sbearer,itfollowedatoncethatsuchdescriptionswerenotnames.This,herecognized,didnotruleoutthetreatmentofsuchdescriptionsasintroducingdenot-ingconceptsintoaproposition.But,heargued,therewasstillaproblem:theproposi-tionexpressedby“ThepresentKingofFranceisbald”oughttobeaboutthepresentKingofFrance.Butthereisnosuchthing;sothetheoryimpliesthatthepropositionisaboutnothing,whichRusselltakestoimplythat“itoughttobenonsense.”But,hecon-tinues,“itisnotnonsense,sinceitisplainlyfalse”(“OnDenoting,”Papers,4:419).Thisargumentiscondensed,butthewaytounderstandthecrucialmovefrombeing“aboutnothing”to“beingnonsense”istoconnectthe“aboutness”thesiswiththethesisthatthetruthorfalsehoodofapropositionwithadenotingconceptdependsonthetruthorfalsehoodofthepropositionsspecifiedbyreferencetothatwhichthefirstpropositionisabout.Forinthelightofthis,apropositionaboutnothingwillbeoneforwhichtherearenosuchpropositionstodetermineitstruthorfalsehood.Soitcanbeneithertruenorfalseitself;butthisisabsurdsincepropositionsareinherentlyeithertrueorfalse.Henceitfollowsthatthesentencewithanemptydescriptionfailsafteralltoexpressaproposition:inwhichcase“itoughttobenonsense.”Wenowhavethetwo“puzzles”which,Russellsaysin“OnDenoting,”anadequatetheoryofdescriptionsmustsolve:(1)Whyisitthatquestionsaboutidentityareoftenofinteresttous–howcanitbethatGeorgeIVwishedtoknowwhetherScottwastheauthorofWaverleybutdidnotwishtoknowthatScottwasScott?(2)Whatproposi-tionsareexpressedbysentenceswithemptydescriptions,suchas“ThepresentKingofFranceisbald,”andhowistheirtruthorfalsehooddetermined?Beforeshowinghowhisnewtheorycansolvethesepuzzles,Russellexplainswhyherejectstheviewthatoneshouldaccountforthesepuzzlesbydistinguishingbetweenthe“meaning”andthe“denotation”ofadescription.HespecificallymentionsFregeinconnectionwiththisviewanditisnaturaltothinkofhimasarguinghereagainstFrege’stheoryofSinnandBedeutungwhichwasofcoursepreciselyintroducedtohandlethefirstofRussell’spuzzlesandseemswellsuitedtohandlethesecond(seeFREGE).Russell’sargumentin“OnDenoting”againstthe“Fregean”positionisnotoriouslyobscure.Hisconclusionisthatthereisan“inextricabletangle”intheaccountthisposi-tionoffersoftherelationbetweenthemeaningandthedenotationofadescription30\nBERTRANDRUSSELL(Papers,4:422);butintruthitisRussell’sowndiscussionwhichcertainlyappearstobeaninextricabletangle.MyownviewisthatRussell’sargumentisvitiatedbythefactthatheadaptsFrege’stheorytohisowndifferentconceptionofaproposition,andtherebycruciallydistortsFrege’sactualposition.(Russellhimselfacknowledgesthedif-ferenceshere:Papers,4:419n.9.)Thiscanbeillustratedbyconsideringaslightlylater,andconsiderablyclearer,discussionbyRussellofthesameissue,in“KnowledgebyAcquaintanceandKnowledgebyDescription,”inwhichheisconsideringthestructureofthepropositionexpressedby“theauthorofWaverleyistheauthorofMarmion”(Papers,6:159).Russellarguesherethat(1)thepropositioninvolvesanidentity;but(2)plainlydoesnotinvolvetheidentityofthemeaningsof“theauthorofWaverley”and“theauthorofMarmion”;so(3)itmustcomprisetheidentityofthedenotationsdeterminedbythemeaningsof“theauthorofWaverley”and“theauthorofMarmion.”ButthesedenotationsarejustScotthimself,sothepropositioninquestionisjustthepropositionthatScottisScott.Yetthewholepointofthemeaning/denotationtheorywastopreservethedistinctionbetweeninterestingandtrivialidentities.ForFregethisargumentsimplygetsoffonthewrongfoot.Thethoughtexpressedby“theauthorofWaverleyistheauthorofMarmion”isindeedathoughtaboutiden-tity;butthethoughtitselfisnotstructuredbytherelationofidentity,butbythesenseemployedhereofthisrelation,andthissensecanthenrelatethesensesexpressedbythetwodescriptionswithoutturningitintoathoughtthatthesesensesarethesame.Russell’sargumentagainstthemeaning/denotationdistinctionasappliedtodescrip-tionsonlyworksinsofarasitdependsonanon-FregeanhybridconceptionofapropositionwhichviolatesFrege’sdistinctionbetweenSinnandBedeutung.ThefactthatRussell’sargumentagainstFregeisafailuredoesnot,ofcourse,vin-dicateFrege’streatmentofdescriptionsasfunctionalexpressionsorundermineRussell’sowntheoryofdescriptions.Thekeytothisnewtheoryis,asRussellputitlater,thatdescriptionsare“incompletesymbols,”thatis,phraseswhich“havenomeaninginisolation”butareonly“definedincertaincontexts”(1910:66).Assuch,forRussell,descriptionssuchas“theman”aretoberegardedasessentiallysimilartotheotherdenotingphrasesdiscussedin“OnDenoting,”e.g.“aman”and“allmen.”Theyhave“nomeaninginisolation”inthesensethatthereisnothing(notevenaconcept)thatistheir“meaning”andthatoccursasaconstituentofthepropositionsexpressedbysentencesinwhichtheyoccur.Insteadtheycontributetothesepropositionsinmorecomplexwaysbyfixingtheirstructure,inthewaythatRussellconceivesoftheroleoftheuniversalquantifierasdescribedabove.ItisthereforenosurprisethatonRussell’snewtheoryofdescriptions,theroleofdescriptionsiselucidatedbyspellingoutthequantificationalstructureofthepropositionsexpressedbysentencesinwhichtheyoccur.ThisturnsouttobeacomplexmattersinceRussellstickstohisinitialassump-tionthatallsuchpropositionsinvolveonlytheuniversalquantifier.ButwecangraspthekeypointsofRussell’snewtheorybyprovidingjustthefirststageofit,whichisthatthepropositionexpressedby“TheauthorofWaverleyisScott”ismoreclearlyiden-tifiedasthepropositionexpressedbythefollowingsentence:Forsomex,(i)xisanauthorofWaverleyand,(ii)forally,ifyisanauthorofWaverleytheny=x,and,(iii)x=Scott.31\nTHOMASBALDWINOncethepropositionisidentifiedinthiswaysolutionstoRussell’stwopuzzlesareimmediate.ThereisnoreasontoidentifythecomplexpropositionthusexpressedwiththepropositionthatScottisScott,forthedescription“theauthorofWaverley”isnotconstruedasgivingrisetoacomplexnameofaconstituentwithinthepropositionwhichturnsouttobejustScott.Sothedistinctionbetweensignificantandtrivialiden-titiesisclearlypreserved.Secondly,theroleofemptydescriptionsiseasilyallowedfor:thepropositionexpressedby“ThepresentKingofFranceisbald”istobeidentifiedasthatexpressedbythesentence:Forsomex,(i)xisapresentKingofFranceand,(ii)forally,ifyisapresentKingofFrancetheny=x,and,(iii)xisbald.Itisthenunproblematicthatthereissuchaproposition,andthatitis“clearlyfalse”asRusselldeclaresthatitshouldbe.Thefactthatthesesolutionsaresostraightforward,anddonotdependontheunder-lyingtheoryofpropositions(thoughtheyareconsistentwithit),showsthatonecanabstractRussell’stheoryofdescriptionsfromthisunderlyingtheory.ThisisinfactwhathaslargelyhappenedtoRussell’stheoryofdescriptions:itistakentorestonthethesisthatdescriptionsarequantifiers,andassuchtheRussellianpositionisbestconceivedofasonethatemploysarestricted“definite”quantifiertoconstruedefinitedescriptions,sothatthelogicalformof“theauthorofWaverleyisScott”canbebettercapturedbyconstruingitasForthexwhoisanauthorofWaverley,x=Scott.OnecantheninterpretRussell’sreductionofthedefinitequantifiertootherquantifiersasamisleadinglyexpressedwayofspellingoutthetruth-conditionsofthissentence.Oncethematterishandledinthisway,thedebatewithFregecanbere-opened,asadebateastowhetherthiswayofconstruingdescriptionsispreferabletoFrege’sapproach,accordingtowhichtheyarecomplexsingularterms,comparabletofunc-tionalexpressionsinmathematics,suchthatthelogicalformof“theauthorofWaverleywasScott”iscapturedbyconstruingitasTheauthorof(Waverley)=Scott.ThisdebatewasfamouslyrevivedbyStrawsoninhisattackonRussell.Strawsonintroducedarangeoflinguisticdataconcerningtheuseofemptydescriptionsinsituationswhichconflictwithour“presupposition”thatdescriptionsarenon-empty,andarguedontheirbasisthataFregeanpositionisinfactpreferabletoRussell’s;inmanycases,Strawsonargued,wetakeitthattheuseofemptydescriptionsissuesinstatementsthatare“neithertruenorfalse”andnot“plainlyfalse”inthewaythatRussellmaintained.Infact,however,thelinguisticdatainthisareaareindeterminate,andmostcontemporarydiscussionsfocusinsteadontherelativemeritsofalternativeaccountsofthelogicalbehaviorofdescriptionsincomplexsentencesinvolvingtemporalmodifiersandcounterfactualconstructions.Evenwhentheseareintroduced,however,theissueremainssurprisinglyopen–somuchsothatitseemstomebesttoconcludethatdefinitedescriptionsblurtheapparentlysharplogicaldistinctionbetweenparticularthoughtsinvolvingasingulartermandgeneralthoughtsinvolvingaquantifier.32\nBERTRANDRUSSELLReturning,however,toRussellhimselfIwantfinallytodiscussthesignificanceofthetheoryofdescriptionswithinhisphilosophygenerally.ThefirstpointisthatRussellfeltthathistheoryshowedthattherewasnoneedtoabandonhisone-dimensionalconceptionofmeaninginfavorofaFregeantheorywithitsall-encompassingdistinctionbetweenSinnandBedeutung.Indeed,aswehaveseen,RussellfeltthathecouldcopebetterthanFregewiththe“puzzles”ofinterestingidentitiesandemptydescriptions.ThishadanimportantimplicationforRussell’streatmentofnames:withnoSinn/BedeutungdistinctionRussellwascommittedtotheviewthatallidentitiesinvolvingjustnamesaretrivialandthattherecannotbemeaningfulemptynames.Soallputativecounterexampleshadtobehandledbysupposingthatthenamesinquestionwerereallyjustdescriptionsunderdisguise.ThescopeofthisthesiswasmassivelyextendedbyapointthatRussellintroducesrightattheendof“OnDenoting”(p.427),namelythatourunderstandingofapropositionisbaseduponour“acquain-tance”withitsconstituents,whicharethemeaningsofthephrasesusedtoexpresstheproposition.Foronceoneadds,asRusselldoeshere,thatwearenotacquaintedwithmatterandthemindsofotherpeople,itfollowsthatourunderstandingofsentencesthatincludeputativenamesofmaterialobjectsandotherpeoplecannotbeachievedbyidentifyingthesethingsasthemeaningsofthenames;insteadweareboundtoreinterpretthenamesasdescriptionsthatinvokepropertiesofthingswithwhichweareacquainted.Ishalldiscussthisradicaldoctrineof“knowledgebyacquaintance”below.WhatIwanttostresshereisjustthatitwasthetheoryofdescriptionsthatmadethisdoctrinetenable,sinceitseemedtoofferawayofescapingthedoctrine’sotherwiseunaccept-ableskepticalimplications.Thetrickherewastosupposethat“logicalanalysis”involv-ingthetheoryofdescriptionscouldsavetheappearancesofcommon-sensebeliefeventhoughitsobviousfoundationshadbeenremovedbythedoctrineoflimitedacquain-tance.ThismovebecamecentraltoRussell’slater“logical-analyticmethodinphiloso-phy”(1914:v),towhichIshallreturnbelow.Anotherimportantpointisthedoctrineof“incompletesymbols,”andinparticularthecentralclaimofthetheoryofdescriptionsthatphraseswhosemeaningatfirstsightseemstoconsistindenotingsomeobjectturnout,after“logicalanalysis,”nottohavesuchameaningatall.Insteadtheirmeaningisgivenonly“incontext”:inthebroadercontextofthesentencesinwhichtheyoccur.ItisstrikingthatFrege’scontextprinci-ple,whichImentionedearlierwhendiscussingtheissueoftheunityofjudgment,hereentersintoRussell’stheory;butofcourseitdoessoonlybecausedescriptionsareacounterexampletoRussell’sfundamentallynon-contextualconceptionofmeaning.ForRusselltheappealtocontextisappropriatepreciselywherethedemandsoflogicconflictwithsuperficialgrammaranditsassociatedconceptionofmeaning.Theideaofsuchaconflicthasbeenadeeplyinfluentialideaintheanalyticaltradition,givingrisetoaveryPlatonistconceptionof“logicalform”assomethingcharacteristicallyveiledbyordinarylanguage.WittgensteinrightlyidentifiedRussell’sseminalroleindevelopingthisconception:Allphilosophyisa“critiqueoflanguage”(thoughnotinMauthner’ssense).ItwasRussellwhoperformedtheserviceofshowingthattheapparentlogicalformofapropositionneednotbeitsrealone.(TractatusLogico-Philosophicus4.0031)33\nTHOMASBALDWINThoughby1914Russelltooktheviewthat“philosophy...becomesindistinguish-ablefromlogic”(in“OnScientificMethodinPhilosophy”)hewouldhaverepudiatedthischaracterizationofphilosophyasjust“critiqueoflanguage,”sinceforhimlogicprimarilyconcernsthelogicalformsof“thevarioustypesoffacts”(Papers,8:65);indeedtowardstheendofhislife,inthe1950s,hewasverycriticalofthewayinwhichphilosophersseemedprimarilyconcernedwithlanguage.Aswehavealreadyseen,andshallseefurtherbelow,Russell’slogicwasalwaysshapedbymetaphysicaltheses(e.g.concerningthenatureofpropositions)anddrivenbyepistemologicalconcerns(e.g.thedoctrineofacquaintance).Nonetheless,hisowntheoryofdescriptions,andtheusestowhichhethenputitinhislogical-analyticprogram,didseemtomanyphilosopherstoshowthatitisthroughthelogicalanalysisoflanguagethatphilosopherscanmakeprogressinresolvingolddebates.Toaconsiderableextentthewholeprojectofanalyti-calphilosophyisfoundeduponthisfaith,andtothatextentRussell’stheoryofdescrip-tionsremains,asRamseycalledit,“aparadigmofphilosophy”(Ramsey1931:263n.).AvoidingthecontradictionRussell’smainconcernintheyearsfollowingthepublicationofThePrinciplesofMathematicswasnotinfactthetheoryofdescriptionsanditsimplications,whichIhavebeendiscussing.Insteadhewasstillpreoccupiedwithhislogicistproject,onwhichhewasnowworkingwiththeCambridgemathematicianand(later)philosopherA.N.Whitehead,andhewas,therefore,confrontedbytheneedtoavoidtheperplexingcontradictionhehaddiscoveredin1901.HisfirstthoughtwasthattheconceptionofanincompletesymbolhehaddevelopedinconnectionwiththetheoryofdescriptionscouldbeputtoworktoshowwhatwaswrongwiththeparadoxicalsetRofthingsthatarenotmembersofthemselves.Russelldevelopedthisthoughtinaningeniouswaybyinterpretingtalkofsetsintermsoftheresultsofsubstitutionswithinpropositionsandthenshowedthat,underthisinterpre-tation,theconditionofself-membershipcannotbecoherentlyexpressed.Regretfully,however,hedecidedthatthisapproachwasnotthewholestorysinceitdidnotresolveparadoxesconcerningpropositions,suchastheliarparadox,which,hefelt,weresocloselyrelatedtohisownparadoxthatthereshouldbeasinglesolutionforthemall.HeturnednexttoanideathataroseinthecourseofadebatewiththeFrenchphilosopher,HenriPoincaré,thattheseparadoxesariseonlybecausetheunderlyingargumenttacitlyinvolvesa“viciouscircle,”inthatsomethingwhichhasbeendefinedintermsofatotalityisthenassumedtobelongtothistotality.ThusRussell’ssetRisdefinedintermsofthesetofsetswhicharenotmembersofthemselves,andthecon-tradictionisthenarrivedatbyconsideringwhetherornotRbelongstothisveryset.Similarlyinthecaseoftheliarparadoxthecrucialmoveisthatwherebythestatementmadebytheliaristakentobeincludedinthescopeoftheliar’sownstatement.So,Russellthought,thewaytoavoidalltheseparadoxesistoadheretothe“viciouscircle”principle,that“Whateverinvolvesallofacollectionmustnotbeoneofthecollection”(1910:37).Thisprinciplegivesrisetoahierarchyof“orders,”sinceanythingdefinedintermsofacollectionofthingsofordernisheldtobeofordern+1andthereforenotacan-didateformembershipofthefirstcollection.Indevelopingtheideafurther,however,34\nBERTRANDRUSSELLRussellwentbacktohisearlierideathattheverydefinitionoftheparadoxicalsetRissomehowill-formed.Sohereturnedtothethoughtthatournormaltalkofsetsinvolvesincompletesymbols,andthusthatsetsareonly“quasi-things”(1910:81).Buthenolongerusedhisearlierinterpretationofthistalkintermsofsubstitutionswithinpro-positions:insteadheintroducedtheconceptionofa“propositionalfunction,”afunc-tionwhosevaluesarepropositions,andproposedawayofinterpretingtalkofsetsintermsofpropositionalfunctions.Underthisinterpretation,talkofa’smembershipofthesetofthingsxsuchthatfxisinterpretedintermsofthetruthofthepropositionthatisthevalueofthepropositionalfunctionf^x(Russell’sstandardnotationforpropositionalfunctions)fortheargumenta(i.e.thepropositionfa).Hencethecondi-tionofself-membershipenteringintothedefinitionofRisinterpretedintermsofthetruthofthepropositionthatisthevalueofapropositionalfunctionappliedtoitselfasargument.Butthis,Russellclaims,isincoherent:hetakesitthattheviciouscircleprin-cipleimpliesthatthepropositionsthatarethevaluesofapropositionalfunctionshouldinnocasebespecifiedbyreferencetothepropositionalfunctionitself,andthusthattheremustbenosuchthingasthevalueforf^xwiththeargumentf^x...Thatistosay,thesymbol‘f(f^x)’mustnotexpressaproposition.(1910:40)Russellreinforcesthispointbyarguingthatalthoughtherearefunctionsoffunc-tions,inallcasesfunctionsmustbeofadifferent“type”fromthatofwhichtheyarefunctions;thusfunctionsofsimpleindividualscannotthemselvesbeindividuals,butmusthavesufficientcomplexitytoyieldcompletepropositionswhenappliedtoindi-viduals.Similarlyfunctionsofthesefunctions,suchasthequantifiers,havetohavethetypeofcomplexityrequiredtoyieldacompletepropositioninthesecases.Hence,hesays,functionscannotbeargumentstothemselves,fortheylackthetypeofcomplex-ityrequiredtoyieldacompletepropositioninthissituation.Thislineofthoughtgeneratesahierarchyoftypes(individuals,functionsofindi-viduals,functionsofsuchfunctions,etc.)differentfromthehierarchyofordersgener-atedbytheviciouscircleprinciple,whichconcernstheorderofdefinitions.Theresultingtheory,the“ramifiedtheoryoftypes,”istheresultofmergingthesetwohier-archies.Thereisanelementofoverkillinthistheory,fortherearenowthreereasonswhythecontradictiondoesnotarise:(1)theviciouscircleprinciplestraightforwardlyimpliesthatRisofahigherorderthanitsmembersandcannotthereforebeamemberofit;(2)RusselltakestheprincipletoimplyalsothatthedefinitionunderlyingR,whenspelledoutintermsofpropositionalfunctionsthatapplytothemselves,isill-formed;(3)Russellalsoinvokesaseparatethesisthatafunctionmusthaveadifferenttypeofcomplexityfromitsargumentsifitistoyieldacompletepropositionasvalue,whichagainimpliesthatthedefinitionunderlyingRisill-formed.Themainaimofthetheory,however,wasnottoavoidthecontradictionbuttofulfillthelogicistprojectofprovidingalogicalfoundationforpuremathematics.ThiswasinawayaccomplishedbyRussellandWhiteheadintheirmassive,thoughincomplete,trilogyPrincipiaMathematica(1910–13).Theyhadfound,however,thattheirtaskwasobstructedbythecomplexitiesoftheramifiedtheory.Forexample,atacrucialpointinthestandardtheoryofrealnumbers(theleastupperboundtheorem),theviciouscircleprincipleisviolated;henceRussellandWhiteheadhad,ineffect,tosetasidethisprinciplebyintroducingtheassumption(the“axiomofreducibility”)thatwherevera35\nTHOMASBALDWINpropositionalfunctionisdefinedintermsofatotalitytowhichitisthenrequiredtobelongthereisawayofdefiningitwithoutreferencetothattotality,intermsof“pre-dicative”functionsthatdonotinvolvereferencetothetotalityinquestion.Astheircriticsobserved,ifRussellandWhiteheadweregoingtohelpthemselvestothisassump-tion,thentheycouldhavesimplifiedthingsagooddealbyformulatingthewholetheoryintermsofpredicativefunctionsinthefirstplace.Indeed,afterdiscussionwithF.P.RamseyRussellandWhiteheadadoptedaproposalofthiskindinthesecondeditiontoPrincipiaMathematica(1927).Theresultingtheory,the“simpletheoryoftypes,”nolongeroffersasolutiontosemanticparadoxessuchastheliarparadox.Butthisisapositivegainsincetheviciouscircleprincipleisanywaynotasatisfactoryresolutionoftheseparadoxes,whichdependonconceptssuchastruthwhosecomplexitiesareseparatefromsettheoryitself.Butthereareotherproblems,whichafflicteventhismodifiedtheory.Onceindivid-ualsandfunctions(orsets)aredividedintoexclusivetypes,therehastobeaseparate,thoughisomorphic,arithmeticforeachtype,anideathatishighlycounterintuitive.Furthermorethevalidityofstandardarithmeticasappliedtoindividualsrequirestheassumptionthatthereisaninfinityofsuchindividuals.AsRussellhimselfrecognized(1919:141),thisassumption,oraxiom,isnotlogical;itisclearlymetaphysical(andmaywellbefalse).Tosaythisistoraisethequestion,centralforthelogicistproject,astowhatlogicis.InPrincipiaMathematicalogicissaidtobethetheoryofformalinference,ofinferencesthatdependmerelyonthelogicalformofthepropositionsinvolved.Thedistinctionbetween“form”and“content”isthencrucial;Russelltakesitthattheidentificationofthelogicalconstants,byreferencetowhichlogicalformisdefined,isonlyamatterofenumeration.Inhisabandoned1913manuscript“TheoryofKnowledge”heappreci-atestheneedforsomedeepertheory,butremarks,“Inthepresentchaoticstateofknowledgeconcerningtheprimitiveideasoflogic,itisimpossibletopursuethistopicfurther”(Papers,7:99).ThisisasurprisingremarkinthelightofallRussell’sworkonlogic.Butthereisnodoubtthathehadfoundtheexperienceofcopingwithhiscon-tradictionachasteningexperience,whichhadtaughthimthateveninlogictherearenosimpleanswers,andthusthatonly“patienceandmodesty,hereasinothersciences,willopentheroadtosolidanddurableprogress”(Papers,8:73).Tocomparelogicwithothersciencesistoinvitethequestionwhetherlogicdiffersfromthemexceptinrespectofitssubjectmatterofformalinference,howeverexactlythatbedefined.Thisquestionisparticularlyappositesinceatleastuntil1911Russellaffirmedthatlogicissynthetic,whichmightsuggestthathealsothoughtitisempirical.Infact,however,heheldthatitisapriori,andrestsuponself-evidentintuitionsconcerningtherelationshipsbetweenlogicalforms,whichareuniversals(1912:ch.X).Whatisthenalittleoddisthathegenerallytakesitthatlogicalinfer-enceisjustamatterofmaterialimplication(1910:8–9),sothatalthoughlogicalinfer-encesmustbeinfacttruth-preservingitisnotrequiredthattheypreservetruthinallpossiblesituations.Hisviewsinthisareaarenoteasilyfittedtogether.Later,presum-ablyundertheinfluenceofhisformerstudentWittgenstein,hedescribeslogicaltruthsas“tautologies,”whichsuggestsamovetoaconceptionofthemasanalytic;butinthe“LecturesonthePhilosophyofLogicalAtomism”(published1918)heisstillverytentative:36\nBERTRANDRUSSELLEverythingthatisapropositionoflogichasgottobeinsomesenseorotherlikeatautol-ogy.Ithasgottobesomethingthathassomepeculiarquality,whichIdonotknowhowtodefine,thatbelongstologicalpropositionsandnottoothers.(Papers,8:211)ModernsettheoriesarebasedontheworkofZermeloandVonNeumannandavoidtheproblemsthatariseforRussellandWhiteheadbysweepingawaytypedistinctions.Thisimpliesthatthereisnothingill-formedabouttheconditionofself-membership,butRussell’sparadoxisavoidedbydenyingthatanywell-formedcondition(orpropo-sitionalfunction,asRussellwouldcallit)determinesaset.Indeeditisstandardtohaveanaxiomof“foundation,”whichrequiresthatthemembershipofasetbefoundedbybeingbasedupon“ur”-elementswhichdonotthemselveshavemembers.Thisaxiom(proposedbyMirimanoffin1917)impliesthattheconditionofnon-self-membershipdoesnotdetermineasetandcapturestheintuitionwhichliesbehindRussell’s“vicious-circleprinciple,”butwithoutobstructingthedevelopmentofasystematicsettheorythatcanbeinterpretedasafoundationformathematics.Thisdoesnot,however,meanthatmodernsettheoryprovidesavindicationofRussell’slogicistproject.Russell’sconceptionofapropositionalfunctionblursthedistinctionbetweenstandardpredicatelogicandsettheory,but,ironically,Russell’sparadoxitselfshowstheimportanceofmaintainingadistinctionhere,andoncethisisdrawnandsettheoryisprovidedwithitsowndistinctiveaxiomsconcerningtheexis-tenceofsets,thereisnogoodreasontocountsettheoryaslogic.Thisdoesnotmeanthatthelogicistprojectisaltogetheruntenable;foronecanabandonsettheoryinfavorofsecond-orderlogicandtrytousethistoprovideafoundationformathematics.Whethertheresultingpositionissatisfactoryremainsdisputed,butevenifitis,itisstillsubjecttotheimplicationsofGödel’sfamousincompletenesstheorem,whichshowsthatarithmetic(andthusmathematics)cannotbecompletelycapturedwithinafor-malizedtheory.Iflogicisjustthetheoryofformalinference,asRussellmaintained,thenGödel’stheoremprovidestheultimaterefutationofhisprojectofshowingthat“allmathematicsisSymbolicLogic.”LogicalatomismWhenRussellwasaskedin1924toprovideapersonalstatementofhisphilosophicalpositionhechosetoentitleit“LogicalAtomism”(1956).Thisisanamethathebegantousein1914(Papers,8:65)andthenusedinhis1918“LecturesonthePhilosophyofLogicalAtomism.”Therationalefortheemphasishereonlogicwillbeobvious;butwhatisthe“atomism”?Whatarethe“logicalatoms”?Theyareatomicfacts.Thereferencehereto“facts”isduetohisabandonmentby1914ofhisearlierMooreanconceptionofpropositions.Hehasnowadoptedaformofthecorrespondencetheoryoftruthaccordingtowhichthetruthofaproposition,nowconceivedofasnormallyalinguisticstructure(thoughRussellalsoallowsforimagis-ticmentalpropositions),isgroundedintheperfectcorrespondenceoflogicallysimplepropositionswithatomicfacts.Ishallexplainhisreasonsforthischangeofmindcon-cerningpropositionsbelow,butsincefactsaresaidtobecomposedofthethingsthatarethemeaningsofthewordsoccurringintheproposition,itturnsoutthatatomicfactsdifferlittlefromold-styletrueelementarypropositions,thepropositionswhose37\nTHOMASBALDWINtruthformedthebasisforthetruthofpropositionswhoseexpressioninvolvesincom-pletesymbols(see“LecturesonthePhilosophyofLogicalAtomism,”Papers,8:175).Atomicfactsarefactsconcerningtheintrinsicqualitiesof,andrelationsbetween,particularindividuals(Papers,8:177).Butsincetheindividualsandpropertiesthatconstituteanyfactwecantalkaboutmustbesuchthattheyarethemeaningsofthewordsweuse,itfollowsthatinpracticetheidentityofatomicfactsisconstrainedbytherequirementsofRussell’stheoryofmeaning.Andthecrucialrequirementhereisthatofouracquaintancewiththethingsinquestion:Aname,inthenarrowlogicalsenseofawordwhosemeaningisaparticular,canonlybeappliedtoaparticularwithwhichthespeakerisacquainted.(Papers,8:178)SinceRussellholdsthatwearenotacquaintedwithordinarythings,suchasPiccadillyandSocrates,butonlywithsensoryparticularswhichare“apttolastforaveryshorttimeindeed”(Papers,8:181),itturnsoutthattheatomicfactswecantalkandthinkaboutdonotdealwiththefamiliarfurnitureoflife,butforthemostpartonlywiththeprivateobjectsofexperience.ItisatthispointthatRussell’slogicalatomismmakescontactwiththedoctrineofknowledgebyacquaintance,whichweencounteredearlier,andtherebybecomesaformofepistemologicalatomism.Thecoreofthatdoctrineisexpressedinhisfunda-mentalepistemologicalprinciple(in“KnowledgebyAcquaintanceandKnowledgebyDescription”)thateverypropositionwhichwecanunderstandmustbecomposedwhollyofconstituentswithwhichweareacquainted.(Papers,6:154)Withwhatthingsareweacquainted?Russell’sapproachisacombinationofempiri-cismandrationalism.Throughperceptionandintrospectionwegainacquaintancewithparticulars:withparticularcolors,noises,feelings,etc.;butthroughreflectionwecanalsogainacquaintancewiththeuniversalsofwhichtheseparticularsareinstances.Thisformofacquaintanceinvolves“conception”(Papers,6:149–50),andiscentraltoourcapacityforintuitiveawarenessofaprioritruths(1912:ch.X).Forourpresentpurposes,thepointonwhichtoconcentrateisouracquaintancewithparticulars.RussellusesMoore’sgeneralterm“sense-data”forthesesensoryparticularsbutemphasizesthatitisforhimanopenquestionwhethersuchparticularsmaynotalsoexistasunperceived“sensibilia.”Sense-datasoconceivedarenot“inthemind,”thoughourawarenessofthemisimmunefromerror.InitiallyRusselltooktheviewthattheyaresubjectivebecausetheirrelativitytotheirsubject’spositionandconditionimpliesthattheycannotbecombinedinasinglepublicworld(1912:ch.I);butin1914heswitchedtotheviewthattheyarephysicalelementswithinprivatespaces,whicharecapableofbeingintegratedintoasinglepublicworldunlesstheybelongmerelytodreamsandhallucinations.Hisaccountofthissupposedintegrationisdifficultandthedetailscannotbediscussedhere,butitisforhimanimportantapplicationofhis“suprememaximinscientificphilosophising”(Papers,8:11)that38\nBERTRANDRUSSELLWhereverpossiblelogicalconstructionsaretobesubstitutedforinferredentities.ThebasicideahereissaidbyRusselltobesimilartothatinvolvedinOckham’srazor,butinfactitismotivatedbytheneedtoescapetheskepticalimplicationsofhistheoryofknowledgebyacquaintance.Forwheretheapparentobjectsofputativeknowledgearethingsofwhichwehavenoacquaintance,Russell’stheoryseemstoimplythatourbeliefscanonlyeverbeamatterofuncertain,speculative,inferenceconcerningcertainkindsofthings“weknownotwhat”whoseexistenceisinsomewayimpliedbytheexis-tenceofthethingswithwhichweareacquainted.Russellthinkshecanavoidthisskepticalresultbysubstitutinganalternativewayofthinking,whichcanstillberep-resentedasawayofthinkingabouttheapparentobjectsofknowledgebutwhich,becauseitis“logicallyconstructed”fromthoughtsofthings(sense-dataanduniver-sals)withwhichwedohaveacquaintance,doesnotrequirespeculationsconcerningtheexistenceofanyfurther“inferredentities.”Thisalternativestrategyshowshowknowledgeispossiblewhilerespectingthefundamentalepistemologicalprinciplequotedearlier.ThusRusselltakesitthatknowledgeoftheexternalworldisachievedonthebasisofacquaintancewithsense-dataprimarilybytheapplicationtothesedataofthelogicalprinciplesinvolvedinthetheoriesofdescriptionsandofclasses(thoughtheintegra-tionofprivatespacesintoasinglepubliconeinvolvesmorethanlogic).Theresultisthattalkandthoughtputativelyaboutpublicphysicalobjectsistobeprovisionallyinterpretedintermsoflogicallycomplexpropositionsaboutclassesofsense-data;butbecauseclassesarethemselves“logicalfictions”thesepropositionsarethemselvestobefurtherinterpretedintermsofpropositionalfunctions.Onlythencanwearriveatanindirectspecificationoftheatomicfacts,involvingonlyitemswithwhichweareacquainted,ofwhichweareaware,andwhichgroundourknowledgeoftheexternalworld.But,inprinciple,suchaspecificationcanbearrivedat;soknowledgeispossible,orsoRussellsupposes.Intruththematterismoretrickythanheallowsbecauseofthe“otherminds”problem.AsRussellacknowledges,theconstructionoftheexternalworldinvolvesapparentreferencetothesense-dataofothers,fortheexternalworldisessentiallysomethingthattranscendsourownsense-data.Butsincewearenotacquaintedwiththesense-dataofothers,Russell’sfundamentalprincipleimpliesthatwecannotunderstandanypropositionsrequiringreferencetothem.Sowehavetostartfromourownsense-dataaloneandbuildoutsimultaneouslytoothermindsandtheexternalworld.ButwhetherthiscanbedoneinawaythatmeetsRussell’srequire-mentsforknowledgeisdoubtful.Russell’sdiscussionofthesemattersconnectsdirectlywiththelogicalpositivistprogramofthe1920sand1930s(especiallyCarnap’sAufbau),though,asweshallseebelow,hehimselfhadmodifiedhisapproachinimportantrespectsbythen.Thatchangewasanindirectresultofadifferentproblem,whichcausedhimtosubstantiallyrethinkhispositionevenwhilehewasstilldevelopinghislogicalatomistprogram.ThebackgroundtothisishisabandonmentofMooreanpropositions.Thereasonforthischangeofmind(in1906)isthatRussellceasedtofinditcrediblethatthereare“objec-tivefalsehoods,”falsepropositionsthatareontologicallyonaparwithtrueproposi-tions,thatis,facts.Aswehaveseen,hecontinuedtoaccepttheexistenceoffacts,and39\nTHOMASBALDWINtothatextenttheoldtheorycontinuedunderanewname.Butsincepropositionshadbeenalsotakentobetheobjectsofjudgment,falseaswellastrue,Russellneededanewtheoryofjudgment,whichfactsalonecouldnotsupply.Hisnewtheory,the“multiple-relation”theory,wasthatwhathadpreviouslybeenconceivedofastheconstituentsofapropositionthatistheobjectofjudgmentshouldnowbeconceivedofastermsofanewmultiple-termrelationconceptionofjudgment;thatis,insteadofthinkingoftheMooreanpropositionexpressedbythesentence“TomjudgesthatAislargerthanB”ashavingthelogicalform:Judges(Tom,thepropositionAislargerthanB)wherethispropositionisitselfacomplexentitysomehowcomposedofA,B,andtherelationofbeinglargerthan,wearetothinkofthesamesentence(nowregardedasapropositionitself,becauseitistobetheprimaryvehicleoftruthandfalsehood)asbeingsuchthat,iftrue,itwouldcorrespondtoafactoftheform:Judges*(Tom,A,B,beinglargerthan)where“judges*”isthemultiple-termrelationthatrelatesthesubjectofjudgment(Tom)withcertainobjectiveterms(A,B,beinglargerthan)insuchawaythat,together,theyconstituteajudgmentthatistrueifandonlyiftheobjectivetermsconstituteafact–thefactthatAislargerthanB.Russellneverintegratedthis“noproposition”theoryintohislogicaltheory.AlthoughitisstatedintheintroductiontoPrincipiaMathematica(PM:43–4),itsimplicationsforhisconceptionofapropositionalfunction,whichhasjustbeendefinedasafunctionwhosevaluesarepropositions,arenotworkedthrough,norareitsimpli-cationsforhissubstitutionaltreatmentofquantifiers.Indeeditisflatlyinconsistentwithhistheoryofdescriptions,sincethecomplexinterweavingofquantifiersandvariablesinthattheorycannotbedecomposedinto“simple”constituentsinthewayrequiredbytheapplicationofthemultiple-relationtheorytojudgmentsinvolvingdescriptions.Anotherdifficultycomesfromthatoldbugbear,theunityofjudgment.Forthetheoryineffectassumesthattheobjectivetermsofthemultiple-termrelation“judges*”canactasasurrogateforthatwhichisjudged,e.g.thatAislargerthanB.Sothechal-lengethatthetermsbythemselvesdonotconstituteacompletejudgmentisonethatcannotbeavoided.Russellneedstoexplainhowanappropriatespecificationoftherighttruth-makingfactisdeterminedsimplybytheobjectivetermsofthemultiple-termrela-tion.OnestandardobjectionisthatthereisnobasisforthedistinctionbetweenjudgingthatAislargerthanBandthatBislargerthanA.Itisarguablethatthiscaninfactbehandledsimplybyattendingtotheorderinwhichtheobjectsoccurastermsoftherelation“judges*”;but,aswithdescriptions,thisstrategywillnotcopewithgeneraljudgmentsinvolvingmultiplequantifiers–e.g.thejudgmentthatallelephantsarelargerthanallmice–wheretheboundvariablesobstructthedecompositionintosimpleconstituentsessentialforRussell’sapproach.Wittgenstein’sobjectiontoRussellcon-cernsarelatedpoint,thatthetheorypermitsonetojudgenonsense.Forunlesssomeconstraintsareplaceduponthetermsof“judges*”thereseemsnothingtoruleoutasimplepermutationoftermstogenerate,say,40\nBERTRANDRUSSELLJudges*(Tom,beinglargerthan,A,B)whichwouldhavetobeasurrogateforTom’s“judgment”thatbeinglargerisAthanB(seeWITTGENSTEIN).Russellmightseektorulethisoutbyplacingtype-restrictionsonthetermsofthemultiple-termrelation;butsincetypedistinctionswereexplainedintermsofthecapacityofthingstoformacompletepropositionRussellcannotappealtothemoncehehasembarkedonhis“no-proposition”perspective.Russellwasshatteredbythisobjection.HehadspentthemonthsofMayandJune1913workingattremendousspeedandinhighspiritsonabookaboutjudgmentandknowledge;butoncehegraspedWittgenstein’spointheabandonedthebook(nowpub-lishedinPapers,7)andfellalmostintodespair.ThreeyearslaterhewrotetoOttolineMorrellaboutthecrisisthiseventhadinduced:DoyourememberthatatthetimewhenyouwereseeingVittozIwrotealotofstuffaboutTheoryofKnowledge,whichWittgensteincriticizedwiththegreatestseverity?Hiscriti-cism,tho’Idon’tthinkyourealizeditatthetime,wasaneventoffirst-rateimportanceinmylife,andaffectedeverythingIhavedonesince.Isawhewasright,andIsawthatIcouldnothopeeveragaintodofundamentalworkinphilosophy.(1968:57)LaterphilosophyRussellwaspreventedfromlapsingintosilencebythepressureofpriorcommit-mentsatthistime,mostnotablytheLowelllectures,whichhedeliveredatHarvardinspring1914(publishedasOurKnowledgeoftheExternalWorld).Indeedhisproductiv-ityduring1914isaremarkabletestamenttohisstrengthofwill.OncetheFirstWorldWarbeganheturnedwithsomerelieffromtheneedtorethinkhisphilosophytopublicoppositiontothewar,thoughby1918hewaskeentoreturntophilosophy.(Attheverytimeinearly1918thathewasstandingtrialforhisanti-warpropagandaandthenappealingagainstthetermsofhissentenceofsixmonths’imprisonmenthewasalsodeliveringthelecturesonthephilosophyoflogicalatomism(Papers,8)Ihavereferredtoabove).Imprisonment,underthecomfortableconditionspermittedtohim,turnedouttoprovidetheconditionsofrelativeisolationthatRussellneededtoachieveafreshstartinhisphilosophy(thoughhealsowrotehisIntroductiontoMathematicalPhilosophy,whichisalucidinformaldiscussionofthemainthemesofPrincipiaMathematica).Thestartingpointforthisnewwork,whichwaspublishedasTheAnalysisofMind(1921),wasWilliamJames’s“neutralmonism.”Russellhadalreadybeenthinkingaboutthisforsometime;James’sclaimhadbeenthatthetraditionaloppositionbetweenmindandmattercouldbetranscendedbysomehowconceivingofthemasjustdifferentwaysofthinkingaboutsomethingintrinsicallyneutral,whichJamescalled“experience.”Russell,notingthesimilaritiesbetweenthisapproachandhisownaccountoftheexter-nalworld,developsasimilaraccountbasedupon“sensations,”whicharelikehisoldsense-data,exceptthathenowholdsthatthefactthattheyarephysicalisnoreasonnottoholdthattheyarenotalsomental(1995b:143–4).Thedetailsoftheconstruc-tionsofthemindandofthephysicalworldthatfollowarecomplexandunpersuasive;butwhatisnonethelessstrikinginthelightofcontemporaryphilosophyofmindisthat41\nTHOMASBALDWINRussellsetsoutapositionthatcombinesontologicalmonismconcerningmindandmatterwithaninsistencethatthenaturallawsregulatingthemarenotreducibleeitherway(1995b:104–5),sothatthepositionisnotareductivemonism.AcentralfeatureofRussell’sanalysisofmindishisattempttodojusticetowhathetakestobetheinsightsofscientificpsychology,andinparticularthebehavioristpsy-chologybeingpropoundedatthistimebyJohnB.Watson.Russellcannotendorsethefullbehavioristposition;hethinksthatbeliefstypicallyinvolvementalimageryinawaythatisincompatiblewithbehaviorism.Buthedoesendorseabroadlybehavioristconceptionofdesire,andindeedrefinesitintoapositionthatisrecognizableasapre-cursorofcontemporaryfunctionalism.Furthermoreheoffersacausalaccountofthecontentoftheimagesthatenterintobeliefsandextendsthisintoagenerallycausalaccountofmeaning.Sohisanalysisofmind,includingmentalcontent,isbasedquitegenerallyuponcausalconsiderations,andthisthenprovideshimwiththematerialsforanewtheoryofjudgmenttoreplacethatwhichWittgensteinhadoverthrown.Hedoesnot,however,takefulladvantageofthisopportunitylargelybecausehestillthinksthatthemeaningofacompletesentenceisconstructedfromtheindependentmeaningsofitsconstituentwords(1995b:273).SoitwaslefttoRamseytothinkthematterrightthroughandpropose,soonafter,atentativefunctionalisttheoryofjudgment.Onthesubjectofknowledge,however,Russellclearlygraspsthepotentialofhisnewcausalconceptionofthemind.HebeginsTheAnalysisofMindbyrejectinghisoldcon-ceptionofacquaintance,andlaterinthebookhereinforcesthisbreakwithhispastbydenyingthatwecanobtainself-evident,certain,knowledgeeitherbyperceptionorbyaprioriintuition(pp.262–6).Inplaceofhisoldconceptionofknowledge,which,henowthinks,cannotruleoutsuch“logicallytenable,butuninteresting”skepticalhypothesesasthattheworldwascreated,withallourputativememories,fiveminutesago(pp.159–60),heoffers“amoreexternalandcausalview”(p.270)ofknowledge.Thisisindeedpreciselytheviewthatisnowfamiliaras“externalist”;andRussellintro-ducesitbymeansofthenow-familiarcomparisonbetweenanaccuratethermometerandsomeonewithreliablytruebeliefs(p.253ff).InTheAnalysisofMindRussell’spresentationofthisexternalistconceptionofknowledgeissomewhattentative.Inhislastmajorworkofphilosophy,HumanKnowledge:ItsScopeandLimits(1948),Russellismuchmoreassuredandsophisticatedinhisdevelopmentofthisconception.HisgeneralaimhereisKantian:heseekstoexplainhowscientificknowledgeispossible,but(unlikeKant)todosowithinabroadlyscientificconceptionofhumanlife.Thetopiconwhichhethendirectsmuchofhisattentionisinduction.ThisisatopichehaddiscussedinTheProblemsofPhilosophy,wherehehadargued,first,thatscientificknowledgeisdependentuponthevalidityoftheinductiveprinciplethatthegreatertheexperienceoftheassociationofpropertiesAandBthelargertheprobabilitythatAandBwillbefoundtobeassociatedinnewsituations,and,secondly,thatsincethisprincipleispresupposedinallreasoningfromexperience,itmustberegardedasaself-evidentaprioritruthcomparabletofunda-mentaltruthsoflogic.InhislaterworkRussellbeginsbyarguingthatthispositionisuntenable,becausetheinductiveprincipleisopentoobviouscounterexamplesifnorestrictionsareplaceduponthepropertiesinvolved.HisargumenthereissimilartothatlatermadefamousbyNelsonGoodman(as“TheNewRiddleofInduction”)(seeGOODMAN)andhetakesfromittheconclusionthatthisisnotanareawithinwhichself-42\nBERTRANDRUSSELLevidenttruthsaretobefound.Instead,heproposes,itisessentialtolookto“scientificcommonsense”anddiscerntheactual“postulatesofscientificinference”(1948:436).Inthelastpartofthebookheundertakesthistaskinthecontextofasophisticatedconceptionofknowledgeassomethingthatcomesindegrees:knowledgeisnotjusttruebelief,buttherearemanykindsofwarrant,somemerelyinvolvingreliablecon-nections,othersinvolvingunderstandingandreflection,whichprovidethehigherdegreesofknowledge.Butintheendeventhehighertypesdependontheexistenceoftheconnectionsthatjustifyprimitivetypesofknowledge.Russell’sconclusionisthat:Owingtotheworldbeingsuchasitis,certainoccurrencesaresometimes,infact,evidenceforcertainothers;andowingtoanimalsbeingadaptedtotheirenvironment,occurrenceswhichare,infact,evidenceofotherstendtoarouseexpectationofthoseothers.Byreflect-ingonthisprocessandrefiningit,wearriveatthecanonsofinductiveinference.Thesecanonsarevalidiftheworldhascertaincharacteristicswhichweallbelieveittohave.(1948:514–15)Onecouldnotaskforaclearerstatementoftheexternalist’sjustificationofinduction,thoughRussellisundernoillusionthatthiswillaltogethersatisfythephilosophicalskeptic.ThislateworkshowsRussellstillcapableoforiginalityattheageof76.Theselatewritingsareoftenneglectedtoday.Butthisisamistake.Forintheselatewritings,hepracticestheprinciplehehadenunciatedin1924that“weshallbewisetobuildourphilosophyuponscience”(1956:339).Asaresult,hiswritingsfromthisperiodconnectdirectlywithcontemporarydebates,sincefromthe1970sonwardsthe“naturalization”ofanalyticalphilosophyhasintroducedintophilosophicaldebatetherequirementofharmonywithscientificknowledgethatRussellhadrecognizedfiftyyearsearlier.Russellisstillourcontemporary.BibliographyWorksbyRussell1903:ThePrinciplesofMathematics,London:Allen&Unwin.1910–13(withA.N.Whitehead):PrincipiaMathematica,Cambridge:CambridgeUniversityPress.1912:TheProblemsofPhilosophy,London:Williams&Norgate.1914:OurKnowledgeoftheExternalWorld,London:OpenCourt.1916:PrinciplesofSocialReconstruction,London:Allen&Unwin.1919:IntroductiontoMathematicalPhilosophy,London:Allen&Unwin.1948:HumanKnowledge:ItsScopeanditsLimits,London:Allen&Unwin.1956:“LogicalAtomism,”inLogicandKnowledge,ed.R.C.Marsh,London:Allen&Unwin,pp.323–43.(FirstpublishedinContemporaryBritishPhilosophy(firstseries),ed.J.H.Muirhead,London:Allen&Unwin,1924,pp.359–83.)1967:TheAutobiographyofBertrandRussell1872–1914,London:Allen&Unwin.1968:TheAutobiographyofBertrandRussell1914–1944,London:Allen&Unwin.1983–2000:TheCollectedPapersofBertrandRussell,in15vols.,London:Allen&Unwin/Routledge.(Themainpaperscitedinthischapterarelistedbelow,andtheirpagenos.inthePapersareusedintextreferences.)43\nTHOMASBALDWIN“KnowledgebyAcquaintanceandKnowledgebyDescription,”invol.6,pp.148–61.(FirstpublishedinProceedingsoftheAristotelianSociety11(1910–11),pp.108–28.)“LecturesonthePhilosophyofLogicalAtomism,”invol.8,pp.160–244.(FirstpublishedinTheMonist28(1918),pp.495–527;29(1919),pp.32–63,190–222,345–80.)“OnDenoting,”invol.4,pp.414–27.(FirstpublishedinMind14(1905),pp.479-93.)“OnScientificMethodinPhilosophy,”invol.8,pp.57–73.(OriginallytheHerbertSpencerLecture,Oxford,November1914.)“TheoryofKnowledge:The1913MSS,”vol.7.“TheRelationofSense-datatoPhysics,”invol.8,pp.3–26.(FirstpublishedinScientia16(1914),pp.1–27.)1995a:MyPhilosophicalDevelopment,London:Routledge.(FirstpublishedLondon:Allen&Unwin,1955.)1995b:TheAnalysisofMind,London:Routledge.(FirstpublishedLondon:Allen&Unwin,1921.)WorksbyotherauthorsGödel,K.(1944)“Russell’sMathematicalLogic,”inThePhilosophyofBertrandRussell,ed.P.A.Schilpp,LaSalle,IL:OpenCourt,pp.123–54.Griffin,N.(1991)Russell’sIdealistApprenticeship,Oxford:ClarendonPress.Hylton,P.(1990)Russell,Idealism,andtheEmergenceofAnalyticPhilosophy,Oxford:ClarendonPress.Monk,R.(1996)BertrandRussell,vol.1,London:JonathanCape.Pears,D.F.(1967)BertrandRussellandtheBritishTraditioninPhilosophy,London:Fontana.Ramsey,F.P.(1931)TheFoundationsofMathematicsandOtherLogicalEssays,ed.R.B.Braithwaite,London:RoutledgeandKeganPaul.Watling,J.(1970)BertrandRussell,Edinburgh:OliverandBoyd.44\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd20013G.E.Moore(1873–1958)ERNESTSOSAReflectingonhislongphilosophicalcareer,G.E.Moorehadthistosay:Idonotthinkthattheworldorthescienceswouldeverhavesuggestedtomeanyphilo-sophicalproblems.Whathassuggestedphilosophicalproblemstomeisthingswhichotherphilosophershavesaidabouttheworldorthesciences.Yetthisphilosophers’philosopherwaslionizedbytheBloomsburyliterati,andhisfirstbook,PrincipiaEthica,isnowincludedbytheModernLibraryBoardamongtheonehundredmostimportantnonfictionbooksofthecentury.Bornin1873toamiddle-classfamilyinaLondonsuburb,MoorewentuptoTrinityCollege,Cambridge,at19.Aftertwoyearsstudyingclassics,heswitchedtophilosophyundertheinfluenceofhisfriendBertrandRussell,butsoonafterthatitwasMoorewholedRussellfromandagainstidealism.With“TheRefutationofIdealism”Mooresetthedirectionthatwastotakethembothtologicalandphilosophicalanalysis,andtofound-ing,alongwithLudwigWittgensteinandthelogicalpositivists,thephilosophicalmove-mentthatcametobeknownas“analyticphilosophy.”Moore’sfocuswasnotjustonthegivingofdefinitionsor“analyses,”however,thoughthatcertainlywascentraltohiswork,asitwasforPlato.Onthecontrary,hemadeitclearthatitisalsoajobforphilosophytogive“ageneraldescriptionofthewholeofthisuniverse,mentioningallthemostimportantkindsofthingswhichweknowtobeinit.”Healsoreflectedlonganddeeplyonwhatknowledgeisandonhowitmightbeattained.ThesereflectionshadanimportantimpactonWittgenstein,whoseOnCertaintyisaresponsetoMooreinepistemology(seeWITTGENSTEIN).AnditwasethicsthatfirstattractedMoore’sintenselyconcentrated,patientattention,yieldingthatgreatfirstbookofhis.Inthatbook,Mooreintroducestheanalysisofconceptsandpropertiesintocompo-nents.Consider,forexample,beingmaleandbeingasibling,whichcometogetherthroughconjunctiontoformthecomplexconceptofbeingabrother.AccordingtoMoore,someconceptsarenotthusanalyzable,however,amongthemthatofbeingyellowandthatofbeinggood(i.e.intrinsicallygood),thelatterofwhichisfundamentaltoethics.Thisfundamentalconceptofethicsissaidtobenotonlysimplebutalsononnatural,andirreducibletoanynaturalproperty.Thusitisnotanalyzableinutilitarianstyleas45\nERNESTSOSAamatterofcausingorcontainingstatesofpleasureorpain.Therearewaysofbeingintrinsicallygoodnotdreamtofbysuchhedonism;beinggoodcannotbeidentifiedwithanyonememberofthepluralityofintrinsicgoods.Moreover,concerninganynaturalpropertyXproposedasidenticalwiththepropertyofbeinggood,thereisalwayssomesuch“openquestion”as:“IshavingXinevitablynecessaryandsufficientforbeinggood?”Evenifweanswersuchaquestionintheaffirmative,onceouranswerisdis-puted,assurelyitwillbeifnotabsolutelytrivial,thatwillshowthepropertyXandthepropertyofbeinggoodnottobeoneandthesameproperty.Forifthesewerethesame,thenourquestionwouldbetantamounttothequestion“IshavingXinevitablynecessaryandsufficientforhavingX?”–thanwhichfewquestionscanbemoretrivial.Surelythisquestioncannotoccasionthesortofcontroversythatalwaysattendstheo-riesofthegood.Mooreputsforwardsomeexcellentissueshere,buthisanswersarenotentirelysat-isfactoryifonlybecausebasedontoofewdistinctions(uncharacteristicallyso).Ataminimum,weneedthedistinctionbetweenmeaninganalysisandphilosophical,meta-physicalanalysis,aswellasthedistinctionbetweenconceptsandproperties.Perhaps“malesibling”canhelpprovideameaninganalysisof“brother,”sothatthequestion“Isabrotheramalesibling?”isindeedtrivial.Bycontrast,“isproductiveofagreaterbalanceofpleasureoverpainthananyalternative”maynotgiveusameaninganaly-sisof“istherightthingtodo.”Butitisleftopenthatitmaygiveagoodphilosophicalanalysisinanycase,ifonethatisnotobvious,notjustamatterofsurfacemeaning.Howthenarewetothinkofphilosophicalanalysiscompatiblywiththefactthat,unlikeakindof(surface)meaninganalysis,itisfarfromtrivialandrequiresreflection?Thishasprovedadifficultandtroublingproblemforanalyticphilosophy,andhasbeennotsomuchsolvedasshelvedundertheheadingof“theparadoxofanalysis.”Thosewhostillcareaboutpiecemealanalysis–whichstillincludesmanyphilosophers,thoughnotallwhocountthemselves“analytic”–havegoodreasontofeelnaggedbythisworry.NotthroughsuchcoolanalyticworkdidMooreattracttheBloomsburies.MostlikelyitwasratherthroughbuckingtheVictorianpenchantforrigidrules.ThesecouldnotwithstandMoore’sprobingintelligenceexceptasrulesofthumb.ButMoorealsodis-tancedhimselffromBentham–Millutilitarianism.Itiscertainlynotjustpleasurethatfundamentallydeservesouradmirationandpursuit.Therearethingsofvarioussortsthatseparatelyhavethatspecialstatusoftheintrinsicallygood;prominentamongthesefigure,first,theenjoymentofcertainhumanrelationsand,second,theappreciationofthingsofbeauty.HereMoore’smilieuandupbringingmayshow.Talktothestarvingmillionsaboutthingsofbeauty,andyouwillbelesssuccessfulthaninBloomsbury.Still,pluralistthatheis,Moorecouldsimplyacceptfurtherintrinsicgoodsthathislistmayhaveoverlooked,withoutanymajorsetbacktohisoverallposition.Anactisright,Mooreadvises,ifandonlyifthereisnobetteravailablealternative,wherethevalueofeachalternativeisafunctionnotonlyofitsownintrinsicmeritbutalsoofthecombinedvalueofitsconsequencesuntoeternity.Thesecombinationsarenotjustbruteadditions,however,butmayinvolvespecialvaluederivingfromthewayinwhichtheycombine,aswhensomeoneisrewardedfordoingwellorgood,orwhensomeoneisunrewardedorpunishedfordoingillorharm.Especiallyinhismetaphysicsandepistemology,Moorejoinedatraditionofcommonsensephilosophy,onetowhichhewasnodoubtinherentlyandantecedentlyreceptive.46\nG.E.MOOREPerhapshecametoknowitsodeeplyduringhisyearsinEdinburgh,betweentheendofhisfellowshipatTrinityin1904andthebeginningofhistenurebackinCambridgeasalecturerin1911.Thefullernameofthattraditionis,afterall,ScottishCommonSense,whichisexplainedmostlybythefactthatitsgreatestearlyproponentwastheScotsmanThomasReid.Inanycase,howeverhemayhavebeenledtothistradition,Mooretooktoitnaturally,andwoulddefenditanddevelopitinhisowninimitablestyle.ForMooreseriousanalysisrequiredakindofpatientilluminationofdetailandnuancewhichisverydifficulttofollowattentively,andimpossiblefornearlyanyonebutMoorehimselftoproduce.Indeedmanyofhisessaysinmetaphysicsandepiste-mologyaremadeupalmostentirelyofsuchminuteanalysis,leadinguptoafewbril-liantinsightsveryquicklystatedneartheend.Take,forexample,hisfamous“ProofofanExternalWorld.”Exceptfortheconcludingfewpages,Mooreisengagedmostlyinfiguringoutingreatdetailwhat“externality”couldmean.Thatheisabletostaywiththatquestionthroughsomanytwistsandturns,andthathedoesnotbothertohidetheanalyticcomplexitiesilluminatedbyhisintelligence,showshisintegrity,butalsomakeshimhardtoread.Ifonestaysuntiltheend,however,therewardcomesintheinsightsoftheconcludingpages.Thisistruenotonlyof“ProofofanExternalWorld,”butalsoof“FourFormsofScepticism,”whereinheisagaindefendinghiscommonsensefromskepticalattack.Thatcommonsensicalviewhadalreadybeenexpoundedinhis“DefenceofCommonSense,”whereinhedescribesvariousfeaturesoftheworldashecommonsensicallybelievesittobe.AtthecenteroftheseontologicalreflectionsMooreinquiredintothenatureofsensoryexperienceanditsrelationtophysicalrealityinawaycharacteristicallyexploratoryandattentivetodetail.Aresense-dataidenticalwithphysicalsurfaces?Aretheyrathernonphysicaldenizensofourmentalworldwhilerepresentativeofphysicalrealitiesbeyondthem?Isphysicalrealityitselftobeviewedassomehowaconstructionfromoranalyzableorreducibletocombinationsofsense-data?Moorelongstruggledwithsuchquestions,buthisworkinthisareaneverreachedclosure.Nevertheless,hefeltcertainenoughofthecoreofhiscommonsense,whateveritscorrectanalysismayturnouttobe,thathewaswillingtogiveitfirmexpressionthroughalistofsomeofitscentralcommitments,amongthemthefollowingtwo:thathehasandhasforsometimehadahumanbody,whichhasbeenincontactwiththesurfaceoftheearth,andthattherehavebeenmanyotherthree-dimensionalthingsatvariousdistancesfromhisbody.Suchpropositionsformafirstgroup.Inasecondgrouparesuchpropositionsasthathehashadexperiencesofvarioussorts,andthathehasobservedvariousthingsinhissurroundingsatthetime,andhashaddreams,andothermentalstates.Finally,inathirdgroupisthepropositionthatthesameistrueofmanyotherhumanbeings.InthatpaperMoorealsoquiteexplicitlyclaims,finally,thatheknowswithcertaintypropositionsinhisfirsttwogroups,andthatthemanyotherhumanbeingsofwhomsimilarthingsaretruealsohavefrequentlyenjoyedsuchcertainknowledge.Theseareofcoursetheclaimsthatsetuphisconfrontationwiththeskeptic,whereMoore’slegendarypatienceandpowersofanalysisareverymuchinevidence.Itisthiswork,inmyjudgment,thatmanifestsadepthofinsightbeyondanythingshownin47\nERNESTSOSAMoore’sotherwork.Notthatwecannotseegapsandproblemsinhindsight.Buthisachievementwasnonethelessrealandimpressive,andimpressedthesingularlyunim-pressionableWittgenstein,fromwhomitelicitedhisownbestworkinepistemology.Inthenextandfinalpartofthisdiscussion,weturntothisworkofMoore’s.Istheexistenceofexternalthingsjustanarticleoffaith?Certainlynot,saysMoore,andoffersusaproof(whichisheresimplified),thusaimingtoremoveKant’s“scandaltophilosophy.”Moore’sproofHereisahand(areal,fleshandbonehand).Therefore,thereisatleastoneexternalthinginexistence.AccordingtoMoore,hisargumentmeetsthreeconditionsforbeingaproof:first,thepremiseisdifferentfromtheconclusion;second,heknowsthepremisetobethecase;and,third,theconclusionfollowsdeductively(“ProofofanExternalWorld,”in1962:144–5).Furtherconditionsmayberequired,butheevidentlythinkshisproofwouldsatisfytheseaswell.AsMooreiswellaware,manyphilosopherswillfeelhehasnotgiven“anysatisfac-toryproofofthepointinquestion”(1962:147).Some,hebelieves,willwantthepremiseitselfproved.Buthehasnottriedtoproveit,anddoesnotbelieveitcanbeproved.Provingthathereisahandrequiresprovingoneisawake,andthiscannotbedone.DoesMooreadequatelyanswertheskeptic?Manyhavedenieditforthereasonthathefailstoruleoutacrucialpossibility:thatourfacultiesareleadingusastray,forexamplethatwearedreaming.Awareofthisobjection,Mooregrants,in“Certainty,”thattoknowheisstandinghemustknowheisawake(“Certainty,”in1962).Thepoint“cutsbothways,”however,andhewouldprefertoconcludethathedoesknowheisawakesincehedoesknowheisstanding.Thishaspersuadednearlynoone.Onthecontrary,somehavethoughthimcom-mittedtoanargument,Mbelow,whichislikeArgumentA,precedingit:ArgumentAA1Thismapisagoodguidetothisdesert.A2Accordingtothemapanoasisliesahead.A3Therefore,anoasisliesahead.ArgumentMM1Mypresentexperienceisaveridicalguidetoreality(andIamnotdreaming).M2MypresentexperienceisasifIhaveahandbeforeme.M3Therefore,here(beforeme)isahand.WhenchallengedonpremiseA1,ourdesertdullardresponds:“ImustknowA1sincetheonlywayIcouldknowA3isthroughargumentA,andIdoknowA3.”Isthisajustcomparison?IsMoore’sresponsetotheskepticrelevantlysimilar?IfMooredependsonargumentMforhisknowledgeofM3,hisresponseseemslikethedullard’s.Thedullardiswrongtorespondashedoes.Hemustsayhowheknowshispremisewithoutpresupposingthathealreadyknowstheconclusion.AndMoorewouldseemcomparablywrongintheanalogousresponsetotheskeptic.InexplaininghowheknowsM1,hemustnotpresupposethathealreadyknowsM3.48\nG.E.MOOREDoesMooredependonargumentMforhisknowledgeofM3?Thereisreasontothinkthathedoesnot,givenhisemphaticacknowledgmentthathecannotproveM3.Afterall,MwouldseemaproofofM3justasgoodasMoore’sown“proofofanexternalworld.”Mooreconcedes,ineffect,thatifhedoesnotknowthatheisnotdreamingthenhedoesnotknowofthehandbeforehim.ButthatisnotnecessarilybecausehetakeshimselftoknowM3onlythroughMoranyothersuchargument.And,inanycase,evenifheisrelyingonsomesuchargument,whichwouldrequiremakingthatcon-cession,thedefenderofcommonsensehasotheroptions.Onemight,afterall,makethatconcessiononlybecauseofthefollowing“principleofexclusion”:PEIfoneistoknowthath,thenonemustexclude(ruleout)everypossibilitythatoneknowstobeincompatiblewithone’sknowingthath.AsMooregrantsexplicitly,thepossibilitythatheisjustdreamingisincompatiblewithhisknowing(perceptually)thathehasahandbeforehim.Andthis,incombinationwithPE,isquitesufficienttoexplainhisconcessionabove.SupposeMooreisnotdependingonargumentMforhisknowledgeofM3.Althoughherecognizeshisneedtoknowheisnotdreaming,supposethatisonlybecauseheacceptsPE,ourprincipleofexclusion.ThenthesortofridiculecastonthedullardismisdirectedagainstMoore.Whatismore,itisnotevenclearthatMooremustknowhowheknowsheisnotdreamingifheistoknowM3.Thatisnotentailedbytheapplicationoftheprincipleofexclusion.AllthatfollowsfromtheapplicationofthatprincipleisthatMooremustknowthatheisnotdreaming,notthathemustknowhowheknowsthis.Infact,however,thehistoricalMooredidrelyonsomethingverymuchlikeargu-mentM(moreonthisbelow).Soishenotafterallexposedtothedamagingcompari-sonwiththedesertdullard?Notatall.Thereseemsnogoodreasonwhy,inrespondingtotheskeptic,Mooremustshowhowheknowsheisnotdreaming.Ofcoursehisresponsetotheskepticwouldbeenhancedifhecouldshowthat.Butitnowseemsnotproperlysubjecttoridiculeevenifheisnottheninapositiontoshowhowheknowsheisnotdreaming.Thequestionheisaddressingiswhetherheknowsthatheisnotdreaming,and,atmost,byextension,whatgroundshemighthaveforhisanswertothatquestion,inansweringwhichhedoesnot,norneedhe,alsoanswerthequestionofhowheknowshimselftobeawakeandnotdreaming.Itmightberepliedthatonecannotknowthathereisahandifone’sbeliefrestsontheunprovedassumptionthatoneisawake.AccordingtoMoore,however,thingswhichcannotbeprovedmightstillbeknown.Besides,eventhoughhecannotprovethatheisawake,hehas“conclusiveevidence”forit.Unfortunatelyhecannotstatehisevidence,andthematterisleftinthisunsatisfactorystateattheendof“ProofofanExternalWorld.”ButMoorehasmoretosayinanotherpaperoftheperiod,“FourFormsofScepticism”(1962:193–223).Therehetakeshimselftoknowforsureaboutthehandbeforehim,andtakesthisknowledgetobebasedonaninductiveoranalogicalargument.Wearetoldthatintrospectiveknowledgeofone’sownsensoryexperiencecanbeimmediate,unlikeperceptualknowledgeofone’sphysicalsurroundings.WhileagreeingwithRussellthatonecannotknowimmediately49\nERNESTSOSAthatoneseesahand,Moorethinks,contraRussell,thathecanknowitforcertain.AndhedisagreeswithRussellmorespecificallyinallowingknowledgeforcertainabouthishandthroughanalogicalorinductivereasoningfrompremisesknownintrospectively.However,itisdoubtfulthatanyallowableformofinference–whetherdeductive,inductive,oranalogical–willtakeusfromthecharacterofourexperiencetothesortofknowledgeofoursurroundingsthatweordinarilyclaim.Familiarskepticalscenarios–dreaming,evildemon,braininavat,etc.–showthatourexperiencepromptsbutdoesnotlogicallyentailitscorrespondingperceptualbeliefs.Experienceasifthereisafirebeforeusdoesnotentailthatthereisafirethere,experienceasifhereisahanddoesnotentailthathereisahand,etc.Perhapswhatisrequiredforone’sbeliefsandexperiencestohavecertaincontentsentailsthatthesecouldnotpossiblybeentirelyfalseormisleading.Indeed,somesuchconclusionfollowsfromcertainexternalistandepistemicrequirementsonone’sjustifiedattributionoffamiliarcontentstoone’sownexperiencesorbeliefs.Butevenifthatmuchisright–whichisstillcontroversial–one’sexperienceorbeliefthathereisahand,oryonderafire,mightstillbewildlyoffthemark.Wecannotdeducemuchofoursupposedknowl-edgeoftheexternalfromunaidedpremisesaboutourexperience.Asforinductiveoranalogicalreasoning,onlyabductivereasoning–inferencetothebestexplanation–offersmuchpromise,butitseemsquestionableasasolutionto1ourproblem.Suppose(1)thatwerestrictourselvestodatajustaboutthequalitativecharacterofourownsensoryexperience,and(2)thatweviewbeliefinacommonsensicalexternalworldasatheorypostulatedtoexplainthecourseofourexperience.Whatexactlyistheproposal?Isitproposedthatwhenordinarilyweacceptthepresenceofahandbeforeus,wedoknow,andknowonthebasisofanabductiveinference;orisitproposedratherthatinsuchcircumstanceswehaveresourcesthatwouldenableustoknowifonlyweusedthoseresourcestomakeeffectiveabduc-tivearguments?Thesecond,moremodest,proposalistoomodest,sinceitleavesourordinaryperceptualbeliefsinapositionlikethatofatheoremacceptedthroughaguessorablunder,onethatwedohavetheresourcestoproveaftermuchhardthought,butonethatwehavenotcomeclosetoprovingatthetimewhenwearejustguessingorblundering.Eventhemodestproposal,moreover,seemsunlikelytosucceed.Couldweformarichenoughsetofbeliefspurelyaboutthequalitativecharacterofoursensoryexperience,onerichenoughtopermitabductiveinferencesyieldingourcommonsenseviewofexternalreality?Thisseemsdoubtfulwhenweconsider(1)thatsuchpuredatabeliefscouldnotalreadypresupposetheexternalrealitytobeinferred,and(2)thatthepos-tulatedcommonsense“theory”ofexternalrealitymustpresumablymeetconstraintsonabductiveinference:forexamplethatthepostulatedtheorybeempiricallytestableandalsosimplerandlessadhocthanalternatives(e.g.Berkeley’s).Theserequirementsplausiblyimplythatourdatamustgobeyonddetachedobservations,andincludesomeacceptablecorrelations.Yetthesecorrelationsareunavailableifwerestrictourselves2tobeliefsaboutthecharacterofourexperience.Mostespeciallyaretheyunavailable,andmostespeciallyisthepostulatedinferenceimplausible,whenourdatabaseisrestricted,asitisbyMoore,tointrospectivelyknownfactsofone’sownthenpresentsubjectiveexperience,andtodirectlyrecalledfactsofone’sownearlierexperience.(If50\nG.E.MOOREdeprivedoftheepistemicresourcesoftestimonyandofretentivememory–exceptinsofarassuchresourcescanbevalidatedbyreason-cum-introspection,whichisnotveryfarifatall–thenthereispreciouslittlewecananylongerseeourselvesasknowing,thusdeprived.)Accordingly,theskeptichasapowerfulcaseagainstMoore’sclaimthatourknowl-edgeoftheexternalisbasedonaninductiveoranalogicalinferencefromsuchinfor-mationaboutourexperience.Itisnotrealistictosupposethatweconsciouslymakesuchinferencesineverydaylife.Itismoreplausibletoconceiveofsuchinferencesasimplicitordispositional,buteventhisstrainsbelief.Besides,evengrantedthatwemakesuchinferencesifonlyimplicitly,dotheyyieldsimplerandlessadhochypothesesthanalternatives?Thatisfarfromclear;nordosuchhypothesesseemempiricallytestableandcrediblesimplyasexplanationsofthepurelyqualitativecharacterofourthenpresentordirectlyrecalledexperience.Havingreachedadeadend,letushavesomesecondthoughtsonMoore’sviewofperceptualbeliefsasinferential.HerehejoinedavenerabletraditionalongwithRussellhimself.Ifperceptualknowledgeisthusmediateandinferential,whatknowledgecanqualifyasimmediateandfoundational?ModernphilosophybeginswithDescartes’s3canonicalanswertothisquestion.Descarteshadtwocircles,notonlythebigfamousoneinvolvingGodasguarantorofourfaculties,butalsoasmalleronefoundinthesecondparagraphofhisMeditationIII,wherehereasonslikethis:IamcertainthatIamathinkingbeing.DoInotthereforealsoknowwhatisrequiredformybeingcertainaboutanything?InthisfirstitemofknowledgethereissimplyaclearanddistinctperceptionofwhatIamasserting;thiswouldnotbeenoughtomakemecertainofthetruthofthematterifitcouldeverturnoutthatsomethingwhichIperceivedwithsuchclarityanddistinctnesswasfalse.SoInowseemtobeabletolayitdownasa4generalrulethatwhateverIperceiveveryclearlyanddistinctlyistrue.Aboutthecogito,IwishtohighlighttheinferencedrawnbyDescartes:SoInowseemtobeabletolayitdownasageneralrulethatwhateverIperceiveveryclearlyanddistinctlyistrue.JustwhatisDescartes’sargumentinsupportofthisgeneralrule?Wouldhisrea-soningtakethefollowingform?1Datum:IknowwithahighdegreeofcertaintythatIthink.2IclearlyanddistinctlyperceivethatIthink,andthatistheonly,oranyhowthebestaccountofthesourceofmyknowledgethatIthink.3SomyclearanddistinctperceptionthatIthinkiswhatexplainswhyorhowitisthatIknowIthink.4Butmyclearanddistinctperceptioncouldnotserveasasourceofthatknowledgeifitwerenotaninfalliblyreliablefaculty.5So,finally,myclearanddistinctperceptionmustbeaninfalliblyreliablefaculty.Themovefrom(1)and(2)to(3)isaninferencetoanexplanatoryaccountthatonemightacceptforthecoherenceitgivestoone’sviewofthingsinthedomaininvolved.5Descartesdoeselsewhereappealtocoherenceatimportantjunctures.Sohemay51\nERNESTSOSAbedoingsohereaswell,althoughquestionsdoariseabouthowDescartesviewscoherence.Doesheacceptthepowerofcoherencetoaddjustifiedcertainty,and,inpar-ticular,wouldheclaiminfallibilityfor(sufficientlycomprehensiveandbinding)coher-6enceashedoesforclearanddistinctintuition?Inanycase,thecomprehensivecoherenceofhisworld-viewwouldbeenhancedbyanexplanationofhowclearanddistinctperceptioncomestobesohighlyreliable,eveninfallible.AndthisisjustwhatDescartesattempts,throughhistheologicalandotherreasoning.Descartescanseethatreasonmighttakehimtoapositionthatissufficientlycomprehensiveandinterlocking–andtherebydefensibleagainstanyforeseeableattack,noholdsbarred,againstanyspecificdoubtactuallypressedorintheoffing,nomatterhowslight.Unaidedreasonmighttakehimtothatposition.Needhegoanyfurther?Whatismore:mightonereachasimilarpositionwhiledispensingwiththetrappingsofCartesiantheologyandevenofCartesianrationalism?ComparenowhowMooremighthaveproceeded:1Datum:Iknowwithahighdegreeofcertaintythathereisahand.2Icanseeandfeelthathereisahand,andthatistheonly,oranyhowthebestaccountofthesourceofmyknowledgethathereisahand.3SomyperceptionthathereisahandiswhatexplainswhyorhowitisthatIknow(withcertainty)thathereisahand.4Butmyperceptioncouldnotserveasasourceofthatdegreeofjustifiedcertainty7ifitwerenotareliablefaculty.5So,finally,myperceptionmustbeareliablefaculty.Moorecouldofcoursegoontosaymoreaboutthenatureoftheperceptionthatassureshimaboutthehand.Hemightstillsaythatsuchperceptioninvolvesanimplicitinfer-encefromwhatisknownimmediatelyandintrospectively,perhapsaninductiveoranal-ogicalinferenceofsomesort.Andthatmightmakehisviewmorecomprehensivelycoherent,butwehavealreadyseenreasonswhypostulatingsuchaninferenceisques-tionable.Sowefocusratheronasecondalternative:Mooremightwelltakeperceivingtoinvolvenoinferenceatall,notevenimplicitinference,butonlytransferoflight,nerveimpulses,etc.,insuchawaythatthecharacterofone’ssurroundingshasadis-tinctiveimpactononeselfandoccasionscorrespondingandreliablebeliefs.Thismightalsoamounteventuallytoacomprehensivelycoherentviewofone’sknowledgeoftheexternalworld.AnditsepistemologicallysignificantfeatureswouldnotdistinguishitinanyfundamentalrespectfromtheprocedurefollowedbyDescartes.ThereareotherwaysofopposingMoorebesidesthatofthetraditionalskeptic.Theseareallbasedinsomewayorotheronakeyrequirementof“sensitivity”forknowledge,oneimposedonanybeliefcandidateforknowledge,asfollows:one’sbeliefthatpamountstoknowledgethatponlyifonewouldnotbelievethatpifitwerenotthecasethatp.Itisinitiallyverytemptingtoacceptthesensitivityideacommontothevariousformsofsensitivity-basedoppositiontoMoore:namely,theskeptical,tracking,relevant-alternative,andcontextualistapproachesthatsharesomeformofcommitmenttothatrequirement.And,giventhisidea,onecanthenarguepowerfullyforthefirstpremiseoftheskeptic’s“argumentfromignorance,”AI,formulablebymeansofthefollowingabbreviations:52\nG.E.MOOREHIamahandlessbraininavatbeingfedexperiencesasifIwerenormallyembodiedandsituated(seePUTNAM).GInowhavehands.HerenowisAI:(i)Idonotknowthatnot-H.(ii)If(i),thenC(below).CIdonotknowthatG.Thatlaysouttheskeptic’sstance.Mooreforhispartgrantstheskepticpremise(ii),but8rejectsCandtherefore(i).RobertNozick’sstanceisdifferent.LikeMoore,herejectsC.Liketheskeptic,heaffirms(i).Sohemustreject(ii),whichhedoesaidedbyhisinde-pendentlysupportedaccountofknowledgeastracking.ItisnotonlyNozickwhorejectsclosureunderknownentailment;sodoestherelevantist,forwhominordertoknowsomefactXyouneednotknow(andoftencannotknow)thenegationofanalterna-tiveknowntobeincompatiblewithX,solongasitisnota“relevant”alternative.HavinggrantedtotheskeptichispremiseA(i),contextualismisabletodefendordi-naryclaimstoknowonlybydistinguishingtheordinarycontextsinwhichsuchclaimsaremadefromthecontextwheretheskepticassertshisdistinctivepremiseinthecourseofgivingargumentAI.Withthisdifferenceincontextcomesadifferenceinstandards;and,becauseofthisdifference,itisincorrecttosayintheskeptic’scontextthatoneknowsG,correctthoughitmaybetosayitinanordinarycontext.Thatresponsetotheskepticfacesaproblem.Moore’sopponentsarguethatsensi-9tivityisnecessaryforcorrectattributionsofknowledge.Despiteitsplausibility,however,seriousobjectionshavebeenpublishedagainstanysuchrequirementofsen-sitivity.Buttheproblemsforsensitivitydonotaffectasimilarrequirementof“safety.”Abeliefissensitiveiffhaditbeenfalse,Swouldnothaveheldit(i.e.itwouldhavebeenfalseonlywithoutSholdingit),whereasabeliefissafeiffSwouldhavehelditonlywithitbeingtrue.Forshort:S’sbeliefB(p)issensitiveiff~pÆ~B(p),whereasS’sbeliefissafeiffB(p)Æp.Thesearenotnecessarilyequivalent,sincesubjunctiveconditionals10donotcontrapose.Andnowweseetheproblemfacedbythecontextualistresponsetotheskeptic:namely,thatanalternativeexplanationisequallyadequateforundisputedcases(undis-puted,forexample,betweenthosewhooptforaMooreanstanceopposingtheskeptic’sdistinctivepremise(i)andthosewhooptforacontextualiststancewhichacceptsit).Accordingtothisalternativeexplanation,itissafetythat(correctattributionof)“knowledge”requires,arequirementviolatedintheordinarycasescited,whereinthesubjectfailstoknow.Onefailstoknowinthosecases,itisnowsaid,becauseone’sbeliefisnotsafe.Supposethisgeneralizestoalluncontentiouscasesadducedbythecon-textualisttofavorhissensitivityrequirement.Supposeinallsuchcasestheconditionrequiredcouldjustaswellbesafetyassensitivity.Andsuppose,moreover,thattheprob-lemsforsensitivitybrieflynotedabovedonotaffectsafety,asIhaveclaimed.Ifso,thenonecannotdifferentiallysupportsensitivityastherightrequirement,soastoinvokeitinsupportoftheskeptic’smainpremise.Hereisthestrikingresult:ifweoptforsafetyastherightrequirementthena11Mooreanstanceisdefensible,andweavoidskepticism.Thatistosay,onedoessatisfy53\nERNESTSOSAtherequirementthatone’sbeliefofnot-Hbesafe:afterall,onewouldbelievethatnot-H(thatonewasnotsoradicallydeceived)onlyifitwastrue(whichisnottosaythatonecouldbelievethatnot-Honlyifitwastrue).Intheactualworld,andforquiteadis-tanceawayfromtheactualworld,uptoquiteremotepossibleworlds,ourbeliefthatwearenotradicallydeceivedmatchesthefactwhetherornotweareradically12deceived.Consider,moreover,theneedtoexplainhowtheskeptic’spremise–thatonedoesnotknowoneselfnottoberadicallymisled,etc.–isasplausibleasitis.Thatrequire-mentmustbebalancedbyanequallyrelevantandstringentrequirement:namely,that13oneexplainhowthatpremiseisasimplausibleasitis.Tomanyofusitjustdoesnotseemsouniformlyplausiblethatonecannoteverbesaidcorrectlytoknowthatoneisnotthenbeingfedexperienceswhileenvatted.Sotheexplanatoryrequirementisinfactrathermorecomplexthanmightseematfirst.And,giventhedistributionofintuitionshere,thecontextualistandtheNozickianstilloweusanexplanation.Interestingly,ourdistinctionbetweensensitivityandsafetymayhelpusmeetthemorecomplexexplanatorydemand,compatiblywiththeMooreanstance,whichIadoptasmyown.Mypreferredexplanationmaybesketchedasfollows.Thosewhofindtheskeptic’sdistinctivepremiseplausibleonthebasisofthesortsofsensitivityconsiderationsfavoredbyopponentsofMooreancommonsensemayperhapsbeconfusingsensitivitywithsafety,andmayonthatbasisassessascorrectaffirmationsofthatpremise.Afterall,therequirementofsafetyiswellsupportedbythesortsofconsiderationsadducedbyMoore’sopponents.Sensitivitybeingsosimilartosafety,soeasytoconfuse,itisnosurprisethatonewouldfindsensitivitysoplausible,enoughtomisleadoneintoassessingascorrectaffirmationsofthatpremise.Theplausibilityoftheskeptic’spremiseisthusexplainedcompatiblywithitsfalsity,whichfitsthestanceoftheMoorean.Ofcourseallwereallyneedinordertoexplaintheplausibilityoftheskeptic’spremiseisthatitclearlyenoughfollowsfromsomethingplausibleenough.Andthesensitivityrequirementmayperhapsplaythatrolewellenoughindependentlyofwhetheritisconfusedwithasafetyrequirement.Butthatwouldstillleavethequestionofwhysensitivityissoplausibleifitisjustfalse.Andheretheremightstillbearoleforsafetytoplay:ifthisrequirementofsafetyisplausiblebecauseitistrueanddefensiblethroughreflection,thenitmaybedeeplyplausibletoussimplythroughourabilitytodiscernthetruefromthefalseinsuchapriorimatters.Compatiblywiththat,someofusmaybemisledintoacceptingtherequirementofsensitivitybecauseitissoeasytoconfusewiththecorrectrequirement,14thatofsafety.IhavewantedtoconveythepoweranddepthofMoore’sthoughtnotonlybydescrib-ingitataloftydistancebutalsobyengagingwithitatclosequarters.DespitethereservationsIhaverecordedonthisorthatpoint,Ihopetohavemadeitclearhowper-suasivelyrightarehisviewsonsomeofthemostdifficultanddisputedissuesinthehistoryofoursubject.Butbeingrightdoesnotaloneconfergreatnessinphilosophy.Fromhisearliestdaysasathinker,Moorewasnotonlyrightbutalsoabletothinkforhimselfinwaysopposedtotheregnantorthodoxies,andtoprevailasamasterdialec-tician.Onemainsourceofhisinfluenceisnowimpossibletocapturefully,however,sinceitresidedinhisvivavocecontributionstotheintellectuallifeofthatgoldenageof54\nG.E.MOORECambridgephilosophy.ThisSocraticsideiswellconveyedbyhisyoungerCambridgecolleague,C.D.Broad,inanobituaryforMoore:Itwasbyhislectures,hisdiscussion-classes,hisconstantandilluminatingcontributionstodiscussionattheCambridgeMoralScienceClubandtheAristotelianSociety,andhisprivateconversationwithhiscolleaguesandpupilsthathemainlyproducedhiseffectsonthethoughtofhistime.Notes1ForRussellthe“commonsensehypothesis”ofindependentphysicalobjectsis“simpler”thanthesuppositionthatlifeisbutadream(asheexplainsinchapterIIofTheProblemsofPhilosophy).ForQuinethe“hypothesisofordinaryphysicalobjects”is“posited”or“pro-jected”fromthedataprovidedbysensorystimulations.“Subtractinghiscuesfromhisworldview,wegetman’snetcontributionasthedifference”(WordandObject(Cambridge,MA:MITPress,1960),p.5).ThatQuine’spositionisdeeplyproblematicisshownbyStroud(TheSignificanceofPhilosophicalSkepticism(Oxford:OxfordUniversityPress,1984),ch.VI).2ThisisarguedbyWilfridSellarsin“Phenomenalism,”inhisScience,Perception,andReality(London:RoutledgeandKeganPaul,1963)(seeSELLARS).3TheshifttodiscussionofDescartesmayseemabrupt;however,whatwefindaboutthenatureofimmediateknowledgeinthatdiscussionhasimportantimplicationsforapositionthatMoorefailedtoexplore.SkepticswhoarewillingtograntDescarteshisimmediateknowledgethroughintrospectionorrationalintuitionwouldneedtoexplainexactlywhyperceptioncouldneveryieldsuchknowledge.(Andwhatofmemory?)ThediscussionofDescartestofollowismeanttohighlightthisissue.4ThePhilosophicalWritingsofDescartes,ed.J.Cottingham,R.Stoothoff,andD.Murdoch(Cambridge:CambridgeUniversityPress,1975),vol.II,p.24.5InhisPrinciplesofPhilosophy(PartIV,art.205)forexample,henotesthatifwecaninterpretalongstretchofotherwiseundecipherablewritingbysupposingthatitiswrittenin“one-offnaturallanguage,”wherethealphabethasallbeenswitchedforwardbyoneletter,etc.,thenthisisgoodreasonforthatinterpretation.Therehealsoarguesforhissci-entificaccountofrealityintermsofcertainprinciplesbyclaimingthat“itwouldhardlyhavebeenpossibleforsomanyitemstofallintoacoherentpatterniftheoriginalprincipleshadbeenfalse”(Cottinghametal.,PhilosophicalWritingsofDescartes,p.290).Ofcourse,ifwejoinDescartesinadoptingthissortofinferencetoanaccountthataidscomprehensivecoherencewewillneedtobeabletodistinguishitrelevantlyfromtherejectedabductiveinferencetoanexternalworldfromintrospectivedataaboutone’sownexperienceanddirectmemoriesaboutone’spastexperiences.Butthereareimportantdifferences:foronething,thepresentCartesianinferenceisnotaninferencetoacausalaccount,onewithdis-cerniblerivalsthatweareunabletoruleoutwithoutviciouscircularity.ButitremainstobeseenwhethertheadditionaltheologicalprojectthatDescartesnextlaunchesisorisnotopentosimilarproblemsasthosethatbesettheabductiveinferencetotheexternalworld,orevenworseproblems.WedonotconsidertheseissueswhicharemattersofdetailbycomparisonwiththemoreabstractepistemologicalstructureofDescartes’sreasoningthatweconsider.6MyattributiontoDescartesistentativebecauseoftheenormousbibliographyonthe“Cartesiancircle.”Indeferencetothatimportanttraditionofscholarship,IdonomorethansuggestthatthereislogicalspaceforaninterpretationofDescartesthatisperhapsmore55\nERNESTSOSAcomplexthanmanyalreadytried,butthatseemscoherentandinteresting.(IammyselfconvincedthatthisisDescartes’sactualposition,anddefendthismorefullyelsewhere.)7HereonewouldreduceDescartes’srequirementofinfalliblecertainty.8RobertNozick,PhilosophicalExplanations(Cambridge,MA:HarvardUniversityPress,1981).9KeithDeRose,“SolvingtheSkepticalProblem,”PhilosophicalReview104(1995),pp.1–52.10Ifwaternowflowedfromyourkitchenfaucet,itwouldnotthenbethecasethatwatersoflowedwhileyourmainvalvewasclosed.Butthecontrapositiveofthistrueconditionalisclearlyfalse.11Imeanthatweinourreflectionandinourdiscussionsinjournalandseminar,avoidskep-ticism;wecansayrighthereandnowthatwedoknowvariousthings,andnotjustthatwesay“Iknow”correctlyinvariouscontextsnotnowourown.12Thissortofexternalistmovehasbeenwidelyregardedasunacceptablycircular,mistakenly,asIargueindetailelsewhere.13WhenIhaveaskedmyclassestovoteonthatpremise,generallyIhavefoundthatthosewhofinditfalseoutnumberthosewhofindittrue,andquiteafewprefertosuspendjudg-ment.Ateverystagepeoplespreadoutinsomesuchpatternofthree-wayagreement-failure.14ForafullerdefenseofaMooreanstanceinepistemologybycomparisonwithalternativeideasontheepistemologymarketplace,seemy“HowtoDefeatOppositiontoMoore,”PhilosophicalPerspectives13;Epistemology13(1999),SupplementtoNoûs,pp.141–55.BibliographyWorksbyMoore1903:PrincipiaEthica,Cambridge:CambridgeUniversityPress.1912:Ethics,London:Williams&Norgate.1922:PhilosophicalStudies,London:RoutledgeandKeganPaul.1953:SomeMainProblemsofPhilosophy,London:Allen&Unwin.1962:PhilosophicalPapers,NewYork:CollierBooks.WorksbyotherauthorsBaldwin,T.(1990)G.E.Moore,LondonandNewYork:Routledge.Fratantaro,S.(1998)TheMethodologyofG.E.Moore,AldershotandBrookfield,VT:AshgatePublishingLtd.Klemke,E.D.(1969)TheEpistemologyofG.E.Moore,Evanston,IL:NorthwesternUniversityPress.Schilpp,P.A.(1968)ThePhilosophyofG.E.Moore,LaSalle,IL:OpenCourt.Stroll,A.(1994)MooreandWittgensteinonCertainty,Oxford:OxfordUniversityPress.White,A.R.(1958)G.E.Moore,Oxford:BlackwellPublishers.56\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd20014C.D.Broad(1887–1971)JAMESVANCLEVECharlieDunbarBroadwasaleadingcontributortoanalyticphilosophyofthetwenti-ethcentury,knownnotsomuchforanystartlinglyoriginaldoctrineshepropoundedasforhisformidablepowersofdistinction,analysis,andargument.BorninLondon,hewaseducatedatDulwichCollegeandCambridge.HeenteredCambridgein1905,firststudyingphysicsandchemistryinthenaturalsciencetriposandthenswitchingtophilosophyinthemoralsciencetripos.TheinfluenceofRussellandMooreatCambridgewasthenverystrongandshowsitselfinBroad’swork(seeRUSSELLandMOORE).HepublishedhisdissertationasPerception,Physics,andRealityin1914.Foraperiodofyearsbeginningin1911heservedasG.F.Stout’sassistantinSt.Andrews,andin1920hewasappointedprofessorattheUniversityofBristol,wherehegavethecourseoflecturesinphilosophyfornaturalsciencestudentsthatbecameScientificThought.In1922hedeliveredtheTarnerLectures,subsequentlypublishedasTheMindandItsPlaceinNature,andwasinvitedtosucceedMcTaggartaslectureratCambridge.AfterMcTaggart’sdeathin1925,heoversawthepublicationofthesecondvolumeofMcTaggart’sTheNatureofExistence,whichservedasthestimulusforwritinghisownmonumentalExaminationofMcTaggart’sPhilosophy.(Thisisthebookofchoiceforanymetaphysicianwhoissentencedtoexileonadesertisland.)From1933untilhisretire-mentin1953hewasKnightbridgeProfessorofMoralPhilosophyatCambridge.HisotherbooksincludeFiveTypesofEthicalTheory(1930)andtwocollectionsofpapers,EthicsandtheHistoryofPhilosophy(1952)andReligion,Philosophy,andPsychicalResearch(1953).AfterhisdeathhisstudentCasimirLewypublishedhiscoursesoflecturesonLeibnizandKant.ThescopeofBroad’sinterestswasvast.Selectedforattentioninthisarticlearefourmaintopics:hisconceptionof“criticalphilosophy,”hiswritingsonsensaandpercep-tion,hisphilosophyoftime,andhisviewsontherelationofmindtomatter.NotcoveredhereareBroad’simportantcontributionstothefollowingareas:induc-tivelogic(hesoughttoidentifyandjustifysomeprincipleabouttheworldthat,iftrue,wouldmakeinductionlegitimate);determinismandfreedom(hearguedthatthenotionof“obligability”ormoralresponsibilityisincompatiblebothwithdeterminismandwithindeterminism,makingitaproblematicconcept);therelevanceofpsychicresearchtophilosophy(heassessedtheevidenceforparanormalphenomenaandidentifiedmeta-physicalprinciplesthatwouldhavetobegivenupiftherealityofsuchphenomena57\nJAMESVANCLEVEbecameestablished);andethics.HisFiveTypesofEthicalTheory,astudyoftheethicalsystemsofSpinoza,Butler,Hume,Kant,andSidgwick,wasawidelyusedtext;healsowroteinfluentialpapersinmetaethics,clarifyingthestatusofnon-naturalistintuitionismofthesortespousedbyMoore,naturalisttheoriesofthe“moralsense”variety,andnon-cognitivistor“interjectional”theories.CriticalversusspeculativephilosophyInadiscussionofthenatureandvalueofphilosophyintheintroductiontoScientificThought,Broaddistinguishedtwobranchesofhissubject,criticalphilosophyandspeculativephilosophy.Thefirsttaskofcriticalphilosophyis“totaketheconceptsthatwedailyuseincommonlifeandscience,toanalysethem,andthustodeterminetheirprecisemeaningsandtheirmutualrelations.”Conceptsripeforsuchanalysisincludetheconceptsofsubstance,cause,place,date,duty,andself.Thesecondtaskofcriticalphilosophyistotestthebeliefsthatweconstantlyassumeineverydaylifeandscience,“resolutelyandhonestlyexposingthemtoeveryobjectionthatonecanthinkofoneselforfindinthewritingsofothers.”Beliefssubjecttosuchcriticalscrutinyincludethebeliefsthatweliveinaworldofobjectsthatareindependentofourknowledgeofthemandthateveryeventhasacause.Wemayemergefromcriticalphilosophywithver-ballythesamebeliefswestartedwith,buttheprocesswillhave“enabledustoreplaceavaguebeliefbyaclearandanalysedone,andamerelyinstinctivebeliefbyonethathaspassedthroughthefireofcriticism.”Hethenwentontocharacterizespeculativephilosophyasfollows:Itsobjectistotakeovertheresultsofthevarioussciences,toaddtothemtheresultsofthereligiousandethicalexperiencesofmankind,andthentoreflectuponthewhole.Thehopeisthat,bythismeans,wemaybeabletoreachsomegeneralconclusionsastothenatureoftheUniverse,andastoourpositionandprospectsinit.Broadnotedthatspeculativephilosophyislesscertaininitsresultsthancriticalphi-losophy,andthatitmustbeaugmentedbycriticalphilosophyifitistobeofanyvalue.Heengagedinbothvarietiesofphilosophyhimself,buthisstrongsuitwascriticalphi-losophy.IthinkitisfairtosaythatmanyanalyticphilosopherswouldciteBroad’sdefinitionofcriticalphilosophyasanexcellentdescriptionofwhattheydoandBroadhimselfasanoutstandingpractitionerofit.Sense-dataandperceptionBroadwasoneoftheleadingexponentsofasense-datumtheoryofperception.Theterm“sense-datum”wasintroducedbyRussellandMoore;Broadhimselfalmostinvariablypreferredtheterm“sensum.”Thoughsometimesusedbroadlytocoverthesensuousaspectofexperienceingeneral(howeveritmaybeanalyzed),theterms“sensedatum”and“sensum”haveforBroadandotherphilosophersofhiseraanarrowerandmoreprecisemeaning.Thenotionofasensumhasapplicationonlyifoneadoptsanact–objectanalysisofsensoryexperience.Toseewhatthismeans,con-siderthevarioustypesofsensoryexperiencearrangedinanorder,startingwiththose58\nC.D.BROADofsight,passingthroughthoseofhearing,taste,andsmell,andendingwithbodilysensationslikeheadache.Atthebeginningoftheseries,Broadclaimed,itseemsplausibletoanalyzeasensationofredintotwocomponents,anactofsensingandaredobject.Attheotherendoftheseries,itdoesnotseemplausibletoanalyzeasensationofheadacheintoanactofsensinganda“headachy”object.Havingaheadacheisnotsensingsomething–itissensingsomehow.Inthemiddleoftheseries(withtasteandsmell),itmaynotbeobviouswhetheronecandistinguishactandobject.Somephilosophershaveassimilatedtheentireseriestooneortheotherendofit,advocatingeitheranact–objectanalysisacrosstheboard(H.H.Price)oranobjectless“wayofsensing”analysisacrosstheboard(ThomasReid).Broadsawnoreasontotreattheentireseriesuniformly;hetookbodilysensationstobeobjectless,butespousedanact–objectanalysisatleastforsight,hearing,andtouch.“Itseemstomemuchmorecertainthat,inasensationofred,Icandistinguishtheredpatchandtheactofsensingit,thanthat,inasensationofheadache,Icannotdistinguishaheadachyobject,andanactofsensingit”(1923:256).Theredpatchthatfiguresastheobject-componentinthesensationofredisthesortofthingBroadmeantbyasensum.ThatisnottosaythatwhenIamseeingaripetomato,thetomatoisasensum.Evenifmyexperienceofatomatowereatotalhallucinationandtherewerenoredphysicalobjectsinmyenvironment,therewouldstillbesomethingredthatIamsensing,andthatsomethingisasensum.Thussensaarenotautomaticallytobeidentifiedwithphysicalobjectsorevenpartsofthem;theirrelationtophysicalthingsisamorecomplexaffairtobediscussedfurtherbelow.Thetheoryofsensamaybeexpoundedfurtherbynotingsomeofthefamiliarfactsitismeanttoexplain.Whenviewingapennyoracoffeecuptiltedawayfrommylineofsight,ImaybecertainthatIamhavingtheexperienceexpressedby“Thislooksellip-ticaltome,”eventhoughIknowthatinfactthepennyorthecupisnotellipticalbutround.Thismuchissupposedtobeafactonwhichallpartiesagree.Thesensumtheoryanalyzesthesituationasinvolving“theactualexistenceofanellipticalobject,whichstandsinacertaincognitiverelationtomeontheonehand,andinanotherrelation,yettobedetermined,totheroundpenny”(1923:237–8).Thisellipticalobjectisasensum.Broadpithilyconveyedtheguidingmotivationforpositingitasfollows:“If,infact,nothingellipticalisbeforemymind,itisveryhardtounderstandwhythepennyshouldseemellipticalratherthanofanyothershape”(p.240).GeneralizingfromwhatBroadsaysaboutthepenny,wemayputtheessentialcoreofthesensumtheoryasfollows:wheneveranyobjectxappearstoasubjectStohaveapropertyF,itdoessobecauseSisdirectlyawareofanitemy(asensum)thatreallydoeshavethepropertyF.Theitemyisthesensum,anditsrelationtoxcannotingeneralbeidentity(sinceifxappearsFwithoutbeingF,yisFandxisnot).(Certainrestrictionsaretobeunderstoodasattachingtothisformulation;“appearsF”isusedphenomenally,notcomparatively,andthevariableFrangesovercolor,shape,anddis-tance.)Itisgenerallyheldthatsensathemselves,unlikephysicalobjects,neverappeartohaveanypropertyFwithoutreallyhavingit.Thisisimplicitinthereasonforposit-ingsensainthefirstplace:ifsensacouldappeartohavepropertiestheydonotreallyhave,wewouldhavetopositasecondtierofsensatobethebearersofthepropertiesapparentlypossessedbysensainthefirsttier.59\nJAMESVANCLEVEWhatabouttheconverseassumption,thatsensahaveonlythepropertiestheyappeartohave?Broaddeniedthis,holdingthatsensamaybemorevariegatedordeter-minatethantheyappeartobe.Thisenabledhimtoavoidtheobjectionthatsensawouldbeindeterminateintheirproperties.Ifasensumappearstobemany-speckledwithoutappearingtobeexactlyn-speckledforsomen,Broaddoesnothavetosaythatthesensumhasspeckleswithouthavinganydefinitenumberofthem.Manywritersassumethatvisualsensamusthaveonlytwodimensions–thattheyareextendedinlengthandbreadth,butaltogetherlackingindepth.Broadarguedtothecontrarythatvisualsensedataareasfullyvoluminousorthree-dimensionalasanyobjectsinphysicalspace.Manywritersassumethatsensaarementalentities.This,too,Broaddenied.Accordingtohim,theyareneithermentalnorphysical,buthavealegineachrealm(1925:184).Forexample,theyarelikephysicalobjectsinhavingspatialqualitieslikeextensionandshape,butlikementalthingsinbeingprivatetoobserversandsensemodalities.Theirprivacy,however,doesnotmeanthatsensaareexistentiallymind-dependent;likeRussell,Broadaccepteditasarealpossibilitythattherecanbeunsensedsensibilia.Sense-datahavebeenoutofvoguefornearlyfiftyyears.Oppositiontothemhasstemmedfromtwomainmotives.First,theyarehardtoaccommodatewithinapurelyphysicalistviewoftheuniverse:iftheexperienceofred,whetherveridicalornot,involvesaliterallyredobject,itishardtoseewithwhatbrainentitiesorprocessesthisredobjectcouldbeidentified.Second,sensamakedifficultiesfordirectrealistaccountsofperception:theyareoftenthoughttoconstitutea“veil”betweenperceiversandthephysicalworld,cuttingusoffnotonlyfromdirectperceptionofphysicalthingsbutknowledgeofthemaswell.What,then,arethealternativestoadmittingsensa?Aradicalalternativeistodeny(withDanielDennettandothers)thatthereisasensuouselementinexperienceatall,inwhichcasetherewouldbenothingforthesensumtheorytoanalyze.Broadwouldhavedismissedthissuggestionasaflagrantdenialofthefacts.Hedid,however,recog-nizetwoalternativeanalysesofthefactsinadditiontothesensumtheory:themulti-plerelationtheoryandthemultipleinherencetheory.Thefirstofthesealternativesismentioned,thoughnotdiscussed,inScientificThought;botharediscussedinTheMindandItsPlaceinNature.Onewaytounderstandthedifferencesamongthethreetheoriesistoseewhateachwouldsayaboutthephenomenonofperceptualrelativity:thefactthatthesameobjectcanappeartohavedifferentpropertiestodifferentobserversorfromdifferentview-points,aswhenwaterfeelshottoonehandandcoldtoanother,oramountainlooksbluefromadistanceandgreencloseup.Itwouldbecontradictory,ofcourse,tosaythatthesamemountainisbothgreenandblue,period.Buttherearethreewaystostatethefactsofperceptualrelativitywithoutcontradiction.First,wecansaythattheincompatiblecolorsinhereindifferentsubjects.Thisiswhatthesensumtheorysays:Isenseonebatchofsensa(blueones)whenIamviewingthemountainfromafarandanotherbatchofsensa(greenones)whenIamstandingonthesummit.Second,wecansaythatthemountainlooksblueasIapproachitonthehighway,thatitlooksgreenwhenIgetthere,andthatonatleastoneoftheseoccasionsitlookstohaveacolorthatnothinginthesituationactuallyhas.ThisiswhatBroadcalledthe“multiplerelationtheoryofappearing”;itholdsthatappearingFisanunanalyzablerelationbetweenan60\nC.D.BROADobject,aproperty,andamind,notinvolvingtheexistenceofanyentitythatreallyisF.Third,wecansaythatthemountainisbluefromhereandthatitisgreenfromthere,avoidingcontradictionbyexpandingthenumberofplacesintherelationofinherence.Thisisthemultipleinherencetheory,inwhichwegiveuptheordinaryviewthattheinherenceofacolorinathingisatwo-termrelationbetweenthecolorandthething;rather,itisathree-termrelationbetweenacolor,athing,andaplaceoraviewpoint.Colorsdonotinhereinobjectssimply,butonlyinobjects(or“regionsofpervasion”)fromplaces(or“regionsofprojection”).Themultiplerelationtheoryhasthecounterintuitiveconsequencethatobjects“canhavequalitieswhicharedifferentfromandinconsistentwiththosewhichtheyseemoncarefulinspectiontohave”(1925:160).Themultipleinherencetheoryinvolvesapuzzlingnewformofinherence;inaddition,ithasthepuzzlingconsequencethatthecolorsofobjectsare“causallyadventitious”tothem,inthesensethattheimmediatecausaldeterminantsofthecolorpervadingaregionlienotinthatregionbutinsomeotherregion,a“regionofprojection”containingasuitablyfunctioningbrain.Broaddidnotthinkeitherofthesetheorieswasdecisivelyrefutable,buthefoundthesensumtheorypreferableonthewhole.Afurtherpossibilityisworthmentioning.OnemayaccepttheverbalformulaBroadusesincharacterizingthemultiplerelationtheory–“anobjectcanappearFwithoutanything’sbeingF”–withouttakingtherelationofappearingFasunanalyzable.That,ineffect,iswhatRoderickChisholmdoes,analyzing“xappearsFtoS”as“xcausesStosenseF-ly.”Heabandonstheact–objectanalysisofsensinginfavorofanadverbialapproach,accordingtowhichtohaveasensationofredissimplytosenseinacertainway.Hethenanalyzestherelationofanobject’slookingredtoSasamatteroftheobject’scausingStosenseredly.ItremainstosaysomethingaboutBroad’sviewsontherelationofsensingtoper-ceivingandofsensatophysicalobjects.WhenIperceivesomething,Idonotmerelysenseasensum;Ialsobelieveinanobject(e.g.abelloracandle)towhichthesensumisrelated.Broaddevotedconsiderableattentiontoanalyzingthisbeliefanditsobject.Heworkedoutelaborateanswerstoquestionslikethese:howdophysicalobjectscausesensa,andhowaretheplaces,dates,durations,shapes,andsizesofphysicalobjectstobedefinedorknownintermsofthecorrespondingfeaturesofsensa?Thecorrespond-ingfeaturesofsensathatgobythesamenamearesometimesliterallythesameandsometimesnot.Sensaandphysicalobjectsbothhaveshapeinthesamesense,buttheydonothavelocationinthesamesense.Sensaareliterallylocatedonlyintheirownspaces(e.g.asensumofcolormaybeinthecenterofone’svisualfield).Theymayalsobeassignedlocationsinphysicalspace,butonlyina“Pickwickian”sense.Roughly,tosaythatavisualsensumsis“in”physicalplacepmeansthis:ifIturnmyheadtobringsintothecenterofmyvisualfieldandthenfollowmynose,Iwillbringmyselfcloserandclosertop,obtainingalongthewayaseriesofsensalikesbutbecominglargerandbrighteruntilIeventuallyadvancebeyondpandthes-likesensadisappear.Broadwasneveraphenomenalist,onewhobelievesthatphysicalobjectsarecom-posed(orlogicallyconstructed)entirelyofsensa.Hebelievedthatphysicalobjectsareheterogeneouscomposites,containingasliteralpartsatomsorwhatevertinierparticlesarerecognizedbythebestscienceofthedayandcontainingasPickwickianpartssensabelongingtothevarioussenserealms.Healsoespousedsomethinglikethe61\nJAMESVANCLEVEtraditionaldistinctionbetweenprimaryandsecondaryqualities,maintainingthatshapesinhereliterallybothinthescientificconstituentsofphysicalobjectsandinsensa,whilecolorsinhereinsensaalone.Hismainreasonfordenyingthatcolorsinhereinphysicalobjectswasthatweneedtorefertotheshapesofphysicalobjectstoexplainwhywesensesensaofvariousshapes,butdonotneedtoassigncolorstophysicalobjectsinordertoexplainwhywesensecoloredsensa.PhilosophyoftimeBroadhadagooddealtosayaboutthenatureofspaceandtime,includinginterpre-tationsinScientificThoughtofEinstein’sSpecialandGeneralTheoriesofRelativity,whichwerethenfairlynewonthescene.Ifocushereonhismorepurelymetaphysicalviewsabouttime,aspresentedbothinScientificThoughtandExaminationofMcTaggart’sPhilosophy.Somephilosophersholdthatonlythepresentisreal;othersholdthatpast,present,andfutureareallequallyreal.InST,Broadadvancedatheoryintermediatebetweenthesetwo,acceptingtherealityofthepresentandthepast,butholdingthat“thefutureissimplynothingatall”(1923:66).Thetimeseriesislikeagrowingline,anditpos-sessesadirectionbecause“freshslicesofexistence”arealwaysbeingaddedtotheforwardendofit.Hedrewfromthistheconclusionthatjudgmentsostensiblyaboutfutureeventsareneithertruenorfalseatthetimeswhentheyaremade,sincethereisnothingtheninexistencetomakethemtrueorfalse(p.73).Broaddistinguishedtwoaspectsoftimeoroftemporalfacts,whichhecalledthe“extensive”(orstatic)andthe“transitory”(ordynamic)aspects.ThedistinctioniscloselyrelatedtoMcTaggart’sdistinctionbetweentheAseriesandtheBseries.Calltherelationsofbeingearlierthan,laterthan,andsimultaneouswith“Brelations”;callthecharacteristicsofpastness,presentness,andfuturity“A-characteristics.”AnAseriesisthenanyseriesofeventsormomentswhosemembershaveA-characteristics,andaBseriesanyserieswhosemembersarerelatedbyB-relations.McTaggartnotedthattruthsinvolvingtheB-relationsarepermanent,whiletruthsinvolvingtheA-characteristicsaretransitory.Aneventthatisearlierthananothereventisalwaysearlierthanit,butaneventthatisfuturewillnotalwaysbefuture:itwillbecomelessandlessremotelyfuture,thenitwillbecomepresent,andfinallyitwillbecomemoreandmorepast.Agreatdivideinphilosophiesoftimeseparatesthosewhoacknowledgethetransi-toryaspectoftimeandthosewhorejectit.Russellandmanyothersdenyit,affirmingthattemporalfactsareexhaustedbythoseinvolvingtheB-relations.Broadupheldit,agreeingwithMcTaggartthatthetransitoryaspectoftimeisessentialtoit.Hedidnot,however,believethateventsbecomepresentinthewaythatmaybesuggestedbyMcTaggart’slanguage,thatis,theeventsarealreadystrungoutandbecomepresentasthepalingsofafencebecomeilluminatedbythepassageofaspotlight.Becomingisnotanalogoustoqualitativechange,inwhichasubjectthatalreadyexistsacquiresanewproperty;rather,tobecomepresentisjustto“become,”inanabsolutesense.Broad’sadherencetothetransitoryaspectoftimeisreflectedinsteadinhisinsistenceontheindispensabilityoftense,fortensedstatementsarepreciselythosethatmaychangetruthvaluewiththepassageoftime.62\nC.D.BROADTheindispensabilityoftense–thethesisthattensedverbscannotbedoneawaywithintheanalysisoftemporaldiscourse–isperhapsBroad’smostimportantthesisinhislaterphilosophyoftime.HeopposedbothRussell’sanalysisoftenseintermsoftense-lesscopulasandB-relationsandMcTaggart’sanalysisoftenseintermsoftenselesscopulasandA-characteristics.AccordingtoRussell,anutteranceofthesentence-type“itisnowraining”meansthatanoccurrenceofrainissimultaneouswiththatveryutterance;inanalogousfashion,anutteranceof“ithasrained”or“itwillrain”wouldmeanthatanoccurrenceofrainisearlierorlaterthanthatveryutterance.Broadexpresseddoubtaboutwhetherthekindofself-referenceinvolvedhereisreallypossi-bleandaboutwhethertenselessverbsareanythingbutaphilosopher’sfiction.Hismainobjection,however,wassimplythatRussell’sanalysisleavesoutthetransitoryaspectoftemporalfacts.Ifanoccurrenceofrainis(tenselessly)simultaneouswithacertainutterance,itisalwayssimultaneouswiththatutterance,makinganyutteranceof“itisnowraining”trueeternallyifitistrueatall.McTaggart’spresuppositionthattenseiseliminableisanessentialpartofhisnoto-riousargumentforthetheunrealityoftime,anargumentthatBroadsubjectedtopen-etratinganalysis.McTaggart,unlikeRussell,believedthattherecouldnotbetimewithoutanA-series:aseriesofeventsormomentsexemplifyingthecharacteristicsofpast,present,andfuture.HiscaseagainsttherealityoftimewasthattheA-seriesinvolvesacontradiction:theA-characteristicsaremutuallyincompatible,yeteachiteminanyA-seriesmusthavethemall.TothistheobviousobjectionisthateacheventhasalloftheA-characteristicsonlysuccessively,andthereisnocontradictioninthat.Aneventthatisnowpresentisnotnowpastandfuture;rather,ithasbeenfutureandwillbepast.ButMcTaggartanticipatedthisobjection,andrepliedthatourattempttoremovethecontradictiononlyraisesitanew.WhenwesaythatShasbeen(willbe,isnow)P,wearesayingthatSisPatamomentofpast(future,present)time.Thustosaythataneventhasbeenfuture,isnowpresent,andwillbepastimpliesthatthereisanA-seriesofmoments.Andthis,McTaggartalleged,bringsbackacontradictionjustliketheoriginalone:everymoment,likeeveryevent,ispast,present,andfuture.ButwhydidMcTaggartthinktherewasacontradictiontobeginwithinsayingthataneventisfuture,present,andpast,acontradictionthatremainsevenifweaddthequalification“successively”?Tosaythataneventissuccessivelyfuture,present,andpastistosay(ifitisnowpresent)thatitwasfutureandwillbepast.AccordingtoBroad,itisatthispointintheargumentthatMcTaggart’sassumptionthattenseiseliminableplaysacrucialrole.Broadarticulatedtheassumptionasfollows:whatismeantbyasentencewithatensedverborcopulamustbecompletelyandmoreaccuratelyexpres-siblebyasentenceinwhichthereisnotensedverborcopula,butonlytemporalpred-icatesandtenselessverbsorcopulas.Tohighlightthefactthatthemoreaccurateexpressionmustbefreeoftense,letususe“be”asatenselesscopula.ThenMcTaggart’sclaimis(e.g.)that“ewasfuture”means“forsomemomentm,ebefutureatm&mbepast.”Well,ifmbepast,itistimelesslyorsempiternallypast.Andthatcontradictstheassumption,inherentinbeliefintheA-series,thateverymomentissometimesfutureandsometimespresent.ThusBroadconcludes:[T]hesourceofMcTaggart’sregressisthat,ifyoutakethe“is”in“tispresent”tobetime-less,youwillhavetoadmitthattisalsopastandfutureinthesametimelesssenseof“is”.63\nJAMESVANCLEVENowthisisimpossible,foritisobviousthattcanhavethesepredicatesonlyinsuccession.If,toavoid,this,yousaythatthe“is”in“tispresent”means“isnow”,youhavenotgotridoftemporalcopulas.Therefore,ifyouarecommittedatallcoststogettingridofthem,youwillnotbeabletorestatthisstage.(1933:314–15)SoBroadinsistedontheineliminabilityoftense.Russell’sattempttoeliminatetenseinfavoroftheB-relationsignoresthetransitoryaspectoftime,andMcTaggart’swayofgettingridofitmakesthetransitoryaspectcontradictory.ThemoralBroaddrewisthatifwewishtodojusticetothetransitoryaspectoftime,wemusttaketenseseriously.Broadalsodiscussedatlengththeontologicalcategoriesofthingandprocess.Oneofthedifferencesisthatthingsendureliterallythroughtime,whereasaprocessper-sistsonlyinvirtueofhavingdistinctpartsorphasesthatexistatvariousmomentswithintheinterval.IfIsay“ThisisthesamechairIsatinyesterday,”Imeanthatlit-erallythesameobjectIsatinyesterdayisherenow,butifIsay“Iamstillhearingthesamebuzzingnoise,”ImeanonlythatIamhearinglaterphasesofaprocesswhoseearlierphasesIheardbefore.Thosewhobelieveinthedynamicaspectoftimecom-monlyholdthatidentitythroughtimeisamatterofthing-likeendurance,whilethosewhoembraceastaticconceptoftimetypicallyholdthatidentitythroughtimeisreallyamatterofprocess-likepersistence.Confoundingexpectationsonthisscore,Broadcombinedhisbeliefinthetransitoryaspectoftimewiththeviewthatthingsaredis-pensableinontologyinfavoroflogicalconstructionsoutofprocesses.Ashesometimesbluntlyputit,athingisjustalongandboringevent.MindandmatterIntheconcludingchapterofMindandItsPlaceinNature,Broadundertooktoclassifythevariouspossiblemetaphysicaltheoriesontherelationofmindtomatter,tosumuptheirstrongandweakpoints,andtodecidebetweenthem.Hisschemeofclassificationyieldedseventeentypesofpossibletheory,whichhethoughtcouldbenarroweddowntothreeorfourbestoptionsandonethatwasmostreasonableoverall.InowgiveasomewhatsimplifieddescriptionofBroad’sschemeandofhisownfavoredalternative,whichhecalled“emergentmaterialism.”SupposetheX-propertiesofanythingfollowwithlogicalormetaphysicalnecessityfromsomeselectionofitsY-properties(ortheY-propertiesofitsparts).ThiscouldhappenbecausetheX-propertiesareidenticalwiththeY-propertiesorareanalyzableintermsofthem.Inthiscase,X-propertiesarereducibletoY-properties.SupposetheX-propertiesofathingarenotreducibletotheY-propertiesofitspartsortherelationsamongthem,butdofollowwithnomologicalnecessityfromtheseY-propertiesandrelations.Inthiscase,X-propertiesareemergentfromtheY-properties.Withthesepreliminarynotionsgranted,wecandefineBroad’snotionofa“differ-entiatingattribute”(orforshort,simplyanattribute):anattributeisahighlygeneralpropertythatisinstantiatedintheuniversewithoutbeingeitherreducibletooremergentfrompropertiesofanyothertype.Broadanalyzedmaterialityastheconjunctionofextension,publicity,persistence,andexistentialindependencefromobservingminds.Heanalyzedmentalityasa64\nC.D.BROADhierarchyofpropertiesrangingfrombaresentienceupthroughthehighercognitive(bothintuitiveanddiscursive)andaffectivecapacities.WecannowarriveatmostofthepositionsinBroad’sschemebyaskingthefollow-ingquestionsabouteachofmaterialityandmentality:Isitinstantiatedintheuniverseornot?Ifso,isitreducible,emergent,orneither?Andifitisreducibleoremergent,withrespecttowhatotherpropertiesisitemergentorreducible?Ifmaterialityandmentalityarebothinstantiatedintheuniverse,butneitherisreducibletooremergentfromanythingelse,thatmakesbothofthemattributesinBroad’ssense,givingusthepositionhecalled“dualism.”Hesubdividedthisaccordingtowhethermaterialityandmentalitycanorcannotinhereinthesamesubstance.Iftheanswerisyes,wehavedualismofcompatibles,thepositionofSpinoza;ifitisno,wehavedualismofincompatibles,thepositionofDescartes.Ifmaterialityisanattributebutmentalityisnot,wehavethefamilyoftheoriesBroadcalled“materialist.”Thissubdividesaccordingtothethreewaysinwhichmen-talitymightfailtobeanattribute.Ifmentalityisnotinstantiatedintheuniverseatall,wehavepurematerialism(orwhatwouldnowadaysbecalledeliminativematerialism).Ifmentalityisreducibletosomethingelse(determinatesofmateriality,presumably),wehavereductivematerialism.Broaddiscussedtwochiefvarietiesofthis,“molarbehaviorism,”accordingtowhichhavingamentalstatejustmeansbehavingincertainways,and“molecularbehaviorism,”accordingtowhichmentalprocessesaretobeidentifiedwithprocessesinthebrainandnervoussystem.Finally,ifmentalityisemer-gent,wehaveemergentmaterialism,accordingtowhichmentalpropertiesemergeasnovelpropertiesofmaterialsystemsthatachieveacertaindegreeofcomplexity.Ifmentalityisanattributebutmaterialityisnot,wegetthefamilyoftheoriesBroadcalledmentalist.Aswithmaterialism,therearethreepossiblevarieties:puremental-ism,reductivementalism,andemergentmentalism.TheactualmentalistsBroadmen-tions–forexampleBerkeley,Leibniz,andMcTaggart–areallofthepurevariety.Itmightbethoughtthatphenomenalismaffordsanexampleofreductivementalism,butmostphenomenaliststurnouttobeeitherpurementalists(becausetheyholdthatnothingintheuniverseexemplifiesallthetraitsrequisiteformateriality)orneutralists(becausetheyreducemattertopropertiesof“neutral”sensedatainthemannertobedescribednext).Finally,ifneithermaterialitynormentalityqualifiesasanattribute,wegetthefamilyoftheoriesBroadclassifiedasneutralist.Somewhatextravagantly,neutralismadmitsofninesubdivisions.Broadsingledouttwoformsofneutralismasespeciallyworthyofattention.First,thereistheviewofSamuelAlexanderinSpace,Time,andDeitythatmindandmatterbothemergefrompurelyspatiotemporalattributes.Second,thereistheviewofRussellinTheAnalysisofMindthatmaterialityisnotstrictlyinstantiatedatall(eventhoughitsvariousrequisitesareseparatelyinstantiated)andthatmental-ityiseitherreducibletooremergentfrompropertiesofsensedatathatarethemselvesneithermentalnorphysical.Broadwentontoarguethatmanyoftheseventeentypesoftheorycanbequitedefinitelyruledout.Purematerialismandthethreevarietiesofneutralismthatsaymentalityisnotinstantiatedcanbeeliminatedimmediately,heclaimed,formentalityatleastseemstobeinstantiated,andifthereareseemings,thereareeventsthatinstan-tiatementality.Hebelievedthatbothvarietiesofreductivematerialismcouldalsobe65\nJAMESVANCLEVEruledout.Againstmolarbehaviorism,hepointedoutthatmanyofthementalstatesweobservewithinourselvesarenotidenticalwiththeirassociatedpatternsofbehavior,andhealsoraisedthedoubtwhethereverymentalstateevenhasapatternofbehaviorcoextensivewithit.Againstmolecularbehaviorism,heraisedtheobjectionthatneuralprocesseshaveproperties(e.g.takingplaceswiftlyorslowly)thatdonotapplytohavingasensationofred.ItremainstosaysomethingaboutthepositionBroadjudgedmostreasonableonthewhole,namely,emergentmaterialism.Becauseitimpliesthatmentalpropertiesarenotreducibletophysicalproperties,thisisaformofwhatissometimescalledpropertydualism,eventhoughnotaformofdualisminBroad’sownsense(whichrequiresthatmentalitybeanattribute).Theideaisthatmentalpropertiesbegintobedisplayedwhenmatterreachesacertainlevelofcomplexity.Theyaredependentonmatterfortheirinstantiationandarewhollydetermined,causallyornomicallyspeaking,bymaterialconfigurations.However,thementalpropertiesofanorganism“couldnot,evenintheory,bededucedfromthemostcompleteknowledgeofthebehaviorofitscompo-nents,takenseparatelyorinothercombinations,andoftheirproportionsandarrange-mentsinthiswhole”(1925:59).Inthisrespect,Broadbelievedmentalpropertiestobelikethepropertiesofchemicalcompoundsandunlikethepropertiesofclocks.Someonewhohadneverseenaclockbeforecouldpredictitsbehaviorfromthelawsofphysicstogetherwithknowledgeoftheclock’spartsandhowtheyareputtogether.Bycon-trast,someonewhohadneverexaminedcommonsaltbeforecouldnotpredictitsprop-ertiesfromthelawsofphysicstogetherwithcompleteknowledgeofthepropertiesofsodiumandchlorine(takenseparatelyandinothercombinations)andhowtheyareputtogetherinthenewcompound.Thereare,ofcourse,psychophysicallawsrelatingmentalpropertiestothephysicalpropertiesoftheirbearers.ButBroadbelievedtheselawstobeultimate“trans-ordinal”laws:lawsnotdeduciblefromlawsalreadyknowntoholdatthelowerlevel,butdis-coverableinsteadonlyafterwehavebecomeacquaintedwithobjectsandpropertiesatthehigherlevel.Heconcededthatthepropertiesofchemicalcompounds,whichheusedasexamplesofemergentproperties,mightturnoutwiththegrowthofourphysicalknowledgeormathematicalcompetencenottobeemergentafterall.Buthethoughtthatthetraditionalsecondaryqualities(whetherconceivedofnaivelyasintrinsicpropertiesofexternalthingsorinmoresophisticatedfashionasappearancestoperceivers)wereinherentlyemergentandthatthelawsconnectingtheirinstantia-tionwithpropertiesofmicrophysicswouldnecessarilybeofthetrans-ordinaltype.Notevenamathematicalarchangelwithmicroscopicalpowersofperception,Broadventuredtoassert,wouldbeabletopredictthatammoniasmellsacridorthattheskylooksblue.BibliographyWorksbyBroad1914:Perception,Physics,andReality,Cambridge:CambridgeUniversityPress.1923:ScientificThought,London:KeganPaul,Trench,Trubner&Co.1925:TheMindandItsPlaceinNature,London:KeganPaul,Trench,Trubner&Co.1930:FiveTypesofEthicalTheory,London:KeganPaul,Trench,Trubner&Co.66\nC.D.BROAD1933:ExaminationofMcTaggart’sPhilosophy,vol.I,Cambridge:CambridgeUniversityPress.1938:ExaminationofMcTaggart’sPhilosophy,vol.II,Cambridge:CambridgeUniversityPress.1952a:EthicsandtheHistoryofPhilosophy,London:RoutledgeandKeganPaul.1952b:Religion,Philosophy,andPsychicalResearch,London:RoutledgeandKeganPaul.1975:Leibniz:AnIntroduction,ed.C.Lewy,Cambridge:CambridgeUniversityPress.1978:Kant:AnIntroduction,ed.C.Lewy,Cambridge:CambridgeUniversityPress.WorksbyotherauthorsMcLaughlin,Brian(1992)“TheRiseandFallofBritishEmergentism,”inEmergenceorReduction?,ed.A.Beckerman,H.Flohr,andJ.Kim,Berlin:DeGruyter.Schilpp,P.A.(ed.)(1959)“ThePhilosophyofC.D.Broad,”vol.XinTheLibraryofLivingPhilosophers,NewYork:Tudor.(ThisvolumecontainsBroad’sautobiography,criticalessaysonBroad’sphilosophybytwenty-oneauthors,Broad’sresponsestothem,andacompletebibliographyofhiswritings.)67\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd20015LudwigWittgenstein(1889–1951)P.M.S.HACKERBackgroundLudwigJosefJohannWittgensteindominatesthehistoryoftwentieth-centuryanalyticphilosophysomewhatasPicassodominatesthehistoryoftwentieth-centuryart.Hedidnotsomuchcreatea“school,”butratherchangedthephilosophicallandscape–notonce,buttwice.Andhissuccessors,withinthebroadstreamofanalyticphiloso-phy,whethertheyfollowedthepathshepioneeredornot,hadtoreorientthemselvesbyreferencetonewlandmarksconsequentuponhiswork.Hecompletedtwodiamet-ricallyopposedphilosophicalmasterpieces,theTractatusLogico-Philosophicus(1921)andthePhilosophicalInvestigations(1953).Eachgaverisetodistinctphasesinthehistoryoftheanalyticmovement.TheTractatuswasasourceofCambridgeanalysisoftheinterwaryears,andthemainsourceofthelogicalpositivismoftheViennaCircle.TheInvestigationswasaprimaryinspirationfortheformofanalyticphilosophythatflourishedinthequarterofacenturyaftertheendoftheSecondWorldWar,withitscenteratOxfordanditscircumferenceeverywhereintheEnglish-speakingworldandbeyond.HetaughtatCambridgefrom1930untilhisprematureretirementin1947.Manyofhispupilsbecameleadingfiguresinthenextgenerationofphilosophers,trans-1mittinghisideastotheirstudents.Wittgenstein’scentralpreoccupationsatthebeginningofhisphilosophicalcareerwerewiththenatureofthoughtandlinguisticrepresentation,oflogicandnecessity,andofphilosophyitself.Thesethemescontinueinhislaterphilosophy,from1929onwards,althoughphilosophyofmathematicsoccupiedhimintensivelyuntil1944andphilosophyofpsychologyincreasinglydominatedhisthoughtfromthelate1930suntilhisdeath.Havingbeentrainedasanengineer,hecametoCambridgein1911,withoutanyformaleducationinphilosophy,toworkwithRussell.Hewaspoorlyreadinthehistoryofthesubject,andintentionallyremainedsoinlateryears,preferringnottobeinfluencedbyothers.HehadreadSchopenhauerinhisyouth,andtracesofTheWorldasWillandRepresentationaredetectableintheTractatusdiscussionoftheselfandthewill.Heacknowledgedtheearlyinfluenceuponhimofthephilosopher-scientistsBoltzmann(inparticular,apparently,ofhisPopuläreSchriften)andHertz(especiallyhisintroductiontoThePrinciplesofMechanics).Apartfromthesefigures,themainstimulitohisthoughtswerethewritingsofFregeandRussellonlogicandthe68\nLUDWIGWITTGENSTEINfoundationsofmathematics.Inlateryears,asheputit,he“manufacturedhisownoxygen.”HecertainlyreadsomeKantwhenhewasprisonerofwarinCassino,someoftheworksofAugustine,Nietzsche,Kierkegaard,andPlato,butdidnotcitetheseas2influencesuponhim.TheonlylaterinfluencesheacknowledgedwereOswaldSpengler,anddiscussionswithhisfriendsFrankRamseyandPieroSraffa.Hisstyleofthoughtandwritingwereidiosyncratic.Hewasabletodigdowntothemostfundamental,andtypicallyunnoticed,presuppositionsofthoughtinagivendomain.Wherephilosophershadpresentedopposingviewsofatopic,anddebatehadlongcontinuedpolarizedbetweenalternatives,forexamplebetweenidealismandrealisminepistemology,ordualismandbehaviorisminphilosophyofmind,orPlatonismandintuitionisminphilosophyofmathematics,Wittgensteindidnotsidewithoneoranotherofthereceivedoptions,butstrovetofindtheagreedpresupposi-tionscommontobothsidesofthevenerabledispute,andthenchallengedthese.Hisinsightsweretypicallywrittendowninhighlycondensedform:oftenasinglesentence,abriefparagraph,orafragmentofanimaginarydialogue.Writingstandardcon-secutiveprosedistortedhisthoughts,and,forthewholeofhislife,hiswritingsweresequencesofremarks,enteredintonotebooks,fromwhichhelaterextractedandorderedthebest.This,togetherwithhisgreatgiftofsimplicityofstyle,richinmetaphor,simile,andilluminatingexample,giveshisphilosophicalwritingpowerandfascination,aswellasformidableinterpretativedifficulty.Inonesense,hehadthemindofanaphorist,forwhatisvisibleonthepageisoftennomorethanthetrajectoryofathought,whichthereaderisrequiredtofollowthrough.Nootherphilosopherinthehistoryofthesubjectsharedhiscastofmindorstyleofthinking.Theclosestinspiritarethephilosophically-mindedaphoristsGeorgChristophLichtenberg(whomhemuchadmired)andJosephJoubert(withwhosewritings,itseems,hewasnotacquainted).Duringhislifetimehepublishedonlyonebook,theTractatus,andonearticle“SomeRemarksonLogicalForm,”writtenfortheMindandAristotelianSocietymeetingin1929.Bythetimeofhisdeath,hehadmoreorlesscompletedtheInvestigations(Part1),andwishedittobepublished.Asfortherest,helefttohisliteraryexecutorsthedecisiononwhatpartsofhisliteraryremainsofmorethantwentythousandpagesof3notesandtypescriptsshouldbepublished.AftertheposthumouspublicationofthePhilosophicalInvestigationsin1953,hisliteraryexecutorseditednumerousvolumesofhisunfinishedtypescriptsandnotesfromallphasesofhisphilosophicalcareer.Notebooks1914–1916consistsofprepara-torynotesfortheTractatus.PhilosophicalRemarkswaswrittenin1929,andrepresentsthestageatwhichthephilosophyoftheTractatuswasstartingtocrumble.PhilosophicalGrammarisaneditorialcompilationfromtypescriptswrittenintheyears1931–4,andsignalsthetransformationofWittgenstein’sthought,abandoningthephilosophyoftheTractatusandarticulatinghisnewmethodsandideas.Halfofitconcernsproblemsinthephilosophyofmathematics,asubjectwhichwasatthecenterofhisinterestsfrom1929until1944.TheBlueandBrownBooksconsistsofdictationstohispupils,givenin1933–5.Itelaborateshisnewphilosophicalmethodsandhistransformedconceptionofphilosophy,andexaminesproblemsinthephilosophyoflanguage,epistemology,metaphysics,andphilosophyofpsychology.TheRemarksontheFoundationsofMathematicsisaselectionfromtypescriptsandmanuscriptswrittenbetween1937and1944.ZettelisacollectionofcuttingsWittgensteinhimselfmade69\nP.M.S.HACKERfromtypescriptswrittenbetween1929and1947,althoughmostoftheremarksdatefromtheperiod1944–7.Thethemesaremainlytopicsinthephilosophyoflanguageandphilosophyofmind.ThefourvolumesofRemarksonthePhilosophyofPsychologyandLastWritingsonthePhilosophyofPsychologyarenoteswrittenbetween1947and1951.OnCertaintyandRemarksonColourwerewrittenattheveryendofhislife,theformerbeinguniqueamonghisworksinitsexclusivefocusonepistemologicalthemes.Apartfromotherminorwritings,forexampleonFrazer’sGoldenBoughoraphorismsandgeneralculturalobservationsjotteddownamidsthisphilosophicalreflectionsandgatheredtogetherinCultureandValue,fivevolumesoflecturenotestakenbyhisstudentshavebeenpublished.ThecompleteNachlassiscurrentlybeingpublishedinelectronicform.Wittgensteinisuniqueinthehistoryofphilosophyastheprogenitoroftwoprofoundlyopposedcomprehensivephilosophies.Tobesure,therearecontinuitiesofthemebetweenthetwo:thenatureoflinguisticrepresentation,oflogicandlawsofthought,oftherelationbetweenthoughtanditslinguisticexpression,oftheinten-tionalityofthoughtandlanguage,ofmetaphysicsandofphilosophyitselfaretopicsexaminedindetailintheTractatusandthenre-examinedinthelaterphilosophy.Therearealsocontinuitiesofphilosophicaljudgment.ManyofthenegativeclaimsintheTractatusarereaffirmedinthelaterworks,inparticularhiscriticismsofFregeandRussell,hisdenialthatphilosophycanbeacognitivediscipline,hisrejectionofpsy-chologisminlogicandoflogicisminthephilosophyofmathematics.AndmanyofthefundamentalinsightsthatinformedtheTractatus,forexamplethatthereisaninternalrelationbetweenapropositionandthefactthatmakesittrue,thatthepropositionsoflogicaresenselessbutinternallyrelatedtoinferencerules,thatthelogicalconnectivesandquantifiersarenotfunctionnames,thatordinarylanguageisingoodlogicalorder,areretainedinthelaterphilosophy.Nevertheless,theinsightsthatarethusretainedundergotransformation,arerelocatedinthewebofourconceptualscheme,aredif-ferentlyelucidated,andquitedifferentconsequencesarederivedfromthem.Ingeneral,thetwophilosophiesrepresentfundamentallydifferentphilosophicalmethodsandwaysofviewingthings.TheTractatusisinspiredanddrivenbyasingleunifyingvision.Itwasintendedtobetheculminationandclosureofthegreatessentialistmetaphysi-caltraditionofwesternphilosophy.Aninsightintotheessentialnatureoftheelemen-tarypropositionwasheldtoyieldacomprehensiveaccountofthenatureoflogicandofthemetaphysicalformoftheworld,thenatureandlimitsofthoughtandlanguage.Anineffablemetaphysicsofsymbolismwasweddedtoanequallyineffablesolipsisticmetaphysicsofexperienceandtoanatomist,realist,ontology.TheTractatusThetwomajorthinkerswhoseworkbothinspiredWittgensteinandconstitutedthemaintargetofhiscriticismswereFregeandRussell.Theyhadrevolutionizedlogic,displacingthesubject/predicatelogicoftraditionalsyllogisticbythefunctiontheoreticlogicbasedonthegeneralizationofthemathematicaltheoryoffunctions.Fregehadinventedthelogicofgenerality,thepredicatecalculus(seeFREGE).Bothphiloso-phersrepudiatedpsychologisminlogicandidealisminmetaphysicsandepistemology,propoundinginsteadformsofrealism.Bothhadtriedtodemonstratethereducibility70\nLUDWIGWITTGENSTEINofarithmetictopurelogic,FregeinTheBasicLawsofArithmetic(1893,1903)andRussell,togetherwithWhitehead,inPrincipiaMathematica(1910).Itwas,aboveall,theirconceptionoflogicthatsettheagendafortheyoungWittgenstein.FregeandRussellthoughtthatlogicwasasciencewithasubjectmatter.Thepropo-sitionsoflogic,theyheld,arecharacterizedbytheirabsolutegenerality.OnFrege’sviewtheyareperfectlygeneralpropositionsconcerningsempiternalrelationsbetweenthoughts(propositions),articulatinglawsoftruthvalidforallthinking.AccordingtoRussell,logicisthescienceoftheperfectlygeneral.Itspropositionsaredescriptionsofthemostgeneralfactsintheuniverse.Henceneitherwouldhaveconsideredasimpletautologysuchas“Eitheritisrainingoritisnotraining”asapropositionoflogic,butwouldhaveconceivedofitasaninstantiationofthelogicalproposition“(p)(pv~p).”Bothtendedtoviewrulesofinference(“lawsofthinking”)asrelatedtotheproposi-tionsoflogic(“lawsoftruth”)somewhatastechnicalnormsspecifyingameanstoanendarerelatedtolawsorregularitiesofnature.ThelawsoftruthaccordingtoFregedescribetheimmutablerelationsbetweenthoughts(propositions)irrespectiveoftheirsubjectmatter;accordingtoRussell,theyarethemostgenerallawsgoverningthefactsofwhichtheuniverseconsists.Accordingly,rulesofinferencearetechnicalnorms,dependentonsuchgenerallaws,ensuringthatifonewishestothinkcorrectly,i.e.inferonlytruthsfromtruths,onewilldoso.Thelogicalsystemsthetwophilosophershadinventedwereaxiomatized,andtheyviewedtheaxiomsasself-evidenttruths.Fregeconceivedofthoughtsandofthetwotruth-valuesaslogicalobjects,andofthenotionsofobject,concept,first-andsecond-levelfunctionasultimatesummagenera,drawingontologicaldistinctions“foundeddeepinthenatureofthings.”Thelogicalconnectiveshethoughttobenamesoflogicalentities,unaryorbinaryfirst-levelfunctionsmappingtruth-valuesontotruth-values,andthequantifierstobenamesofsecond-levelfunc-tions.Russellheldthattermssuchas“particular,”“universal,”“relation,”“dualcomplex,”arenamesoflogicalobjectsor“logicalconstants”signifyingthepureformswhicharethesummageneraoflogic,theresiduefromaprocessofgeneralizationwhichhasbeencarriedouttoitsutmostlimits.Weunderstandsuchexpressions,hethought,onthebasisof“logicalexperience”orintuition.Bothphilosophersheldnaturallan-guagetobelogicallyimperfect,containingvagueandambiguousexpressionsornameswithoutreference,andhence,Fregethought,allowingtheformationofsentenceswithoutatruth-value.Theyviewedtheirownnotationsaslogicallyperfectlanguages.Fromthepost-Wittgensteinianperspective,FregeandRussellwereradicallymistakenaboutthenatureoflogicaltruths(conceivingofthemasessentiallygeneral),aboutthenatureoflogicalnecessity,aboutthecontentoflogicaltruths,aboutthestatusoftheaxiomsoflogic,aboutthecharacterofthelogicalconnectivesandquantifiers,andabouttherelationbetweenthetruthsoflogicandrulesofinference.Ifweareanycleareronthesemattersthanthey,itislargelyduetoWittgenstein.IntheTractatus,WittgensteinacceptedsomeofthesalientdoctrinesofRussellandFrege.Likethem,headopteda(different)variantofmetaphysicalrealismintheTractatusontologyofsimplesempiternalobjects,ofcomplexes,andoffacts.Heacceptedunreflectivelytheassumptionthatthefundamentalroleofwordsistonameentities(althoughthisrolewasdeniedtologicaloperatorsandtocategorialexpressions)andofsentencestodescribehowthingsareinreality.Hethoughtthattheremustbeaconnectionofmeaningbetweenwordsandtheentitiestheyname,thatlanguage71\nP.M.S.HACKERacquirescontentbymeansofsuchaconnectionwithreality.Heagreedwiththeirantipsychologisminlogic.HeacceptedFrege’sdemandofdeterminacyofsense,althoughunlikeFrege,hethoughtthatthevaguenessofnaturallanguagewasmerelysuperficialandanalyzableintodisjunctionsofdeterminatepossibilities.And,likeFrege,Russellandmanyothers,heassumedthatthelogicalconnectivesandquantifiersaretopic-neutral.Someofthesecommitmentshewaslatertoabandon,othershereinterpreted.UnlikeFregeandRussell,Wittgensteinheldthatordinarylanguageisingoodlogicalorder.Forlogicisaconditionofsense,andinsofarassentencesofordinarylanguageexpressasense,conveythoughts,theyareingoodorder–anyappearancetothecon-trary(e.g.vagueness)beingafeatureofthesurfacegrammarofexpressions,whichwilldisappearonanalysis.Insofarastheyfailtoexpressasense,theyareill-formedpseudo-sentences.Henceitisnotthetaskofphilosophytodevisealogicallyideallan-guage,althoughdevisingalogicallyperspicuousnotationwillenablethephilosophertolaybarethetruelogicalformsofthoughts,whichareobscuredbythesurfacegrammarofordinarylanguage.AccordingtotheTractatusthefundamentalfunctionoflanguageistocommunicatethoughtsbygivingthemexpressioninperceptibleform.Theroleofpropositions(sentenceswithasense)istodescribestatesofaffairs,whichmayormaynotobtain.Ifthestateofaffairsdepictedbyapropositionobtains,thenthepropositionistrue,otherwiseitisfalse.Propositionsarecomposedofexpressions.Logicalexpressionsapart,theconstituentexpressionsinapropositionareeitherana-lyzable,definablebyanalyticdefinitionorparaphrase,orunanalyzable.Unanalyzableexpressionsaresimplenames,whicharerepresentativesofsimpleobjects.Thesimpleobjectsarethemeaningsofthenames.Hencenameslinklanguagetoreality,pinningthenetworkoflanguageontotheworld.Nameshaveameaningonlywhenusedasrepresentatives,andtheyaresousedonlyinthecontextofaproposition.Theelemen-tary(logicallyindependent)propositionisaconcatenationofnamesinaccordancewithlogicalsyntax.Itdoesnotnameanything,paceFrege(whothoughtsentencesnametruth-values)andRussell(whothoughttheynamecomplexes),butdepictsa(pos-sible)stateofaffairs,whichisisomorphictoitgiventherulesofprojection,andassertsitsexistence.Thenamesinanelementarypropositionmustpossessthesamecombi-natorialpossibilitiesinlogicalsyntaxasthemetaphysicalcombinatorialpossibilitiesoftheobjectsinrealitythataretheconstituentsofthestateofaffairsrepresented.Thelogicalsyntaxthatunderliesanypossiblemeansofrepresentationmirrorsthelogico-metaphysicalformsofreality.PaceFregeandRussell,theassertionsignhasnologicalsignificance.UnlikeFrege,whothoughtthattherewerealternativeanalysesofpropo-sitions,andunlikeCarnap,who,inthe1930s,thoughtthatwecanchoosebetweendifferentlogics,Wittgensteinthoughtthatanalysisisuniqueandthatinlogictherearenooptions.ThemetaphysicsoftheTractatuswasrealist(asopposedtonominalist),pluralist(asopposedtomonist),andatomist.Thesempiternalobjectsthatconstitutethesubstanceofallpossibleworldsincludepropertiesandrelationsofcategoriallydistincttypes.ItisfarfromclearwhatkindsofthingsWittgensteinhadinmind,buttheyarearguablysuchitemsasminimallydiscriminableshadesofcolor,tones,etc.aswellasspatio-temporalpointsinthevisualfield.Objectsaresimple(thisismirroredbythelogicalsim-plicity,i.e.unanalyzability,oftheirnames).Theyhaveinternalandexternalproperties.72\nLUDWIGWITTGENSTEINTheirinternalpropertiesconstitutetheir(essential)form:theircombinatorialpossibil-itieswithotherobjects(thisismirroredbythelogico-syntacticalcombinatorialpos-sibilitiesoftheirnames).Differentobjectsbelongingtothesameontologicalcategory(e.g.differentshadesofcolor)haveacommonform(namely,color).Theexternalprop-ertiesofobjectsareaccidental:theircontingentconcatenationswithotherobjectstoformactualstatesofaffairs.Astateofaffairsisapossiblecombinationofobjects(e.g.thatsuchandsuchaspatiotemporalpointisacertainshadeofsuchandsuchacolor).Theobtainingornon-obtainingofastateofaffairsisafact(hencetherearepositiveandnegativefacts).Elementarystatesofaffairsare“atomic”or“independent,”thatis,eachsuchstateofaffairsmayobtainornotobtainwhileallotherelementarystatesofaffairsthatobtainremainthesame.Thisisreflectedbythelogicalindependenceoftheelementaryproposition,whichhasnoentailments.ThemetaphysicsofexperienceintheTractatuswasapparentlyaformofempiricalrealismandtranscendentalsolipsism(cf.Kant’sempiricalrealismandtranscendentalidealism).Theempiricalselfthatisstudiedbypsychologyisnotanobjectencounteredinexperience,buta(Humean)col-lectionofexperiences.Themetaphysicalself,whichistheconcernofphilosophy,isalimitofexperience.Itisthewillingself,thebearerofgoodandevil.Sentencesareexpressionsofthoughts.Butthoughtitselfisakindoflanguage,composedofthought-constituents.Theformofathoughtmustmirrortheformofrealitynolessthanaproposition.Naturallanguageisnecessaryforthecommunica-tionofthoughtsbutnot,itseems,forthinking–whichcanbeeffectedinthe“languageofthought.”Itismentalprocessesofthinkingandmeaningthatinjectcontentintothebarelogico-syntacticalformsoflanguage.Whatpinsanameontoanobjectinrealitythatisitsmeaning(Bedeutung)isanactofmeaning(meinen)bythenameofthatobject.Whatrendersalicitconcatenationofsignsalivingexpressionofathoughtistheemploymentofthemethodofprojection,whichisthinkingthesenseofthesentence,i.e.meaningbythesentencesuchandsuchastateofaffairs.Hencetheintentionalityofsignsisderivedfromthe(intrinsic)intentionalityofthinkingandmeaning(meinen).TheTractatusaccountoftheintentionalityofthoughtandlanguageisinformedbytheinsightthatthoughtandpropositionalikeareinternallyrelatedtothefactthatmakesthemtrue.Thethoughtorpropositionthatpwouldnotbethethoughtorpropo-sitionitiswereitnotmadetruebythefactthatpandmadefalsebythefactthatnotp.Whatonethinkswhenonethinkstrulythatpispreciselywhatisthecase,andnotsomethingelse(suchasaFregeanGedanke),whichstandsinsomerelationtowhatisactuallythecase.Butwhatonethinkswhenonethinksfalselythatpisnotwhatisthecase(sincewhatonethinksdoesnotobtain).Yetonedoesnotthinknothing.Indeed,whatonethinksisthesame,nomatterwhetheronethinkstrulyorfalsely.Thepicturetheoryofthoughtandpropositionprovidedalogico-metaphysicalexplanationofhowitispossibletosatisfythedemandsconsequentupontheseinternalrelations.Itattemptstoexplainhowitispossibleforathoughttodeterminewhatstateofaffairsinrealitywillmakeittrue,howitispossibleforthecontentofathoughttobepreciselywhatisthecaseifitistrueandyettohaveacontentevenifitisfalse,howitispossiblethatonecanreadofffromathought,inadvanceofthefacts,whatwillmakeittrue,andhowitispossibleforthe“meresigns”oflanguagetobeintentional,i.e.foranametoreachuptotheveryobjectitselfofwhichthenameisthenameandforthe73\nP.M.S.HACKERsentencetodescribetheverystateofaffairstheexistenceofwhichwillmaketruethepropositionexpressed.Everyrepresentationisapictureofapossibility.Apropositionorthoughtisalogicalpicture,whosesimpleconstituentsnamesempiternalobjectswithdeterminateform.Thereisametaphysicalharmonybetweenlanguageandthoughtontheonehandandrealityontheother;forwhenonethinkstrulythatp,whatisthecaseisthatp;andwhenonethinksfalselythatp,whatonethinksispreciselywhatisnotthecase.This“pre-establishedharmony”isorchestratedbyametaphysicsofsymbolism.Onlysimplenamescanrepresentsimpleobjects.Simplenameshaveameaningbutnosense.Relationstooareobjects,andonlyrelationscanrepresentrelations;henceinthepropo-sition“aRb,”itisnot“R”thatrepresentstherelationthatastandsintob,butratheritisthat“R”standstotherightof“a”andtotheleftof“b”(inthisnotation).Onlyfactscanrepresentfacts,andsentences–intheirsymbolizingcapacity–arefacts,whichareusedtodescribehowthingsare.Foritisthefactthattheconstituentnamesarearrangedastheyare(inaccordancewithlogicalsyntax)thatsaysthatthingsarethusandso.Sentenceshaveasensebutnomeaning.Thepossiblestatesofaffairsinrealityaredeterminedbythelanguage-independentcombinatorialpossibilitiesofobjects.Everyelementarypropositiondepictsapossiblestateofaffairs.Itistrueifthepossibilitydepictedobtains,falseifitdoesnot.Itisoftheessenceofthepropositionwithasensetobebipolar,i.e.tobecapableofbeingtrueand4capableofbeingfalse.Thismirrorsthemetaphysicaltruththatitisofthenatureofstatesofaffairsthattheyeitherobtainorfailtoobtain.Thesenseofapropositionisitsagreementanddisagreementwiththeexistenceandnon-existenceofstatesofaffairs.Forthepropositionthatpagreeswiththefactthatpanddisagreeswiththefactthatnot-p.Whatonethinkswhenonethinksthatpisapossibility,apossibilitywhichisactualizedifone’sthoughtistrueandisnotifone’sthoughtisfalse.Henceonecanreadoffapropositionorthought(whichisakindofproposition)whatmustbethecaseforittobetrue,andwhatonethinkswhenonethinksthatpispreciselywhatisthecaseifone’sthoughtistrueandwhatisnotthecaseifone’sthoughtisfalse,andistheverysamethoughtnomatterwhetheritistrueorfalse.Thelogicalconnectivesarenotnamesoffunctions,butrathersignifytruth-functionaloperationsonpropositions.Thequantifiersareconstruedasoperatorsuponapropositionalfunction(e.g.“fx”)whichisalogicalprototypecollectingallproposi-tionsofacertainform(whosevaluesareallthosepropositionsobtainedbysubstitut-inganameforthevariable),hencegeneratinglogicalsumsorproductsofsuchsetsofpropositions.Allpossiblemolecularpropositionscanbegeneratedbytruth-functionaloperationsuponelementarypropositions.Hencealllogicalrelationsaredeterminedbytruth-functionalcombinationsofpropositions.Amolecularpropositionpentailsanotherpropositionqifandonlyifthesenseofqiscontainedinthesenseofp,i.e.ifthetruth-groundsofpcontainthetruth-groundsofq.Thevariousoperatorsareinter-definable,andreducibletothesingleoperationofjointnegation,namely“not...andnot...”Amongthetruth-functionalcombinatorialpossibilitiesofagivennumberofelementarypropositions,therewillalwaysbetwolimitingcases(1)inwhichthepropo-sitionsaresoconjoinedastobetrueirrespectiveofthetruth-valuesoftheconstituentpropositionsand(2)falseirrespectiveoftheirtruth-values.Theformerisatautologyandthelatteracontradiction.Thesearethepropositionsoflogic.Sincetheyare,respec-74\nLUDWIGWITTGENSTEINtively,trueandfalseirrespectiveofhowthingsare,theyarewhollywithoutanycontent,andsaynothingabouthowthingsareinreality.Sobycontrastwithothermo-lecularpropositionswhicharetrueundercertainconditions(i.e.forcertainassign-mentsoftruth-valuestotheirconstituents)andfalseunderothers,thepropositionsoflogicareunconditionallytrueorfalse.Hencetheyaresaidtobesenseless,tohave,asitwere,zero-sense.Alltautologiessaythesamething,namelynothing.Butdiffer-enttautologiesmayneverthelessdiffer,foreverytautologyisaformofaproof(sinceeverytautologycanberewrittenintheformofamodusponens),anddifferenttautologiesrevealdifferentformsofproof.Itisamarkofthepropositionsoflogic,Wittgensteinheld,thatinasuitablenotationtheycanberecognizedassuchfromthesymbolalone.Heinventedaspecialnotationtodisplaythis,hisT/Fnotation.Insteadofwritingmolecularpropositionsbymeansofsymbolsforlogicalconnectives,heusedtruth-tablesaspropositionalsigns.Hereitisimmediatelyperspicuousfromthesignalonewhetherapropositionisatautology,andifso,itisvisiblyevidentthatitcannotbefalse.Itisequallyevidentwhetheronepropositionfollowsfromanother,i.e.whetherthetruth-groundsofonecontainthoseofanother.Thisshowed,hethought,thenatureofthepropositionsoflogicandtheircategorialdifferencefromempiricalpropositions.ThisconceptionoflogicaltruthmadeclearhowmisleadingwastheFrege/Russellaxiomatizationoflogic,withitsappealtoself-evidencefortheaxioms.Theiraxiomswerenotprivilegedbytheirself-evidence.Theyweretautologiesnolessthantheirtheorems.Theywerenot“essentiallyprimitive,”norwereFrege’sandRussell’stheo-remsessentiallyderivedpropositions,for“allthepropositionsoflogicareofequalstatus,”namelytautologiesthatsaynothing.Hencetoo,contrarytoFregeandRussell,thepropositionsoflogichavenosense,anddescribenothing.Inanimportantsense,thepropositionsoflogichavenosubjectmatter,andlogicismisconstruedasthescienceofthemostgenerallawsoftruthorofthemostgeneralfactsintheuniverse.Consequently,thepropositionsoflogicdonotconstitutethefoundationsfortheelaborationoftechnicalnormsofthinkingonthemodeloftherelationbetweenlawsofnatureandtechnicalnormsforachievingdesiredends.Rather,everytautologyisinternally(notinstrumentally)relatedtoaruleofinferenceorformofproof.TheconceptionoflogicintheTractatuswasstillflawed.Butitsflaws,whichWittgensteinwaslatertoexpose,didnotsignificantlyaffectthecriticismsoftheFregeanandRussellianconceptionsoflogic.AccordingtotheTractatustheonly(effable)necessityislogicalnecessity.Everywell-formedpropositionwithasensemustbebipolar.Whatphilosophershadhithertoconceivedofascategorial(orformal)concepts,suchasobject,property,relation,fact,proposition,color,number,etc.are,Wittgensteinargued,expressionsforforms,whicharerepresentedbyvariables,ratherthanbynames.Hencetheycannotoccurinafullyanalyzedpropositionwithasense.Onecannotsaythat,forexample,oneisanumber,thatredisacolor,orthatAisanobject,forsuchpseudo-propositionsemployaformalconceptasifitwereagenuineconcept,andtheyarenotbipolar.Hencesuchmetaphysicalpronouncements(whichattempttodescribenon-logicalnecessities)arenonsense–ill-formedconjunctionsofsigns.Butwhatsuchpseudo-propositionstrytosayisactuallyshownbygenuinepropositionswhichcontainnumberwords,colornames,orothernamesofobjects.Itisshownbyfeaturesoftheexpressionsinsuchpropositions,namelyby75\nP.M.S.HACKERtheformsoftheexpressions–theiressentialcombinatorialpossibilities.Thesearerepresentedbythevariableofwhichthemeaningfulnamesaresubstitution-instances.AnimmediateconsequenceofthisisthatmostofthepropositionsoftheTractatuswhichdelineatethenecessaryformsoflanguageandrealityarenonsense.HenceWittgenstein’spenultimateremarkinthebook:“Mypropositionsserveaselucidationsinthefollowingway:anyonewhounderstandsmeeventuallyrecog-nizesthemasnonsensical,whenhehasusedthem–assteps–toclimbupbeyondthem.”Hencetoo,theconceptionofphilosophyadvocatedforthefutureisnotthepracticeexhibitedinthebook.TheTractatusconsistslargelyofsentencesthatareneitherbipolarpropositionsnortautologies.Theyattempttodescribetheessenceoftheworld,oflan-guage,andoflogic,andoftheessentialrelationsbetweenthem.Butthisisanattempttosaytheverythingsthatcannotbesaidinlanguage,butarerathershownbylan-guage.Whatisthusshownisindeedineffable.Hencemetaphysics,theattempttodis-closetheessentialnaturesofthings,isimpossible.Oncethecorrectlogicalpointofviewhasbeenachieved,oncetheworldisseenaright,thetaskoftheTractatusiscompleted.Thetaskoffuturephilosophyisanalysis:clarificationofphilosophicallyproblematicpropositionswhichwillelucidatetheirlogicalformsorclarifywhyandwhere(inthecaseofputativemetaphysicalpropositions)theyfailtoaccordwiththerulesoflogicalgrammar.Futurephilosophywillnotbeatheory,norwillitpropounddoctrinesorattainknowledge.Itwillbeanactivityoflogicalclarification.Philosophy,thuscon-ceived,isacritiqueoflanguage.TheroleoftheTractatusinthehistoryofanalyticphilosophyInsixrespectstheTractatusintroducedthe“linguisticturn”inphilosophy.First,itsetthelimitsofthoughtbysettingthelimitsoflanguage:byelucidatingtheboundariesbetweensenseandnonsense.Thisputlanguage,itsformsandstructures,atthecenterofphilosophicalinvestigation.Second,thepositivetaskforfuturephilosophywasthelogico-linguisticanalysisofsentences.Thelogicalclarificationofthoughtsistoproceedbytheclarificationofpropositions–sentenceswithasense.Third,thenegativetaskoffuturephilosophywastodemonstratetheillegitimacyofmetaphysicalassertionsbyclarifyingthewaysinwhichattemptstosaywhatisshownbylanguagetransgressestheboundsofsense.Fourth,theTractatusattemptedtoclarifytheessentialnatureofthepropositionalsignbyelucidatingthegeneralpropositionalform,thatis,bygiving“adescriptionofthepropositionsofanysign-languagewhatsoeverinsuchawaythateverypossiblesensecanbeexpressedbyasymbolsatisfyingthedescription,andeverysymbolsatisfyingthedescriptioncanexpressasense,providedthatthemeaningsofnamesaresuitablychosen.”Fifth,thelogicalinvestigationofphenomena,theunfoldingoftheirlogicalforms,whichwasnotundertakeninthebook,istobeeffectedbylogicalanaly-sisofthelinguisticdescriptionsofthephenomena.(Thefirstmovesincarryingoutthistaskweretakeninthe1929paper“SomeRemarksonLogicalForm,”whereuponthewholeprojectcollapsed.)Forthelogicalsyntaxoflanguageisandmustbeisomorphicwiththelogico-metaphysicalformsoftheworld.Sixth,thegreatestachievementofthebook,asseenbytheViennaCircle,wasitselucidationofthenature76\nLUDWIGWITTGENSTEINoflogicalnecessity.Thiswaspatentlymadebyaninvestigationofsymbolism.Thatonecanrecognizethetruthofalogicalpropositionfromthesymbolalonewasheldtocontaininitselfthewholephilosophyoflogic.Manyoftheseclaimswerelatertoberepudiated.Buttheyheraldedthelinguisticturn,whichwasexecutedbytheViennaCircle,and,inadifferentway,byWittgensteinhimselfinhislaterphilosophy,andbyOxfordanalyticphilosophy.TheTractatuswasaparadigmofanalyticphilosophyinitsheroicorclassicphaseintheinterwaryears.ItwasthemajorinspirationofCambridgeanalysisandoflogicalpositivism.Itsprogram,asunderstoodbothinCambridgeandinVienna,committedonetothemethodoflogico-linguisticanalysisofcomplexexpressionsintotheirsimpleunanalyzablecon-stituents.Itencouragedtheprogramofreductiveanalysisanditsmirrorimage,logicalconstruction.Itcleavedtothethesisofextensionality,holdingallnon-extensionalcon-textstobeeithereliminable,merelyapparent,orillicit.Itrepudiatedtheintelligibilityofputativelysyntheticaprioripropositions,insistedthattheonlynecessityislogicalnecessityanddeniedanysensetothepropositionsoflogic.HenceitseemedtoprovidethefoundationsforwhattheViennaCirclehailedtriumphantlyas“consistentempiri-cism,”foritdeniedthatpurereasonalonecanattainanyknowledgeoftheworld.Itheldmetaphysicstobenonsense(theCircleavertedtheirgazefrom,orquicklycon-demnedandpassedover(Neurath),orattemptedtocircumvent(Carnap),itspara-doxicalineffabilityclaims).Anditallocatedtophilosophyasuigenerisanalyticroleandastatuswhollydistinctfromthatofscience.Schlick,theleadingfigureintheCircle,wentsofarastocharacterizetheTractatusas“theturningpointinphilosophy,”thedeepestinsightintowhatthetaskandstatusofphilosophyshouldbe.Wittgenstein’sinfluenceupontheViennaCirclewassecondtonone.Indeed,theprincipleofverificationitselfwasderivedfromconversationswithWittgensteinin1929/30,andreadbackintotheTractatus.MembersoftheCirclespenttwoacademicyearsreadingthroughthebooklinebyline,abandoningsomeofitsclaimsandaccept-ingothers.Theyabandonedthepicturetheoryoftheproposition,thedoctrineofshowingandsaying,andmostoftheontologyoflogicalatomism.Butwhattheyacceptedwascrucial:theaccountofthenatureandlimitsofphilosophy,theconcep-tionoflogicandlogicalnecessity,andtheprogramofthelogicalanalysisoflanguage(seeAYER,CARNAP,HEMPEL,andQUINE).Theseideas,interpretedandsometimesseriouslymisinterpreted,werepivotaltotheirwork.Themostimportantmisinterpretationcon-cernedtheTractatusaccountoflogic.MembersoftheCircleagreedwiththecriticismsoftheFregeanandRussellianmisconceptionsofthenatureoflogic,andwelcomedtheviewthatthepropositionsoflogicarevacuous(senseless).Buttheygaveaconven-tionalistinterpretationtoWittgenstein’saccountoflogicwhichwasfarremovedfromhisconception.Theythoughtofthelogicalconnectivesasarbitrarysymbolsintroducedtoformmolecularpropositions,whereasWittgensteinhadarguedthattheyareessen-tiallygivenbythemereideaofanelementaryproposition.Whereheviewedthetruthsoflogicasflowingfromtheessentialbipolarityoftheproposition,theyconceivedofthemasfollowingfromthetruth-tabulardefinitionsofthelogicalconnectives–henceastrueinvirtueofthemeaningsofthelogicaloperators.Alogicaltruththereforewasheldtobethelogicalconsequenceofconventions(definitions).Wittgenstein,bycon-trast,hadarguedthatthesenselesstruthsoflogicreflectthelogicalstructureoftheworld.Logic,farfrombeingdeterminedbyconvention,istranscendental.Inthe77\nP.M.S.HACKER1930s,whenheturnedtoreconsiderhisearlierconception,WittgensteinnotonlyreformulatedhisviewsbutalsovehementlycriticizedtheconventionalismoftheCircle.Farfromfollowingfromthemeaningsofthelogicalconnectives,thetruthofthepropo-sitionsoflogic,heargued,isconstitutiveoftheirmeanings.ThecollapseoftheTractatusvisionAlreadyintheTractatusWittgensteinhadtakennoteofthefactthatdeterminatesofadeterminable,e.g.redandgreen,aremutuallyexclusive:ifAisredallover,itfollowsthatitisnotgreen(orblueoryellow,etc.)allover.Atthetime,hethoughtthatthisshowedthat“Aisred”isnotanelementaryproposition,andthatitsentailmentswould,onanalysis,beclarifiedasfollowingfromitstruth-functionalcompositionoutofele-mentarypropositions.Whenhereturnedtophilosophyafterahiatusofadecade,herealizedthatthiswasmisconceived.Thereareirreduciblelogicalrelationsofexclusionorimplicationwhicharedeterminednotbytruth-functionalcomposition,butbytheinnerstructureofelementarypropositions.Hetriedtobudgetforthisbyabandoningthetopicneutralityofthelogicalconnectivesanddrawinguptruth-tablesspecifictothe“propositionalsystem”(i.e.thesystemofdeterminatesofadeterminable)towhichagivenelementarypropositionbelongs.Inthecaseofcolor,theconjunctionof“Aisredallover”and“Aisgreenallover”isnonsense.Hencethetruth-valueassignment“TT”mustbeexcludedfromsuchconjunctionsbyaspecialruleofsyntax.Butthiscon-cession,herapidlyrealized,spellsthedeath-knellforthephilosophyoflogicalatomism,andstrikesattheheartoftheTractatus.Fortheindependenceoftheelementarypropo-sitionwasthepivotuponwhichturnedthewholeconceptionoflogicandtheinef-fablemetaphysicsofthebook.Withoutit,theideathatthelogicofpropositionsdependsonlyuponthebipolarityoftheelementarypropositioncollapses.ThesignificanceoftheT/Fnotationasrevealingtheessentialnatureoflogicalpropositionsandrelationsevaporates,preciselybecausetherearelogicalrelationsthatdependupontheinnerstructureofelementarypropositions.Sincethelogicaloperatorsarenottopicneutral,separatetruth-tableswouldhavetobedrawnupforeachpropositionalsystem.Theideathatthereisageneralpropositionalform,accordingtowhicheverypropositionisaresultofsuccessiveapplicationstoelementarypropositionsoftheoperationofjointnegationmustlikewiseberelinquished.Sotoomustthethoughtthatgeneralitycanbeanalyzedintologicalsumsandproducts,andthatthequantifierscanbegivenauniformtopic-neutralanalysis.AsthelogicaltheoryoftheTractatuscollapsed,sotoodidthemetaphysics.Itwaswrongtosaythattheworldconsistsoffactsratherthanofthings.Rather,adescrip-tionoftheworldconsistsofstatementsoffacts,notofanenumerationofthings.Butthestatementofafactjustisatruestatement.Onecannotpointat,butonlypointout,afact.Andtopointoutafactjustistopointoutthatthingsarethusandso,thatis,tomakeatrueassertion.Factsarenotconcatenationsofobjects.Unlikeconcatenationsofobjects,andunlikestatesofaffairs,factshavenospatiotemporallocation.Thefactthatacircleisredisnotcomposedofrednessandcircularityconcatenatedtogether,sincefactsarenotcomposedofanythinganddonothave“constituents.”Theproposi-tionthatpisonly“madetrue”bythefactthatpinthesenseinwhichbeingabache-lormakesoneunmarried.Allitmeansisthatthepropositionthatpistrueif,infact,78\nLUDWIGWITTGENSTEINthingsareasitsaystheyare.Theconceptionofabsolutelysimplesempiternalobjectswasincoherent.Forthenotionsofsimplicityandcomplexityarerelative,notabsolute.Tocallspatiotemporalpoints,properties,orrelations“objects”isamisuseoflanguage.Whathadappearedtobeobjectsthathadtoexistareinfactsampleswhichweemployinexplainingthemeaningsofcertainostensivelydefinedexpressionsinthelanguage.Assuch,theybelongtothemeansofrepresentation,not(likethepostulated“objects”oftheTractatus)towhatisrepresented.Asthemetaphysicscollapsed,sotoodidthepicturetheory,theconceptionofiso-morphismbetweenlanguageandreality,andtheaccountofintentionality.Whathadseemedlikeaninternalrelationbetweenthepropositionthatpandthefactthatpwhichmakesittruewasnomorethantheshadowcastuponrealitybyanintra-grammaticalrelationbetweentheexpressions“thepropositionthatp”and“thepropositionmadetruebythefactthatp.”Thereisaninternalrelationhere,butitisforgedinlanguage–inthegrammaticalrulethatpermitstheinter-substitutionoftheseexpressions–notbetweenlanguageandreality.Henceitwasmistakentothinkthatrealitymusthaveacertainmetaphysicalformwhichmustbereflectedinthelogico-syntacticalformsoflan-guageinorderforthisinternalrelationtoobtain.Theintentionalityofthoughtandproposition,whichhadseemedtodemandapre-establishedmetaphysicalharmonybetweenlanguageandreality,isfullyexplainedbyreferencetointra-grammaticalcon-nectionsbetweenexpressions.Thethoughtorexpectationthatitwillbethecasethatpdoesnot“anticipatereality”;rather,onlywhatsatisfiesthedescription“itisthecasethatp”willbecalled“thefulfillmentoftheexpectationthatitwillbethecasethatp.”Ofcourseonecan“readoff”fromthethoughtwhatwillmakeittrue,sincetheexpressionofthethoughtcontainsthedescriptionofthestateofaffairstheobtainingofwhichiscalled“theconfirmationofthethought.”Ofcoursewhatonethinks,whenonethinksthatp,iswhatisthecasewhenone’sthoughtistrue,butthisisnotastrangeformofidentityorcoincidencebetweenashadowypossibilityandanactuality.Rathertheques-tion“Whatisbeingthought?”and“Whatisthecase?”herereceivethesameanswer.ThemetaphysicsofsymbolismoftheTractatuswasinfactamythologyofsymbol-ism.Themeaningofanameisnotanobjectofanykind.WhatislegitimateabouttherolewhichtheTractatussimpleobjectwasinvokedtofulfillisinfactplayedbydefiningsamplesusedinostensivedefinitions,e.g.ofcolorwords.Butthesamplepointedatintheostensivedefinition“Thisisblack”ispartofthemeansofrepresentation,tobeusedasanobjectofcomparisonandstandardofcorrectapplicationoftheword“black.”Namesderivetheirmeaningsnotfromobjectsintheworldwhichtheyrepresent,butfromexplanationsofmeaning,ofwhichostensivedefinitionsarebutonetype.Butitisatbestvacuoustoclaimthatallnonlogicaltermsarenames.Thereareindefinitelymanygrammaticallydifferentkindsofexpressions,whichfulfilldifferentrolesinalan-guageandhavedifferentuses,givenbytheexplanationsoftheirmeanings,whichareineffectrulesfortheiruse.InthesenseinwhichtheTractatusclaimedthatthereisaconnection–ameaning-endowingconnection–betweenlanguageandreality,thereisnosuchconnection.Itwasmistakentosupposethatapropositionalsignisafact,thatonlyfactscanrepresentfacts,orthatonly“simplenames”canrepresentsimpleobjects.Farfromthelogicalsyntaxoflanguagehavingtomirrorthelogicalformsofthings,thedifferentgrammarsofdifferentlanguagesareautonomous.Theyowenohomagetoreality.Theydonotreflectlanguage-independentmetaphysicalpossibilities,79\nP.M.S.HACKERdeterminedbytheessentialnatureofobjectsrepresented,butratherthemselvesdeter-minelogicalpossibilities,i.e.whatitmakessensetosay.Empiricalpropositionsareindeedcharacteristically(althoughnotuniformly)bipolar,buttheconceptofapropo-sitionisafamilyresemblanceconcept:therearemanydifferentkindsofproposition,whicharenotcharacterizedbyanessentialnature,butbyoverlappingsimilarities.TheconceptoflogicalformwhichhadinformedtheTractatusischimerical.Forparaphraseintoacanonicalnotation(asinRussell’stheoryofdescriptions)isnotananalysisofwhatisalreadypresentintheparaphrasedpropositionorthoughtbutaredescriptioninadifferentformofrepresentation.Logicalformisnoreflectionofthelogico-metaphysicalformsofreality,sincethereisnosuchthing.AlreadyintheTractatusWittgensteinhadrejectedthelogicisminthephilosophyofmathematicswhichFregeandRussellhadendeavoredunsuccessfullytoprove.Hedeniedthatnumberswerelogicalobjectsorreducibletoclasses.Mathematicalpropositions,heclaimed,arenotdescriptionsofpossiblestatesofaffairs.Noraretheybipolar.Theyare,ineffect,nonsensicalpseudo-propositions;theydonothaveasenseconsistingintheiragreementanddisagreementwiththeexistenceandnonexistenceofstatesofaffairs.Rather,theyaresubstitution-rulesforthetransformationofoneempiricalpropositionconcerningmagnitudesorquantitiesorspatialrelations,etc.intoanother,andexpressionsofrulesarenotpropositions.Inthe1930shewroteextensivelyaboutthefoundationsofmathematics.Itisnotpossibleheretodomorethanindicatebrieflythegeneraltrajectoryofhisthought.Hedidnotrejectlogicisminordertoembracewhatseemedtobetheonlyalternatives,namelyintuitionismandformalism.Hisfundamentalclaimisradical.Withtheliberalizationinhisconceptofaproposition,hewasnowwillingtospeakofmathematicalpropositions.Nevertheless,theyareradicallyunlikeempiricalpropositions,andequallyunlikelogicalones.Mathematicsisasystemofinterlockedpropositions.AsalreadyimpliedintheTractatus,thefundamentalroleofthissystem(butnotofeverypropositionwithinit)istoconstituterulesforthetransformationofempiricalpropositions.Anarithmeticequa-2tion,suchas25=625,isarulelicensingthetransformationofsuchanempiricalpropositionas“Thereare25boxeseachcontaining25marbles”intotheproposition“Thereare625marbles.”Atheoremofgeometryisanormofrepresentation:arulepermittingthetransformationofempiricalpropositionsaboutshapes,distances,orspatialrelations.Differentgeometriesarenotdifferenttheoriesaboutempiricalspace,whichmightturnouttobetrueorfalse.Noraretheydifferentuninterpretedcalculi.Rather,theyaredifferentgrammarsforthedescriptionofspatialrelations.Proofbymathematics(e.g.inengineering)iswhollydifferentfromproofinmathematics.Whileamathematicalpropositionisarule,unlessitisanaxiom,itisnotstipulated,butproducedaccordingtorulesbyaproof.Herewemustdistinguishproofswithinaproofsystem,e.g.acomputation,whichisjust“homework,”asWittgensteinputit,fromproofswhichextendmathematicsbyextendingaproofsystem.Proofsthatextendmathematicscreatenewinternalrelations,modifyingexistingconceptsbylinkingthemwithconceptswithwhichtheywerehithertounconnected,orconnectingthemwithconceptsinnewways–thuslicensingnoveltransformationsofappro-priateempirical(orothermathematical)propositions.Mathematicsisconceptformation.Thepropositionsofmathematicsdeterminetheconceptstheyinvoke.Whatweconceiveofasmathematicalnecessityisatbestadistortedreflectionoftheinter-80\nLUDWIGWITTGENSTEINnalrelationswithinaproofsystem.Mathematicsisahumancreation,inventedratherthandiscovered.ThePhilosophicalInvestigationsDismantlingtheTractatuspreoccupiedWittgensteinintheearly1930s.Graduallyanewmethodandawhollydifferentconceptionoflanguage,oflinguisticmeaning,andoftherelationbetweenlanguageandrealityemerged.ItbecameclearthathisneglectofquestionsinthephilosophyofpsychologyintheTractatus,whichhehadtakentobelicensedbytheanti-psychologismhetookoverfromFrege,wasunwarranted.Fortheconceptsoflinguisticmeaningareboundupwiththeconceptsofunderstanding,thinking,intending,andmeaningsomething,andthesepivotalnotionsdemandphilo-sophicalelucidation.Thenewmethodalsoledtoanewconceptionofphilosophyitself–relatedto,butstillimportantlydifferentfrom,theconceptionofphilosophyadvocatedintheTractatus.Thatinturnledtoadifferentcriticismofmetaphysics.SuccessiveeffortstocomposeabooklayingforthhisnewideasculminatedinthecompositionofthePhilosophicalInvestigations,Part1,whichwasvirtuallycompletedby1945/6.Itishismasterwork.Despitesomecontinuitiesofthemeandnegativeconceptions,itstandsinstarkcontrastnotonlytothesibyllinestyleoftheTractatusbutabovealltoitsspirit.WheretheTractatusstroveforasublimeinsightintothelanguage-independentessencesofthings,theInvestigationsproceededbyaquietweigh-ingoflinguisticfactsinordertodisentangleknotsinourunderstanding.TheTractatuswaspossessedbyavisionofthecrystallinepurityofthelogicalformsofthought,lan-guage,andtheworld,theInvestigationswasimbuedwithasharpenedawarenessofthemotleyoflanguage,thedeceptiveformsofwhichleadusintoconfusion.TheTractatusadvocatedconceptualgeology,hopingtodisclosetheineffableessencesofthingsbydepthanalysisoflanguage,theInvestigationspracticedconceptualtopography,aimingtodissolvephilosophicalproblemsbyapatientdescriptionoffamiliarlinguisticfacts.TheTractatuswastheculminationofatraditioninwesternphilosophy.TheInvestigationsisvirtuallywithoutprecedentinthehistoryofthought.Wittgenstein’slaterwork,ashehimselfsaid,isnotmerelyastageinthecontinuousdevelopmentofphilosophy,butconstitutesa“kink”inthedevelopmentofthoughtcomparabletothatwhichoccurredwhenGalileoinventeddynamics;itwas,inasense,anewsubject,anheirtowhatusedtobecalled“philosophy.”Anewmethodhadbeendiscovered,andforthefirsttimeitwouldnowbepossiblefortheretobe“skillful”philosophers–whowouldapplythemethod.ThetransitionfromtheTractatustohislaterphilosophy,ashewrotewhenhisnewideasweredawningin1929,isthetransi-tionfromthemethodoftruthtothemethodofmeaning.ItisatransitionfromWesensschau–putativeinsightsintothenatureoressenceofthings–totheclarifica-tionofconceptualconnectionsinthegrammarofourlanguages,withthepurposeofdisentanglingknotsinourthought.TheconceptionofphilosophyadvocatedintheInvestigationshasnoprecedent,althoughitis,inaqualifiedsense,anticipatedbytheTractatusprogramforfuturephilosophy.Thephilosophyoflanguageisequallywithoutancestors:itisneitheraformofidealisttelementationallinguistictheory(onthemodelofclassicalempiricismordeSaussure)noraformofbehavioristlinguistictheory,itisneitherarealisttruth-conditionalsemanticsnoraformof“anti-realist”semantics.The81\nP.M.S.HACKERphilosophyofmindrepudiatesbothdualismaswellasmentalismontheonehandandlogicalbehaviorismaswellasphysicalismontheother.ThecritiqueofmetaphysicsrestsneitheronHumeanorverificationistgrounds,nordoesitresembletheKantiancritiqueoftranscendentmetaphysics.ItisnowonderthatWittgenstein’slaterphilosophyhasbeensofrequentlymisunderstoodandmisinterpreted,foritcannomorebelocatedonreceivedmapsofphilosophicalpossibilitiesthantheNorthStarcanbelocatedonaterrestrialglobe.TheInvestigationsopenswithaquotationfromSt.Augustine’sautobiographyinwhichherecountsthemannerinwhichheassumesthathehadlearnttospeak.Theseunselfconscious,nonphilosophicalreflectionsseemedtoWittgensteintocrystallizeanimportantproto-pictureoflanguage,apre-philosophicalconceptionofitsroleandfunction,whichinformsamultitudeofphilosophicaltheories.Accordingtothispicturetheessentialroleofwordsistonamethings,andtheessentialroleofsentencesistodescribehowthingsare.Wordsareconnectedtothingsbymeansofostension.Thisproto-picture,whichisakintoanunnoticedfieldofforceunconsciouslymouldingtheshapeofsophisticatedphilosophicaltheories,isonerootofextensivemisconceptionsinphilosophyoflanguage,logic,mathematics,andpsychology.Itisamutedleitmotifrunningthroughthebook,andcombatingtheinfluenceofthispictureisoneofthecentraltasksofthebook.Forwearepronetothinkthatcorrespondingtoeveryname,orcorrespondingtoeverynameonanalysis,theremustexistsomething:thatnounsnameobjects,adjectivesnameproperties,verbsnameactions,thatpsychologicalexpressionssuchas“pain”namepsychologicalobjects,and“believe,”“want,”“intend,”“think,”etc.namepsychologicalstatesorprocesses,numberwordsnamenumbers,andlogicalconnectivesnamebinaryrelations.Weareinclinedtobelievethateverydeclarativesentencedescribessomething:thatlogicalpropositionsdescriberela-tionsbetweenthoughts,thatmathematicalpropositionsdescriberelationsbetweennumbers,thatwhatweconceiveofasmetaphysicalpropositionsdescribenecessaryrelationsbetweenultimatecategoriesofbeing,thatpsychologicalpropositionsinthefirst-persondescribestatesofmind,andsoon.Butthisisillusion.PhilosophyoflanguageThephilosophyoflanguageoftheInvestigationshasadestructiveandaconstructiveaspect.ItsdestructiveaspectisconcernedwithunderminingtheconceptionofanalysisthathadinformedtheTractatusand,moreremotely,hascharacterizedphil-osophyatleastsincetheCartesianandempiricistprogramsofanalysisintosimplenaturesandintosimpleideasrespectively.Itaimstodestroytheconceptionofalanguageasacalculusofmeaningrulesandtheideathatthemeaningorsenseofasentenceiscomposedofthemeaningsofitsconstituentwordsandderivablefromthem,giventheirmodeofcombination.Hencetoo,itcombatstheidealofdeterminacyofsense,andthethoughtthatallexpressionsareeitherdefinablebyanalyticdefinitionorareindefinablesandhenceexplainedbyanostensivedefinition,conceivedofaslinkinglanguagewithrealityandlayingthefoundationsoflanguageinsimpleobjectsgiveninexperience.Ithasalreadybeennotedthattheconceptsofsimpleandcomplexarerelative.HencewhetheranAiscomplexorsimplehastobedeterminedbyreferencetocriteriaof82\nLUDWIGWITTGENSTEINsimplicityandcomplexitylaiddownforAs–iftherearesuchcriteria.Butwecom-monlyconfusetheabsenceofanycriteriaofcomplexity(sincenonehavebeenlaiddown)withthesatisfactionofcriteriaofsimplicity.Wearepronetothinkthatanexpressioniscomplexifitisdefinedbyanalyticdefinition,andsimpleifitisexplainedbyostension.Butanalyticandostensivedefinitionsareneitherexclusivenorexhaus-tive.Wecanexplainwhat“circle”meansbysayingthatthisisacircle,orbysayingthatacircleisalocusofpointsequidistantfromagivenpoint.Andwecanexplainwhatwordsmeanbycontextualparaphrase,contrastiveparaphrase,exemplification,byaseriesofexamplestogetherwithasimilarityrider,bygesture,andsoon.Themeaningofawordisnotanobjectforwhichawordstandsorofwhichitisthename.Rather,itiswhatisgivenbyanexplanationofmeaning,andanexplanationofmeaningisarulefortheuseoftheexplanandum–astandardofcorrectnessforitsapplication.Toaskforthemeaningofawordistoaskhowitistobeused.Indeed,themeaningofawordis(or,morecautiously,isdeterminedby)itsuse.Ostensivedefinitionisonelegitimatemannerofexplainingthemeaningsofsomewords.Itisnotespeciallyprivileged:asargued,itdoesnot“connectlanguagewithreality”orlaythefoundationsoflanguage;itisonlyonerulefortheuseofthewordinquestion,anditisascapableofbeingmisunderstoodasanyotherexplanationofmeaning.Manyexpressionsdonothavenecessaryandsufficientconditionsofappli-cation.Amongthesearefamily-resemblanceconcepts,suchas“game,”whichareexplainedbyaseriesofexamplesandasimilarityrider.(Evenifsomeonecomesupwithasharpdefinitionof“game,”thatdefinitionisnottherulebyreferencetowhichwehavebeenapplyingtheword“game”andbyreferencetowhichwewouldhavejusti-fiedouruseoftheword.)Indeed,manyofthepivotalconceptsinphilosophy,suchas“language,”“proposition,”“number,”“rule,”“proof,”aswellasmanypsychologicalconcepts,arefamily-resemblanceconcepts.Theirextensionisnotdeterminedbycommonproperties,butbyoverlappingsimilarities–likethefibersinarope.Sincenumerouskindsofexpressionarenotexplainedintermsofnecessaryandsuf-ficientconditionsofapplication,theideathatvaguenessisonlyasurfacegrammaticalfeatureoflanguageorthatitmustbeanimperfectioninlanguageisawry.TheFregeandemandfordeterminacyofsensewasincoherent.Fordeterminacyofsenseisnotmerelytheabsenceofvagueness,buttheexclusionoftheverypossibilityofvagueness:theexclusion,byacompleteexplanationofmeaning,ofeverypossibilityofdoubtineveryconceivablecircumstance.Butthereisnosuchthing.Thereisnoabsolutecon-ceptionofcompleteness.Theconceptsofcompleteandincompletearebothrelativeandcorrelative.Acompleteexplanationofmeaningisanexplanationwhichmaybeinvokedasastandardofapplicationinallnormalcontexts.Relativetothatstandard,explanationsmaybejudgedtobecompleteorincomplete.Butwehavenosingleidealofexactness;whatcountsasexactorvaguevariesfromcontexttocontext.Moreover,vaguenessisnotalwaysadefect(“Iaskhimforabreadknife,”Wittgensteinmocked,“andhegivesmearazorbladebecauseitissharper”),anditsoccurrenceisnotlogi-cally“contagious.”Theideathatthesenseofasentenceisafunctionofthemeaningsofitsconstituentsandtheirmodeofcompositionisadistortedstatementoftheplatitudethatifonedoesnotknowwhatthewordsofasentencemeanordoesnotunderstandthewayinwhichtheyarecombined,thenonewillnotunderstandwhatissaid.Thesupposition83\nP.M.S.HACKERthatwhatasentencemeansfollowsfromanexplanationofwhatitswordsmean,togetherwithaspecificationofitsstructure,errswithregardtobothmeaningandunderstanding.Themeaningofasentenceisnomorecomposedofthemeaningsofitspartsthanafactiscomposedofobjects.Thedistinctionsbetweensenseandnon-sensearenotdrawnonceandforallbyreferencetocircumstance-invariantfeaturesoftype-sentences,butbyreferencetocircumstance-dependentfeaturesoftheuseoftoken-sentences.Sentencesofpreciselythesameformmayhaveverydifferentuses.Indeed,theformsofsentences,nomatterwhetherinnaturallanguageortranslatedintoacanonicalnotationofthepredicatecalculus,concealratherthanrevealtheiruse.Moreover,understandingasentenceisnotaprocessofderivingitsmeaningfromanything.Littleremainsofanalysisaspreviouslyunderstood.Philosophicalproblemsaremisunderstandingscaused,amongotherthings,bymisleadinganalogiesbetweenformsofexpressionswithdifferentuses.Someofthesecanbedissolvedbyparaphrase,asexemplifiedbyRussell’sTheoryofDescriptions(seeRUSSELL).Butitwasanillusionthatthereisanythinglikeafinalanalysisoftheformsofourlanguage,letalonethatanalysisrevealsthelogicalstructureoftheworld.Insteadofanalysisasclassicallyconceived,whatisneededisadescriptionoftheusesofwordsthatwillilluminatephilosophicalconfusion,andarearrangementoffamiliarrulesfortheuseofwordswhichwillmakethegrammaroftherelevantexpressionssurveyable.Forthemainsourceofphilosophicalpuzzlementandofmisguidedphilosophicaltheoriesisourfailuretocommandaclearviewoftheuseofwordsandourconsequententanglementinthenetworkofgrammar.Connectiveanalysis(thetermisStrawsonianratherthanWittgensteinian),thatis,adescriptionoftheconceptualconnectionsandexclusionsinthewebofwords,andtherapeuticanalysis(seebelow)replacereductiveanalysis.Asentenceiscompletelyanalyzed,inthenewsense,whenitsgrammarislaidoutcompletelyclearly.Alanguageismisrepresentedifitisconceivedtobeacalculusofrules.Moreilluminatingistheideathatitisamotleyoflanguagegames.Languageisindeedrule-governed,intheloosemannerinwhichgamesare.Usingsentencesiscompara-bletomakingmovesinagame,andalanguagecanbefruitfullyviewedasamotleyoflanguagegames.Theuseoflanguageisinterwovenwiththelivesandpracticesofspeakers,andispartlyconstitutiveoftheirformoflife.Trainingandteachingunderpinthemasteryofalanguage,andthesepresupposesharedreactiveandbehavioralpropensitieswithinalinguisticcommunity.Wordsareliketools,andthediversityoftheiruseisasgreatasthatofdifferenttools:hencemaskedbyconceivingofthemasessentiallynamesofthings,andconcealedbytheirgrammaticalform.Thegreatesterrorofphilosophersofhisday,Wittgensteinremarked,istoattendtotheformsofexpressionsratherthantotheiruses.Evendeclarativesentencesareusedforendlesslydiversepurposes,ofwhichdescribingisonlyone,andnon-declarativesentencesaremisrepresentediftakentobeanalyzableintoaforce-indicativecom-ponent(e.g.anassertionsignorinterrogativesign)andadescriptive,truth-valuebearing“sentence-radical.”Moreover,theconceptofdescriptionisitselfnon-uniform,fordescribingasceneisaltogetherunlikedescribingadream,describingtheimpres-sionsomethingmadeisunlikedescribingtheitemthatmadetheimpression,anddescribingwhatoneintendsisaltogetherunlikedescribingtheexecutionofone’s84\nLUDWIGWITTGENSTEINintention.Thesearelogicallydistinctkindsofdescriptions,withdistinctkindsofgroundsandconsequences.Understandingisakintoanability,notastatefromwhichperformanceflows.Thecriteriaforlinguisticunderstandingareofthreegeneralkinds:correctuse,givingacorrectexplanationofmeaningincontext,andrespondingappropriatelytotheuseofanexpression.Viewingexplanationsofmeaningasrulesfortheuseofwords,theuseofwordsasrule-following,andunderstandingasthemasteryofthetechniqueoftheuseofwordsrequiresthattheseconceptsbetightlyinterlocked.Andsotheyare.Thereisaninternalrelationbetweenaruleandwhatcountsascompliancewithit,whichismanifestnotonlyintheinterpretationsonemightgiveoftherule,butaboveallinthepracticeofactinginaccordancewithit,andinthecriticalpracticesofteachingthemeaningsofexpressions,ofcorrectingmisapplicationsandmistakenexplanationsofmeaning.Meaningisdeterminedbyuse,itisgivenbyanexplanationofmeaning,anditiswhatisunderstoodwhenthemeaningofanexpressionisunderstood.Noteverydifferenceinuseisadifferenceinmeaning,buteverydifferenceinmeaningisadiffer-enceinuse.Wittgenstein’slaterphilosophyoflanguageisguidedbythisseriesofcon-ceptualconnections,theramificationsofwhichheexploredindetail.PhilosophyofmindAgainstprevailingtradition,Wittgensteinchallengedtheinner/outerpictureofthemind,theconceptionofthementalasa“world”accessibletoitssubjectbyintrospec-tion,theconceptionofintrospectionasinnerperception,theideathatthecapacitytosayhowthingsarewithus“inwardly”isaformofknowledge(letaloneaparadigmofself-knowledge),thethoughtthathumanbehavioris“barebodilymovement,”thenotionthatvoluntaryactionisbodilymovementcausedbyactsofwill,thesuppositionthatexplanationofhumanbehaviorintermsofreasonsandmotivesiscausal,andthepervasiveinfluenceoftheAugustinianpictureoflanguagethatdisposesonetothinkthatpsychologicalexpressionsareuniformlyoreventypicallynamesofinnerobjects,events,processes,orstates.Hisphilosophyofmindandofactioncanbeseenasprovidingarigorousphilosophicalunderpinningforthehermeneuticinsistenceontheautonomyofhumanisticunderstandinganditscategorialdifferentiationfromunderstandinginthenaturalsciences.Psychologicalexpressionsarenotnamesofentitieswhicharedirectlyobservableonlybythesubject,andavowalsoftheinnerarenotdescriptionsofsomethingvisibleonlyinaprivatepeepshow.Itisalltooeasytothinkofpsychologicalexpressionsasnamesofinnerentities,andhenceofassigningthemmeaningbyprivateostensivedefi-nition.Wittgenstein’s“privatelanguagearguments”areaimedatthismisconception.Therecanbenoinner,private,analogueofpublicostensivedefinition.Sensationscannotfulfilltheroleofsamples.Soapaincannotserveasadefiningsamplefortheapplicationoftheword“pain.”Concentratingone’sattentionuponone’spainisnotakindofpointing.Rememberingasensationpresupposesandsocannotexplainthemeaningofasensation-name,andthememoryofasensationcannotserveasanobjectofcomparisonfortheapplicationofasensation-word.Thereisnosuchthingasapplyinganexpressioninaccordancewitharulewhichisinprincipleincommunica-bletoanyoneelse.Buttheideaofdefiningasensationwordbyreferencetoa85\nP.M.S.HACKERsensation,conceivedofasprivateandintendedtofunctionasadefiningsampleinanostensivedefinitionwouldbesuchapseudo-rule–forwhichtherecouldbenocrite-rionofcorrectapplication.Whateverseemedtoonetoberightwouldberight,andthatmeansthatthereisnosuchthinghereasrightorcorrect.Indeed,theverynotionofprivacywhichinformsCartesianandempiricistconcep-tionsofthementalismisconceived.Thementalwastakentobeprivateintwosenses:privatelyownedandepistemicallyprivate.Pains,forexample,wereheldtobeprivatelyowned,i.e.onlyIcanhavemypain,anotherpersoncannothavemypainbutonlyaqualitativelyidenticalone.AndonlyIcanreallyknowthatIhaveapain,otherscanonlysurmisethatIdo.Bothoftheseclaimsaremisconceived.Tohaveapainisnottoownanything,anymorethantohaveabirthdayoratraintocatch.Thedistinctionbetweennumericalandqualitativeidentity,whichappliestosubstances,nomoreappliestopains(ormentalimages,thoughtsorfeelings)thanitdoestocolors.IfAisredandBisred,thenAandBarethesamecolor;sotoo,ifNNhasathrobbingheadacheinhisrighttempleandMMhasathrobbingheadacheinhisrighttemple,thenNNandMMhavethesameheadache–neithernumericallythesame,norqualitativelythesame,butjustthesame.TothinkthatwhatdifferentiatesmypainfromyoursisthatIhavemineandyouhaveyoursistotransformtheownerofthepainintoadistin-guishingpropertyofthepain–whichisasabsurdasclaimingthattwochairscannothavethesamecolor,sincethecolorofthischairbelongstothischairandthecolourofthatchairbelongstothatchair.Theconceptionofepistemicprivacyisequallyawry.Farfromthe“inner”beingafieldofcertainempiricalknowledgepossessedbythesubject,whichisbetterknownthan,andprovidesthefoundationsfor,otherkindsofempiricalknowledge,first-person,present-tensepsychologicalutterancesarenotgenerallyexpressionsofknowl-edgeatall.“IknowIaminpain”iseitheranemphaticorconcessiveassertionthatIaminpain,orphilosophers’nonsense.Insuchcases,ignorance,doubt,mistake,misidentification,misrecognitionareruledoutbygrammar:wehavenouseforsuchformsofwordsas“Imaybeinpain,orImaynot–Iamnotsure,Imustfindout.”Butwemistakethegrammaticalexclusionofignorance,doubt,etc.,forthepresenceofknowledge,certainty,correctidentification,andrecognition.Whereastheytooareexcludedassenselessinsuchcasesaspain,andtheuseoftheepistemicoperatorsinothercaseshasadistinctivemeaning;“Idon’tknowwhatIwant”or“IdonotknowwhatIbelieve”arenotexpressionsofignorancebutofindecision.IdonotneedtolookintomymindtofindoutwhatIwantorbelieve,buttomakeitup.IfIdonotknowwhatIbelieveaboutX,Ineedtoexaminetheevidence,notmystateofmind.Theutterances“Iaminpain,”“I’mgoingtoV,”“IwantG”arestandardlyemployedasexpressionsoravowals(ratherthandescriptions)ofpain,intention,ordesire,andtheutteranceisacriterionforotherstoascribetothespeakertherelevantpsychologicalpredicate.Acriterionfortheinnerislogically(conceptually),asopposedtoinductively,goodevidence(justification)forascribingtoanotheranappropriatepsychologicalpredicate.Painandpainbehavior,ordesireandconativebehavior,arenotanalogically,induc-tively,orhypotheticallyconnected.Rather,cryingoutincircumstancesofinjury,assuaginganinjuredlimb,avoidingthecauseofinjury,etc.arenon-inductivegroundsforpain-ascriptions.Graspingtheconceptofpaininvolvesrecognizingsuchcriteriaas86\nLUDWIGWITTGENSTEINgroundsforascriptionofpaintoanother.Thecriteriaforascriptionofpsychologicalpredicatesarepartlyconstitutiveoftherelevantconcepts.Psychologicalutterancesoravowalsoftheinnerare(incertaincases)learntextensionsofprimitivebehaviorthatmanifeststheinner.Forexample,anavowalofpainisgraftedonto,andisapartialreplacementof,agroanofpain;andwhileanutteranceofpainisasgroundlessasashriekofpain,ittooconstitutesacriterionforthird-personascriptions.Itismisguidedtosuggestthatwecanneverknowwhetheranotherisinpain.Onthecontrary,weoftenknowwithcompletecertainty.Whensomeoneseverelyinjuredscreamswithpain,justtrytodoubtwhetherhereallyisinpain!Self-knowledgeisahardwonachievement,notgainedmerelybyhavingtoothache,wantingorthinkingthisorthat,andbeingabletosayso.Indeed,othersoftenknowandunderstandusbetterthanwedoourselves.Themindisnotasubstance.Itisnotidenticalwiththebrain.Itisnotaprivatespace,inwhichmentalobjectsareparaded,disclosedtointrospectivevision.Thereis,tobesure,suchathingasintrospection,butitisnotinnerperception.Ratheritisaformofreflectiononone’spast,onereasonsandmotives,affectionsandattitudes.Thethird-personpronounrefersneithertothemindnortothebody,buttotheperson,thelivinghumanbeing.Thefirst-personpronounfunctionsquitedifferently;herereferencefailure,misidentification,misrecognition,andindeterminacyofreferencearestan-dardlyexcluded.“I”isatbestadegenerate,limitingcaseofareferringexpression,asatautologyisalimitingcaseofapropositionwithasense.Psychologicalpredicatesareneitherpredicableofthebodynorofitsparts.Itissenselesstoascribetothebrainpredicatesapplicableonlytothewholecreature,e.g.thinking,believing,wanting,orintending.Forthecriteriaforthethird-personascrip-tionofsuchpredicatesaredistinctiveformsofbehaviorofthecreatureinthestreamoflife,andthereisnosuchthingasabrainmanifestingthoughtorthoughtlessness,belieforincredulity,desireoraversion,intentionorinadvertenceinwhatitdoes.Hencetoo,itmakesnosensetoascribethoughtorthoughtlessness,understanding,misunder-standingorfailureofunderstandingtomachines.Thoughtisessentiallyboundupwiththesentient,affective,andconativefunctionsofabeingthathasawelfare,iscapableofdesiringandsuffering,cansetitselfgoalsandpursuethem,andcanhopetosucceedorfeartofailinitsprojects.Humanbehaviorthatconstitutescriteriafortheascriptionofpsychologicalpredi-catesisnot“barebodilymovement,”fromwhichweinferanalogicallyorhypotheticallytheirinnerstateorwhichweinterpretasaction.Onthecontrary,weseethepaininthefaceofthesufferer,hearthejoyinthevoiceofajoyfulperson,perceivetheaffectioninthelooksoflovers.Pain,pacebehaviorists,isnotpain-behavior,anymorethanjoyisthesameasjoyousbehaviororlovethesameasalovinglook.Butthe“inner”isnothiddenbehindthe“outer”;itmaysometimesbeconcealedorsuppressed(oritmayjustnotbemanifested).Butifitismanifested,thenitinfusesthe“outer,”whichisnotbarebodilymovement,buttheactionsandaffectivereactionsoflivingsentientbeingsinthestreamoflife.Thesearenottypicallydescribablesaveintherichvocabularyofthe“inner.”Humanactionisnotmovementcausedbyactsofwill.Therearesuchthingsasactsofwillandgreateffortsofwill,buttheyareunusual,andarenotcausalantecedentsofaction.Thereissuchathingaswillpower,butthatisamatteroftenacityrather87\nP.M.S.HACKERthanapsychicanalogueofmusclepower.Voluntaryactionsarenotactions,letalonemovements,precededbyanactofwill.Wantingandwillingarenotnamesofmentalactsorprocesses,and“HeV’edbecausehewantedto”doesnotgiveacausalexplana-tionofhisaction;onthecontrary,ittypicallyprecludesone.Voluntarymovementisactionforwhichitmakessensetoaskforagentialreasons,whichapersoncandecidetoperform,trytoexecute,orbeorderedtodo.Itismarkedbylackofagentialsurprise,andtheagentcanbeheldresponsibleforit.Areasonforactionorforbeliefisapremiseinreasoning.Henceitisnomorecausallyrelatedtotheactionforwhichitisareasonthanthereasonsforabeliefarecausallyrelatedtotheconclusionwhichtheysupport.Aperson’sreasonisgivenbyspecifyingthereasoninghewentthroughantecedentlytoactingorthereasoninghecouldhavegonethroughandiswillingtogiveexpostactu.Reasons,unlikecauses,justifyorpurporttojustifythatforwhichtheyarereasons.Aperson’savowalofareasonforhisaction,unlikehistypicalassertionofacauseofsomeevent,isnotahypothesis.Unliketheassertionofacause,inthestandardcaseofanavowalofareason,thereisnoroomformistake.Whatmakestheconnectionbetweenthereasonandtheactionistheagent’savowalitself.Inavowingareason,theagenttypicallytakesresponsibilityforhisactionviewedundertheaspectoftheavowedreason.ThecritiqueofmetaphysicsandnatureofphilosophyTheTractatusprogramforfuturephilosophyadvocatedanon-cognitiveconceptionofphilosophy,denyingthattherecouldbeanyphilosophicalpropositions,afortiorianyphilosophicalknowledge.Philosophyshouldbeanactivityofelucidationbyanalysis.Althoughphilosophywasdeprivedofthepossibilityofstatingessentialtruthsaboutthenaturesofthings,theseverytruthswereheldtobeshownbythewell-formedpropo-sitionsofalanguage,andarrivingatacorrectlogicalpointofviewwouldincludeapprehensionandappreciationofwhatcannotbesaidbutshowsitself(includingtruthsofethicsandaesthetics).Thelaterconceptionofphilosophyadheredtotheradicalnon-cognitivism,butrejectedthedoctrineoflinguisticallymanifestineffabilia.Thereareindeednophilo-sophicaltruths.Whatappearassuch,andwhatwereconstruedbytheTractatusasanattempttosaywhatcanonlybeshown,areineffectexpressionsofrulesfortheuseofexpressionsinthemisleadingguiseofmetaphysicaldescriptionsofthenatureofthings.Sotheportentous,apparentlymetaphysical,claimthattheworldconsistsoffactsnotofthingsamountstothegrammaticalstatementthatadescriptionoftheworldconsistsofastatementoffactsandnotalistofthings.Andthatinturnisjustarulefortheuseoftheexpression“adescriptionoftheworld.”Insofarasmetaphysicsisconceivedtobethequestforknowledgeofthenecessaryformsandstructuresoftheworldorofthemind,itischimerical.Allthatcanbegleanedfromthesebarrenfieldsaregrammaticalpropositions,thatis,expressionsofrulesfortheuseofwords.Therearenosuchthingsas“necessaryfacts,”andsentencessuchas“redisacolor,”or“spaceisthreedimensional”areineffectrules.Ifsomethingissaidtobered,thenitcanbesaidtobecolored;ifsomethingisinspace,thenitslocationisgivenbythreecoordinates;andsoon.Similarly,apparentlysyntheticaprioritruths,suchas“Blackisdarkerthanwhite”or“Redismorelikeorangethanitislikeyellow,”arenotinsights88\nLUDWIGWITTGENSTEINintolanguage-independentnecessitiesintheworld,butexpressionsofrulesthatarepartlyconstitutiveofthemeaningsoftheconstituentexpressions.Foranyorderedpairofsampleswhichcanbeusedtodefine“black”and“white”ostensivelycanalsobeusedtodefinetherelation“darkerthan.”Soifaisblackandbiswhite,itfollowswithoutmoreadothataisdarkerthanb.Ifaisred,borange,andcyellow,thenaismoreakintobincolorthantoc–oneneednotlooktosee.Theapparentlymetaphysicalpropositionisinfactaninferencerule,whichispartlyconstitutiveofthemeaningsoftheconstituentterms.Whatappeartobedescriptionsofobjectivenecessitiesintheworldaremerelytheshadowscastbytherulesfortheuseofcolorpredicatesandrelations.Similarly,“cannot”and“must”inputativelymetaphysicalpropositionsmaskrulesfortheuseofwords.“Youcannottravelbackintime”or“Youcannotcountthroughallthecardinalnumbers”looklike“Anironnailcannotscratchglass,”buttheyarenot.Experienceteachesthatironcannotscratchglass.Butitisnotexperiencethatteachesthatonecannottravelintime,rather,itisgrammarthatstipulatesthattheformofwords“Itravelledbacktolastyear”hasnouse;nothingcountsastravellingbackwardsintime.“Cannot”inmetaphysicsisnotabouthumanfrailty,butisanexpressionofaconvention.“Youcannotcountthroughallthecardinalnumbers”isanexpressionofagrammaticalrulewhichexcludesthephrase“countingthroughallthecardinalnumbers”fromthelanguage.Itdoesnotsaythatthereissomethingwecannotdo,butratherthatthereisnosuchthingtodo.Similarly,“must”inmetaphysicssignifiesnotanobjectivenecessityinreality,butacommitmenttoaformofrepresen-tation.“Everyeventhasacause”isatrueorfalseempiricalgeneralization.“Everyeventmusthaveacause”isanexpressionofacommitmentnottocallanything“anevent”unlessithasa(knownorunknown)cause.Therearenotheoriesinphilosophy,fortherecanbenothinghypothetico-deductiveaboutthedeterminationoftheboundsofsense,norcanitbemerelyprobablethatsuchandsuchaphilosophicalpronouncementmakesnosense.Andwedonotneedtowaituponfutureconfirmationtodeterminewithcertaintythatitmakesnosense.Hencetoo,thereisnophilosophicalknowledgecomparabletoknowledgeinthesciences.Ifanyoneweretoadvancethesesinphilosophy,everyonewouldagreewiththem:forexample,“Canonesteptwiceintothesameriver?”–“Yes.”Indeed,therearenoexpla-nationsinphilosophyinthesenseinwhichthereareinthesciences,forthemethodsofphilosophyarepurelydescriptive,andnotmethodsofhypothesisformation.Thepurifiednon-cognitivismoftheInvestigationshastwoaspects.Ontheonehand,philosophyisaquestforasurveyablerepresentationofasegmentofourlanguagewiththepurposeofsolvingordissolvingphilosophicalperplexity.Ontheotherhand,philosophyisacurefordiseasesoftheunderstanding.Philosophicalproblemsareconceptual,henceaprioriandnotempirical.Theycanbeneithersolvednoradvancedbynewinformationorscientificdiscoveries,althoughscientificdiscoveriesmay,andoftendo,raisefreshconceptualpuzzlesandgeneratenewconfusions.Conceptualprob-lemsmayconcernnovelconcept-formationorexistingconceptualstructuresandrelations.Theformerareexemplifiedbymathematics,thelatterbyphilosophy.Thetaskofphilosophyistoresolveconceptualquestionsarisingoutofourexistingformsofrepresentation,toclarifyconceptualconfusionsthatresultfromentanglementinthewebofthegrammarofourlanguage.Philosophyisnotacontributiontohuman89\nP.M.S.HACKERknowledgebuttohumanunderstanding–anunderstandingofourformsofrepre-sentationandtheirarticulations,anoverviewoftheformsofourthought.Themainsourceofphilosophicalpuzzlementandofmisconceivedphilosophicaltheoriesisourfailuretocommandaclearviewoftheusesofwords.Thegrammarofourlanguageislackinginsurveyability,forexpressionswithverydifferentuseshavesimilarsurfacegrammars:“Imeant”looksakinto“Ipointed,”“Ihaveapain”to“Ihaveapin,”“Heisthinking”appearsakinto“Heistalking,”“tohaveamind”lookslike“tohaveabrain,”“2isgreaterthan1”seemsakinto“JackistallerthanJill.”Hencewemisconstruethemeaningsofexpressionsinourphilosophicalreflections.Wethinkofmeaningsomethingorsomeoneasamentalactoractivityofattachingsignstoobjects,takepaintobeakindofobjectinalienablypossessedbythesufferer,imaginethatthemindisidenticalwiththebrain,assumethatstatementsofnumericalinequalitiesaredescriptions,andsoon.Whatisneededisaperspicuousrepresentationofthesegmentofgrammarthatbearsontheproblemwithwhichweareconfronted.Itenablesustoseedifferencesbetweenconceptsthatareobscuredbythemisleadinglysimilargrammaticalformsofexpressions.Forthisnonewdiscoveriesarenecessaryorpossible–onlythedescriptionofgrammar,theclarificationandarrangementoffamiliarrulesfortheuseofwords.Wemustremindourselvesofwhatwealreadyknowperfectlywell,namelyhowexpres-sions,theuseofwhichwehavealreadymastered,areindeedused.Tobesure,theserulesmustthenbearrangedinsuchamannerastoshedlightuponthedifficultyinquestion.Therulesthatconcernthephilosopheraredifferentfromthosethatconcernthegrammarian,andtheorderingofrulesbythephilosopherisverydifferentfromtheorderingsoughtbythegrammarian,fortheirpurposesarequitedistinct.Aperspicu-ousrepresentationofafragmentofgrammarwillenableustofindourwayaroundtherelevantpartofthegrammaticalnetworkwithoutstumblingintoconceptualconfu-sion.Inphilosophy,unlikeinthesciences,alltheinformationisalreadyathand–inourknowledgeofourlanguage.Theproblemsofphilosophy,unlikethoseofscience,arecompletelysolvable.Failuretosolvethemisduetophilosophers’failuretoarrangethegrammaticalfactsinsuchawaythattheproblemsdisappear.Complementarytotheconceptionofphilosophyasthequestforasurveyablerep-resentationofsegmentsofourlanguagethatgiverisetoconceptualperplexityandcon-fusionistheconceptionofphilosophyastherapeutic.Thephilosopher’streatmentofaquestionislikethetreatmentofanillness.Oneshouldnottrytoterminateadiseaseofthought,eitherbydogmatismorbythesubstitutionofatechnicalconceptfortheprob-lematiconethatcausesconfusion(asCarnapdidwithhismethodof“explication”),forslowcureisallimportant.Everydeepphilosophicalconfusionhasmanydifferentroots,andeachmustbedugupandexamined.Everydeeplymisconceivedanswertoaphilo-sophicalproblemthatmesmerizesusandholdsusinavicehasmanyfacets,andeachmustbeseparatelysurveyed.Wittgensteinsometimescomparedhisnewmethodsofphilosophicalclarificationwithpsychoanalysis.Philosophicaltheoriesarelatentnon-sense;thetaskofthephilosopheristotransformthemintopatentnonsense.Likethepsychoanalyst,thephilosopheraimstogivetheafflictedinsightintotheirownunder-standingandmisunderstanding.Philosophyiscategoriallydistinctfromthesciences.Sincethereisnophilosophicalknowledgeandtherearenolicittheoriesinphilosophy,therecanbenoprogressinthe90\nLUDWIGWITTGENSTEINsenseinwhichthereisinthesciences.Forthereisnoaccumulationofknowledge,nogenerationofeverricherexplanatorytheories,norefinementofinstrumentationmakingpossibleevermoreaccuratemeasurementandobservation.Buttherecanbeprogressinanothersense,namelyinclarificationofconceptualstructures,indrawing,refiningandsharpeningdistinctions,indestroyingconceptualillusionsandineradi-catingconceptualconfusions.However,sincethereisnowayofpredictingfutureformsofentanglementintheweboflanguage,thetaskofphilosophyneverends.Wittgenstein’splaceinpostwaranalyticphilosophyIsWittgenstein’slaterphilosophyaformofanalyticphilosophy?Theconceptofana-lyticphilosophyisneithersharplydefinednoruncontested.Ifonetakestheconceptofanalysisnarrowly,connectingitprimarilywithdecompositionalanalysis,withreduc-tionandlogicalconstruction,thenonewillbeinclinedtoassociateanalyticphilosophyprimarilywithavarietyofformsofphilosophythatflourishedinthefirsthalfofthetwentiethcentury.Onewillalsobepronetoassociatethemovementwithaprofoundinterestin,andingeniousphilosophicaluseof,thecalculiofformallogic,and,insomecases,inthedevisingofformalorsemi-formallanguagestoreplacetheapparentlydefectivenaturallanguagesforphilosophicalpurposes.MooreandRussell,theyoungWittgensteinoftheTractatus,theCambridgeanalystsoftheearlyinterwaryears,andthelogicalpositivistswillthenbeone’sparadigmaticanalyticphilosophers.Thusconstrued,itisclearthatitwouldbeatbestmisleadingtocharacterizethelaterWittgensteinasananalyticphilosopheratall.Butitwouldbeperversetoconstrueana-lyticphilosophythus.Theterm“analyticphilosophy”wasalatecomeruponthescene,andtheOxfordphilosophersofthepostwarerahadnoqualmsincharacterizingtheirworkasanalyticphilosophyandtheirmethodsasconceptualanalysis.Thisdidnotimplythattheywerededicatedtoreductiveanalysisandlogicalconstruction.Indeed,theyrepudiatedthem.Whatitimpliedwasaloosersenseof“analysis”:thedescriptionoftheconceptualconnectionsandarticulationsofsalientelementsinourconceptualscheme.Inthissense,tobesure,thelaterWittgensteinwasananalyticphilosopher,andsaidasmuch.For,heclaimed,apropositionis“fullyanalyzed”whenitsgrammarhasbeencompletelylaidbare.Takeninthisbroadersense,analyticphilosophycon-tinuedafter1945inanewanddistinctiveform.ItwasdominatedbyOxfordratherthanCambridgephilosophers,althoughWittgenstein’sphilosophy,transmittedtoOxfordlargelybywordofmouthbefore1953,wasaprimaryinfluenceuponthem(seeANSCOMBE,FOOT,MALCOLM;cf.AUSTIN,RYLE,STRAWSON).Thispostwarphaseofanalyticphilosophylastedforaquarterofacentury.Itwasnota“school”and,unliketheViennaCircle,issuednomanifestos.Itwasunitedbyitsconceptionofphilosophyasanaprioriconceptualinvestigation,contributingtohumanunderstandingratherthantohumanknowledge,hencewhollyunlikethesci-ences.Therewasconsensusthatthemethodicalexaminationoftheuseoftherelevantwordsisasinequanonofanyseriousphilosophicalinvestigation.Analyticphilosophyoflanguageflourished,asdidanalyticepistemology;sotoodidanalyticphilosophyofpsychologyandphilosophyofaction.PathspioneeredbyWittgensteinwerefollowedandrefined.Butotherbranchesofanalyticphilosophy,suchasanalyticjurisprudence,analyticaesthetics,analyticphilosophyofhistoryandthesocialsciences,whichhad91\nP.M.S.HACKERbeenoflittleornoconcerntohim,werealsodeveloped,ofteninamannerwhichborethemarksofhisinfluence.Thisphaseofanalyticphilosophywanedinthe1970s,andWittgenstein’sinfluencedeclined.Whethertheformsofphilosophythatsucceededitaretobecountedasyetanotherphaseofanalyticphilosophyorassymptomsofitsfinaldemiseissomethingthatwillbecomecleareronlywiththepassingoftime.Whatis,however,clear,isthatWittgensteindominatedtheformsofanalyticphilosophyfromthe1920suntilthe1970s,ineradicablyimpressinghisthoughtupontwentieth-centuryphilosophy.Notes1Forexample,AliceAmbrose,ElizabethAnscombe,MaxBlack,RichardBraithwaite,KarlBritton,PeterGeach,AustinDuncan-Jones,CasimirLewy,MargaretMacDonald,NormanMalcolm,G.A.Paul,RushRhees,StephenToulmin,JohnWisdom,GeorgHenrikvonWright.2Theotherinfluencesuponhisthoughtwhichhecitedretrospectivelyin1931wereKarlKraus,AdolfLoos,PaulErnst,andOttoWeininger.InlateryearshemademuchuseofJames’sThePrinciplesofPsychology,whichheviewedasausefulsourceofinterestingphilo-sophicalconfusions–hencenotsomuchaninfluenceuponhisownideasasastimulustocriticism.3TothismustbeaddedalargequantityofdictationshegavetoFriedrichWaismannfortheprojectedjointworkLogik,Sprache,PhilosophiewhichwasintendedasthefirstvolumeoftheViennaCircle’sseriesSchriftenzurWissenschaftlichesWeltauffassung,thatvolumeitself,pub-lishedinEnglishunderthetitlePrinciplesofLinguisticPhilosophy,andWaismann’snotesofconversationswithWittgensteinpublishedunderthetitleWittgensteinandtheViennaCircle.4ThiscontrastswiththeFregeanandRussellianconceptionofthenatureoftheproposition.Fregeheldthatpropositionsofnaturallanguagemaylackatruth-value,althoughtheyexpressasense.Inhislogicallyideallanguage,Begriffsschrift,everypropositionmustbebivalent(butnotbipolar),i.e.eithertrueorfalse.Russellheldpropositionstobebivalent.BibliographyWorksbyWittgenstein1953:PhilosophicalInvestigations,ed.G.E.M.AnscombeandR.Rhees,trans.G.E.M.Anscombe,Oxford:BlackwellPublishers.1958:TheBlueandBrownBooks,Oxford:BlackwellPublishers.1961a:Notebooks1914–16,ed.G.H.vonWrightandG.E.M.Anscombe,trans.G.E.M.Anscombe,Oxford:BlackwellPublishers.1961b:TractatusLogico-Philosophicus,trans.byD.F.PearsandB.F.McGuinness,London:RoutledgeandKeganPaul.1967:Zettel,ed.G.E.M.AnscombeandG.H.vonWright,trans.G.E.M.Anscombe,Oxford:BlackwellPublishers.1969:OnCertainty,ed.G.E.M.AnscombeandG.H.vonWright,trans.D.PaulandG.E.M.Anscombe,Oxford:BlackwellPublishers.1970:LecturesandConversationsonAesthetics,Psychology,andReligiousBeliefs,compiledfromnotesbyY.Smythies,R.Rhees,andJ.Taylor,ed.C.Barrett,Oxford:BlackwellPublishers.1971:ProtoTractatus:AnEarlyVersionofTractatusLogico-Philosophicus,ed.B.F.McGuinness,T.Nyberg,andG.H.vonWright,trans.D.F.PearsandB.F.McGuinness,London:RoutledgeandKeganPaul.92\nLUDWIGWITTGENSTEIN1974:PhilosophicalGrammar,ed.R.Rhees,trans.A.J.P.Kenny,Oxford:BlackwellPublishers.1975:PhilosophicalRemarks,ed.R.Rhees,trans.R.HargreavesandR.White,Oxford:BlackwellPublishers.1976:Wittgenstein’sLecturesontheFoundationsofMathematics,Cambridge1939,ed.C.Diamond,Sussex:HarvesterPress.1978:RemarksontheFoundationsofMathematics,ed.G.H.vonWright,R.Rhees,andG.E.M.Anscombe,trans.G.E.M.Anscombe,rev.ed.,Oxford:BlackwellPublishers.1979a:LudwigWittgensteinundderWienerKreis,shorthandnotesrecordedbyF.Waismann,ed.B.F.McGuinness,Oxford:BlackwellPublishers,1967.(Englishtranslation,WittgensteinandtheViennaCircle,Oxford:BlackwellPublishers.)1979b:Wittgenstein’sLectures,Cambridge1932–35,fromtheNotesofAliceAmbroseandMargaretMacDonald,ed.AliceAmbrose,Oxford:BlackwellPublishers.1980a:CultureandValue,ed.G.H.vonWrightincollaborationwithH.Nyman,trans.P.Winch,Oxford:BlackwellPublishers.1980b:RemarksonthePhilosophyofPsychology,vol.I,ed.G.E.M.AnscombeandG.H.vonWright,trans.G.E.M.Anscombe,Oxford:BlackwellPublishers.1980c:RemarksonthePhilosophyofPsychology,vol.II,ed.G.H.vonWrightandH.Nyman,trans.C.G.LuckhardtandM.A.E.Aue,Oxford:BlackwellPublishers.1980d:Wittgenstein’sLectures,Cambridge1930–32,fromtheNotesofJohnKingandDesmondLee,ed.DesmondLee,Oxford:BlackwellPublishers.1982:LastWritingsonthePhilosophyofPsychology,vol.I,ed.G.H.vonWrightandH.Nyman,trans.C.G.LuckhardtandM.A.E.Aue,Oxford:BlackwellPublishers.1988:Wittgenstein’sLecturesonPhilosophicalPsychology1946–47,notesbyP.T.Geach,K.J.Shah,andA.C.Jackson,ed.P.T.Geach,HemelHempstead:HarvesterWheatsheaf.1992:LastWritingsonthePhilosophyofPsychology,vol.II,ed.G.H.vonWrightandH.Nyman,trans.C.G.LuckhardtandM.A.E.Aue,Oxford:BlackwellPublishers.1993:LudwigWittgenstein:PhilosophicalOccasions1912–1951,ed.JamesKlaggeandAlfredNormann,Indianapolis:Hackett.WorksbyotherauthorsBaker,G.P.andHacker,P.M.S.(1980–96),AnAnalyticCommentaryonthePhilosophicalInvestigations,4vols,Oxford:BlackwellPublishers;vol.1,Wittgenstein:MeaningandUnderstanding(BakerandHacker);vol.2,Wittgenstein:Rules,GrammarandNecessity(BakerandHacker);vol.3,Wittgenstein:MeaningandMind(Hacker);andvol.4,Wittgenstein:MindandWill(Hacker).Glock,H.J.(1996)AWittgensteinDictionary,Oxford:BlackwellPublishers.Hacker,P.M.S.(1986)InsightandIllusion:ThemesinthePhilosophyofWittgenstein,rev.edn.,Oxford:OxfordUniversityPressandBristol:ThoemmesPress.Kenny,A.J.P.(1973)Wittgenstein,London:PenguinBooks.93\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd20016RudolfCarnap(1891–1970)SAHOTRASARKARRudolfCarnap,pre-eminentmemberoftheViennaCircle,wasoneofthemostinflu-entialfiguresoftwentieth-centuryanalyticphilosophy.TheViennaCirclewasrespon-sibleforpromulgatingasetofdoctrines(initiallyinthe1920s)whichcametobeknownaslogicalpositivismorlogicalempiricism.Thissetofdoctrinesprovidesthepointofdepartureformostsubsequentdevelopmentsinthephilosophyofscience.ConsequentlyCarnapmustberegardedasoneofthemostimportantphilosophersofscienceofthiscentury.Nevertheless,hismostlastingpositivecontributionswereinthephilosophyoflogicandmathematicsandthephilosophyoflanguage.Meanwhile,hissystematicbutultimatelyunsuccessfulattempttoconstructaninductivelogichasbeenequallyinfluentialsinceitsfailurehasconvincedmostphilosophersthatsuchaprojectmustfail.Carnapwasbornin1891inRonsdorf,nearBremen,andnowincorporatedintothe1cityofWuppertal,inGermany.Inearlychildhoodhewaseducatedathomebyhismother,AnnaCarnap(néeDörpfeld),whohadbeenaschoolteacher.From1898,heattendedtheGymnasiumatBarmen,wherethefamilymovedafterhisfather’sdeaththatyear.Inschool,Carnap’schiefinterestswereinmathematicsandLatin.From1910to1914CarnapstudiedattheuniversitiesofJenaandFreiburg,concentratingfirstonphilosophyandmathematicsand,later,onphilosophyandphysics.AmonghisteachersinJenawereBrunoBauch,aprominentneo-Kantian,andGottlobFrege,afounderofthemoderntheoryofquantificationinlogic.BauchimpresseduponhimthepowerofKant’sconceptionthatthegeometricalstructureofspacewasdeterminedbytheformofpureintuition.ThoughCarnapwasimpressedbyFrege’songoingphilo-sophicalprojects,hisreal(andlasting)influenceonlycamelaterthroughastudyofhiswritings(seeFREGE).Carnap’sformalintellectualworkwasinterruptedbetween1914and1918whilehedidmilitaryserviceduringWorldWarI.Hispoliticalviewshadalreadybeenofamildlysocialist/pacifistnature.Thehorrorsofthewarservedtomakethemmoreexplicitandmoreconscious,andtocodifythemsomewhatmorerigorously.SpaceAfterthewar,CarnapreturnedtoJenatobeginresearch.HiscontactswithHansReichenbachandotherspursuingphilosophyinformedbycurrentsciencebegan94\nRUDOLFCARNAPduringthisperiod.In1919hereadWhiteheadandRussell’sPrincipiaMathematicaandwasdeeplyinfluencedbytheclarityofthoughtthatcouldapparentlybeachievedthroughsymbolization.Hebegantheconstructionofaputativeaxiomsystemforaphysicaltheoryofspace-time.Thephysicists–representedbyMaxWien,headoftheInstituteofPhysicsattheUniversityofJena–wereconvincedthattheprojectdidnotbelonginphysics.Meanwhile,Bauchwasequallycertainthatitdidnotbelonginphi-losophy.ThisincidentwasinstrumentalinconvincingCarnapoftheinstitutionaldif-ficultiesfacedinGermanyofdoinginterdisciplinaryworkthatbridgedthechasmbetweenphilosophyandthenaturalsciences.Italsoprobablyhelpedgeneratetheatti-tudethatlaterledthelogicalempiriciststodismissmuchoftraditionalphilosophy,espe-ciallymetaphysics.Bythispointinhisintellectualdevelopment(theearly1920s)Carnapwasalreadyacommittedempiricistwho,nevertheless,acceptedboththeana-lyticityoflogicandmathematicsandtheFrege–Russellthesisoflogicismwhichrequiredthatmathematicsbeformallyconstructedandderivedfromlogic.FacedwiththislackofenthusiasmforhisoriginalprojectinJena,Carnapaban-donedittowriteadissertationonthephilosophicalfoundationsofgeometry,whichwassubsequentlypublishedasDerRaum(1922).Afundamentallyneo-Kantianwork,itincludedadiscussionof“intuitivespace,”determinedbypureintuition,independentofallcontingentexperience,anddistinctfrombothmathematical(orabstract)spaceandphysicalspace.However,incontrasttoKant,Carnaprestrictedwhatcouldbegraspedbypureintuitiontosometopologicalpropertiesofspace;metricpropertiesandeventhedimensionalityofspacewereregardedasempiricalmatters.InagreementwithHelmholtzandMoritzSchlick(aphysicist-turned-philosopher,andfounderoftheViennaCircle–seebelow),thegeometryofphysicalspacewasalsoregardedasanempiricalmatter.Carnapincludedadiscussionoftheroleofnon-EuclideangeometryinEinstein’sGeneralRelativityTheory.Bydistinguishingbetweenintuitive,mathe-matical,andphysicalspaces,Carnapattemptedtoresolvetheapparentdifferencesbetweenphilosophers,mathematicians,andphysicistsbyassigningthedisputingcampstodifferentdiscursivedomains.Inretrospect,thismoveheraldedwhatlaterbecamethemostsalientfeaturesofCarnap’sphilosophicalwork:tolerancefordiversepointsofview(solongastheymetstringentcriteriaofclarityandrigor)andanassign-mentoftheseviewpointstodifferentrealms,thechoicebetweenwhichistoberesolvednotbyphilosophicallysubstantive(forinstance,epistemological)criteriabutbyprag-maticones.TheconstructionistphaseDuringthewinterof1921,CarnapreadRussell’sOurKnowledgeoftheExternalWorld(1914).AccordingtoCarnap’sintellectualautobiography(1963a),thisworkledhim,between1922and1925,tobegintheanalysisthatculminatedinDerlogischeAufbauderWelt(1967),whichisusuallyregardedasCarnap’sfirstmajorwork.ThepurposeoftheAufbauwastoconstructtheeverydayworldfromaphenomenalistbasis.Thisis2anepistemologicalchoice(§§54,58).Carnapdistinguishedbetweenfourdomainsofobjects:autopsychological,physical,heteropsychological,andcultural(§58).Thefirstoftheseconsistsofobjectsofanindividual’sownpsychology;thesecondofphysi-calentities(Carnapdoesnotdistinguishbetweeneverydaymaterialobjectsandthe95\nSAHOTRASARKARabstractentitiesoftheoreticalphysics);thethirdconsistsoftheobjectsofsomeotherindividual’spsychology;andthefourthofculturalobjects(geistigeGegenstände),whichincludehistoricalandsociologicalphenomena.FromCarnap’spointofview,“[a]nobject...iscalledepistemicallyprimaryrelativetoanotherone...ifthesecondoneisrecognizedthroughthemediationofthefirstandthuspresupposes,foritsrecognition,therecognitionofthefirst”(§54).Autopsychologicalobjectsareepistemicallyprimaryrelativetotheothersinthissense.Moreover,physicalobjectsareepistemicallyprimarytoheteropsychologicalonesbecausethelattercanonlyberecognizedthroughthemediationoftheformer:anexpressiononaface,areadinginaninstrument,etc.Finally,heteropsychologicalobjectsareepistemicallyprimaryrelativetoculturalonesforthesamereason.ThemaintaskoftheAufbauisconstruction,whichCarnapconceivesofasthecon-verseofwhatheregardedasreduction(whichisfarfromwhatwasthen–orisnow–conceivedofas“reduction”inAnglophonephilosophy):anobjectis‘reducible’toothers...ifallstatementsaboutitcanbetranslatedintostatementswhichspeakonlyabouttheseotherobjects....Byconstructingaconceptfromotherconcepts,weshallmeantheindicationofits“constructionaldefinition”onthebasisofothercon-cepts.Byaconstructionaldefinitionoftheconceptaonthebasisoftheconceptsbandc,wemeanaruleoftranslationwhichgivesageneralindicationhowanypropositionalfunc-tioninwhichaoccursmaybetransformedintoacoextensivepropositionalfunctioninwhichanolongeroccurs,butonlybandc.Ifaconceptisreducibletoothers,thenitmustindeedbepossibletoconstructitfromthem.(§35)However,constructionandreductionpresentdifferentformalproblemsbecause,exceptinsomedegeneratecases(suchasexplicitdefinition),thetransformationsinthetwodirectionsmaynothaveanysimpleexplicitrelationtoeachother.Thequestionofreducibility/constructibilityisdistinctfromthatofepistemicprimacy.Inanimportantinnovationinanempiricistcontext,Carnaparguesthatboththeautopsychologicalandphysicaldomainscanbereducedtoeachother(inhissense).Thus,attheformallevel,eithercouldserveasthebasisoftheconstruction.Itisepistemicprimacythatdictatesthechoiceoftheformer.Carnap’stask,ultimately,istosetupaconstructionalsystemthatwillallowtheconstructionoftheculturaldomainfromtheautopsychologicalthroughthetwointer-mediatedomains.IntheAufbau,thereareonlyinformaldiscussionsofhowthelasttwostagesofsuchaconstructionaretobeexecuted.Onlytheconstructionofthephysicalfromtheautopsychologicalisfullytreatedformally.AsthebasicunitsoftheconstructionalsystemCarnapchosewhathecalls“elementaryexperiences”3(Elementarerlebnisse)(elex).Thesearesupposedtobeinstantaneouscross-sectionsofthestreamofexperience–oratleastbitsofthatstreaminthesmallestperceivableunitoftime–thatareincapableoffurtheranalysis.TheonlyprimitiverelationthatCarnapintroducesis“recollectionofsimilarity”(Rs).(Intheformaldevelopmentofthesystem,RsisintroducedfirstandtheelexaredefinedasthefieldofRs.)TheasymmetryofRsiseventuallyexploitedbyCarnaptointroducetemporalordering.Sincetheelexareelementary,theycannotbefurtheranalyzedtodefinewhatwouldberegardedasconstituentqualitiesofthem,suchaspartialsensationsorintensity96\nRUDOLFCARNAPcomponentsofasensation.Hadtheelexnotbeenelementary,Carnapcouldhaveused“properanalysis”todefinesuchqualitiesbyisolatingtheindividualsintoclassesonthebasisofhavingacertain(symmetric)relationshipwitheachother.Carnapdefinestheprocessof“quasi-analysis”tobeformallyanalogoustoproperanalysisbutonlydefin-4ing“quasi-characteristics”or“quasi-constituents”becausetheelexareunanalyzable.Quasi-analysisbasedontherelation“partsimilarity”(Ps),itselfdefinedfromRs,isthecentraltechniqueoftheAufbau.Itisusedeventuallytodefinesenseclassesand,then,thevisualsense,visualfieldplaces,thespatialorderofthevisualfield,theorderofcolorsand,eventually,sensations.Thusthephysicaldomainisconstructedoutoftheautopsychological.Carnap’saccountsoftheconstructionbetweentheothertwodomainsremainpromissorysketches.Carnapwasawarethattherewereunresolvedtechnicalproblemswithhiscon-structionofthephysicalfromtheautopsychological,thoughheprobablyunderesti-matedtheseriousnessoftheseproblems.Thesystematicproblemsarethatwhenaqualityisdefinedasaclassselectedbyquasi-analysisonthebasisofarelation:(1)two(different)qualitiesthathappenalwaystooccurtogether(say,redandhot)willneverbeseparated;and(2)qualityclassesmayemergeinwhichanytwomembersbearsomerequiredrelationtoeachotherbuttheremayyetbenorelationthatholdsbetweenallmembersoftheclass.Carnap’sresponsetotheseproblemswasextra-systematic:inthecomplicatedconstructionofourworldfromourelex,hehopedthatsuchexamples5wouldneveroronlyveryrarelyarise.Nevertheless,becauseoftheseproblems,andbecausetheotherconstructionsarenotcarriedout,theattitudeoftheAufbauistentativeandexploratory:theconstructionalsystemispresentedasessentially6unfinished.Bythispointofhisintellectualdevelopment,CarnaphadnotonlyfullyendorsedthelogicismofthePrincipia,butalsotheformthatWhiteheadandRussellhadgiventologic(thatis,theramifiedtheoryoftypesincludingtheaxiomsofinfinityandreducibil-ity)inthatwork.However,Poincaréalsoemergesasamajorinfluenceduringthisperiod.Carnapdidconsiderableworkontheconceptualfoundationsofphysicsinthe1920sandsomeofthiswork–inparticular,hisanalysisoftherelationshipbetweencausaldeterminationandthestructureofspace–showsstrongconventionalistatti-tudes(Carnap1924;seealsoCarnap1923and1926).ViennesepositivismIn1926,atSchlick’sinvitation,CarnapmovedtoViennatobecomeaPrivatdozent(instructor)inphilosophyattheUniversityofViennaforthenextfiveyears.AnearlyversionoftheAufbauservedashisHabilitationsschrift.HewaswelcomedintotheViennaCircle,ascientificphilosophydiscussiongrouporganizedby(andcenteredaround)Schlick,whohadoccupiedtheChairforPhilosophyoftheInductiveSciencessince1922.InthemeetingsoftheViennaCirclethetypescriptoftheAufbauwasreadanddiscussed.WhatCarnapseemstohavefoundmostcongenialintheCircle–besidesitsmembers’concernforscienceandcompetenceinmodernlogic–wastheirrejectionoftraditionalmetaphysics.Overtheyears,besidesCarnapandSchlick,theCircleincludedHerbertFeigl,KurtGödel,HansHahn,KarlMenger,OttoNeurath,andFriedrichWaismann,thoughGödelwouldlaterclaimthathehadlittlesympathyforthe97\nSAHOTRASARKARanti-metaphysicalpositionoftheothermembers.ThemeetingsoftheCirclewerechar-acterizedbyopen,intenselycritical,discussionwithnotoleranceforambiguityoffor-mulationorlackofrigorindemonstration.ThemembersoftheCirclebelievedthatphilosophywasacollectiveenterpriseinwhichprogresscouldbemade.Theseattitudes,evenmorethananycanonicalsetofpositions,characterizedthephilosophicalmovement,initiallyknownaslogicalpositivismand,later,aslogicalempiricism,thatemergedfromtheworkofthemembersoftheCircleandafewothers,especiallyReichenbach.However,besidesrejectingtraditionalmetaphysics,mostmembersoftheCircleacceptedlogicismandasharpdistinctionbetweenanalyticandsynthetictruths.Theanalyticwasidentifiedwiththeapriori;thesyntheticwiththeaposteriori.A.J.Ayer,whoattendedsomemeetingsoftheCirclein1933(afterCarnaphadleft–seebelow)returnedtoLondonandpublishedLanguage,TruthandLogicin1936(seeAYER).ThisshortbookdidmuchtopopularizetheviewsoftheViennaCircleamongAnglophonephilosophersthoughitlacksthesophisticationthatisfoundinthewritingsofthemembersoftheCircle,particularlyCarnap.UnderNeurath’sinfluence,duringhisViennayears,Carnapabandonedthephe-nomenalistlanguagehehadpreferredintheAufbauandcametoacceptphysicalism.Theepistemicallyprivilegedlanguageisoneinwhichsentencesreportingempiricalknowledgeoftheworld(“protocolsentences”)employtermsreferringtomaterialbodiesandtheirobservableproperties.FromCarnap’spointofview,thechiefadvan-tageofaphysicalistlanguageisitsintersubjectivity.Physicalism,moreover,camehand-in-handwiththethesisofthe“unityofscience,”thatis,thatthedifferentempiricalsciences(includingthesocialsciences)weremerelydifferentbranchesofasingleunifiedscience.Todefendthisthesis,ithadtobedemonstratedthatpsychologycouldbebasedonaphysicalistlanguage.Inanimportantpaperonlypublishedsomewhatlater,Carnap(1934b)attemptedthatdemonstration.Carnap’sadoptionofphysicalismwasfinal;heneverwentbacktoaphenomenalistlanguage.However,whathemeantby“physicalism”underwentradicaltransformationsovertheyears.Bytheendofhislife,itmeantnomorethantheadoptionofanon-solipsisticlanguage,thatis,oneinwhichintersubjectivecommunicationispossible(Carnap1963b).IntheViennaCircle,Wittgenstein’sTractatuswasdiscussedindetail.CarnapfoundWittgenstein’srejectionofmetaphysicsconcordantwiththeviewshehaddevelopedindependently.PartlybecauseofWittgenstein’sinfluenceonsomemembersoftheCircle(thoughnotCarnap),therejectionofmetaphysicstooktheformofanassertionthatthesentencesofmetaphysicsaremeaninglessinthesenseofbeingdevoidofcog-nitivecontent.Moreover,thedecisionwhetherasentenceismeaningfulwastobemadeonthebasisoftheprincipleofverifiability,whichclaimsthatthemeaningofasen-tenceisgivenbytheconditionsofits(potential)verification.Observationtermsaredirectlymeaningfulonthisaccount.Theoreticaltermsonlyacquiremeaningthroughexplicitdefinitionfromobservationterms.Carnap’smajorinnovationinthesediscus-sionswithintheCirclewastosuggestthateventhethesisofrealism–assertingthe“reality”oftheexternalworld–isalsomeaningless,apositionnotsharedbySchlick,Neurath,orReichenbach.Problemsgeneratedbymeaninglessquestionsbecamethecelebrated“pseudo-problems”ofphilosophy(Carnap1967).Wittgenstein’sprincipleofverifiabilityposedfairlyobviousproblemsinanyscien-tificcontext.Nouniversalgeneralizationcaneverbeverified.Perhapsindependently,98\nRUDOLFCARNAPKarlPopperperceivedthesameproblem(seePOPPER).Thisledhimtoreplacetherequirementofverifiabilitywiththatoffalsifiability,thoughonlyasacriteriontodemarcatesciencefrommetaphysics,andnotasonealsotobeusedtodemarcatemean-ingfulfrommeaninglessclaims.Itisalsounclearwhatthestatusoftheprincipleitselfis,thatis,whetheritismeaningfulbyitsowncriterionofmeaningfulness.Carnap,aswellasothermembersoftheViennaCircleincludingHahnandNeurath,realizedthataweakercriterionofmeaningfulnesswasnecessary.Thusbegantheprogramofthe“liberalizationofempiricism.”TherewasnounanimitywithintheViennaCircleonthispoint.Thedifferencesbetweenthemembersaresometimesdescribedasthosebetweenaconservative“right”wing,ledbySchlickandWaismann,whichrejectedtheliberali-zationofempiricism,andtheepistemologicalanti-foundationalismthatisinvolvedinthemovetophysicalism;andaradical“left”wing,ledbyNeurathandCarnap,whichendorsedtheoppositeviews.The“left”wingalsoemphasizedfallibilismandpragmat-ics;Carnapwentfarenoughalongthislinetosuggestthatempiricismitselfwasaproposaltobeacceptedonpragmaticgrounds.ThisdifferencealsoreflectedpoliticalattitudesinsofarasNeurath,andtoalesserextent,Carnapviewedscienceasatoolforsocialreform.Thepreciseformulationofwhatcametobecalledthecriterionofcognitivesignifi-cancetookthreedecades.(SeeHempel1950andCarnap1956and1961.)Inanimpor-tantpairofpapers,“TestabilityandMeaning,”Carnap(1936,1937a)replacedtherequirementofverificationwiththatofconfirmation;atthisstage,hemadenoattempttoquantifythelatter.Individualtermsreplacesentencesastheunitsofmeaning.Universalgeneralizationsarenolongerproblematic;thoughtheycannotbecon-clusivelyverified,theycanyetbeconfirmed.Moreover,in“TestabilityandMeaning,”theoreticaltermsnolongerrequireexplicitdefinitionfromobservationalonesinordertoacquiremeaning;theconnectionbetweenthetwomaybeindirectthroughasystemofimplicitdefinitions.Carnapalsoprovidesanimportantpioneeringdiscussionofdispositionpredicates.ThesyntacticphaseMeanwhile,in1931,CarnaphadmovedtoPrague,whereheheldtheChairforNaturalPhilosophyattheGermanUniversityuntil1935when,undertheshadowofHitler,heemigratedtotheUnitedStates.TowardstheendofhisViennayears,asubtlebutimpor-tantshiftinCarnap’sphilosophicalinterestshadtakenplace.Thisshiftwasfromapredominantconcernforthefoundationsofphysicstothatforthefoundationsofmathematicsandlogic,eventhoughheremainedemphaticthatthelatterwereimpor-tantonlyinsofarastheywereusedintheempiricalsciences,especiallyphysics.InViennaandbefore,followingFregeandRussell,Carnapespousedlogicisminitsconventionalsense,thatis,asthedoctrinethatheldthattheconceptsofmathematicsweredefinablefromthoseoflogicandthetheoremsofmathematicswerederivablefromtheprinciplesoflogic.IntheaftermathofGödel’s(1931)incompletenesstheo-rems(seeTARSKI,CHURCH,GÖDEL),however,Carnapabandonedthistypeoflogicismandopted,instead,fortherequirementthattheconceptsofmathematicsandlogicalwayshavetheircustomary,thatis,everydayinterpretationinallcontexts.Healsobegantoadvocatedastrongconventionalismregardingwhatconstituted“logic.”99\nSAHOTRASARKARBesidesthephilosophicalsignificanceofGödel’sresults,whatimpressedCarnapmostaboutthatworkwasGödel’sarithmetizationofsyntax.Downplayingthedistinc-tionbetweenanobjectlanguageanditsmetalanguage,Carnapinterpretedthisproce-dureasenablingtherepresentationofthesyntaxofalanguagewithinthelanguageitself.AtthispointCarnaphadnotyetacceptedthepossibilityofsemanticseventhoughhewasawareofsomeofTarski’sworkandhadhadsomecontactwiththePolishschooloflogic.Inthiscontext,therepresentationofthesyntaxofalanguagewithinitselfsuggestedtoCarnapthatallpropertiesofalanguagecouldbestudiedwithinitselfthroughastudyofsyntax.ThesepositionswerecodifiedinCarnap’smajorworkfromthisperiod,TheLogicalSyntaxofLanguage(Carnap1937b).TheEnglishtranslationincludesmaterialthathadtobeomittedfromtheGermanoriginalowingtoashortageofpaper;theomittedmaterialwasseparatelypublishedinGermanaspapers(Carnap1934a,1935).Conventionalismaboutlogicwasincorporatedintothewell-knownPrincipleofTolerance:Itisnotourbusinesstosetupprohibitionsbuttoarriveatconventions[aboutwhatconstitutesalogic]....Inlogic,therearenomorals.Everyoneisatlibertytobuilduphisownlogic,i.e.,hisownformoflanguage,ashewishes.Allthatisrequiredisthat,ifhewishestodiscussit,hemuststatehismethodclearly,andgivesyntacticrulesinsteadofphilosophicalargu-ments.(1937b:51–2;emphasisintheoriginal)Logic,therefore,isnothingbutthesyntaxoflanguage.InSyntax,thePrincipleofToleranceallowsCarnaptonavigatetheongoingdisputesbetweenlogicism,formalism,andintuitionism/constructivisminthefoundationsofmathematicswithoutabandoninganyinsightofinterestfromtheseschools.Carnapbeginswithadetailedstudyoftheconstructionoftwolanguages,IandII.ThelastfewsectionsofSyntaxalsopresentafewresultsregardingthesyntaxofanylanguageand7alsodiscussthephilosophicalramificationsofthesyntacticpointofview.LanguageI,whichCarnapcalls“definite,”isintendedasaneutralcoreofalllogi-callyinterestinglanguages,neutralenoughtosatisfythestricturesofalmostanyintui-tionistorconstructivist.Itpermitsthedefinitionofprimitiverecursivearithmeticandhasboundedquantification(forallxuptosomeupperbound)butnotmuchmore.Itssyntaxisfullyconstructedformally.LanguageII,whichis“indefinite”forCarnap,isricher.ItincludesLanguageIandhassufficientresourcesfortheformulationofallofclassicalmathematicsandis,therefore,non-constructive.Moreover,Carnappermitsdescriptivepredicatesineachlanguage.Thus,theresourcesofLanguageIIarestrongenoughtopermit,inprinciple,theformulationofclassicalphysics.Theimportantpointisthat,becauseofthePrincipleofTolerance,thechoicebetweenLanguagesIandIIor,forthatmatter,anyothersyntacticallyspecifiedlanguage,isnotbasedonfactualconsiderations.Ifonewantstousemathematicstostudyphysicsinthecustomaryway,LanguageIIispreferablesince,asyet,non-constructivemathematicsremainsnecessaryforphysics.ButtheadoptionofLanguageII,dictatedbythepragmaticconcernfordoingphysics,doesnotmakeLanguageIincorrect.ThiswasCarnap’sresponsetothefoundationaldisputesofmathematics:bytolerancetheyaredefinedoutofexistence.100\nRUDOLFCARNAPThepricepaidifoneadoptsthePrincipleofToleranceisaradicalconventionalismaboutwhatconstituteslogic.Conventionalism,alreadyapparentinCarnap’sadmissionofbothaphenomenalistandaphysicalistpossiblebasisforconstructionintheAufbau,andstronglypresentintheworksonthefoundationsofphysicsinthe1920s,hadnowbeenextendedinSyntaxtologic.Asaconsequence,whatmightbeconsideredtobethemostimportantquestioninanymathematicalorempiricalcontext–thechoiceoflanguage–becamepragmatic.Thistrendofrelegatingtroublesomequestionstotherealmofpragmaticsalmostbyfiat,therebyexcusingthemfromsystematicphilosophicalexploration,becameincreasinglyprevalentinCarnap’sviewsastheyearswenton.Syntaxcontainedfourtechnicalinnovationsinlogicthatareofsignificance:(1)adefinitionofanalyticitythat,aswaslatershownbyS.C.Kleene,mimickedTarski’sdefi-nitionoftruthforaformalizedlanguage;(2)Carnapconstructedaproof,indepen-dentlyofTarski,thattruthcannotbedefinedasasyntacticpredicateinanyconsistentformalizedlanguage;(3)aruleforinfiniteinduction(inLanguageI)thatlatercametobecalledtheomegarule;and(4),mostimportant,ageneralizationofGödel’sfirstincompletenesstheoremthathascometobecalledthefixed-pointlemma.Withrespectto(4),whatCarnapprovedisthat,inalanguagestrongenoughtopermitarithmeti-zation,foranysyntacticpredicate,onecanconstructasentencethatwouldbeinter-pretedassayingthatitsatisfiesthatpredicate.Ifthechosenpredicateisunprovability,onegetsGödel’sresult.BesidesthePrincipleofTolerance,themainphilosophicalcontributionofSyntaxwasthethesisthatphilosophyconsistedofthestudyoflogicalsyntax.GivinganewtwisttotheViennaCircle’sclaimthatmetaphysicalclaimsweremeaningless,Carnaparguesandtriestoshowbyexamplethatsentencesmakingmetaphysicalclaimsareallsyntacticallyill-formed.Moreover,sincethearithmetizationprocedureshowsthatallthesyntacticrulesofalanguagecanbeformulatedwithinthelanguage,eventherulesthatdeterminewhatsentencesaremeaninglesscanbeconstructedwithinthelan-guage.Allthatisleftforphilosophyisastudyofthelogicofscience.But,asCarnapputsit:“Thelogicofscience(logicalmethodology)isnothingelsethanthesyntaxofthelanguageofscience....Tosharethisviewistosubstitutelogicalsyntaxforphilosophy”(1937b:7–8).TheclaimsofSyntaxarefarmoregrandiose–andmoreflamboyant–thananythingintheAufbau.SemanticsInthelate1930sCarnapabandonedthenarrowsyntacticismofSyntaxand,undertheinfluenceofTarskiandthePolishschooloflogic,cametoacceptsemantics.Withthismove,Carnap’sworkentersitsfinalmaturephase.Forthefirsttime,heacceptedthattheconceptoftruthcanbegivenmorethanpragmaticcontent.Thereupon,heturnedtothesystematizationofsemanticswithcharacteristicvigor,especiallyafterhisimmi-grationtotheUSwherehetaughtattheUniversityofChicagofrom1936to1952.InhiscontributiontotheInternationalEncyclopediaofUnifiedScience,in1939,onthefoundationsoflogicandmathematics,thedistinctionsbetweensyntactic,semantic,andpragmaticconsiderationsregardinganylanguagearefirstpresentedintheirmatureform.101\nSAHOTRASARKARIntroductiontoSemantics,whichfollowedin1942,developssemanticssystematically.InSyntaxCarnaphaddistinguishedbetweentwotypesoftransformationsonsen-tences:thoseinvolving“themethodofderivation”or“d-method”;andthoseinvolvingthe“methodofconsequence”or“c-method.”Bothoftheseweresupposedtobesyn-tacticbutthereisacriticaldistinctionbetweenthem.Theformerallowsonlyafinitenumberofelementarysteps.Thelatterplacesnosuchrestrictionandis,therefore,more“indefinite.”Termsdefinedusingthed-method(“d-terms”)include“derivable,”“demonstrable,”“refutable,”“resoluble,”and“irresoluble”;thecorresponding“c-terms”are“consequence,”“analytic,”“contradictory,”“L-determinate,”and“syn-thetic.”AftertheconversiontosemanticsCarnapproposedthatthec-methodessentiallycapturedwhatsemanticsallowed;thec-termsreferredtosemanticconcepts.Thussemanticsinvolvesakindofformalization,thoughonethatisdependentonstrongerinferencerulesthanthesyntacticalones.Inthissense,asChurch(1956:65)hasperceptivelypointedout,Carnap(andTarski)reducesemanticstoformalrules,thatis,syntax.Thusemergestheinterpretationofdeductivelogicthathassincebecomethetextbookversion,socommonlyacceptedthatishasbecomeunnecessarytorefertoCarnapwhenoneusesit.ForCarnap,thesemanticmovehadanimportantphilo-sophicalconsequence:philosophywasnolongertobereplacedjustbythesyntaxofthelanguageofscience;rather,itwastobereplacedbythesyntaxandthesemanticsofthelanguageofscience.Carnap’smostoriginal–andinfluential–workinsemanticsisMeaningandNecessity(1947),wherethebasisforanintensionalsemanticswaslaiddown.LargelyfollowingFrege,intensionalconceptsaredistinguishedfromextensionalones.Semanticalrulesareintroducedandtheanalytic/syntheticdistinctionisclarifiedbyrequiringthatanydefinitionofanalyticitymustsatisfythe(meta-)criterionthatanalyticsentencesfollowfromthesemanticalrulesalone.BynowCarnaphadfullyacceptedthatsemanticconceptsandmethodsaremorefundamentalthansyntacticones:theretreatfromtheflamboyanceofSyntaxwascomplete.ThemostimportantcontributionofMeaningandNecessitywasthereintroductionintologic,inthenewintensionalframe-work,ofmodalconceptsthathadbeenignoredsincethepioneeringworkofLewis(1918).IntheconcludingchapterofhisbookCarnapintroducedanoperatorforneces-sity,gavesemanticrulesforitsuse,andshowedhowothermodalconceptssuchaspossibility,impossibility,necessaryimplication,andnecessaryequivalencecanbedefinedfromthisbasis.Bythispoint,Carnaphadbeguntorestricthisanalysestoexactlyconstructedlan-guages,implicitlyabandoningevenadistanthopethattheywouldhaveanydirectbearingonnaturallanguages.Theproblemwiththelatteristhattheirambiguitiesmadethemunsuitedfortheanalysisofsciencewhich,ultimately,remainedthemoti-vationofallofCarnap’swork.Nevertheless,Carnap’sdistinctionbetweentheanalyticandthesyntheticcameunderconsiderablecriticismfrommany,includingQuine(1951),primarilyonthebasisofconsiderationsaboutnaturallanguages.ThoughphilosophicalfashionhaslargelyfollowedQuineonthispoint,atleastuntilrecently,Carnapwasneveroverlyimpressedbythiscriticism(Stein1992).Theanalytic/syn-theticdistinctioncontinuedtobefundamentaltohisviewsand,inarejoindertoQuine,Carnaparguedthatnothingpreventedempiricallinguisticsfromexploringintensionsandtherebydiscoveringcasesofsynonymyandanalyticity(Carnap1955).102\nRUDOLFCARNAPCarnap’smostsystematicexpositionofhisfinalviewsonontologyisalsofromthisperiod(1950a).Acleardistinctionismaintainedbetweenquestionsthatareinternaltoalinguisticframeworkandquestionsthatareexternaltoit.Thechoiceofalinguis-ticframeworkistobebasednotoncognitivebutonpragmaticconsiderations.Theexternalquestionof“realism,”whichostensiblyreferstothe“reality”ofentitiesofaframeworkinsomesenseindependentofit,ratherthantotheir“reality”withinitaftertheframeworkhasbeenaccepted,isrejectedasnon-cognitive.Thisappearstobeananti-“realist”positionbutisnotinthesensethat,withinaframework,Carnapistol-erantoftheabstractentitiesthatbothernominalists.Theinterestingquestionbecomesthepragmaticone,thatis,whatframeworksarefruitfulinwhichcontexts,andCarnap’sattitudetowardstheinvestigationofvariousalternativeframeworksremainscharacteristicallyandconsistentlytolerant.Carnapcontinuedtoexplorequestionsaboutthenatureoftheoreticalconceptsandtosearchforacriterionofcognitivesignificance,preoccupationsofthelogicalempiri-ciststhatdatebacktotheViennaCircle.In1956hepublishedadetailedexpositionofhisfinalviewsregardingtherelationbetweenthetheoreticalandobservationalpartsofascientificlanguage(Carnap1956).Thispaperemphasizesthemethodologicalandpragmaticaspectsoftheoreticalconcepts.Italsocontainshismostsubtle,thoughnothislast,attempttoexplicatethenotionofthecognitivesignificanceofatermandthusestablishclearlytheboundarybetweenscientificandnonscientificdiscourse.However,thecriterionheformulatesmakestheoreticaltermssignificantonlywithrespecttoaclassofterms,atheoreticallan-guage,anobservationlanguage,correspondencerulesbetweenthem,andatheory.Relativizationtoatheoryiscriticaltoavoidingtheproblemsthatbesetearlierattemptstofindsuchacriterion.Carnapprovesseveraltheoremsthataredesignedtoshowthatthecriteriondoescapturethedistinctionbetweenscientificandnonscientificdiscourse.ThiscriterionwascriticizedbyRoozeboom(1960)andKaplan(1975)butthesecriti-cismsdependonmodifyingCarnap’soriginalproposalinimportantways.AccordingtoKaplan,Carnapacceptedhiscriticismthoughthereisapparentlynoindependentconfirmationofthatfact.However,Carnap(1961)didturntoadifferentformalism(Hilbert’se-operator)inhislastattempttoformulatesuchacriterionandthismayindi-catedissatisfactionwiththe1956attempt.Ifso,itremainsunclearwhy:thatattemptdidmanagetoavoidthetechnicalproblemsassociatedwiththeearlierattemptsofthelogicalempiricists.InductivelogicFrom1941onwardsCarnapalsobeganasystematicattempttoanalyzetheconceptsofprobabilityandtoformulateanadequateinductivelogic(alogicofconfirmation),aprojectthatwouldoccupyhimfortherestofhislife.Carnapviewedthisworkasanextensionofthesemanticalmethodsthathehadbeendevelopingforthelastdecade.ThisunderscoresaninterestingpatterninCarnap’sintellectualdevelopment.Untilthelate1930sCarnaponlyviewedsyntacticcategoriesasnon-pragmaticallyspecifiable;questionsoftruthandconfirmationwereviewedaspragmatic.Hisconversiontosemanticssawtherecoveryoftruthfromthepragmatictothesemanticrealm.Now,confirmationfollowedtruthdownthesamepathway.103\nSAHOTRASARKARInLogicalFoundationsofProbability(1950b),hisfirstsystematicanalysisofprob-ability,Carnapdistinguishedbetweentwoconceptsofprobability:“statisticalprobabil-ity,”whichwastherelevantconcepttobeusedinempiricalcontextsandgenerallyestimatedfromtherelativefrequenciesofevents,and“logicalprobability,”whichwastobeusedincontextssuchastheconfirmationofscientifichypothesesbyempiricaldata.Thoughthelatterconcept,usuallycalledthe“logicalinterpretation”ofprobabil-itywentbacktoKeynes(1921),Carnapprovidesitsfirstsystematicexplication.Logicalprobabilityisexplicatedfromthreedifferentpointsofview(1950b:164–8):(1)asaconditionalprobabilityc(h,e)whichmeasuresthedegreeofconfirmationofahypothesishonthebasisofevidencee(ifc(h,e)=r,thenrisdeterminedbylogicalrela-tionsbetweenhande);(2)asarationaldegreeofbelieforfairbettingquotient(ifc(h,e)=r,thenrisafairbetonhifecorrectlydescribesthetotalknowledgeavailabletoabettor);and(3)asthelimitofrelativefrequenciesinsomecases.AccordingtoCarnap,thefirstofthese,whichspecifiesaconfirmationfunction(“c-function”),istheconceptthatismostrelevanttotheproblemofinduction.Intheformaldevelopmentofthetheory,probabilitiesareassociatedwithsentencesofaformalizedlanguage.InFoundations,Carnapbelievedthatauniquemeasurec(h,e)ofthedegreeofconfirmationcanbefoundandheevenproposedone(namely,Laplace’sruleofsuccession)thoughhecouldnotproveitsuniquenesssatisfactorily.Hisgeneralstrat-egywastoaugmentthestandardaxiomsoftheprobabilitycalculusbyasetof“conventionsonadequacy”(1950b:285),whichturnedouttobeequivalenttoassumptionsabouttherationalityofdegreesofbeliefthathadindependentlybeenproposedbybothRamseyanddeFinetti(Shimony1992).Inalaterwork,TheContinuumofInductiveMethods(1952),usingtheconventionsonadequacyandsomeplausiblesymmetryprinciples,Carnapmanagedtoshowthatallacceptablec-functionscouldbeparameterizedbyasingleparameter,arealnumber,lŒ[0,•].Thetroubleremainedthatthereisnointuitivelyappealingaprioristrategytorestrictltosomepreferablyverysmallsubsetof[0,•].Atonepoint,Carnapevenspeculatedthatitwouldhavetobefixedempirically.Unfortunately,somehigher-orderinductionwouldthenberequiredtojustifytheprocedureforitsestimationand,potentially,thisleadstoinfiniteregress.Carnapspent1952–4attheInstituteforAdvancedStudyatPrincetonwherehecontinuedtoworkoninductivelogic,oftenincollaborationwithJohnKemeny.Healsoreturnedtothefoundationsofphysics,apparentlymotivatedbyadesiretotraceandexplicatetherelationsbetweenthephysicalconceptofentropyandanabstractconceptofentropyappropriateforinductivelogic.Hisdiscussionwithphysicistsprovedtobe8disappointingandhedidnotpublishhisresults.In1954CarnapmovedtotheUniversityofCaliforniaatLosAngelestoassumethechairthathadbecomevacantwithReichenbach’sdeathin1953.Therehecontinuedtoworkprimarilyoninductivelogic,oftenwithseveralcollaborators,overthenextdecade.Thereweresignificantmodificationsofhisearlierattemptstoformulateasys-9tematicinductivelogic.ObviouslyimpressedbytheearlierworkofRamseyanddeFinetti,Carnap(1971b)returnedtothesecondofhisthree1950explicationsoflogicalprobabilityandemphasizedtheuseofinductivelogicindecisionproblems.Moreimportantly,Carnap,in“ABasicSystemofInductiveLogic”(1971a,1980)finallyrecognizedthatattributingprobabilitiestosentenceswastoorestrictive.Ifa104\nRUDOLFCARNAPconceptualsystemusesrealnumbersandreal-valuedfunctions,nolanguagecanexpressallpossiblecasesusingonlysentencesorclassesofsentences.Becauseofthis,henowbegantoattributeprobabilitiestoeventsorpropositions(whicharetakentobesynonymous).Thisfinallybroughtsomeconcordancebetweenhisformalmethodsandthoseofmathematicalstatisticiansinterestedinepistemologicalquestions.Propositionsareidentifiedwithsetsofmodels;however,thefieldsofthesetsaredefinedusingtheatomicpropositionsofaformalizedlanguage.Thus,thoughprobabilitiesaredefinedasmeasuresofsets,theystillremainrelativizedtoaparticularformalizedlanguage.Becauseofthis,andbecausethelanguagesconsideredremainrelativelysimple(mostlymonadicpredicatelanguages)muchofthisworkremainssimilartotheearlierattempts.BythispointCarnaphadabandonedthehopeoffindingauniquec-function.Instead,hedistinguishedbetweensubjectiveandobjectiveapproachesininductivelogic.Theformeremphasizesindividualfreedominthechoiceofnecessaryconven-tions;thelatteremphasizestheexistenceoflimitations.ThoughCarnapcharacteristi-callyclaimedtokeepanopenmindaboutthesetwoapproaches,hisemphasiswasonfindingrationalaprioriprincipleswhichwouldsystematicallylimitthechoiceofc-functions.Carnapwasstillworkingonthisprojectwhenhediedon14September1970.Hehadnotfinishedrevisingthelastsectionsofthesecondpartofthe“BasicSystem,”bothpartsofwhichwereonlypublishedposthumously.Towardstheendofhislife,Carnap’sconcernforpoliticalandsocialjusticehadledhimtobecomeanactivesupporterofanAfrican-AmericancivilrightsorganizationinLosAngeles.AccordingtoStegmüller(1972:lxvi),the“lastphotographwehaveofCarnapshowshimintheofficeofthisorganization,inconversationwithvariousmembers.Hewastheonlywhiteinthediscussiongroup.”ThelegacySomedecadesafterCarnap’sdeathitiseasiertoassessCarnap’slegacy,andthatoflogicalempiricism,thanitwasinthe1960sand1970swhenanewgenerationofana-lyticphilosophersandphilosophersofscienceapparentlyfeltthattheyhadtorejectthatworkaltogetherinordertobeabletodefinetheirownphilosophicalagendas.ThisreactioncanitselfbetakenasevidenceofCarnap’sseminalinfluencebut,neverthe-less,itisfairtosaythatCarnapandlogicalempiricismfellintoaperiodofneglectinthe1970sfromwhichitonlybegantoemergeinthelate1980sandearly1990s.MeanwhileitbecamecommonplaceamongphilosopherstoassumethatCarnap’sprojectshadfailed.Diagnosesofthisfailurehavevaried.Forsomeitwasaresultofthelogicalempiri-cists’allegedinabilitytoproduceatechnicallyacceptablecriterionforcognitivesignifi-cance.Forothers,itwasbecauseofQuine’sdictaagainsttheconceptofanalyticityandtheanalytic/syntheticdistinction.SometookPopper’sworktohavesupersededthatofCarnapandthelogicalempiricists.ManyviewedKuhn’sseminalworkonscientificchangetohaveshownthattheprojectofinductivelogicwasmisplaced(seeKUHN);they,andothers,generallyregardedCarnap’sattempttoexplicateinductivelogictohavebeenafailure.Finally,anewschoolof“scientificrealists”attemptedtoescapeCarnap’sargumentsagainstexternalrealism.105\nSAHOTRASARKARTherecanbelittledoubtthatCarnap’sprojectoffoundinginductivelogichasfaltered.Heneverclaimedthathehadgonebeyondpreliminaryexplorationsofpos-sibilitiesandthoughtherehasbeensomeworksince,byandlarge,epistemologistsofsciencehaveabandonedthatprojectinfavoroflessrestrictiveformalisms,forinstance,thoseassociatedwithBayesianorFisherianstatistics.But,withrespecttoeveryothercasementionedinthelastparagraph,thesituationisfarlessclear.IthasalreadybeennotedthatCarnap’sfinalcriterionforcognitivesignificancedoesnotsufferfromanytechnicaldifficultynomatterwhatitsotherdemeritsmaybe.Quine’sdictaagainstana-lyticitynolongerappearaspersuasiveastheyoncedid(Stein1992);Quine’sprefer-enceforusingnatural–ratherthanformalized–languageintheanalysisofsciencehasprovedtobecounterproductive;andhisprogramofnaturalizingepistemologyisyettoliveuptoanyinitialpromisethatitevermighthavehad.Putnam’s“internalrealism”isbasedonandrevivesCarnap’sviewsonontologyandKuhnisperhapsnowbetterregardedashavingcontributedsignificantlytothesociologyratherthantheepis-temologyofscience.However,tonotethatsomeofthetraditionallyfashionableobjectionstoCarnapandlogicalempiricismcannotbesustaineddoesnotshowthatthatworkdeservesaposi-tiveassessmentonitsown.Wearestillleftwiththequestion:what,exactly,didCarnapcontribute?Theanswerturnsouttobesurprisinglystraightforward:thetextbookpictureofdeductivelogicthatwehavetodayistheonethatCarnapproducedintheearly1940safterhecametoacknowledgethepossibilityofsemantics.Thefixed-pointlemmahasturnedouttobeanimportantminorcontributiontologic.Thereintroduc-tionofmodallogicintophilosophyopenedupnewvistasforKripkeandothersinthe1950sand1960s(seeKRIPKE).Carnap’sviewsonontologycontinuetoinfluencephilosopherstoday.Moreover,eventhoughtheprojectofinductivelogicseemsunsalvageabletomostphilosophersitishardtodenythatCarnapmanagedtoclarifysignificantlythewaysinwhichconceptsofprobabilitymustbedeployedintheempiricalsciencesandwhytheproblemofinductivelogicissodifficult.But,mostofall,Carnaptookphilosophytoanewlevelofrigorandclarity,accompaniedbyanopen-mindedness(codifiedinthePrincipleofTolerance)that,unfortunately,hasnotbeen10widelysharedinanalyticphilosophy.Notes1BiographicaldetailsarefromCarnap1963a.2ReferencestotheAufbauaretosections;thispermitsthesimultaneoususeoftheGermanandEnglisheditions.3AnexcellentdiscussionofCarnap’sconstructionistobefoundinGoodman1951,ch.5.4Thus,ifanelexisbothcincolorandtintemperature,cortcanbedefinedasclassesofeveryelexhavingcortrespectively.However,tosaythatcortisaqualitywouldimplythatanelexisanalyzableintosimplerconstituents.Quasi-analysisproceedsformallyinthisway(asifitisproperanalysis)butonlydefinesquasi-characteristicsthusleavingeachelexunanalyzable.5Goodman(1951)alsoprovidesaveryluciddiscussionoftheseproblems.6SomerecentscholarshiphasquestionedwhetherCarnaphadanytraditionalepistemologi-calconcernsintheAufbau.Inparticular,Friedman(e.g.1992)haschampionedtheviewthatCarnap’sconcernsinthatworkarepurelyontological:theAufbauisnotconcernedwiththe106\nRUDOLFCARNAPquestionofthesourceorstatusofourknowledgeoftheexternalworld;rather,itinvestigatesthebasesonwhichsuchaworldmaybeconstructed.(See,also,Richardson1998.BothFriedmanandRichardson–aswellasSauer(1985)andHaack(1977)longbeforethem–emphasizetheKantianrootsoftheAufbau.)Ifthisreinterpretationiscorrect,thenwhatexactlytheAufbauowestoRussell(andtraditionalempiricism)becomesuncertain.However,asPutnam(privatecommunication)haspointedout,thisreinterpretationgoestoofar:thoughtheprojectoftheAufbauisnotidenticaltothatofRussell’sexternalworldprogram(forreasonsincludingthosethatFriedmangives),thereissufficientcongruencebetweenthetwoprojectsforCarnaptohavecorrectlybelievedthathewascarryingoutRussell’sprogram.Inparticular,theformalconstructionsoftheAufbauareanecessaryprerequisiteforthedevelopmentoftheepistemologythatRussellhadinmind:onemustbeabletoconstructtheworldformallyfromaphenomenalistbasisbeforeonecansuggestthatthisconstructionshowsthatthephenomenaarethesourceofourknowledgeoftheworld.Moreover,thisrein-terpretationignorestheepistemologicalremarksscatteredthroughouttheAufbauitself,includingCarnap’sconcernfortheepistemicprimacyofthebasishebeginswith.7Sarkar(1992)attemptsacomprehensiblereconstructionofthenotoriouslydifficultformalismofSyntax.8ThesewereeditedandpublishedbyAbnerShimony(asTwoEssaysonEntropy,1977)afterCarnap’sdeath.9SeeCarnapandJeffrey1971andJeffrey1980.AnexcellentintroductiontothispartofCarnap’sworkoninductivelogicisHilpinen1975.10ForcommentsonearlierversionsofthisessaythanksareduetoJustinGarson,CoryJuhl,AlMartinich,andItaiSher.BibliographyWorksbyCarnap1922:DerRaum,Berlin:vonReutherundReichard.1923:“ÜberdieAufgabederPhysikunddieAnwendungdesGrundsatzesderEinfachsteit,”Kant-Studien28,pp.90–107.1924:“DreidimensionalitätdesRaumesundKausalität:EineUntersuchungüberdenlogischenZusammenhangzweierFiktionen,”AnnalenderPhilosophieundphilosophischenKritik4,pp.105–30.1926:PhysikalischeBegriffsbildung,Karlsruhe:Braun.1934a:“DieAntinomienunddieUnvollständigkeitderMathematik,”MonatsheftefürMathematikundPhysik41,pp.42–8.1934b:TheUnityofScience,London:KeganPaul,Trench,TrubnerandCo.(Originallypublished1932.)1935:“EinGültigskriteriumfürdieSätzederklassischenMathematik,”MonatsheftefürMathematikundPhysik42,pp.163–90.1936:“TestabilityandMeaning,”PhilosophyofScience3,pp.419–71.1937a:“TestabilityandMeaning,”PhilosophyofScience4,pp.1–40.1937b:TheLogicalSyntaxofLanguage,London:KeganPaul,Trench,TrubnerandCo.(OriginallyLogischeSyntaxderSprache,Vienna:Springer,1934.)1939:FoundationsofLogicandMathematics,Chicago:UniversityofChicagoPress.1947:MeaningandNecessity:AStudyinSemanticsandModalLogic,Chicago:UniversityofChicagoPress.1950a:“Empiricism,Semantics,andOntology,”RevueInternationaledePhilosophie4,pp.20–40.107\nSAHOTRASARKAR1950b:LogicalFoundationsofProbability,Chicago:UniversityofChicagoPress.1952:TheContinuumofInductiveMethods,Chicago:UniversityofChicagoPress.1955:“MeaningandSynonymyinNaturalLanguages,”PhilosophicalStudies7,pp.33–47.1956:“TheMethodologicalCharacterofTheoreticalConcepts,”inTheFoundationsofScienceandtheConceptsofPsychologyandPsychoanalysis,ed.H.FeiglandM.Scriven,Minneapolis:UniversityofMinnesotaPress,pp.38–76.1961:“OntheUseofHilbert’se-OperatorinScientificTheories,”inEssaysontheFoundationsofMathematics,ed.Y.Bar-Hillel,E.I.J.Poznanski,M.O.Rabin,andA.Robinson,Jerusalem:MagnesPress,pp.156–64.1963a:“IntellectualAutobiography,”inThePhilosophyofRudolfCarnap,ed.P.A.Schilpp,LaSalle,IL:OpenCourt,pp.3–84.1963b:“RepliesandSystematicExpositions,”inThePhilosophyofRudolfCarnap,ed.P.A.Schilpp,LaSalle,IL:OpenCourt,pp.859–1013.1967:TheLogicalStructureoftheWorldandPseudoproblemsinPhilosophy,Berkeley:UniversityofCaliforniaPress.(Originallypublished1928.)1971a:“ABasicSystemofInductiveLogic,PartI,”inStudiesinInductiveLogicandProbability,ed.R.CarnapandR.C.Jeffrey,vol.1,Berkeley:UniversityofCaliforniaPress,pp.33–165.1971b:“InductiveLogicandRationalDecisions,”inStudiesinInductiveLogicandProbability,ed.R.CarnapandR.C.Jeffrey,vol.1,Berkeley:UniversityofCaliforniaPress,pp.5–31.1977:TwoEssaysonEntropy,Berkeley:UniversityofCaliforniaPress.1980:“ABasicSystemofInductiveLogic,PartII,”inStudiesinInductiveLogicandProbability,ed.R.C.Jeffrey,vol.2,Berkeley:UniversityofCaliforniaPress,pp.8–155.1971(withJeffrey,R.C.)(eds.):StudiesinInductiveLogicandProbability,vol.1,Berkeley:UniversityofCaliforniaPress.WorksbyotherauthorsAyer,A.J.(1936)Language,TruthandLogic,London:Gollancz.Church,A.(1956)IntroductiontoMathematicalLogic,vol.1,Princeton,NJ:PrincetonUniversityPress.Friedman,M.(1992)“EpistemologyintheAufbau,”Synthese93,pp.191–237.Gödel,K.(1931)“ÜberformalunentscheidbareSätzederPrincipiaMathematicaundverwandterSysteme1,”MonatsheftefürMathematikundPhysik38,pp.173–98.Goodman,N.(1951)TheStructureofExperience,Cambridge,MA:HarvardUniversityPress.Haack,S.(1977)“Carnap’sAufbau:SomeKantianReflections,”Ratio19,pp.170–5.Hempel,C.G.(1950)“ProblemsandChangesintheEmpiricistCriterionofMeaning,”RevueInternationaledePhilosophie11,pp.41–63.Hilpinen,R.(1975)“Carnap’sNewSystemofInductiveLogic,”inRudolfCarnap,LogicalEmpiricist:MaterialsandPerspectives,ed.J.Hintikka,Dordrecht:Reidel,pp.333–59.Jeffrey,R.C.(ed.)(1980)StudiesinInductiveLogicandProbability,vol.2,Berkeley:UniversityofCaliforniaPress.Kaplan,D.(1975)“SignificanceandAnalyticity:ACommentonSomeRecentProposalsofCarnap,”inRudolfCarnap,LogicalEmpiricist:MaterialsandPerspectives,ed.J.Hintikka,Dordrecht:Reidel,pp.87–94.Keynes,J.M.(1921)ATreatiseonProbability,London:Macmillan.Lewis,C.I.(1918)ASurveyofSymbolicLogic,Berkeley:UniversityofCaliforniaPress.Quine,W.V.(1951)“TwoDogmasofEmpiricism,”PhilosophicalReview60,pp.20–43.Richardson,A.(1998)Carnap’sConstructionoftheWorld,Cambridge:CambridgeUniversityPress.108\nRUDOLFCARNAPRoozeboom,W.(1960)“ANoteonCarnap’sMeaningCriterion,”PhilosophicalStudies11,pp.33–8.Sarkar,S.(1992)“‘TheBoundlessOceanofUnlimitedPossibilities’:LogicinCarnap’sLogicalSyntaxofLanguage,”Synthese93,pp.191–237.Sauer,W.(1985)“Carnap’s‘Aufbau’inKantianischerSicht,”GrazerPhilosophischeStudien23,pp.19–35.Shimony,A.(1992)“OnCarnap:ReflectionsofaMetaphysicalStudent,”Synthese93,pp.261–74.Stegmüller,W.(1972)“HomagetoRudolfCarnap,”inPSA1970,ed.R.C.BuckandR.S.Cohen,Dordrecht:Reidel,pp.lii–lxvi.Stein,H.(1992)“WasCarnapEntirelyWrong,AfterAll?,”Synthese93,pp.275–95.109\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd20017KarlPopper(1902–1994)W.H.NEWTON-SMITHBorninVienna,KarlPopperstudiedattheUniversityofViennafrom1918to1922,afterwhichhebecameapprenticedtoamastercabinetmaker,AdalbertPosch.Inhisintellectualautobiography,PopperreportedthathelearnedmoreaboutepistemologyfromPoschthanfromanyotherofhisteachers.In1925heenrolledintheCityofVienna’snewPedagogicInstitutetoworkonthepsychologyofthoughtanddiscovery.However,hisintereststurnedtomethodologyandin1928heobtainedhisdoctorateforathesisonmethodologicalproblemsinpsychology.WhileteachingmathematicsandphysicsinasecondaryschoolhewrotehisLogikderForschung,whichwaspublishedin1934,appearinginanEnglishtranslationin1959asTheLogicofScientificDiscovery.In1937hewenttoNewZealandasalecturerinphilosophyatCanterburyUniversityCollege.WhiletherehewrotehisinfluentialworksThePovertyofHistoricismandTheOpenSocietyanditsEnemies.AppointedReaderandsubsequentlyProfessorinLogicandScientificMethodattheLondonSchoolofEconomicsin1946heremainedthereforthebalanceofhisacademiccareer.Untilhisdeathin1994hecontinuedtopublishprolifically.ThedistinctivefeatureofPopper’sphilosophyofscienceishisattitudetoinduction.LikeHumeheheldthatnoinductiveinferenceiseverrationallyjustified.Findingthat1millionrandomlyselectedsamplesofsodiumburnwithayellowflameprovidesnoreasonatall,accordingtoHumeandPopper,forthinkingthatallpiecesofsodiumwillburnwithayellowflame.Whatwewouldnormallycountasevidenceforsuchahypothesisdoesnotevengivefalliblegroundsforthinkingitismoreprobablytruethanfalse.Inductivearguments,argumentsinwhichthepremisesdonotentailtheconclu-sionbutpurporttosupportit,simplyhavenorationalforce.ConsequentlyPoppersoughttorelyentirelyondeductiveargumentation.Whilewecanneverhavetheleastpositivereasonforthinkingthatahypothesisistrueorprobablytrue,wecanuseadeductiveargumenttoshowthatitisfalse.Forgiventhatwehaveobservedoneblackswanwecandeductivelyinferthatitisfalsethatallswansarewhite.ThisisthecruxofPopper’sphilosophyofscience.Itisonlytherejectionofbeliefsorhypothesesthatcanhavethesanctionofreason(butseebelow).Humeneversoughttopersuadeustoabandoninduction.Forhim,itispartofournaturetoproceedinductively.Customandhabitcarryusforwardwherereasonfails.Toputthepointanachronistically,forHumeweare“hard-wired”toinduct.Itis110\nKARLPOPPERsimplyabemusingfeatureofthehumanconditionthatourinductiveproceduresdonothavethesanctionofreason.ButforPopper,ontheotherhand,wedonotorshouldnotproceedinductively.Andheclaimsthatgoodscientistsneverdoso.ThePopperianscientist,equippedwithafertileimagination,simplymakesaboldconjectureandthebolderthebetter.Hethenseekstorefutethatconjecturebyobservationandexperi-mentation.Ifacontraryinstanceisfoundtheconjectureisfalsifiedandhencerejected.Inwhichcasethescientiststartsagainwithanewconjecture.Ifaconjectureisnotfalsifiedinatest,ithasbeen“corroborated.”Corroboration,asdefinedbyPopper,doesnotprovideanyreasonforthinkingthatthehypothesishasanylikelihoodofholdinginthefuture.Itissimplyareportthatithasnotyetfailed.Criticshavewonderedwhy,inthiscase,weshouldtrustevenahighlycorroboratedhypothesis.Clearlywedosowhen,forexample,wetrustourfatetoairplanesdesignedonthebasisofaeronauticaltheories.TheanswerforPopperisthatwehavenoreasonatalltodoso.Weproceedonblindfaith!Criticsalsoobjectthatrejectingahypothesisinthefaceofacontraryinstanceisitselfadisguisedformofinduction.Forinsodoingweareassumingthatthefuturewillbelikethepast:whatfailedonMondaywillalsofailonTuesday.Inutterlyrejectinganythingotherthandeductivejustification,Poppercommittedhimselftoaverystrongformoffallibilism,accordingtowhichnotonlycanwenothavecertainknowledgeinscienceorineverydaylife,wecanhavenopositivereasons,howeverweak,forholdingthatparticularbeliefsinscienceorineverydaylifeareevenmorelikelytobetruethanfalse.SomereadersofPoppermaywellhavefailedtoseetheextremeconsequencesofhisfallibilism.Foritappliesalsotothebeliefsweformaboutwhatweobserve.Consequently,itfollowsthatwecanhavenorationalgroundsforclaimingtohavediscoveredthatahypothesishadbeenfalsified.Forwecanhavenomorereasonforrejectingahypothesisthanwehaveforourbeliefthatwehaveobservedacounterexample.Thatbeingso,Popper’sfallibilismamountstoanextremeformofskepticism.Wecanhavenoreasonsforthinkingthatanyempiricalpropositionistrue;norcanwehaveanyreasonsforthinkingthatitisfalse.MuchofwhatPopperwrotehasplausibilityonlyifwesetasidethisextremeconsequenceofhispositionandthisIwilldoinmuchofwhatfollows.FalsificationprovidedPopperwithhiscriterionforthedemarcationofsciencefromnon-scienceorpseudo-science.Hedescribedthiscriterionastheverycenterofhisphi-losophyofscience.ImpressedpositivelybythesuccessofEinsteinandnegativelybywhathetooktobethefailureofFreudandMarx,helookedforthehallmarkofthesci-entificandthoughthehadfounditinfalsification.Atheoryisscientificjustincaseitmakespredictionsthatcouldinprinciplebeobservednottoobtain.Iftheydonotobtainthetheoryisrefuted.AccordingtoPopperpsychoanalysisrulednothingoutandhencecouldnotbefalsifiedandwasnotscientific.HeheldthatMarxismwasoriginallyfalsi-fiable.However,inthefaceofnegativeinstances,Marxismwasrevisedsoastobecomeimmunefromrefutation.InPopper’sterminologyanysuchunfalsifiabletheoriesaremetaphysicalratherthanscientific.Unfortunately,muchofwhatwecountasscienceturnsoutnottobefalsifiable.Atheory,suchasquantummechanics,whichmakesonlyprobabilisticpredictions,isnotfalsifiable.Considerthehypothesisthattheprobabilitythatthiscoinwilllandheadsonthenexttossisp.Thecoinmaylandtailsanynumberoftimeswithoutfalsifyingthathypothesis!Poppersoughttoavoidthisdifficultybyadoptingamethodologicalrulethatwouldrejectthishypothesisifaftersomenumber111\nW.H.NEWTON-SMITHNoftrialstheresultsdivergedsignificantlyfromp.CriticshavenotbeensatisfiedthatthereisanyreasonablewayoffixinganappropriatevalueforNwithouttheuseofinductiveargumentation.Popperdidcometoappreciatethattherearemetaphysicalelementsingoodscien-tifictheoriesandthatmetaphysicaltheoriessuchasDarwinismhadimportantbene-ficialinfluencesonscientificdevelopment.HavingrecognizedthisPoppershiftedsome-whatandsoughttoevaluatemetaphysicaltheoriesaswell.Thisistobedonebycon-sideringwhetherthetheorysolvesproblems,whetheritspurportedsolutionscanbeexaminedcriticallyandwhetheritsolvestheproblemsbetterthanrivaltheories.Thisinturngeneratesamoregeneraldemarcationcriterionfordistinguishingbetweenwhathereferredtoascriticizableversusnon-criticizabletheories.Itisnodoubtastepforwardtoconsiderthemeritsoftheoriesingeneralwithoutspecialregardtowhethertheyarescientificornot.Butitmeansabandoningwhatoncehadprideofplacewithinhisphilosophyofscience:ahallmarkofthescientific.EvenFreudandMarxmeettheconditionofbeingcriticizabletheories,asPopper’sownwritingsmakemanifest.Themethodofscienceistoproposeboldtheories,thebolderthebetter.Thescien-tistthenseekstorefutethem.Devotingoneselfmerelytofindingoutthattheoriesarefalsedoesnotseemaveryedifyingvocation.Iftherewereonlyafinitenumberoftheo-riesinanybranchofscience,onecouldtakecomfortinthefactthatwitheachrejec-tion,theprobabilityofthenexttheoryselectedbeingtruewouldincrease.Butunhap-pilythereareaninfinitenumberofrivaltheories.APopperianscientistexpectsthatevenhismostcherishedtheorywilleventuallybefalsified.Ifallhecaneverfindisthatatheoryisfalse,whatpositiveglosscanheputonhisscientificendeavors?ForPopperthescientisthopestohavetheories–falsetheories–thatareeverbetterapproxima-tionstothetruth.Thesearetheorieswith,inhiswords,“increasingtruthlikenessorverisimilitude.”ThemovefromNewtontoEinsteinwasprogressive,becausewhileNewtonsaidsometruethingsabouttheworldandsomefalsethings,Einsteinsaidmoretruethingsandfewerfalsethings.WecanpicturethisasNewtongettingacertainper-centageofhisclaimsrightandEinsteinscoringahigherpercentage.Somefuturesci-entistcanhopeforahigherscoreyet.TheideathattheaimofscienceisnottruthpersebutevermoreapproximatelytruetheorieshasattractedadherentsincludingmanywhorejectPopper’saccountofscientificmethod.UnfortunatelyPopper’sowntechni-caldefinitionofverisimilitudeprovedunsatisfactory.Itturnedoutthatonhisdefini-tionalltheoriesotherthantrueoneshavethesamedegreeoftruthlikeness.Popper’sapproachhasinspiredmuchfurtherworkonthisnotionbutatpresentnosatisfactoryexplicationoftruthlikenesshasbeenforthcoming.Popperisconvincedthatscienceisgeneratingevermoretruthlikefultheories.Butassomeonewhoavoidsallinductiveargumentationhehastoregardthisbeliefasirra-tional.IntuitivelywemightarguethatthefactthatEinsteinpassesmoreteststhanNewtongivesusreasonablegroundsforthinkingthatEinstein’stheoryismoretruth-likefulthanNewton’s.Butthisisaninductiveargument.Theconclusionisreasonableonlyifweassumethattheareaoftheuniversewehaveexploredtodateisarepresen-tativesampleoftheentireuniverse.PerhapsitisjustalocalpeculiaritythatEinsteinfaresbetterthanNewton.Popperhimselfistemptedbysucharguments.AtonepointheclaimedthatwecouldargueforthegreatertruthlikefulnessofEinsteinoverNewtononthegroundsthatit112\nKARLPOPPERwouldbe“ahighlyimprobablecoincidenceifatheorylikeEinstein’scouldcorrectlypredictveryprecisemeasurementsnotpredictedbyitspredecessorsunlessthereis‘sometruth’init”(Schlipp1974:1192–3).Thisargumenthasmanyadherentspar-ticularlyamongthosewhoadvocaterealismaboutscientifictheories.Butitisaformofinductionknownasinferencetothebestexplanation.WeareinvitedtoinferthatEinstein’stheoryhasmoretruthinitthanNewton’stheoryonthegroundsthatthatassumptionprovidesthebestexplanationofthegreaterpredictivesuccessofEinstein’stheory.Respectableasthisargumentmaybe,itisnotopentoonewhorejectsallbutdeductiveargumentation.OnemightfeelthatinallfairnessweshouldallowPopperjustthisonelittleinductivemove.Hehimselfconcedesinthepassagequotedabovethat“theremaybea‘whiff’ofinductivismhere.”Butifaninductivemoveislegitimatehere,whynotelsewhereaswell?Onceweallowinductionarole,Popperlosesclaimtoourattention.Forwhatmadehisphilosophyofscienceuniqueandtherebyinterestingwasitsexplicitandtotalrejectionofinduction.Butwithoutinduction,hisbeliefinscien-tificprogressisirrational.Philosophersofsciencearedividedonmanyissues.ButtheyarealmostunanimousinrejectingaPopperianaccountofscience.WhetherornotwehaveasatisfactoryanswertoHume’sskepticismaboutinduction,ittakescouragetodenythatscientistsproceedinductively.Scientistsmakeaninductivemovewhentheyconcludethatthereissomeprobabletruthinthetheoryoftheelectrononthegroundsthatthattheoryexplainswhytelevisionswork.Eventhescientistwhoconcludesmoremodestlythatthereareatleastgoodreasonstothinkthatthetheoryoftheelectronwillgivesuc-cessfulobservationalpredictionsinthefutureisassumingthelegitimacyofinduction.Popper’sfalsificationisttheoryofscienceisitselffalsifiedbyscientificpractice.Hisgrandexperimentinofferinganon-inductivisttheoryofscienceservesonlytoheightenourappreciationofthedeep-seatedcommitmenttoinduction.Popperarguedforanumberofphilosophicalpositionsquiteindependentofhisfal-sificationism.Forinstance,hewasapassionatedefenderofthefreedomofthewill.Hearguedagainstthedeterminist’sthesisthatthefutureisfixedbythepaststatesoftheuniversetogetherwiththelawsofnature.Hisstrategywastoseektoshowthatnotallfutureeventsinvolvinghumanagentscanbescientificallypredicted.However,criticshavebeenunabletoseehowaninabilityeveninprincipletopredictsomefuturehumanactionsshowsthatthoseactionsmaynotbedeterminednonetheless.Moreisneededtoestablishfreedomthananinabilitytopredict.Inadditionhehasurgedacontrover-sialthree-partontology.Hepositsaworldofphysicalobjects(“world1”),aworldofsubjectiveexperiences(“world2”),andaworldofthe“objectivecontentsofthought”(“world3”).“World3”isreminiscentoftheworldofabstractPlatonicobjectsthatsomephilosophershavefeltdriventopostulate.ButwhatisperhapsquiteuniquetoPopperisthethoughtthatthisworld,initiallycreatedbyus,takesonanautonomywherebyitactsinaquasi-causalwayontheobjectsof“world2”andevenof“world1.”Fewphilosophershavebeenwillingtofollowhiminthislavishpostulationofanunfalsifi-abletheoryofcausallyactiveabstractobjects.HistoryislikelytorememberPoppermoreasaculturalfigurethanaphilosopherinthenarrowAnglo-Saxonsenseoftheterm.ForthroughhisThePovertyofHistoricism(1944)andhisTheOpenSocietyanditsEnemies(1945),hemaywellhavehadmoreinfluenceinthesocialandculturalspheresthananyothertwentieth-century113\nW.H.NEWTON-SMITHphilosopher.HisinitialaimwastoprovideanintellectuallydecisivecritiqueofMarxandMarxism.TheresultingpolemicbecameahugelyinfluentialattackontotalitarianismingeneralandaglimpseofaninspiringifonlyvaguelysketchedUtopia:theOpenSociety.PoppersawbothfascismandCommunismasrestingonapernicioushistoricism,avisionofhistorymovinginevitablytosomefixedfinaldestination.Popper’shistoricistthinkshecandetectbyintuitiveobservationhistoricaltrendsthathemistakenlytakestobeironlawsandnotmeretrendsthatcanbereversed.PoppercharacterizesMarxism,hisparadigmofhistoricism,asbadhistoricalprophecycombinedwiththeinjunction“Helptobringabouttheinevitable!”(1976:35).EvengrantingPopper’sassumptionaboutthenegativerolethatabeliefinhistoricismhasplayedandevengrantingthecogencyofhisargumentsagainstit,hispositiondoesnotreallyaddressthegeneralproblemoftotalitarianism.Forunfortunatelytherearetotalitarianregimesthathavecomeaboutasaresultofforcesotherthanbeliefinhistoricism.Popper,incontrasttothehistoricists’policyofwaitingfortheinevitable,grand-scale,socialchange,advocates“piecemealsocialengineering.”Weshouldmakesmallexperimentaladjustmentsinoursocialinstitutions;thisheillustrateswithsuchexam-plesastheintroductionofanewsalestax.Wethenobservetheresultsofthetest;findoutourerrorsandlearnfromourmistakes.ForPopperthisisexplicitlyanextensiontothesphereofpoliticsofthescientificmethod,althoughironicallytheboldnessthatcharacterizedtheidealscientist’sconjecturesistobereplacedbycautioussmall-scaleconjectures.Thebadmovesareexposed;newonesaretriedintheirplace.Oursocialengineerisurgedtoundertakea“systematicfightagainstdefinitewrongs,againstcon-creteformsofinjusticeorexploitation,andavoidablesufferingsuchaspovertyorunemployment”(1957:91).Inthesocialandpoliticalsphere,thenotionofscientificrationalityorreasonable-nessisliberalizedtogiveawidernotionofrationalityorreasonablenessasopennesstocriticism:anattitudeofreadinesstolistentocriticalargumentsandtolearnfromexpe-rience.AcommitmenttorationalityinthissenseisanecessarycoreofPopper’slightlysketchedvisionofanopensociety.Anopensocietyisademocraticonethatpromotescriticismanddiversitywithoutrepressionorirreconcilablesocialdivisions,avoidsvio-lence,andencouragestoleration.Criticalpublicdiscussionwiththeparticipationofallisthemeanswherebythoseinanopensocietyseektoarriveataconsensusonsocialandpoliticalissues.Whilethedetailsareslight,Poppermakesclearhistouchinglynaiveviewthatthe“freeworld”isareasonableapproximationtohisidealofanopensociety,having“verynearly,ifnotcompletely,succeededinabolishingthegreatestevilswhichhavehithertobesetthesociallifeofman”(1963:370).Thecoreoftheideaofanopensocietyissupposedtolieintherecognitionofourfallibility.Popper’sapproachisreminiscentofMill,whoarguedinOnLibertythatwehaveaninterestinpromotingdiversityofopinion.Beliefsmaybeincorrect.Ifwearewrong,ourbesthopeincorrectingourselvesliesincriticaldiscussionwiththosewhodisagree.Andifweareinfactcorrect,thecriticaldiscussionwillbringoutmoreclearlythecontentsofandgroundsforourbeliefs.ButitishardtoseehowanyargumentofthischaractercangiveusPopper’sfullconceptionofanopensociety.Forinstance,aninterestincriticaldiscussionislogicallycompatiblewithamajoritymaintainingaclassofslaveswhoseroleinlifewastoprovideuswithcriticalcomment!Attractiveasthe114\nKARLPOPPERnexusofvaluesembodiedintheopensocietyare,theydonotseemderivablejustfromtherefutationofhistoricismtogetherwiththerecognitionofourfallibility.Whilesocietywouldnodoubtbenefitfrommorecriticaldialogue,itisarguablethatthesta-bilityofsocialinstitutionsrequiresthatnoteverythingisopentodebateatalltimes.Thereareserioustensionsbetweenthemethodofpiecemealsocialengineeringandtheadvocacyofanopensociety.Ontheonehand,thismethodmayonlybeviableforthosewhohavealreadyestablishedanopensociety.ThetransitionsinAlbaniaandRomania,forexample,fromaclosedtoanopensocietywerenotachievedbypiecemealsocialengineeringandprobablycouldnothavebeenachievedbyanythingotherthanrevolution.Ontheotherhand,itisnotclearthatallproblemsfacingopensocietiescanbesolvedthroughpiecemealsocialengineering.Globalizationpresentsproblemstheresolutionofwhichwillrequireinternationalregulatorymechanismsandthesewillnotcomeaboutthroughpiecemealsocialengineering.Forallthedeficienciesinargumentationandconception,Popper’srhetorichasbeenmuchusedandusedtolargelybeneficialeffect.ThePovertyofHistoricismandTheOpenSocietyweretextswidelyreadinsamizdatunderCommunismandcertainlyhadaninspirationalrolefordissidentsseekingtoopentheirsocieties.AtthesametimethesetextswerewidelyusedbywesternEuropeansocialdemocraticpartiesofthelefttodis-tancethemselvesfromtheCommunistPartiesofEurope.Progresstosocialequalitywastobeachievedbypiecemealsocialengineering,notrevolutionarychange.AndinChinaintheperiodjustbeforeTiananmenSquaretheliberalwingoftheCommunistPartyusedPopperexplicitlyintheiranalysisofthemistakesoftheCulturalRevolutionandinadvocatingopen,criticaldiscussionofsocialissues.Popperisnodoubtrightinencouragingmorecriticaldiscussionofsocialandpoliticalissues.Unfortunatelyhisownphilosophicalsystem,withwhichheseekstounderpinthisencouragement,islimited.Hisutterrejectionofinduction,hisfallibilism,severelyrestrictsthescopeofrationalcriticism.Giventhatonlydeductiveargumentationhasrationalforce,theonlyintellectuallyjustifiedcriticismthatcanbemadeofanypositioninscienceorinsocietyisthatthepositionislogicallyincoherent.ButsuchalimitedformofcriticismisunlikelygreatlytoassistinthesolutionofthesocialillsthatconcernedPopper.BibliographyWorksbyPopper1945:TheOpenSocietyanditsEnemies,London:RoutledgeandKeganPaul.1957:ThePovertyofHistoricism,London:RoutledgeandKeganPaul.1959:TheLogicofScientificDiscovery,London:Hutchinson.1963:ConjecturesandRefutations,London:RoutledgeandKeganPaul.1972:ObjectiveKnowledge,Oxford:OxfordUniversityPress.1976:UnendedQuest:AnIntellectualAutobiography,London:Collins.WorksbyotherauthorsNewton-Smith,W.H.(1981)TheRationalityofScience,London:RoutledgeandKeganPaul,ch.III.O’Hear,A.(1980)KarlPopper,London:RoutledgeandKeganPaul.115\nW.H.NEWTON-SMITHO’Hear,A.(ed.)(1995)KarlPopper:PhilosophyandProblems,Cambridge:CambridgeUniversityPress.Raphael,F.(1998)Popper,London:Phoenix.Schilpp,P.A.(ed.)(1974)ThePhilosophyofKarlPopper,LaSalle,IL:OpenCourt.116\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd20018GilbertRyle(1900–1976)AVRUMSTROLLGilbertRyleandhisjuniorcolleague,J.L.Austin,weretheleadingfiguresofpost-WorldWarIIOxfordphilosophy.Thoughtheiraimsandmethodsweredifferent(seebelow),botharecorrectlycharacterizedas“ordinarylanguagephilosophers.”UnlikeAustin,whopublishedonlysevenpapersinhislifetime,Rylewasaprolificwriter.Muchofwhatweknowabouthispersonallifederivesfromself-referencesinhisnumerousbiographi-calsketchesandreviews,andespeciallyfromhisautobiography.Inthesevariousessayshedescribeshisinteractionswith,andassessmentsof,theforemostphilosophersofthetime,amongthemWittgenstein,Moore,Collingwood,Carnap,Prichard,H.H.Price,andAustin.HisautobiographyistobefoundinRyle,editedbyO.P.WoodandG.Pitcher(1970).Althoughitisonlyfifteenpageslong,itiswittilyself-deprecating,devastatinginitsdepictionofthestateofphilosophyinOxfordinthe1920sand1930s,packedwithinformation,andinstructivewithrespecttohisphilosophicaldevelopment.AboutOxfordphilosophyhesays:Duringmytimeasanundergraduateandduringmyfirstyearsasateacher,thephilo-sophicalkettleinOxfordwasbarelylukewarm.IthinkthatitwouldhavebeenstonecoldbutforPrichard,whodidbringintohischosenandrathernarrowarenasvehemence,tenacity,unceremoniousness,andaperverseconsistencythatmadeourhacklesriseasnothingelseatthattimedid.TheBradleianswerenotyetextinct,buttheydidnotcomeoutintotheopen.IcannotrecollecthearingonereferringmentionoftheAbsolute.TheCookWilsonianswerehankeringtogainsaytheBradleiansandtheCroceans,butweregivenfewopenings.PragmatismwasstillrepresentedbyF.C.S.Schiller,butashistaste-lessjocositiesbeatvainlyagainstthesnubbingprimnessesofhiscolleagues,eventhispunysparkwaseffectuallyquenched...SoonOxford’shermeticallyconservedatmospherebegantosmellstuffyeventoourselves.Abouthimselfhestatesthatinhismid-twentieshedecidedthatphilosophyessen-tiallyinvolvesargumentation,andthereforethat“thetheoryandtechnology”ofrea-soningneededtobestudiedbyanywould-bephilosopher.SincenothingofthatsortwasavailableinOxfordhe“wentallCambridge,”andseriouslybegantostudyRussell;but,ashefranklyadmits,withmarginalqualifications:Havingnomathematicalability,equipmentorinterest,Ididnotmakemyselfevencom-petentinthealgebraoflogic;nordidtheproblemofthefoundationsofmathematics117\nAVRUMSTROLLbecomeaquestionthatburnedinmybelly.MyinterestwasinthetheoryofMeanings–horridsubstantive!–andquitesoon,Iamgladtosayinthetheoryofitsseniorpartner,Nonsense.Ilaboureduponthedoublets–SenseandReference,IntensionandExtension,ConceptandObject,PropositionsandConstituents,ObjectivesandObjects,FactsandThings,FormalConceptsandRealConcepts,ProperNamesandDescriptions,andSubjectsandPredicates.ItwasinRussell’sPrinciplesofMathematicsandnotinhisPrincipiaMathematica,inhisMeinongarticlesandhis“OnDenoting,”thatIfoundthepack-iceoflogicaltheorycracking.ItwasupthesecracksthatWittgensteinsteeredhisTractatus.Hisinterestsinthetheoriesofmeaningandreferenceweretodominatetheremain-derofhiscareer,anddifferentiatehisversionofordinarylanguagephilosophyfromAustin’s.Austin’smainconcernswereintheutterancesthatconstitutepromises,warnings,recommendations,admonishments,counsels,andcommands,–i.e.inso-called“speechacts”(seeAUSTIN)–whereasRylesawhistaskasthatofdistinguishinglocutionsthatmakesensefromthosethatdonot.InasuccinctpassageRyleexplainsthedifferencebetweenhistaskandAustin’s.Anexaminermightposetwoquestions:(1)WhycannotatravellerreachLondongradually?(2)Whyis“Iwarnyou...”thebeginningofawarning,but“Iinsultyou”notthebeginningofaninsult?OnsixdaysoutofsevenQuestion1wouldbeRyle’sfavourite;Question2,Austin’s.Eachofuswouldthink–wrongly–thatthereisnotmuchrealmeatintheunfavouredquestion.Buttheirmeatsareofsuchentirelydisparatekindsthattheepithet“linguistic”wouldapplyintotallydifferentways(1)totheanswer-sketch,“Adverbslike‘gradually’won’tgowithverbslike‘reach’forthefollowingreason...”;(2)totheanswer-sketch“Toinsultistosaytosomeoneelsepejorativethingswithsuchandsuchanintention,whiletowarnistosay...”Anti-nonsenserulesgovernimpartiallysayingsofalltypes.“Reachgradually”willnotdoinquestions,commands,counsels,requests,warnings,complaints,promises,insults,orapologies,anymorethanitwilldoinstatements.Epimenidescanteaseusinanygrammaticalmood.Toanenquiryintocategorialrequirements,referencestodifferencesofsaying-typeareirrelevant;toanenquiryintodifferencesbetweensaying-types,referencestocategory-requirementsareirrelevant.Infelicitiesandabsurditiesarenotevencongeners.AsRylepointsoutthesedifferentapproacheswerenotincompetition,butratherrepresentedtwoparallelpathsthat“informalphilosophy”couldlegitimatelytakeindealingwithphilosophicalproblems.Amongthosewhoemphasizedthesense/non-sensedistinctionwereWittgenstein,Moore,J.T.Wisdom,O.K.Bouwsma,andNormanMalcolm.Austin’sfocusonspeechactswaslatertoinfluencetheworkofPaulGrice,ZenoVendler,JohnSearle,andA.P.Martinich.And,ofcourse,therearemanyphiloso-phers,includingRyleandAustinandsomeofthosejustmentioned,inwhichbothapproachesplayconcurrentroles.Inthetwentyyearsbetween1927and1947,Rylehadpublishedmorethanthirtyarticles,reviews,andcriticalnotices,butnobooks.HisfirstventureintothislargerformatwasTheConceptofMind(1949).Apartfromcollectionsofhisessayshewastopublishonlytwootherbooksinhislifetime,Dilemmasin1954,andPlato’sProgressin1966.Intheformerbook,Rylediscussessixtensions(dilemmas)thatarenotcounter-118\nGILBERTRYLEvailingformaltheoriesbutratheropposing“platitudes.”Eachisananalogueofaclas-sicalphilosophicalperplexity,suchasthefreewillproblem.Thus,“Incardgamesandattheroulette-tableitiseasytosubsideintotheframeofmindoffancyingthatourfortunesareinsomewayprearranged,wellthoughweknowthatitissillytofancythis.”Ryleshowsbyasubtle,piecemealanalysisofthelinguisticidiomsinwhichtheopposingplatitudesareframedhowtheapparentdilemmaisfactitiousandcanbedis-solved.Plato’sProgressisanentirelydifferentkindofbook.ItisahistoricalanalysisinwhichRyletriestogiveadifferentportrayalofPlato’scareer.Itisaprovocativetrea-tisethatquestionsthecommonviewthatAristotlewasPlato’spupil,andthatgivesnewdatingstothePlatonicdialogues.Thoughthesemonographsareexcitingpieces,andwellworthseriousstudy,theydonotmatchthepoweranddepthoftheConceptofMind.Ithastwoaspects:anegative,deflationaryoneandapositive,constructiveone.Thetwoapproachesaretiedtogetherbyanattackonacertainpictureofthehumanmindanditsrelationshiptothehumanbody.Rylegivesdifferentnamestothispicture:Hecallsitthe“OfficialDoctrine,”the“CartesianModel,”“Descartes’Myth,”the“GhostintheMachine,”andthe“Para-MechanicalHypothesis.”Thenegativeattackistoshowthatthispictureisincoherent;thepositivecontributionistogiveanaccurateaccount(notapicture)oftherelation-shipbetweenmindandbody.Thepositiveaccountisdetailed.Itdealswiththeentirerangeofthemental:thewill,knowing,emotions,dispositionsandoccurrences,self-knowledge,sensation,observation,imagination,andtheintellect.Thebookisthusatreasure-houseofdetaileddescriptionsofallthemajorfeaturesofmentation.WhatistheOfficialDoctrineheisouttodestroy?Thisdoctrine,hecontends,isgivenitscanonicalformulationbyDescartes,butitsantecedentsaremucholder.Itiswidelyacceptedbyphilosophers,psychologists,religiousteachers,andmanyordinaryper-sons.Itholdsthateveryhumanbeingisbothamindandabodythatareordinarilyharnessedtogether,butthatafterthedeathofthebodythemindmaycontinuetoexistandfunction.Humanbodiesareinspaceandaresubjecttothemechanicallawsofphysics,chemistry,andbiology.Thebodyisapublicobjectandcanbeinspectedbyexternalobservers.Butmindsareimmaterial,andarenotinspace,noraretheiroperationssubjecttomechanicallaws.Themindisanentity,tobesure,butanimma-terialandinvisibleonethatinhabitsamechanicalbody.ThisiswhyRylecallsitthe“ghostinthemachine.”ItisrescogitansinDescartes’parlance.Itisthethingthatthinks,deliberates,decides,wills,andopines.Eachmindisprivate,i.e.onlyeachpersoncantakedirectcognizanceofthestatesandprocessesofhisorherownmind.Apersonthuslivesthroughtwocollateralhistories;oneconsistingofwhathappenstohisbody,theothertowhathappenswithinhismind.Thefirstispublic,thesecondprivate.TheCartesianpicturethusdependsontheinternal/externaldistinction.Thisleadstotheproblemofhowthemindinfluencesbodilyaction.Sincethemindiscon-struedasnonphysicalandnonspatialhowdoesone’sactofwill,say,leadtoamove-mentofone’slegs,i.e.tothesortofthingcalledwalking,forinstance?Moreover,howarewetoaccountfortheknowledgewepresumewehaveofthemindsofothers?IftheCartesianmodeliscorrect,observerscannotknowwhatistakingplaceinthemindofanother,sincetheyareinprinciplecutofffromanysortofdirectcognitiveawarenessofthatperson’smentalstatesorprocesses.Theonlydirectknowledgeanyhumanhasisofhisorherownmentalfunctions.119\nAVRUMSTROLLAsplausibleasthisviewmayseem,itisabsurdaccordingtoRyle.Itisonebigmistakeandamistakeofaspecialkindthathecallsa“categorymistake.”Toillustratewhathemeansbya“categorymistake,”Ryleoffersthreeexamples.Hereisanabbreviatedversionofthefirstofthese:AforeignervisitingOxfordorCambridgeforthefirsttimeisshownanumberofcolleges,libraries,playingfields,museums,scientificdepartmentsandadministrativeoffices.Hethenasks“ButwhereistheUniversity?IhaveseenwherethemembersoftheCollegeslive,wheretheRegistrarworks,wherethescientistsexperimentandtherest.ButIhavenotyetseentheUniversityinwhichresideandworkthemembersofyourUniversity.”...HismistakelayinhisinnocentassumptionthatitwascorrecttospeakofChristChurch,theBodleianLibrary,theAshmoleanMuseumandtheUniversity,tospeak,thatis,asif“theUniversity”stoodforanextramemberoftheclassofwhichtheseotherunitsaremembers.HewasmistakenlyallocatingtheUniversitytothesamecategoryasthattowhichtheotherinstitutionsbelong.(1949:16)Ryle’spointisthatthissortofmistakeismadebypeoplewhodonotknowhowtoemploytheconceptofauniversity.Thatis,theirpuzzlearisesfromaninabilitycorrectlytousecertainitemsintheEnglishvocabulary.AccordingtoRyle,theOfficialDoctrinearisesfromacategorymistakeanalogoustothepreceding.Itassumesthatmindsbelongtothesamecategoryasbodiesinthesensethatbotharerigidlygovernedbydeterministiclaws.Thehumanbodyworksaccordingtomechanicalprinciples:theheartisapump,theveinsarepipes,andtheflowofbloodisdeterminedbythepres-suresthataredescribedinfluidmechanics.Thesystemisthusanassemblageofinter-actingpartsthatconsistoffluids,solids,andelectricalforces,allofwhichoperateaccordingtothelawsofmechanics.Alltheseforcesusuallyworktosomedesiredend,suchasmovingbloodfromonepartofthebodytoanother.Mindsalsoworkinanalogousways.WhenIamhungry,amentalstate,adesire,actsonmybodyandinitiatesthosemovementsofhandsandfingersthatallowmetopickupandtransferfoodtomymouth.Accordinglymindsmustbegovernedbydetermin-isticlaws.Butmindsarenonmaterial.Theyarenotcomposedofsolids,fluids,andelectricalforces.Sotheirlaws,thoughdeterministic,arenon-mechanical.TheseRylecalls“para-mechanical.”TheOfficialDoctrineinvokesthemastheanaloguesofthemechanicallawsthatgovernthebehaviorofphysicalentities.Buttheconceptofapara-mechanicallawisabsurd.Therearenosuchthingsasimmateriallevers,valves,andpumps.Valves,levers,andpumpsaresolidentitiesthatoperatetoeffectphysicalmovements.Toinvoketheimmaterialanaloguesofsuchentitiestoexplainmentalactivityisthustomakeacategorymistake,i.e.toapplytheconceptsofmechanicalforcesandlawstoadomainwheretheyhavenogrip.Themistakearisesbecausephilosophersdonotknowhowtoemploytheordinaryepithetsweusefordescribingmentalactivity.Philosophersarethuslikethepersonwhodoesnotknowhowtoemploytheconceptofauniversity.Itisthispara-mechanicalmodelthatRyleattacksinhisbook.Itsexistenceindicatesthatthesetheoristsdonotknowhowtowieldthesetofconceptsthatcharacterizeourmentalfunctions.ThealternativeheofferstotheOfficialDoctrineisadetaileddescriptionofhowmentalconceptsareusedineverydaylife.Ashesays:“Thephilosophicalarguments120\nGILBERTRYLEwhichconstitutethisbookareintendednottoincreasewhatweknowaboutmindsbuttorectifythelogicalgeographyoftheknowledgewhichwealreadypossess.”Ryleisthusremindingusofwhatwehavealwaysknown,andalsoremindingushowphilo-sophicalconceitscanblindustothefamiliar.Hisdescriptionofthe“logicalgeography”ofmentalconceptsisthusareminderofhowweemploytheseconceptswhenwearenotdoingphilosophy.Sinceanysuchemploymentisenormouslycomplex,its“logicalgeography”willbelengthy,detailed,andspecific.Here,bywayofillustration,isasegmentofamuchlongerspecimenoflogicalgeography:ItistruethatthecobblercannotwitnessthetweaksthatIfeelwhentheshoepinches.ButitisfalsethatIwitnessthem.ThereasonwhymytweakscannotbewitnessedbyhimisnotthatsomeIronCurtainpreventsthemfrombeingwitnessedbyanyonesavemyself,butthattheyarenotthesortsofthingsofwhichitmakessensetosaythattheyarewit-nessedorunwitnessedatall,evenbyme.Ifeelorhavethetweaks,butIdonotdiscoverorpeeratthem;theyarenotthingsthatIfindoutaboutbywatchingthem,listeningtothem,orsavouringthem.Inthesenseinwhichapersonmaybesaidtohavehadarobinunderobservation,itwouldbenonsensetosaythathehashadatwingeunderobservation.Theremaybeoneorseveralwitnessesofaroad-accident;therecannotbeseveralwitnesses,orevenonewitness,ofaqualm.(1949:205)ThispassageisagoodexampleofRyle’swayofexorcizingtheghostinthemachine.TheOfficialDoctrinepresupposesthatonehasprivilegedaccesstoaprivaterealmconsistingofone’sownsensations,thoughts,andmentalstates;andthatsuchanaccessconsistsintheobservationofone’ssensationsandstates.Buttosaythatoneisobservingsomethingimpliesthatoneisusingone’seyes,orcertainkindsofobservationalaidssuchtelescopes,stethoscopes,andtorches.One’seyes,andtheseinstruments,canbeusedfortheobservationofplanets,heart-beats,andmoths.Butwedonotknowwhatitwouldbeliketoapplythemtofeltsensationsortoassertseriouslythatwe“observeourpains.”SincetheOfficialDoctrinepresupposesthereissuchapara-mechanicalanalogueasobserving,itcanbeshowntobeaspeciesofnon-sensebycomparingitsrequirementswithouractualuseofsuchmentalconceptsas“tweaks”and“qualms.”Whatthecomparisonrevealsisacategorymistake.Theconceptofobservationappliestothephysicaldomaininawayitlogicallycannotapplytothemental.JustasonelogicallycannotreachLondongradually,soonecannot“observe”one’sachesandpains.Ryle’slineofreasoningthroughouttheworkisthustoshowthattheoristshaveincorrectlywieldedtheordinaryconceptsthatdescribehumanmentallife.TheConceptofMindcreatedasensationwhenitappearedin1949.ForatleastadecadeafteritspublicationitwasthesinglemostdiscussedbookinAnglo-Americanphilosophy.Nearlyeveryperiodicalcarriedlongarticlesaboutit.Itwastranslatedintoahostofforeignlanguages,wastaughtinvirtuallyeverymajorwesternuniversity,andwithinashorttimeseeminglyhadachievedthestatusofaphilosophicalclassic.Yetadecadelaterithadfallenintoobscurity,andsubsequentlyithashardlybeenreferredtoatall.Whathappenedtooccasionsuchacollapse?Itisespeciallypuzzlinggiventhatthebookwasofsuperbphilosophicalquality,waselegantlywritten,introducedmanyoriginalandpowerfuldistinctions,andwasthefirststudytoshowindetailhowthe121\nAVRUMSTROLLphilosophyoflanguageandthephilosophyofmindaretiedtogether.Inthislastrespect,itwasabellwetherforworkthatwastobedevelopedthirtyyearslater.Thereareseveralpossibilitiestoexplainwhathappened.OnefactoristhatfouryearslaterWittgenstein’sposthumousPhilosophicalInvestigationsappeared(seeWITTGEN-STEIN).ItcoveredmuchthesameterritoryasRyle’sstudyandingreaterdepth.Asbril-liantasRyle’sbookwasitpaledincomparisontothepowerandinsightofWittgenstein’s.SophilosophersturnedfromRyletoWittgenstein.Itwasthelatterandnottheformerwhowasnowread:Rylehadsimplygoneoutoffashion.Thereisasecondfactor.Ryleclaimedthatinthisworkhewas“chartingthelogicalgeography”ofthemanyconceptsusedinspeakingaboutthehumanmind.Andthoughthiswasclearlyanaptdescriptionitwasalsopatentthathisworkhadastrongverificationistthrust.Rylefrequentlyandincrucialpassagesspeaksaboutthetestabil-ityofpropositionsaboutmentalconcepts.Forexample,hestates:“For,roughly,themindisnotthetopicofsetsofuntestablecategoricalpropositions,butthetopicofsetsoftestablehypotheticalandsemi-hypotheticalpropositions”(p.46).SomecriticshavethusemphasizedthatRyle’saimistocorrectwhatotherphilosophershavesaidaboutthemethodsofverifyingstatementsinvolvingmentalconcepts,ratherthantryingtoexplicatetheseconceptsthemselves.Thepositivists,ofcourse,identifiedthemeaningofastatementwiththemethodofitsverification,andinmanyplacesintheConceptofMindRyleseemstopresupposethatindescribinghowcertainpropositionsinvolvingmentalconceptsaretobetestedheisexplicatingthemeaningofthoseconcepts.Thebookwasthuseventuallyassessedasasophisticatedformoflogicalpositivism,aviewwhichhadlostitsinfluencebythe1950s.Ryle’sworkwassweptawaywiththerestofthismovement.Itsbehaviorismwasathirdfactor.Rylestatesthattogivereasonsforacceptingorrejectingstatementscontainingmentalconceptswillalwaysinvolvehypotheticalstate-mentsaboutovertbehavior.Inrespondingtothequestion,“Whatknowledgecanonepersongetoftheworkingsofanothermind?”Ryleanswersthatitis“howweestab-lish,andhowweapply,certainsortsoflaw-likepropositionsabouttheovertandthesilentbehaviorofpersons.Icometoappreciatetheskillandtacticsofachessplayerbywatchinghimandothersplayingchess”(p.169).AlthoughRylealwaysdeniedthathewasreducingmindtobehavior,andassertedinsteadthatchartingthe“logicalgeog-raphy”ofmentalconceptswasaphilosophicallyneutralendeavor,hisdetailedanaly-sesseemedtomanyphilosopherstoleaveoutonefundamentalcharacteristicofthemind,theinward,feltqualityofmentalexperience.Forthesephilosopherssuchmentalactivitiesasdeliberatingorconjecturing,orsuchstatesasbeinginpain,weredistinctfrombehavior.Onecould,forexample,beinpainwithoutevincingitinanymodeofbehavior.Andevenifoneweretoevinceit,thepainitselfwasnottobeidentifiedwiththebehaviorinquestion.Apainisnotagrimace.SoevenifRylewerecorrectinarguingthatmentalactivitywasexercisedinvariousintersubjectivesituationsitdidnotfollowthatthebehaviorsoexhibitedwasidenticalwiththementaleventsinquestion.UnlikeRyle,whominimizedinternalexperience,Wittgensteinemphasizedandacknowledgedtheexistenceofsuchphenomena.Hispointwasthatoneshouldnotidentifythemwithsuchfeaturesasmeaning,expecting,thinking,andsoforth.AndthispositionwasseentobemorecompellingthanRyle’s.IntheendthismayhavebeenthedecisivefactorintheeclipseofRyle’sreputation.122\nGILBERTRYLEBibliographyWorksbyRyle1931–2:“SystematicallyMisleadingExpressions,”ProceedingsoftheAristotelianSociety32,pp.139–70.1945:“PhilosophicalArguments,”InauguralLectureasWaynfleteProfessorofMetaphysicalPhilosophy,Oxford.(ReprintedinRyle1971,vol.2,pp.194–211.)1945–6:“KnowingHowandKnowingThat,”ProceedingsoftheAristotelianSociety46,pp.1–16.1949:TheConceptofMind,London:Hutchinson.1950–1:“Heterologicality,”Analysis11,pp.61–9.1954:Dilemmas,TheTarnerLectures,Cambridge:CambridgeUniversityPress.1961:“Use,Usage,andMeaning,”ProceedingsoftheAristotelianSociety,supp.vol.35,pp.223–30.1966:Plato’sProgress,Cambridge:CambridgeUniversityPress.1971:CollectedPapers,vols.1and2,London:Hutchinson.WorksbyotherauthorsLyons,W.E.(1980)GilbertRyle:AnIntroductiontohisPhilosophy,Brighton:Harvester.Wood,O.P.andPitcher,G.(eds.)(1970)Ryle,GardenCity,NY:AnchorBooks.123\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd20019AlfredTarski(1901–1983),AlonzoChurch(1903–1995),andKurtGödel(1906–1978)C.ANTHONYANDERSONAlfredTarskiTarski,borninPoland,receivedhisdoctorateattheUniversityofWarsawunderStanislawLesniewski.In1942,hewasgivenapositionwiththeDepartmentofMathematicsattheUniversityofCaliforniaatBerkeley,wherehetaughtuntil1968.UndoubtedlyTarski’smostimportantphilosophicalcontributionishisfamous“semantical”definitionoftruth.Traditionalattemptstodefinetruthdidnotusethisterminologyanditisnoteasytogiveaprecisecharacterizationoftheidea.Theunder-lyingconceptionisthatsemanticsconcernsmeaningasarelationbetweenalinguisticexpressionandwhatitexpresses,represents,orstandsfor.Thus“denotes,”“desig-nates,”“names,”and“refersto”aresemanticalterms,asis“expresses.”Theterm“sat-isfies”islessfamiliarbutalsoplausiblybelongsinthiscategory.Forexample,the2number2issaidtosatisfytheequation“x=4,”andbyanalogywemightsaythatAristotlesatisfies(orsatisfied)theformula“xisastudentofPlato.”Itisnotquiteobviousthatthereisameaningof“true”whichmakesitasemanti-calterm.Ifwethinkoftruthasapropertyofsentences,asdistinguishedfromthemoretraditionalconceptionofitasapropertyofbeliefsorpropositions,itturnsouttobecloselyrelatedtosatisfaction.Infact,Tarskifoundthathecoulddefinetruthinthissenseintermsofsatisfaction.ThegoalwhichTarskisethimself(Tarski1944,Woodger1956)wastofinda“mate-riallyadequate”andformallycorrectdefinitionoftheconceptoftruthasitappliestosentences.Tobemateriallyadequateadefinitionmust“catchholdoftheactualmeaningofanoldnotion,”ratherthanmerely“specify[ing]themeaningofafamiliarwordusedtodenoteanovelnotion”(Woodger1956:341).Again,indiscussingthematerialadequacyofsomeofhisotherdefinitions,Tarskiwrites,“Nowthequestionarisesofwhetherthedefinitionsjustconstructed(theformalrigourofwhichraisesnoobjection)arealsoadequatematerially;inotherwordsdotheyinfactgraspthecurrentmeaningofthenotionasitisknownintuitively?”(Woodger1956:128–9).Todeterminewhetherornotaproposeddefinitionofacertainconceptismateriallyadequate,Tarksithinksthatwemustfirstformulateacriterionofmaterialadequacyforsuchadefinition:apreciseconditionwhichthedefinitionmustmeetandwhichwillguaranteethatthedefinednotionisfaithfultotheoriginalintuitiveconception.Of124\nALFREDTARSKI,ALONZOCHURCH,ANDKURTGÖDELcourse,whetheraproposedconditionreallyguaranteessufficientconformitytotheoldnotionissubjecttocriticalreview.Therequirementofformalcorrectnessmeansthattheproposeddefinitionmustbenon-circularandthatitmustmeetotherlogicalconstraintsonacceptabledefinitions.Oneofthetraditionalrequirementsisthatadefinitionmustnotdefinesomethingintermsofthingswhicharelessclearthanit.Tarksievenmaintainsthatitmustbespeci-fiedwhichpreviouslyadoptedtermsaretobeusedingivingthedefinitionandrequiresthattheformalstructureofthelanguageinwhichthedefinitionistobegivenbepre-ciselydescribed.Thesearerigorousconstraints.Themotivatingideaseemstobethatonlyundersuchconditionscanwehopetoprovethematerialadequacyandformalcorrectnessofadefinitionoftruth.TarskiproposesasacriterionofmaterialadequacyforadefinitionoftruththatthedefinitionshallhaveaslogicalconsequencesallinstancesofSchema(T):(T)Xistrueifandonlyifp,where“X”isreplacedbyanameofanarbitrarysentenceofthelanguageinquestionand“p”isreplacedbythatverysentence(orbyasentencewithexactlythesamemeaning).Thenameinquestionmustbeaquotation-marknameoratleastanamewhichnecessarilydesignatesthesentence.AnappropriateinstanceofSchema(T)isthussuchathingas:(S)“Snowiswhite”istrueifandonlyifsnowiswhite.Ontheleft-handsideofthis“ifandonlyif”thereoccursanameofacertainsentence–whichnameisconstructedbyenclosingthesentenceinquestioninquotationmarks.Thenusingthatnametomentionthesentence,thepropertyofbeingtrueispredicatedofthesentence.Ontheright-handsideoftheequivalence,theverysentence,whichisnamedontheleftandsaidtheretobetrue,isused.Thethingmayappeartobeatrivi-alityandperhapsthatisalltothegood.Thecondition,afterall,issupposedtoconstrainanadequatedefinitioninsuchawaythatsatisfyingthisconditionguaranteesthatthedefinitioncatchesholdoftheactualmeaningoftheterm“true.”NotecarefullythatSchema(T)isnotTarski’sdefinitionoftruth.ThatadefinitionshouldimplyallinstancesofSchema(T)isthecriterionofadequacyforthedefinition.ButTarskidoesseemtothinkthatalltheinstancesof(T)togethercompletelycapturethemeaningof“true.”Ifwecouldformaninfiniteconjunction,connectingalltheinstanceswith“and,”wewouldhaveacompletespecificationofthesemanticalcon-ceptionoftruth.Thisisnotanacceptableprocedureaccordingtotheusualrulesofdefinition,butacorrectdefinitionwouldbeobtainedifwecouldsomehowachievethesameeffect.Nowtheconditionswhichhavealreadybeengivenforanacceptabledefinitionoftruthrequirethatthelanguageinvolvedbespecifiedquiteprecisely.Naturallanguagesdonothave,oratleastwedonotknow,ruleswhichdetermineexactlywhatitsexpres-sionsare;forexample,thesentencesofEnglisharenotpreciselyspecified.Ifweignorethisandsetasourtasktogiveadefinitionoftruthforanaturallanguage,sayEnglish,weencounteraparadox.NopredicateofasufficientlyexpressivelanguagesuchasEnglishcanhavethepropertythatitvalidateseveryinstanceofSchema(T).Andthis125\nC.ANTHONYANDERSONissowhetherthepredicateisdefinedornot.Theproofofthisappealstotheinfamousliarantinomy(orparadox).Inaverysimpleversiontheantinomy(something“con-trarytolaw”)goeslikethis.Consider(A)Aisnottrue.Thatis,considerthesentence“Aisnottrue,”whichsentencewehavedecidedtoname“A.”NowSchema(T)implies:(1)“Aisnottrue”istrueifandonlyifAisnottrue.ButobservingthatthesentenceAistheverysentence“Aisnottrue”,wemayassert:(2)A=“Aisnottrue.”Iftwothingsareidentical,thentheyshareallthesameproperties.So,substitutingtheleft-handsideof(2)fortheright-handsidein(1),weget:(3)AistrueifandonlyifAisnottrue.Inthepropositionalcalculus,thishastheform:(4)P∫~P,“Pifandonlyifnot-P”andthisisequivalenttotheexplicitcontradiction:(5)P&~P,“Pandnot-P.”Somethingmustgive.Ifweareunwillingtogiveuptheusuallawsoflogic,since(2)isundeniable,itappearsthatwemustalterormodifySchema(T),ourcriterionallegedlydeterminedbytheverymeaningof“true.”Tarskiconcludes,somewhathastily,thatordinarylanguageisinconsistent.TheconceptoftruthmustconformtoSchema(T),butifwehavesuchsentencesasA,wearriveatacontradiction.Theproblem,saysTarski,isthatnaturallanguagesareseman-ticallyclosed,thatis,theycontainwithinthemselvesthetermsandmachineryfordoingtheirownsemantics.Forexample“istrueinEnglish”isitselfapredicateofEnglish.Wemust,hesays,giveourdefinitionoftruthinametalanguageforthelanguagewhosesen-tencesareinquestion.Ametalanguageisalanguagewhichwemayusetotalkaboutanotherlanguage.Forexample,inabookwritteninEnglishwhichexplainsthegrammarandmeaningoftheGermanlanguage,themetalanguageisEnglish.Thelan-guagebeingstudiediscalledtheobjectlanguage:inthecaseofthisexample,German.Further,claimsTarski,wemustconfineourattentiontoformalizedlanguageswhich,unlikenaturallanguage,neednotbesemanticallyclosedandwhichareotherwisepreciselyspecified.Withtheseprovisos,Tarskiproceedstoshowthatdefinitionsoftruthcanbegivenforobjectlanguageswhichdonotcontainsemanticalterms.Hismethodofdefi-nitionhasthestrikingqualitythatthedefinition,giveninametalanguage,doesnotitselfuseanysemanticalterms.Becauseoftheliarantinomyandotherconundrumsinvolvingsemanticalnotions,Tarskiconsidereditimportanttogivethedefinitioninsuchawaythatnosemanticaltermsarepresupposedasprimitiveorunderstoodwithoutdefinition.126\nALFREDTARSKI,ALONZOCHURCH,ANDKURTGÖDELToseehowthedefinitionwouldbegivenforaverysimpleformalizedlanguage,letussupposethatwehavejusttwopredicates:“R,”meaningisred,and“S,”meaningissquare.Inaddition,supposethatthelanguagecontainsavariablex;asignfornega-tion,say“-”;forconjunction,“&”;andanotationforuniversalquantification,“"”meaningForevery.Thus,forexample,wecanwrite“"x-S(x)”for“Foreveryobjectx,xisnotsquare”or,morenaturally,“Nothingissquare.”Weassumethatourmetalanguagecontainsthemeansofexpressingatleasttheverysamenotionsastheobjectlanguage.HereweareusingabitofEnglishasmeta-languagesothatwehavethewords“isred”tomeanthesameasthepredicate“R”inthesimpleformalizedlanguage.Nowletsomedomainofobjectsbeselectedasthecol-lectionofthingswewillbetalkingabout.Onecanthendefinesatisfactionfortheobjectlanguage:(1)Anobjectsatisfies“R(x)”ifandonlyifitisred.(2)Anobjectsatisfies“S(x)”ifandonlyifitissquare.jk(3)Anobjectsatisfiesanegation-fifandonlyifitdoesnotsatisfyf.jk(4)Anobjectsatisfiesaconjunctionoftheformf&yifandonlyifitsatisfiesfanditsatisfiesy.jk(5)Anobjectsatisfiesauniversalquantification"¥fifandonlyifeveryobject(inthedomain)satisfiesf.Herefandyareformulaeoftheformalizedlanguage.Theseareexpressionswhichwehavenotreallydefinedbutwhichincludesuchthingsas“R(x)”(“xisred”),“-[S(x)&R(x)]”(“Itisnotthecasethatxissquareandxisred”),aswellassentencessuchas“-"x-S(x)”(“Itisnotthecasethatforeveryx,xisnon-square,”i.e.“Somethingissquare”).Thisdoesn’tlooklikeadefinition,butinfactitreallydoescompletelyexplainthemeaningof“satisfies”asitappliestooursimplelanguage.Usingthesedefinitionalrulesoncomplicatedexpressionswecanproceedstepbysteptosimplerexpressionsuntilwegetdowntocasescoveredby(1)and(2).Anditmaylookasifwehavesomekindofviciouscircularity.Forexample,wehaveused“and”(inthemetalanguage)todefinesatisfactionforexpressions(oftheobjectlanguage)containing“&.”Buttheappear-anceisdeceptive.Wehaveassumedthatwhateverwecansayintheobjectlanguage,wecansayinthemetalanguage,butnotnecessarilyviceversa.Thisassumptiondoesnotintroduceanylogicalorphilosophicaldifficultyintothedefinition.Finallywedefinetruth:Asentencefistrueifandonlyifeveryobject(inthedomain)satisfiesit.Again,wehaven’treallydefinedthesentencesofourobjectlanguage,buttheywillbeexpressionsinwhichnooccurrencesofthevariableare“dangling.”Forexample,“-"x-[R(x)&S(x)]”(“Somethingisredandsquare”)isasentence,asopposedtoaformulasuchas“R(x)”(“xisred”).Herethevariablexisjustaplaceholder,indefinitelyindicatingsome-thingorother,butnodefinitething.Itisnotobviousthatthisdefinitionactuallyconformstothecriterionofmaterialadequacy.Butitdoes.ItcanbeprovedthateveryinstanceofSchema(T),confinedtosentencesofourobjectlanguage,isaconsequenceofthisdefinition.Thewholethingmayseemtrivial,butitisreallyquiteamazingthatinanappropriatemetalanguagetruthcanbedefinedwithoutappealingtoanysemanticalnotions.Thismeansthatit127\nC.ANTHONYANDERSONhasbeendefinedintermsofthingswhichareclearer:theyarejusttheconceptsoflogictogetherwiththeconceptsoftheobjectlanguage.ItremainstomentionTarski’sworkonthenotionoflogicalconsequence(Woodger1956:409–20).This,likethenotionoftruth,wasusedinanintuitivewaybylogiciansandphilosophersbeforeTarski,butitwasthelatterwhomadethenotionprecise.Consideronceagainoursimpleformalizedlanguage.Donotselectaparticulardomainandparticularmeaningsfor“S”and“R.”Rather,contemplateanyarbitraryinterpretationofthelanguage–anydomainofobjectswhatsoeverandanyappropriatemeaningsforthesesymbols.Thelogicalsymbols“-,”“&,”andsoon,aretoretaintheiroriginalmeaningsthroughout.Foranysuchspecification,wecanexplaintruthunderthatinterpretationalongthelinesusedabovefortheparticularinterpretationwewereconsidering.Supposethatinsomeinterpretationaparticularsentence,say“"x[S(x)&R(x)],”comesouttrue.Theninthatinterpretationcertainothersentenceswillcomeouttrueaswell.Forexample,“"xS(x)”and“"xR(x).”Infact,thiswillalwayshappen.Ifaninterpretationmakesourexamplesentencetrue,thatinterpretationwillalsomakethesetwosentencestrue.Insuchacase,Tarskisaysthatthelattertwosentencesarelogicalconsequencesofthefirstsentence.Ingeneral,asentenceyisalogicalconsequenceofasentencefifandonlyifeveryinterpretationwhichmakesfcomeouttruealsomakesycomeouttrue.Andasentenceisdefinedtobealogicalconsequenceofacollection,orset,ofsentencesifeveryinterpretationwhichmakeseverysentenceinthesetcomeouttruealsomakesthesentenceinquestioncomeouttrue.Finally,Tarskidefinesasentencetobelogicallyvalidifitcomesouttrueundereveryinterpretation.Theimportanceofsuchadefinitionisthatwecannowstrictlydefinewhatitisforsomethingtobeavalidargumentinourlanguage.And,ofcourse,thestudyofvalidargumentsisattheveryheartofthedisciplineoflogic.Usingthesedefinitionswecanthenprovethatcertainsystemsoflogicalrulesare“complete”inthesenseofbeingadequatetotheirintendedpurposeofcapturingallvalidinferencesexpressibleinthelanguage.Forexample,certainformulationsoffirst-orderlogic,thelogicofsuchnotionsasand,not,or,if,...then,not,someall,andthelikewereprovedcompletebyKurtGödel,tobediscussedbelow.Thesetwothings,hisdefinitionoftheconceptoftruthforformalizedlanguagesandhisexplicationoftheconceptoflogicalconsequenceareTarski’sdistinctivephilo-sophicalcontributions.Theyaresubstantialindeed.AlonzoChurchIn1927,ChurchreceivedhisPh.D.fromPrinceton,wherehetaughtfrom1929to1967.Thereafter,hetaughtatUCLAuntil1990.Hewasalong-timeeditoroftheJournalofSymbolicLogic,whichhehelpedtofound.Church’sphilosophicalcontribu-tionslargelyconcernquestionsaboutthefoundationsoflogicandmathematics,espe-ciallytheirontology,andtopicsinthephilosophyoflanguageandintherelatedareaofintensionallogic.Church’sthesisisahypothesisconcerningtheidentificationofthemechanicallycomputableorcalculablefunctionsdiscussedbelowinconnectionwithGödel’sincom-pletenesstheorem.Churchproposedasaprecisemathematicalanalysisoftheideaof128\nALFREDTARSKI,ALONZOCHURCH,ANDKURTGÖDELsuchfunctionsthattheybeidentifiedwiththelambda-definablefunctions.Thislatternotionistootechnicaltobeexplainedindetailhere.AlanTuringindependentlypro-posedtheidentificationofthemechanicallycomputablefunctionswithfunctionscom-putableinprinciplebyapreciselydefinablesortofabstract“machine,”nowcalledaTuringmachine.ThisidentificationturnedouttobeequivalenttoChurch’sthesis.Thatis,theclassoflambda-definablefunctionsisexactlythesameastheclassoffunctionscomputablebyaTuringmachine.Otherattemptstoanalyzethenotioninquestionhavealwaysledtothesameclassoffunctions.Theidentificationoftheclassofmechani-cally(“algorithmically”)calculablefunctionswiththeclassoflambda-definableorTuringmachine-computablefunctions(the“Church–Turingthesis”)isnowalmostuniversallyaccepted.Church’stheorem,tobecarefullydistinguishedfromChurch’sthesis,isatheoremofmathematicallogictotheeffectthatthereisnoeffective(=mechanical)procedurefordecidingwhetherornotaformulaoffirst-orderlogicisvalid.ChurchwasaPlatonistor,ashepreferred,arealistabouttheentitiesapparentlydescribedandstudiedbymathematicsandlogic.Numbersandothermathematicalentitiesare,hebelieved,objectivelyexisting,mind-independentobjectsandmathemat-icsitselfconsistsoftruthsaboutthesethings.Logicseemstorequire,ifformulatedinfullgenerality,propositions,properties,and“individualconcepts.”Thesekindsofthings,usuallycalledintensionalentities,aresupposedtobeabstract,real,andobjectiveenti-tiessuitabletobethemeaningsofexpressionsinvariouslanguages.Propositions,forexample,areclaimedtobethemeaningsofdeclarativesentences,thesameforsyn-onymoussentences,whetherinasinglelanguageorintwoormoredifferentlanguages.Church’sgeneralmethodologicalviewpointabouttheformalscienceswasakindof“hypothetico-deductiverationalism.”Accordingtothisview,intuitionsorfeelingsofself-evidenceprovideinitialsupportforassumptionsaboutabstractentities.Thetheo-riesofthesearetobeformalized,statedusingthepreciselanguageandterminologyofsymboliclogic,andtheresultsaretobeevaluatedusingthesortsofcriteriacommontoscientificproceduresingeneral.Onewayweevaluatetheoriesisbydeducingconse-quencesandtherebydeterminingwhethertheyareadequatetoaccountforthedata.IntheformalsciencesChurchtookthedatatoincludetheacceptedfactsofmathe-maticsandlogic.ManyofChurch’sphilosophicalcontributionsappearinreviewsintheJournalofSymbolicLogic.Hisrelativelyfewpapersdevotedexplicitlytophilosophicaltopicsusuallyconcernedquestionsaboutmeaningandrelatedtopicsinthephilosophyoflan-guage.Therearealsoargumentsagainstnominalismasitissometimesespousedinconnectionwithmathematics,logic,orsemantics.Asasampleofthelatter(Church1950),consideranominalistattempttogiveananalysisofcertainstatementsapparentlyaboutpropositions.Supposeitisclaimedthatsuchasentenceas(1)“Senecasaidthatmanisarationalanimal”istobeanalyzedas:(2)“ThereisalanguageS¢suchthatSenecawroteasasentenceofS¢wordswhosetranslationfromS¢intoEnglishis‘Manisarationalanimal’.”Thismayalreadyseemexcessivelycomplicated,butsimplerattemptstoanalyzestatementsaboutassertionsothattheyconcernsuchrelativelyconcretethingsassentencesaresubjecttoeasyrefu-tation.Tobringoutclearlythat(2)willnotdoasananalysisof(1),Churchusesthe“translationtest,”aprocedurewhoseinventionisusuallyattributedtoC.H.Langford.129\nC.ANTHONYANDERSONIfwetranslate(1)intoGerman,weget(1¢)“Senecahatgesagt,dasderMenscheinvernünftigesTiersei”.Intranslating(2)intoGerman,notecarefullythattheword“English”mustbetranslatedas“Englisch”(notas“deutsch”)andthequotationwhichformspartof(2)istobetranslatedas“Manisarationalanimal”(notas“DerMenschisteinvernünftigesTier”).Thislattertranslation,callit(2¢),certainlywouldnotconveyanythingliketheinformationwhichwouldbeconveyedtoaGermanspeaker(whospokenoEnglish)by(1¢).Thus,arguesChurch,(1¢)isnotanacceptableanalysisof(1).Thebasicideaoftheobjection,whichcanbeseenevenwithoutappealingtotransla-tion,isthat(1)doesnotsayanythingaboutanyparticularlanguage(andsoneitherdoesitstranslation(1¢)),whereas(2)makesspecificreferencetoEnglish.AphilosophicalargumentwhichhasaquitesurprisingconclusionisgivenbyChurch(Church1956:24–5)asamorepreciseversionofreasoningofferedbyGottlobFrege.Theconclusionoftheargumentisthatsentencesdenotetruth-values,truesen-tencesdenotingTruth(orTheTrue)andfalsesentencesdenotingFalsehood(orTheFalse)!Putlikethis,thethesisseemsquiteincredible,evenunintelligible.Whysupposethatsentences“denote”anythingatall?Andwhat,wemayask,arethesealleged“objects,”TruthandFalsehood?Thesearegoodquestions,buttheessentialpointofChurch’sargument(andFrege’sbeforehim)couldbestatedlikethis:thetruthorfalsityofasentenceistheonlythingthatstandstothesentenceasthedenotationofa(complex)namestandstoitsparts.Toseethistakesuchasentenceas(a)“SirWalterScottistheauthorofWaverley.”Ifwereplace“theauthorofWaverley”byanexpres-sionwhichdenotesthesame,“themanwhowrotetwenty-nineWaverleyNovelsalto-gether,”wegetanewsentence:(b)“SirWalterScottisthemanwhowrotetwenty-nineWaverleynovelsaltogether.”Ifwearesupposingthatthe“denotation”ofasentence,whateveritis,isunchangedifadenotingpartisreplacedbyanotherwiththesamedenotation,thenthisnewsentencemusthavethesamedenotationastheoriginal.Further,itisplausible(Churchclaims)thatthesentence,(c)“Thenumber,suchthatSirWalterScottwrotethatmanyWaverleyNovelsaltogetheristwenty-nine,”issocloseinmeaningto(b)astohavethesame“denotation”(again,withoutyetassumingthatweknowwhatthisis).Butnowletusreplacethedenotingexpression“Thenumber,suchthatSirWalterScottwrotethatmanyWaverleyNovelsaltogether”in(c)byanexpressionwiththesamedenotation,namely;“ThenumberofcountiesinUtah”(whichisinfacttwenty-nine).Wethengetasentencewhichissupposedtohavethesamedenotationas(c),(d)“ThenumberofcountiesinUtahistwenty-nine”(againassumingthatasentencedoesnotchangeitsdenotationifadenotingpartisreplacedbyanotherwiththesamedenotation).Nowcompareouroriginalsentence(a)“SirWalterScottistheauthorofWaverley”with(d)“ThenumberofcountiesinUtahistwenty-nine.”Bythereasoningjustexplained,thesetwosentencesmusthavethesame“denotation.”Buttheonlymeaning-relevantfeaturewhichtheyseemtohaveincommonisthatbotharetrue.Alittlereflectiononsuchexamplespointstotheconclusionthattheonlythingthatcanbeexpectedtoremaininvariantundersuchsubstitutionsisthetruthorfalsityoftheoriginalsentence.Soif“denotation”hasananalogforsentences,itwillhavetobethetruth-values,truthandfalsity,whichmaybeseenasmathematicalabstractions.(Comparethemathematicalabstractionofnumbers,asobjects,fromcollectionsorfromconceptsofcollections.)TheChurch–Fregeargumentheremaynotbeconclusive,130\nALFREDTARSKI,ALONZOCHURCH,ANDKURTGÖDELbuttheanalogyuncoveredisstrikinganditmaywellbeausefultheoreticalassump-tionforsemanticsthatsentences“denote”truth-values.(SeeAnderson1998forfurtherdiscussion.)Church’smostimportantphilosophicalideasarecontainedinhisworkonthefoun-dationsofintensionallogic(Church1951,1973,1974).Philosophersandlogicianscontrastintensionandextension,butitisbynomeanseasytogiveaclearcharacteri-zationofthesenotions.Inthecaseofsentences,Churchwouldmaintainthatthesense,orintension,ofthesentenceisthepropositionwhichitexpressesandthedenotation,asalreadyexplained,isthetruth-valueofthesentence.Logicasstandardlytaughtinphilosophyandmathematicsdepartmentsmakesnosignificantdistinctionbetweensentenceswiththesametruth-value;argumentswhichturnonfinerdistinctionsofmeaningaresimplynottreated.Similardistinctionsholdbetweenthesetofthingsofwhichapredicateistrue,theextensionofthepredicate,andthepropertyconveyedbythepredicate,itsintension.Again,adistinctionbetweenthemeaning,strictlyso-called,ofanexpressionsuchas“ThepresentpresidentoftheUS”(itsintension)andwhatitstandsfor,theactualperson,isneeded.Herewemightsay,againwithChurch,thatthemeaningoftheexpressioninthestrictsenseistheconceptthatitexpresses,itsinten-sion,butwhatitdenotesorstandsfor,thepersonor,moregenerally,theobject,isitsextension.So,asalreadyexplained,Churchcallsthepropositionexpressedbyasentenceitssenseandthetruth-valuethatitstandsforitsdenotation.Predicateshavepropertiesastheirsensesandsetsastheirdenotations,andindividualexpressions(e.g.descriptivenames)havecertainconceptsastheirsensesandwhattheystandforastheirdenota-tions.Therelationshipthatholdsbetweenthesenseofanexpressionandwhatitdenotesletuscalltheconceptrelation,andsymbolizeitbythecapitalGreekletterD(delta).Thenpropositionsareconceptsoftruth-values,propertiesareconceptsofsets,andindividualconceptsareconceptsoftheindividualthingsthattheconceptschar-acterize.Generalizingourterminology(asChurchdoes),callanythingthatiscapableofbeingthesenseofsomeexpressionaconcept.TheintensionallogicthatChurchenvisionedwouldhavetwokindsofintensionalaxioms:logicalprinciplesaboutDandprinciplesthatwouldspecifytheessentialchar-acteristicsofpropositionsandothercomplexconcepts.Inconnectionwiththelatter,Churchtookittobeespeciallyimportanttohaveaxiomswhichgive,orcorrespondto,criteriaofidentityforcomplexconcepts.Acriterionofidentityinthepresentcaseisaprinciplethatdeterminestheidentityordifferenceofthecomplexconceptsexpressedbydifferentsentences(orpredicatesordescriptivenames)intermsofsomeknownrela-tionbetweenthesentences(complexexpressions)themselves.Anexamplewouldbetheprinciplethattwosentencesexpressthesamepropositionifandonlyiftheyarelogi-callyequivalent;inourexample,thatis,theyhavethesametruth-valuenecessarily,oronlogicalgroundsalone.Wehavealreadyexplainedthatafunctionofnumbersisacorrelationofacertainkind.Thus,squareorsquaringissaidtobeafunctionfromnumberstonumbers.Ingeneral,anycorrelationbetweenthethingsintwocollectionsiscalledafunction.Generally,afunctionisjustanyconceivablecorrelationbetweenthethingsinonecol-lectionandthethingsinanother(or,possibly,thesame)collection;itisallowedthattwoormorethingsinthefirstcollectionbecorrelatedwiththesamethinginthesecond.131\nC.ANTHONYANDERSONAnameofafunctionhasbothasenseand(ingeneral)adenotation.Thesenseisthereforeaconceptofthefunctiondenotedbythatname.Forexampletheexpression“Thesquaringfunction”denotesthefunctionthattakeseachnumberinto2itssquareandithasasitssensewhatisconveyedby“x.”Thecombinationofanameofafunctionwiththenameofsomethingtowhichthefunctionisapplied(anargumentofthefunction)willalsohaveasense:acomplexsenseinvolvingthesenseofthefunctionnameandthesenseofthenameoftheentitytowhichthefunctionisbeingapplied.Theimportanceofthisideaappearsintheobservationthatanycomplexexpressionmaybeconstruedasbeingbuiltupfromafunctionexpression,togetherwithexpres-sionsforoneormoreargumentstowhichthefunctionisapplied.Nowletuswrite“D(X,Y)”tomeanthatXisaconceptofY.TheaxiomswhichChurchtooktogovernthedelta-relationare:(C1)ForeveryX,Y,andZ,ifD(X,Y)andD(X,Z),thenY=Z.(C2)ForeveryFandF1,ifD(F1,F),thenforeveryXandX1,ifD(X1,X),thenD(F1X1,FX).(C3)ForeveryFandF1,ifforeveryXandX1,D(X,X1)impliesthatD(F1X1,FX),thenD(F1,F).(C1)saysthatanythingwhichisaconceptofsomethingisaconceptofexactlyonething.In(C2)and(C3),FisanyfunctionandFXistheresultofapplyingthatfunctiontoanargumentX;thatis,FXistheentitywhichiscorrelatedwithXbythefunction.WhereF1andX1areconcepts,wehavejustwritten“F1X1”forthecomplexconceptthatresultswhentheconceptF1iscombinedwiththeconceptX1.Intheseterms,(C2)amountstotheclaimthatifanexpressiondenotingafunctioniscombinedwithanexpressiondenotinganargument(insomepossiblelanguage),thenthesenseofthecomplexexpressionistheresultofcombiningthesenseofthefunctionnamewiththesenseoftheargumentname.Theproposedaxiom(C3)ismoreproblematic.Tounderstandandacceptit,onereallymustgoalongwithahypothesisthatChurchproposestosimplifythelogicofthesystem.Churchassumesthataconceptofafunctioncanbetakentobeafunctionfromcon-ceptstoconcepts.Thisisfineforaxiom(C2),whichisthenjustunderstoodinsuchawaythatcombiningaconceptofafunctionwithaconceptofanargumentisnothingmorethanapplyingacertainkindoffunctiontoacertainkindofargument.Butaxiom(C3)ismuchbolder.Itamountstotheclaimthatanyfunctionfromconceptstoconceptssatisfyingacertainconditionisaconceptofacertainfunction.Itsays:ifafunctionappliedtoaconceptofanargumentalwaysyieldsaconceptoftheoutputofsomefunc-tionappliedtotheargumentthusconcepted,thenthefunction(fromconceptstocon-cepts)isaconceptofthefunctionfromobjectstoobjects.Thisaxiomleadstovariousdifficulties,whichcannotbeexplainedhere(seeAnderson1998).Itisfairtosaythateventhebasicprinciplesofintensionallogic,asChurchconceivedit,arestillnotsettled.Intensionalprinciplesofthesecondsort–thosesupposedtoindividuatecomplexconcepts–arealsostillproblematic.Churchproposedthreeheuristicprinciplestoguidetheformulationofsuchaxioms:(A)thatlogicallyequivalentexpressionsexpressthesameconcept,(B)thatexpressionsthathaveexactlythesame132\nALFREDTARSKI,ALONZOCHURCH,ANDKURTGÖDELsyntacticalstructureandwhosecorrespondingpartshavethesamemeaningsexpressthesameconcept,and(C)expressionsthatcanbeobtainedfromoneanotherbyapply-ingthelogicaloperationoflambdaconversionexpressthesameconcept.Lambdacon-versionisalogicallyvalidtransformationofexpressions,whichwedonotattempttoexplainhere.Theideabehind(A)workswellif,butonlyif,oneisdealingwithreasoninginvolv-ingnofinerdistinctionsofmeaningthanareinvolvedinargumentsturningonmodal-ity:necessity,possibility,impossibility,andsimilarconceptions.Suggestion(C)appealstothetechnicalnotionoflambda-conversionwhichisverydifficulttomotivatefromaphilosophicalpointofview.Handsdown,thenotionurgedin(B)isthemostpromis-ing.Church(whoagreedwiththeassessmentjustoffered)triedtoimplementthisapproachseveraltimesinhispublishedwork,buttechnicalandlogicaldifficultiesstillblockthewayofasatisfactorytheory.ItisfairtosaythatthisprojecttowhichChurchcontributedfundamentalandimportantwork,toestablishacomprehensiveandadequategeneralintensionallogic,hasnotyetbeencompleted.Buthissuccessfulphilosophicalcontributionsareimpressiveindeed.KurtGödelKurtGödelreceivedhisdoctoratein1930attheUniversityofVienna.HeemigratedtotheUnitedStatesin1940andsoonafterwardsbecameamemberoftheInstituteforAdvancedStudiesatPrinceton,NewJersey,untilhisdeath.Gödel,likeTarskiandChurch,isbestknownasalogician.Buthislogicaldiscoveriesareofprofoundsignifi-canceforpartsofphilosophy:thephilosophyoflogicandmathematics,epistemology,and(perhaps)thephilosophyofmind.Inaddition,inlateryearsGödelconcentratedonphilosophicalquestionsandmadestrikinglyoriginalsuggestionsastotheirsolu-tion,includinganimprovedversionofAnselm’sfamousontologicalargumentfortheexistenceofGod,aselaboratedbyLeibniz.Gödel’smostfamousdiscoveriesarehistwoincompletenesstheorems(Gödel1931).Herewewillgiveoutlinesofmodernizedversionofhisproofs.Supposethatthearith-meticofthenaturalnumbers(0,1,2,andsoon)isformulatedasanaxiomaticsystem.Thelanguageusedisapreciselyspecifiedsymbolic,orformalized,languagewithaxiomsstatingthebasicpropertiesofthenaturalnumbersandrulesofinferencestatingwhichsentencesmaybecorrectlyinferredfromothers.Asequenceofsentencesbeginningwithaxiomsandconstructedbyapplyingtherulesofinferenceissaidtobeaproofofthelastsentenceinthesequence,whichlatterissaidtobeatheoremofthesystem.Gödelobservedthatwecanassignnumbers(nowcalled“gödelnumbers”)tothesyntacticalentitiesoftheaxiomaticsystem.Thatis,onecancorrelatenumberswithsymbols,withcomplexexpressions,andevenwithsequencesofexpressionssuchasproofs.Thesenumbersareassignedtosymbolsoftheobjectlanguageinthemetalanguage.Butthesenumbersarepartofthesubjectmatteroftheobjectlanguagetheoryitself.Thisdone,wehaveasortofindirectwayoftalkingaboutexpressionsandsequencesofexpressionsoftheformaltheorywithinthetheoryitself.Bytalkingaboutthegödel133\nC.ANTHONYANDERSONnumbersofexpressionsandsequencesofexpressions,wecansimulate,ormodel,talkabouttheexpressionsandsequences.Gödelthenprovedthatiftheformalsystemofarithmeticmeetscertainminimalcon-ditionsofadequacy,thenthesetofgödelnumbersofthesentencesprovableinthesystem(thetheoremsofthesystem)canbedefinedwithinthatsystem.Theconditionofminimaladequacyisthatthesystemofarithmeticbecapableofexpressingcertainfunctionsofnaturalnumbers.Afunctionofnaturalnumbersisjustacorrelationbetweennumbersandnumbers,orbetweenpairsofnumbersandnumbers,or...andsoon.Intuitively,thefunctionsofnaturalnumbersthatmustbeexpressiblearethosewhosevaluesforgivenargumentscanbe“effectivelycalculated”:calculatedbymeansofanalgorithmorrecipe,mechanically(andmindlessly)computedinamanneravail-abletoacomputingmachine.Next,itcanbeprovedthatthesetofgödelnumbersoftruesentences(“true”beingdefinedinthemannerofTarski)isnotdefinableinarithmetic.Theproofusesanargu-mentwhichparallelsthereasoningoftheliarantinomy(!)butwhichislogicallyunex-ceptionable.Theconclusionisthatsuchasystemofarithmeticcannotcontainordefineitsowntruthpredicate(asappliedtogödelnumbersassurrogatesforsentences).Butifthesetofgödelnumbersofprovablesentencesisdefinableinarithmeticandthesetofgödelnumbersoftruesentences(ofarithmetic)isnotdefinableinthatsystem,thenthetwosetshavetobedifferent.Therefore,eithersomesentenceprovableinarith-meticisnottrueorsometruesentenceofarithmeticisnotprovable.Wecannotaccepttheformer,atleastifwehavechosenasystemofaxiomswhichwecanseetobetrueofthenaturalnumbers.Weconcludethatsometruesentenceofarithmeticisnotatheoremofarithmetic.Anyformalsystemofarithmeticmeetingreasonableconditionsofadequacywillbeincomplete.ThisisessentiallyGödel’sfirstincompletenesstheorem.Gödel’sactualproofdidnotproceedinthewaywehavedescribed.Rather,heassumednotthattheaxiomsofarithmeticaretrue(andthattherulesofinferencepre-servetruth)butonlythatarithmeticisconsistent:itisnotpossibletoderiveanactualcontradictionfromtheaxiomsusingtherulesoflogic.(Really,heassumedthatarith-meticisomega-consistent,astrongerassumptionthanconsistencywhichweneednotexplainhere.)ThenGödelshowedhowtoconstruct,giventhatarithmeticisconsistent,aparticularsentenceGwhichissuchthatneitherGnoritsnegation-G(“not-G”)isatheorem.Theproofproceedsinsuchawaythatwecould,withsufficientpatienceandlongevity,actuallywritedownatruesentenceofarithmeticwhich,ifarithmeticisformallyconsistent,cannotbeprovedinarithmetic.Andonecanseethatifthecon-sistencyofarithmeticisaccepted,thesentenceGisthetruebutunprovableone,asopposedto-G.ThissentenceGinvolvesjustquiteordinaryarithmeticalconceptssuchas“plus”and“times”togetherwiththeusuallogicalconcepts“and,”“not,”“some,”“all,”“equals,”andsoon.Itisworthnoticingthat,contrarytovariouspopularexpositions,Gödel’soriginalproofdoesnotinvolveself-referenceinanysense.Thetruebutunprovablesen-tenceGdoesnot“say”thatit,itself,isunprovable.Itisasentenceentirelyaboutnaturalnumbersandtheirpropertiesandrelations.Butitisasentencethatsimulatessuchaself-referentialsentenceinthesensethatitistrueifandonlyifitisnotprovable.Well,sowhat?Itisnaturaltosuggestthattheaxiomsofarithmeticwithwhichwebeganarejustnotadequateandthatsomenewaxiomsmustbeadded.ItisasifEuclid’s134\nALFREDTARSKI,ALONZOCHURCH,ANDKURTGÖDELgeometryhadbeenformulatedwithouttheParallelPostulate.Onewouldsimplyhavetoaddit,orsomeequivalent.However,ifyouconsiderthedetailsofGödel’sproof,itisevidentthatasystemobtainedbyaddinganyfinitenumberofaxiomswillstillbesubjecttotheproof–althoughtheunprovablebuttruesentencewillbedifferent.Indeed,evenaddinginfinitelymanynewaxiomsinany“effective”waydoesnotevadetheproof.Wemustconcludethatnothingthatwouldcountasaformalsystemcancontainallthetruthsofarithmetic.(Ifyouthinkthattobeatruthofarithmeticistobeprovableinarithmetic,thenthisresultwillbequitedifficulttocomprehend.Butoneconclusionwecandrawfromtheseconsiderationsisthatthisidentificationcannotbecorrect.)Thismuchisalreadyquitestartling.ThegoalofmathematicianssinceEuclidhasbeentospecifycertainbasictruthsofmathematicsandtojustifyallothersbydeductionfromthese.Gödelprovedthatthisgoalisunattainable!Nomatterwhatformal(alias“axiomatic”)systemisproposed,therewillbetruthsofarith-metic(themostbasicpartofmathematics)whichthesystemcannotprove.Ofcoursetoshowthiswithmathematicalprecisionrequiresthatwepreciselydefine“axiomaticsystem”or“formalsystem,”butthiscanbedoneinawaythatisundeniablycorrect.Gödel’ssecondincompletenesstheorembuildsonthefirst.Usingthetechniqueofpseudo-self-referencementionedabove,onecanfindasentenceofany(minimallyad-equate)formalsystemofarithmeticwhich“says”thatthesystemitselfisconsistent.Callthissentence“Consis.”NowtheproofofGödel’sfirstincompletenesstheoremcanbemimickedwithinarithmetictoproduceaproofoftheconditionalsentence:“IfConsis,thenG.”SupposeitwerepossibletoprovewithinarithmeticthesentenceConsiswhichmirrorsthepropositionthatarithmeticisconsistent.Thenitwouldbepossibletoprove(bymodusponens)theGödelsentenceG.Butwealreadyknow,fromthefirstincompletenesstheorem,thatifarithmeticisconsistent,Gcannotbeprovedtherein.Ifwesuppose,aswearecertainlyentitledtodo,thatthetheoremsofarithmeticaretrue,thenarithmeticisconsistent.Weconcludethatthesentencewhich(inanindirectsense)expressesthatarithmeticisconsistent,cannotitselfbeprovedinarithmetic.And,ofcourse,thesentence“expressing”thatarithmeticisinconsistentwillnotbeatheoremeither.Ittoo,likethesentenceG,isundecidableinarithmetic:canneitherbeprovednorrefutedtherein.WehavebeenatsomepainstodispeltheimpressionthatGödel’sproofsliterallyinvolveself-reference.Whatthendoes,forexample,thesentence“expressing”theconsistencyofarithmeticlooklike?Well,likethesentenceG,itiswrittenentirelyinthelanguageofarithmetic(involving“plus,”“times,”and“equals,”forexample).Infactitcanbeseenasexpressingacertainmathematicalclaimaboutapolynomial.LetP(x,y)beapolynomialinvolvingjustthetwoindicatedvariablesandintegralcoefficients.Thenthestatement“Foreveryy,thereisanx,suchthatP(x,y)=0”maybetrueorfalse,dependingonthedetailsofthepolynomial.Problemsofthissort,astowhetherornotsuchastatementiscorrect,arecalled“diophantine”problems.Moregenerally,supposewehavenvariablesx1,x2,...,xnandmvariablesy1,y2,...,ym,andapolynomial(withintegralcoefficients)involvingthese,say“P(x1,...,xn,y1,...,ym).”Thenthestatement“expressing”theconsistencyofarith-meticisoftheform:135\nC.ANTHONYANDERSONForeveryy1,...,ym,therearex1,...,xn,suchthatP(x1,...,xn,y1,...,ym)=0.Thisisapurelyarithmeticalstatementand,onewouldhavesupposed,aclaimthatarithmeticaltechniquesshouldbeabletosettle,yeaornay.Buttheycannot.(Thedetailsoftheseproofsareratherdifficulttograsp.ThebestintroductiontothemisSmullyan1957.Forprecisedetails,BoolosandJeffrey1989isexcellent.)Thephilosophicalimportoftheseresultsismorecontroversial.Wehavealreadyobservedthatacertainmathematicalprogramistherebyprovedimpossible:thatofdeducingallofmathematicsfromanaxiomaticbasis.Ifwethinkofmetaphysicsasincludingallnecessarytruths,mathematicsnotexcluded,thenthegoal,perhapsmostcloselyassociatedwithSpinoza,ofanaxiomaticdevelopmentofmetaphysicsisthusalsoprovedimpossibletoachieve.Gödel’sownphilosophicalspeculationsontheimportofhistheoremsaremostclearlyarticulatedinhisCollectedWorks(1995:304–23).Letusconfineourattentiontoarithmeticandspeakofthetruesentencesthereofas“objectivemathematics.”Humanbeings,usingourpresentlyacceptedarithmeticalassumptions,cancertainlyprovesomeofthesesentences.Gödelcallsthemathematicaltruthsthathumanbeingsarecapableofdemonstrating“subjectivemathematics”(perhapsnotthebestchoiceofterminologysincetheseareallobjectivelytrueandindeedknowabletobesuch).Someofthesemayrequireaxiomsaboutthenumberswhicharepresentlyunknown,butwhichcaninprinciplebeseentobeevidentbyhumanmathematicians.NowconsideragainGödel’ssecondincompletenesstheorem.Thetheoremsofaformalsystemofarithmeticcompriseasetofsentencesthatcouldbemechanicallygenerated,oneaftertheother.AccordingtoGödel’ssecondincompletenesstheorem,anysuchsystemwillbeunabletoprove(generate)thearithmeticalsentencethat“expresses”itsownconsistency.Recallthatthiswasacertainsentenceexpressingadiophantinearithmeticalproblem.Gödeldrawsthefollowingconclusionfromthis:Eithermathematicsisincompletableinthissense,thatitsevidentaxiomscanneverbecomprisedinafiniterule,thatistosaythehumanmind(evenwithintherealmofpuremathematics)infinitelysurpassesthepowersofanyfinitemachine,orelsethereexistabsolutelyunsolvablediophantineproblemsofthetypespecified.(1995:310)Tosaythattheevidentaxioms(andrulesofinference)canbe“comprisedinafiniterule”isequivalenttothepossibilityoftheformulationofarithmeticasawholeasaformalsystem.Andthislatteramountstothepossibilityofbeinggeneratedinamachine-likefashion.Tosaythatadiophantineproblemisabsolutelyunsolvablemeansthatneitherthestatementnoritsnegationwilleverbeatheoremofsubjectivemath-ematics.ThereisnodoubtthatthisconclusionactuallydoesfollowfromGödel’sproof,togetherwiththementionedanalysisofafinitemechanicalprocedure.Gödelhimselfthinks,andargues,thatthesecondoptionisincorrect–therearenoabsolutelyunsolv-ablearithmeticalproblems–buthisargumentsforthisconclusionarenotairtightandmaybereasonablydoubted.GödelalsohadanimprovementonAnselm’sontologicalargumentfortheexistenceofGod,especiallyasitwasdevelopedbyLeibniz.Leibnizobservedthattheone136\nALFREDTARSKI,ALONZOCHURCH,ANDKURTGÖDELthingthatneedstobeprovedtocompleteAnselm’sproofisthattheexistenceofGodispossible.Anumberofcommentatorsontheargument,includingKant,haveobservedthatallthatisreallyestablishedbyAnselmisthatifGodexists,thenHenecessarilyexists.Now,giventhenatureoftheproof,wemayfurtherconcludethatthiscondi-tionalitselfisnecessary.So,wemayconcludefurther,byastandardprincipleofmodallogic,thatifitispossiblethatGodexists,thenitispossiblethatitisnecessarythatHeexists.(Thestandardprincipleofmodallogicinquestionisthis:ifpnecessarilyimpliesq,thenifpispossible,thenqispossible.)IfwefurthersupposethatwecansomehowprovethatitispossiblethatGodexists(givenacertaindefinitionof“God”),thenitfollowsthatitisnecessarythatitispossiblethatGodexists.Butaccordingtooneplau-siblesystemofmodallogic(standardlycalled“S5”),itthenfollowsthatitisnecessarythatGodexists!ButcanweprovethatitispossiblethatGodexists?Gödelthoughtthatwecanandthataversionoftheontologicalargumentisthencogent.Hisargumentforthis,whichbearssomeresemblancetoLeibniz’sargumentforthesameconclusion,usestheideaofapositiveproperty.Gödeldoesn’treallysayveryclearlywhatthisconceptioninvolves,butheremarksthatithastwopossibleinterpretations,“positiveinthemoral-aestheticsense”andpositiveinthesenseofinvolvingonly“pureattribution.”AbeingisdefinedtobeGod-likeifithaseverypositiveproperty.Thenapropertyisdefinedtobeanessenceofanentityxifxhasthatpropertyanditentailseveryotherpropertythatxhas.Anentitynecessarilyexists,bydefinition,ifeveryessenceofitisnecessarilyexemplified.Finally,Gödelassumesthefollowing“axioms”abouttheseconcepts:1Apropertyispositiveifandonlyifitsnegationisnotpositive.2Anypropertyentailedbyapositivepropertyisitselfpositive.3ThepropertyofbeingGod-likeispositive.4Ifapropertyispositive,itisnecessarilypositive.5Thepropertyofnecessarilyexistingispositive.FromtheseitfollowsthatitispossiblethataGod-likebeingexistsand,byessentiallytheargumentexplainedearlier,thatsuchabeingthereforeexists,andindeedneces-sarilyso.Therearesomeproblemswiththeargument,nottheleastofwhichistheobscurityofthenotionofapositiveproperty.Foranablediscussionoftheargumentanditsallegeddefects,seeAdams1970.GödelalsothoughtthatEinstein’sTheoryofRelativityhasimplicationsforidea-lism,inparticularthatitsupportssomeideasofImmanuelKant.Hearguesthatthereisconsiderablereasontobelievethat“timeisunreal”(1951:555–62).Essentiallytheargumentisthis:Iftimeandchangearesomethingreal,thentheremustbesuchathingasanobjectiveandabsolutelapseoftime.ButtheTheoryofRelativity,awell-confirmedscientifictheory,seemstodenythatthereissuchanobjectivelapse.Gödelconsidersvariousobjectionsandexplainstherelevanceofatechnicalcon-tributionofhistothattheory.Thatwork,bytheway,seemstoimplythepossibilityof“timetravel”!Insum,wemaysaythatGödel’smainworkinlogicisofprofoundphilosophicalsignificanceandthathisotherphilosophicalworkcertainlydeservesfurthercarefulstudy.137\nC.ANTHONYANDERSONBibliographyWorksbyChurch,Gödel,andTarskiChurch,A.,1950:“OnCarnap’sAnalysisofStatementsofAssertionandBelief,”Analysis10,pp.97–9.1951:“AFormulationoftheLogicofSenseandDenotation,”inStructure,MethodandMeaning,ed.P.Henle,H.M.Kallen,andS.K.Langer,NewYork:LiberalArtsPress.1956:IntroductiontoMathematicalLogic,vol.I,Princeton,NJ:PrincetonUniversityPress.1973:“OutlineofaRevisedFormulationoftheLogicofSenseandDenotation(PartI),”Noûs7,pp.24–33.1974:“OutlineofaRevisedFormulationoftheLogicofSenseandDenotation(PartII),”Noûs8,pp.135–56.Gödel,K.,1931:“ÜberformalunentscheidbareSätzederPrincipiaMathematicaundverwandterSystemeI,”MonatsheftefürMathematikundPhysik38,pp.173–98.1951:“Aremarkabouttherelationshipbetweenrelativitytheoryandidealisticphilosophy,”inAlbertEinstein:Philosopher-Scientist,ed.P.A.Schilpp,NewYork:TudorPublishingCompany,pp.555–62.1986:CollectedWorks,vol.I,ed.S.Feferman,J.Dawson,andS.Kleene,NewYork:OxfordUniversityPress.1995:CollectedWorks,vol.III,ed.S.FefermanandJ.Dawson,NewYork:OxfordUniversityPress.Tarski,A.,1944:“TheSemanticConceptionofTruthandtheFoundationsofSemantics,”PhilosophyandPhenomenologicalResearch4,pp.341–76.1956a:“OnDefinableSetsofRealNumbers,”inJ.H.Woodger,LogicSemantics,Metamathematics:Papersfrom1923to1938byAlfredTarski,Oxford:ClarendonPress,pp.110–42.1956b:“OntheConceptofLogicalConsequence,”inWoodger,pp.409–20.1956c:“TheConceptofTruthinFormalizedLanguages,”inWoodger,pp.152–278.WorksbyotherauthorsAdams,R.M.(1995)“Introductorynoteto*1970,”inKurtGödel:CollectedWorks,vol.III,ed.S.Feferman,etal.,Oxford:OxfordUniversityPress,pp.388–402.Anderson,C.A.(1998)“AlonzoChurch’scontributionstophilosophyandintensionallogic,”BulletinofSymbolicLogic4,pp.129–71.Boolos,G.S.andJeffrey,R.C.(1989)ComputabilityandLogic,3rdedn.,Cambridge:CambridgeUniversityPress.Smullyan,R.(1957)“Languagesinwhichselfreferenceispossible,”JournalofSymbolicLogic22,pp.55–67.Woodger,J.H.(trans.)(1956)LogicSemantics,Metamathematics:Papersfrom1923to1938byAlfredTarski,Oxford:ClarendonPress.138\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200110F.P.Ramsey(1903–1930)BRADARMENDTFrankPlumptonRamseymadelastingcontributionstophilosophy,logic,mathematics,andeconomicsinanastonishinglyshortperiod.Heflourishedduringthe1920satCambridgeUniversity,andheinteractedwiththemanynotablefiguresthere,includ-ingRussell,Moore,Keynes,andWittgenstein.Hewasbynomeansaminorfigureamongthisgroup.Hisworkmakesitclear,infact,thathewasatleasttheirintellec-tualequal,andperhapsmore,ajudgmentquiteconsistentwithopinionsexpressedbyhiscontemporaries.Ramseyheldanappointmentinmathematics,buthismainmathematicalinterestswereinitsfoundations.Hepublishedjustonepieceofrealmathematicalwork,anine-pagefirstsectionofhisinvestigationofadecisionprocedureforaspecialcaseoffirst-orderpredicatelogic(“OnaProblemofFormalLogic,”inRamsey1931).Whathepresentedthereasapreliminarytoolhassincebeenrecognizedasanimportantresult–Ramsey’sTheorem–andistheoriginofanowthrivingbranchofmathematics,RamseyTheory.Healsopublishedtwopapersineconomics,oneontaxation,andanotheronsaving,thathavealsocometoberegardedasimportantpioneeringcontri-butionstotheirsubjects(inRamsey1978).Ramseyhadwidephilosophicalinterests.HecriticizedandrevisedthelogicalsystemofPrincipiaMathematica,simplifieditstheoryoftypes,anddistinguishedbetweenthelogicalandsemanticparadoxes.Hegaveaproto-functionalistaccountofbelief,togetherwitharedundancytheoryoftruth.Hedevelopednowinfluentialaccountsofpartialbelief,reasonablebelief,probability,andknowledge.HedevelopedanaccountofcausalitycloselyrelatedtoHume’sthathasstronglyinfluencedimportantcontempo-raryaccountsoflawsofnature,andhemadeaproposalforrepresentingthecontentofscientifictheoriesviawhatarenowknownasRamseysentences.Thisisadistin-guishedcollectionoforiginalcontributionstophilosophy;itisastonishinginlightofthetragicfactthatRamseydiedshortlybeforehistwenty-seventhbirthday,inJanuary1930.ThecircumstancesofRamsey’sshortlife,andthetimeittookformanyofhisvariousprojectstoinfluenceothers,putusinanunusualpositioninstudyingandassessinghiswork.Ramseywasextremelyproductive,andhisworkalwayscontainsremarkableinsightandoriginality.Buthistimewasshort,andprobablynoneofhiseffortsapproachtherefinementofthetreatmentsthatwecanimaginehewouldhaveproduced,hadhelivedlonger.139\nBRADARMENDTFoundationsofmathematicsThoughhelaterbecameaconverttofinitism,Ramsey’swell-knownworkonthefoun-dationsofmathematicsexploresanddefendslogicism,theviewthatmathematicsis1partoflogic.RamseyadoptedthelogicofWittgenstein’sTractatusanduseditsaccountsofpropositionsandtautologytocriticizeandimproveonthesystemofPrincipiaMathematica(PM).RussellandWhiteheadusedtheramifiedtheoryoftypestoavoidavarietyofpotentialcontradictions,includingtheparadoxfoundbyRussellthatunderminedthepreviouslogicisttheoryofFrege(seeFREGEandRUSSELL).PM’sramifiedtypetheoryinvolvedadoublehierarchyof(1)typesofclassesandproposi-tionalfunctions,and(2)ordersofpropositionalfunctionswithineachtype.Bothhier-archiesweremotivatedby,andaccordingtoRamseysloppilydeducedfrom,aviciouscircleprinciple:itsideaisthatclassesmaynotincludethemselvesasmembers,norcanpropositionalfunctionsmeaningfullyapplyto,orquantifyover,themselves.Withitsarrangementofclassesintotypesaccordingtotheirmembership,thesystemofPMrejectedtheparadoxicalsetofallsetsnotmembersofthemselves.Withitsinsist-encethatmeaningfulpropositionalfunctionsbedefinedtoapplyonlytofunctionsoflowerorder,thesystemavoidedfurthercontradictions,suchasRichard’sandGrelling’sparadoxes.RamseycriticizedPMforrecognizingonlyanimpoverishedrangeofclasses,namelythosedefinablethroughitscomprehensionaxiom,buthehadnoobjectiontothetypehierarchyitimposedonclasses.Hethoughttheramifiedhierarchyofordersamongfunctionsofagiventypewasflawedintwoways,however.Itmissedtherealnatureoftheparadoxesitsoughttodefuse,anditforcedintothesystemofPManonlogicalaxiom,therebyunderminingthelogicistproject.Theaxiominquestionwastheaxiomofreducibility,assertingthatforanypropositionalfunctionofhigherorderthereisanextensionallyequivalentfunctionoflowestorder.Withramifiedtypestheaxiomisneeded,forinstance,toguaranteethatupperboundsofsetsofrealnumberswillthemselvesberealnumbers,ratherthandistinctentitieswithdefiningcharacteristicshavingahigherorderthandothedefiningcharacteristicsofthereals.WithouttheaxiomPMcouldnotcaptureimportantpartsofmathematics(calculusandanalysis,forexample),butwithit,Ramseyargued,PMmadeuseofanaxiomlackinglogicalnecessity,andsofaileditsattempttogroundmathematicsinlogicalone.RamseycitesPeanofornoticingthatRichard’sparadoxislinguisticratherthanmathematical,butheistheonerememberedfordrawingageneraldistinctionbetweenlogicalandsemanticparadoxes.ThecontradictionthreatenedbyRussell’ssetwouldoccurwithinamathematicalsystem,butnotsowithmanyotherparadoxes.Theliarparadox,Richard’sparadoxandGrelling’sparadox(whichRamseyattributestoWeyl)“occurnotinmathematics,butinthinkingaboutmathematics...[they]couldnotbeconstructedwithoutintroducingtherelationofwordstotheirmeaningorsomeequivalent”(1990:184,200).PMmadethemistakeoftreatingthesemanticparadoxesaslogicalones,bringinginramifiedtypetheory,andwithit,thereducibilityaxiom.RamseyreconceivedpropositionsandlogicalnecessityalongthelinesoftheTractatus,andheprovidedarevised,simplertheoryoftypes,nowapplyingtothevarioussymbolicexpressionsofpropositions,forwhichtheaxiomofreducibility140\nF.P.RAMSEYwasnotneeded.“Formepropositionsinthemselveshavenoorders;theyarejustdifferenttruth-functionsofatomicpropositions–adefinitetotality,dependingonlyonwhatatomicpropositionsthereare.Ordersandillegitimatetotalitiesonlycomeinwiththesymbolsweusetosymbolizethefactsinvariouslycomplicatedways”(1990:211–12).ThelogicistprojectwasdoomedbyworkdoneafterRamsey’sdeath,butitseemsRamseyhimselfabandoneditbeforethen.Hehadjokedofpreservingmathematicsfromthe“BolshevikmenaceofBrouwerandWeyl,”buthelatermadeextensivenotesonintuitionistmathematics,andBraithwaitereportsthathewasconvertedtothatviewneartheendofhislife(1931:xii,1990:219,1991a:197–220).BeliefandtruthWhatistruth?ForRamsey,whotakestruthtobefirstapropertyofbeliefsandjudg-ments,andonlyderivativelyapropertyofsentences,thisbecomesthequestion,Whatisitforabelieftobetrue?Hisansweristhatabeliefistruewhenitisabeliefthatp,andp.Heregardedthisasentirelyobvious;thedifficultpartinanalyzingthetruthofbeliefisnotwiththeconceptoftruth,butwiththeanalysisofwhatitistobelievethatp.Tosaythatitistruethattheearthisroundamountstosayingthatifanyoneweretobelievethattheearthisround,theirbeliefwouldbetrue.Whichistosay“nomorethanthattheearthhasthequalityyouthinkithaswhenyouthinkitisround,i.e.thattheearthisround”(1990:38–9,1991b:7–13).Tosay,“Everythinghebelievesistrue”amountstonomorethan“Forallp,ifhebelievesp,thenp.”Thelattersoundsoddinordinarydiscourse;ourgrammaticalhabitsdemandaverbandpushustowardtackingan“istrue”ontotheendofthelattersentence,whichthenwouldhardlybeenlight-ening.Buttheverbandtheclauseareunnecessary;averbisalreadypresentinanyofthebeliefshemayhave.Ramseyregardedhisaccountoftruthasaqualifiedcorrespondencetheory,butithascometobeknownasaredundancytheory.Ashemadeclear,ifinordertogiveanaccountofwhatitistobelievethatp,hewereforcedtorelyontheconceptoftruth,thennotmuchprogresswouldhavebeenmade.Sowhataccountsforthecontentsofbeliefs,judgments,andassertions?Ramseysketchedwhatheregardedasapragmatistaccount,influencedbyRussellandPeirce,thatlookstothecausalpropertiesofourmentalstates.Theprimarytargetofhisaccountisoccurrentlinguisticallyexpressedbelief,consciouslyassertedordenied;thecontentsofdisposi-2tionalbeliefsarederivativefromthecontentsofoccurrentthoughts.Therelevantmentalstatesaresilentorspokenlinguisticutterancesaccompaniedbyfeelingsofassentordenial,andtheircontentsaregivenbycausalpropertiestheybeartoothermentalstatesandtotheworld.Beyonddescribingthegeneralpicture,Ramseydevotedmostofhisattentiontoexplaininghowsuchcausalpropertiesexhibitpatternscorrespondingtothelogicalstructuresofthepropositionalcontentsbornbythestates.Thekeypointisthathisaccountofbeliefgenerateshisaccountofmeaning,ratherthantheotherwayaround.“TheessenceofpragmatismItaketobethis,thatthemeaningofasentenceistobedefinedbyreferencetotheactionstowhichassertingitwouldlead,or,morevaguelystill,byitspossiblecausesandeffects”(1990:51).141\nBRADARMENDTReasonablebelief,probability,andknowledge3“TruthandProbability”waswrittenin1926andpublishedposthumously(1931).Itisbestknownforitstreatmentofpartialbeliefandsubjectiveprobability.Beyondthis,however,Ramseypresentedaviewoflogicconstruedasthescienceofrationalthought,4whichhedividedintotwoparts,thelogicofconsistencyandthelogicoftruth.Atthecoreoftheviewsnowknownasprobabilism,Bayesianism(inepistemologyordecisiontheory),andsubjectivism(aboutprobability)istheideaofpartialbelief,andthereisstillnobetterintroductiontoitthanthethirdsectionof“TruthandProbability.”Weholdsomeofourbeliefsmorestronglythanweholdothers.Whatisitthatvariesamongthebeliefsheldwithdifferentstrengths?Differentbeliefswillhavedifferentcontents,ofcourse,andsometimesthecontentofabeliefmaybeaboutthestrengthofabelief,aswhenIbelievethatyoustronglybelievethatp,ordoubtthatq.Imighthaveaviewaboutthestrengthofmyownbelief,butthatview(asecond-orderbelief,wemightsay)isdistinctfromitssubject(mybeliefthatp,withwhateverstrengthithas).Soifitisnotinthecontentofabelief,whereandwhatisthestrengthofabelief?Andifwecananswerthat,canwegoontomakesystematicsenseofthedegreesofstrengththatbeliefscananddohave?Perhapswecandolittlebetterthantosaythatondifferentoccasionswearecertain,orconfident,orthinkthatmaybe...,andsoon.ButRamsey’sanswertothefirstproblem(Whatisstrengthofbelief?)illuminatesthepointofdistinguishingdifferentdegreesofstrength,anditmakesroomforarangeofstrengthsricherandmoresystematicthanistherangeindicatedbyourordinaryreports.Theessentialideaisthatthemostfruitfulwaytothinkaboutthedegreesofstrengthtowhichweholdbeliefs(degreesofbelief,forshort),istoattendtothewaysbeliefsguideusinourchoosingandacting.Adegreeofbeliefisacausalpropertyofit,“whichwecanexpressvaguelyastheextenttowhichwearepreparedtoactonit”(1990:65).Ramseyproposed,first,thatthereisaunivocalwayinwhichapropositionpentersintoaperson’sdeliberations,sothatitmakessensetospeakofthedegreetowhichsheholds5p.Ifwearewillingtoacceptthat,thenwecanlooktoaperson’sdeliberationsandpotentialchoicesforindicationsofp’sinfluenceonthem,i.e.forindicationsofherdegreeofbeliefinp.Indoingthis,weareexploringameasurementproblem,andRamseywaswellawareofboththetheoreticaldifficultiesandthepracticalcomplica-tionsthataccompanyasolutiontoit.Onbothscores,hedrewananalogytomeasure-mentinphysicalscience.Onthetheoreticalside,justasthelengthofatimeintervalbetweentwoeventsdependsinrelativisticphysicsonexactlyhowitismeasured,somaytheinfluenceofabeliefonchoice.Sojustaswhenweusetheideaoftimeintervals,whenweusetheideaofdegreesofbeliefweshouldkeeptrackofhowweproposetomeasurethem,but“formanypurposeswecanassumethatthealternativewaysofmeasuringitleadto[approximately]thesameresult”(1990:63;seealsop.68).Onthepracticalside,theintertwiningofdifferentphysicalinfluencesandthedisturbancesintroducedbymeasurementprocessesdonotundermineallourattemptstounderstandandquantifyphysicalphenomena.Soitisatleastnotobviousthatsimilarpracticalcomplicationsinstudyingbeliefandchoicewilldefeatourattemptstodoso.Ramseyfirstgaveafamiliar,ifpiecemeal,account.Asheputit,“Theold-establishedwayofmeasuringaperson’sbeliefistoproposeabet,andseewhatarethelowestodds142\nF.P.RAMSEYwhichhewillaccept.ThismethodIregardasfundamentallysound;[thoughinexactandnotgeneral]”(1990:68).Theoddsonewillofferonabet(ratiosofitspotentialpayoffs)indicateone’sdegreeofbelief.IfIamwillingtoofferhighoddsasIdefendp,mydegreeofbeliefinpishigh;ifIofferonlylowodds(demandhighoddsfrommyopponent)mydegreeofbeliefislow.Aconditionaldegreeofbeliefinp,giventhatq,isindicatedbytheoddsIwouldplaceonawageronpwhichonlypaysoffinthecir-cumstancethatqistrue.Ofcourseweundertakewagersinfrequentlycomparedtothefrequencywithwhichwehavebeliefs,andthebettingscenarioisquiteartificialasamodelforthevarietyofchoiceswemake.Beliefsandtheirstrengthsaredispositions,though,andthemeasurementsproposedviawageringareindicatorsofthem,notthedegreesofbeliefthemselves.Andinawidersense,Ramseysaid,allourliveswearebetting:“Wheneverwegotothestationwearebettingthatatrainwillreallyrun,andifwehadnotasufficientdegreeofbeliefinthisweshoulddeclinethebetandstayathome”(1990:79).Whatdegreesofbeliefoughtwetohave?Thelogicofconsistencyrequiresthatdegreesofbeliefobeytherulesofprobability.Aproposofthebettingscenario,Ramseystated–andseemstohaveunderstoodbetterthanmost–whathasbecomeknownastheDutchbookargument.Degreesofbeliefthatviolateprobabilitywouldguideapersontowardbettingarrangementsguaranteedtoyieldaloss(acertainloss,accord-ingtohisownvalues)andRamseytookthistobeanindicationofinconsistencyinthepartialbeliefs.ThemostremarkablepartofRamsey’streatmentofpartialbeliefishisgeneraliza-tionofthebettingscenariousedsofar.Generalizetheideaofabetonptoagamblehavingtheform,aifp,bifnot,whereaandbrepresentstatesoftheworldbearingvalueandobtainingaccordingtowhetherornotpdoes.Somegambleswillbefavoredoverothers,dependinguponthevaluesofaandb,andonp.Ramseyshowedthatifaperson’spreferencesamongarichsetofthesegamblesiswellarranged,accordingtostatedprinciples(transitivity,forexample,isone),thentherearenon-arbitrarymea-surementsofallthevaluesandoftheprobabilitiesofallthepropositionsp.Thesemea-surementsareattributabletothatperson’svaluesandbeliefs,andcanbetakentobethevaluesanddegreesofbeliefthatguidehischoices.Thisresultisanearlyforerun-nerofsimilardemonstrationsgivenbymanylatereconomistsandphilosophers,resultsthatareusuallytakenasfoundationsforutilitytheoryordecisiontheory.Ramseyemphasizedtheimportanceoftheresultforanaccountofpartialbelief,thoughthistheoryyieldsboth.ImportantlatertheoriesthatfollowhiminthisareSavage’sand6Jeffrey’s.Beyondthedictatesofconsistency,whatdegreesofbeliefarereasonable?WithrepeatedacknowledgmentofPeirce’sinfluence,Ramseyconceivedofthelogicoftruthalongpragmatistlines.Thebestapproachistoaskaboutthereasonabilityofthehabitsbywhichwearriveatandholdourbeliefs.Alwaysfullybelievethetruthisnotbadadvice,butitisnotaveryusefulrecommendationeither.Norisitusefulasastandardforthesortofgeneralhabitsopentohumans,habitsthatsooftenyieldpartialbeliefsratherthancertainties.Amoreappropriatestandardjudgesthehabitsaccordingtohowcloselytheirpartialbeliefscorrespondtotherateatwhichthebeliefsaretrue.Thatis,thehabitshouldyieldapartialbeliefwhosestrengthcorrespondstothefrequencywithwhichrelevantlysimilarbeliefsaretrue.Ramseyusedanillustrationinvolvingabelief143\nBRADARMENDTaboutthewholesomenessofyellowtoadstools;wecanusethebeliefsofweatherfore-casters.Aforecasterdoeswellwhenitrains70percentofthetimeinwhichherdegreeofbeliefis7/10.Ramsey’sworkontheconsistencyofpartialbeliefiswellknown.Itisworthemphasizingthatheattachedasmuchimportancetothissecondstandardcallingforalignmentofdegreesofbeliefwithfrequenciesoftruth(and,further,thathewasawareofthecomplexitiesofdevelopingit,e.g.,inidentifyinghabitsandspeci-fyingwhatclassesofcasesarerelevant).Toreturntotheaction-guidingnatureofpartialbelief:[B]eliefofdegreem/nisthesortofbeliefwhichleadstotheactionwhichwouldbebestifrepeatedntimesinmofwhichthepropositionistrue....Itisthisconnectionbetweenpartialbeliefandfrequency,whichenablesustousethecalculusoffrequenciesasacal-culusofconsistentpartialbelief.Andinasensewemaysaythatthetwointerpretationsaretheobjectiveandsubjectiveaspectsofthesameinnermeaning,justasformallogiccanbeinterpretedobjectivelyasabodyoftautologyandsubjectivelyasthelawsofcon-sistentthought.(1990:84)Ramseysaidlessaboutknowledgethanaboutreasonablebelief.Heregardedknowl-edgeastrue,certainbeliefproducedbyareliableprocess,andsoitappearsheconsid-eredittheextremecaseoffullyheldtruebelief,backedupbyaprocessthattendstoproducesuchbeliefs.Thisisanaturalextensionofhissuggestionforevaluatingourhabitsofbelief,especiallyiftherequiredreliabilitymatchesthestrengthofthebelief(certainty).Thoughhismoststraightforwardandexplicitstatementsdemandcertainty,theyareaccompaniedbydiscussionsoffallibilismandbyfurtherremarksthatsoftenthedemandtonear-certainty,practicalcertainty,orconviction“justaminutefractionshort”ofcertainty(1990:110–11,1991b:62–4).HeagreeswithRussell’sviewthat“allourknowledgeisinfectedwithsomedegreeofdoubt,”andinaparagraphonMoore’sparadoxandtheparadoxofthepreface,heshiftsfromtalkofcertaintytotalkofbeingnearlycertain.Itislikelyamistaketooveranalyzehisbriefunpublishedremarks,anditisdifficulttodeterminehowcloselyRamsey’saccountofknowledgeistiedtohisaccountofreasonablebelief.Itisclear,though,thatheendorsedareliable-processaccountofknowledge,andheisrememberedbycontemporaryepistemologistsfordoingso.Laws,causality,andtheoriesInthelasttwoyearsofhislifeRamseyworkedseriouslyoncausality,lawsofnature,andtheformalstructureofscientifictheories.Theseveralpapersonthesetopicsarenotsofinishedashisearlierwork,andinplacesindicaterapidlyevolvingviews.Thisisalsotheperiodinwhichhewasatworkonhisbookontruthandmovingawayfromlogicismtoanintuitionistviewofmathematics.Ramsey’sviewofcausalitywasnotverydistantfromHume’s.Inhisbrief1928paperonlawandcausality(1990:142–3),hesuggeststhatthedifferencebetweenuni-versalsoflawanduniversalsoffact(betweenlawlikeandaccidentalgeneralizations,wemightsay)liesintheirdistinctrolesinoursystemofknowledge.Ifweknewevery-thingandorganizedourknowledgeinadeductivesystemthatstroveforsimplicity,the144\nF.P.RAMSEYgeneralaxiomsofthesystemwouldbethefundamentallawsofnature.Theyandthegeneralizationsderivablefromthemwithoutreferencetofactsofexistencearethe“statementsofcausalimplication.”Andthoughreallywedonotknoweverything,wedotendtoorganizeourknowledgeinadeductivesystem,regarditsaxiomsaslaws,andregardasundiscoveredlawsthefutureaxiomsweexpecttoariseaswelearnmore.Ramseysoonrevisedthisview,butitsinfluencepersistsintheworkofmorerecent7philosophers,notablyinDavidLewis’sbest-systemaccountoflaws(seeLEWIS).Ramsey’srevisedtreatmentisinthe1929paper,“GeneralPropositionsandCausality.”Theviewthereisthatcausallawsdonotgettheirforcebybeingsimplefun-damentalgeneralizationsinanaxiomaticsummaryofourknowledge.Theircausalforceliesinourtrustingthemasguidesinourinferencesaboutparticularevents.Causalgeneralizations“arenotjudgmentsbutrulesforjudging‘IfImeetaf,Ishallregarditasay.’Thiscannotbenegatedbutitcanbedisagreedwithbyonewhodoesnotadoptit.”Anassertionofacausallawisanassertionnotofaproposition,butofaformulafromwhichwederivepropositionsaboutparticularevents.Itscausalchar-acterliesinthetemporalorderingoftheeventsaboutwhichitlicensesourjudgments(ydoesnotprecedef).Thespecialimportanceweattachtojudgmentswiththatorder-ingistraceabletotheimportanceofforward-lookingjudgmentsinourthinkingabout8theinfluenceouractionsmayhaveontheworld.Inthecourseofdiscussingconditionalsinthispaper,Ramseysuggeststhattheacceptabilityofaconditionalgoesbytheacceptabilityofitsconsequentaftertheantecedentishypotheticallyaddedtoone’sbeliefs:Iftwopeoplearearguing“Ifpwillq?”andarebothindoubtastop,theyareaddingphypotheticallytotheirstockofknowledgeandarguingonthatbasisaboutq;...Wecansaythattheyarefixingtheirdegreesofbeliefinqgivenp...(1990:155)IncontemporaryworkonconditionalsthisideahasbecomewidelyknownastheRamseytestfortheacceptabilityofaconditional.Inanother1929paper,Ramseyaddressestheformalstructureofscientifictheories(“Theories,”inRamsey1990).Heisparticularlyinterestedinthequestionofthecontentoftheoreticalassertions,andhowsuchcontentisrelatedtotheobservationalassertionsonwhichthetheoryisbuilt.Oneideafordemonstratingthedispensabilityoftheoreticaltermsistoshowthattheyareexplicitlydefinableintermsoftheobservationalassertions,andfurther,thatthedefinitionscanbeinvertedsothatanythingweexpressintheoreticallanguagecanalsobeexpressedwithoutrecoursetothetheoreticalterms.Ramseyworksallthisoutforaverysimpleexample;evenforthatexample,theresultsarecomplexandextremelycumbersome.Worse,ashepointsout,themethodofexplicitdefinitionscreatestheproblemthatadditionaldatanotseriouslyinconsistentwiththetheorywillneverthelessfalsifythetheoryunlessadjustmentsaremadetothemeaningsofitsterms.Isthereanotheralternative?Ramseyoffersone.Conjoinallthesentencesofthe(first-order)theory,replacetheoccurrencesofeachdistincttheoreticaltermwithasecond-ordervariable,andintroduceforeachdistinctvariableasecond-orderexistentialquantifierthatbindsitsoccurrences.TheresultisnowknownastheRamseysentenceofthetheory.Itcontainsonlyobservationalterms,isentailedbythefirst-ordertheory,anditentailsthe145\nBRADARMENDTsameparticularobservationsentencesasthefirst-ordertheory.ThisdevicehassincebeenusedbyHempel,Carnap,andmanyothersintreatmentsofthecontentandmeaningoftheories,whethertheconcernis,asabove,withtheobservationalcontentofscientifictheoriesorwith,forexample,therelationbetweenmentalandneuro-physicaltheories.Notes1L.Wittgenstein,TractatusLogico-Philosophicus,London:Routledge,1922.Ramseycon-tributedtothe1922Englishtranslationandmadecriticismsthatledtochangesinthe1933secondedition.WittgensteinmentionsRamsey’sinfluenceintheprefacetohisPhilosophicalInvestigations.2Dispositionallinguisticbelief,thatis;Ramseyalsoentertainedtheideathatachickenhasbeliefs,tobeunderstoodasrelationsbetweenitspotentialbehaviorandcircumstancesinwhichthebehaviorisappropriate.Thisresemblestheanalysishegivesofdispositionalstrengthsofbelief(seebelow);Ramsey1990:40.3SeverallaternotesexpressingafterthoughtsandfurtherideasareinRamsey1931,1991a.4Theorganizationof“TruthandProbability”wasthestartingpointforaplannedbook.MuchoftheunfinishedmanuscriptforthebookhasbeenpublishedasRamsey1991b.5Noticethattheideaofasingledegreeofbeliefisforaperson,atatime,amongacollectionofotherbeliefs.Adegreeofbeliefmayverywellchangeforavarietyofreasons,includingobservations,thepassageoftime,orchangesinotherbeliefs.6L.J.Savage,TheFoundationsofStatistics[1954],2ndedn.,NewYork:Dover,1972;R.C.Jeffrey,TheLogicofDecision[1965],2ndedn.,Chicago:UniversityofChicagoPress,1983.7D.Lewis,PhilosophicalPapers,vol.II,Oxford:OxfordUniversityPress,1986.Lewisacknowl-edges“followingtheleadof(ashorttemporalsegmentof)Ramsey,”p.xi.8Ramseycallstheserulesforjudgingvariablehypotheticals,ofwhichcausallawsareanimportantkind.Ramsey1990:149,157–9.BibliographyRamsey’sbest-knownandmostinfluentialworkisfoundinthecollectionsof1931,1978,and1990;theirphilosophicalcontentsareverysimilar.HissurvivingpapersareintheRamseyCollectionhousedattheUniversityofPittsburgh’sHillmanLibrary.Afascinatinggroupofthesepapers,almostallotherwiseunpublished,appearinthecollectiondated1991a.Asubstantialmanuscriptontruth,alsofromtheRamseyCollection,waspublishedin1991b.ThecontentsofMellor(1980)aremoreinspiredbyRamseythancenteredonhiswork.BibliographiesofRamsey’sworkappearinRamsey1931,1978,1990,andinSahlin1990;allofthebookslistedbelowcontaininterestingintroductionsorprefaces.WorksbyRamsey1931:TheFoundationsofMathematicsandOtherLogicalEssays,ed.R.B.Braithwaite,London:RoutledgeandKeganPaul.1978:Foundations,ed.D.H.Mellor,AtlanticHighlands,NJ:HumanitiesPress.1990:PhilosophicalPapers,ed.D.H.Mellor,Cambridge:CambridgeUniversityPress.1991a:NotesonPhilosophy,Probability,andMathematics,ed.M.C.Galavotti,Naples:Bibliopolis.1991b:OnTruth,ed.N.RescherandU.Majer,Dordrecht:Kluwer.146\nF.P.RAMSEYWorksbyotherauthorsMellor,D.H.(ed.)(1980)ProspectsforPragmatism:EssaysinMemoryofF.P.Ramsey,Cambridge:CambridgeUniversityPress.Sahlin,N.(1990)ThePhilosophyofF.P.Ramsey,Cambridge:CambridgeUniversityPress.147\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200111CarlG.Hempel(1905–1997)PHILIPKITCHERIntroductionCarlGustavHempelwasoneofagroupofphilosophersfromCentralEuropewhoemigratedtotheUnitedStatesinthe1930sandwhoprofoundlymodifiedthecharacterofAmericanphilosophy.TogetherwithRudolfCarnap,ErnestNagel,andHansReichenbach,Hempelwascentraltothetransitionfromlogicalpositivismtologicalempiricism.Hiswritingsnotonlysettheagendaforphilosophyofscienceinthemiddledecadesofthetwentiethcenturybutalsocontinuetoshapethisimportantfieldofphilosophy.EducatedinBerlin,Hempelwasinfluencedbyearlytwentieth-centuryattemptstoapplytheconceptsandtechniquesofmathematicallogictotheempiricalsciences,pio-neeredbyReichenbachinGermanyandCarnapandhisco-workersinVienna.TheViennaCirclehadhopedtodiagnosemosttraditionalphilosophicaldiscussionsastreat-mentsofpseudo-problems,byformulatingandapplyingaprecisecriterionofcognitivesignificance.Theenvisagedcriterionwasintendedtopickout,asmeaningful,state-mentsoflogicandmathematicsconceivedasanalytictruths,andthosenon-analyticstatementsthatadmitofempiricaltest.Thesciences,paradigmaticallythephysicalsci-ences,weretocountasmeaningfulbecausetheysatisfiedthelattercondition,whereasthestatementsoftraditionalphilosophyweretobeexposedasneitheranalytictruthsnorsusceptibletoempiricaltest,andconsequentlydevoidofcognitivesignificance.Duringthe1930s,logicalpositivistsmadesuccessiveattemptstomakethecriterionofcognitivesignificancesufficientlysharptoperformtheplannedsurgery.Hempelpar-ticipatedactivelyinthesediscussions,examiningvariousproposalsandidentifyingdif-ficulties.Hiseffortsculminatedinsyntheticessaysthatarguedfortheimpossibilityofthepositivistprojectanddrewtheblueprintforlogicalempiricism,thephilosophicalstructurewithinwhichhewouldbuildsubstantivepositions.CognitivesignificanceHempelneverwaveredinhiscommitmenttotheideathat“thegeneralintentoftheempiricistcriterionofmeaningisbasicallysound”(1965:102),thatis,thatmean-ingfulempiricalstatementsarethosethatadmitofexperientialtest.Hebelieved,148\nCARLG.HEMPELhowever,thatattemptstostatepreciselogicalcriteriaofempiricalsignificanceencountersystematicdifficulties.Ifweinsistthatempiricalstatementsarethosethatadmitofconclusiveverification,thenthecriterionwilldebaruniversalstatements(includingthelawsofthevarioussciences);toproposethatempiricalstatementsmustallowconclusivefalsificationisequallyhopeless,sincethiswoulddenymeaning-fulnesstoexistentialclaims.Norcanweproposetoformulateacriterionofmeaning-fulnessbyrequiringdefinabilityofalltermsinsomelanguagewhosenonlogicalvocabularycanbelearnedinapplicationtoobservationalentitiesandproperties,forthespecialtermsthatplaysofruitfularoleinthephysicalsciences,expressionslike“statefunction”and“covalentbond”cannotbegivenexplicitdefinitions.Hempelcon-cludesthattherequirementofmeaningfulnessmustimitatethewayinwhichscien-tistsextendtheresourcesofeverydaylanguage,towitbysystematicallyconnectingsentencesinwhichunfamiliarexpressionsappearwithsentenceswhosevocabularyisunproblematic.MyreconstructioncondensesawealthofsubtlepointsthatHempelarticulateswithcharacteristiclucidity,butitbringsouttwoimportantaspectsofthecritiqueofformalcriteriaofcognitivesignificance.First,Hempelisnotsimplyconcernedwithaphilo-sophicalenterprisethatintendstotranscendwhataretakentobesteriledisputes,but,inaddition,toprovideanassessmentofmovementsinthenaturalandsocialsciencesthatcallforempiricalconditionsfortheapplicationofbitsoftheoreticalvocabulary.Heoffersadefinitiverebuttaltotheoperationalistdemandthatnewly-introducedtermsmustbeassociatedwithanexperientialprocedureforapplyingthem(seeHempel1965:ch.5).Second,thatrebuttaldependsontakingtheunitofappraisaltobe“sentencesformingatheoreticalsystem”(1965:117).Explicitlyacknowledgingthatthismakescognitivesignificance,“amatterofdegree”(ibid.),heoffersfourmainwaysinwhichtheoreticalsystemsshouldbeevaluated:byscrutinizingtheconnectionsamongtheoreticaltermsandtheconnectionstostatementscouchedinobservationalvocabulary;byconsideringtheexplanatoryandpredictivepowerofthesystem;byappraisingthesimplicityofthesystem;andbyassessingthedegreeofconfirmationbyempiricalevidence.Forthepositivisteagertoprickthepretensionsoftraditionalmetaphysicsortheoperationalistconcernedthatsomeareaofscienceshouldberigorouslydeveloped,thisconcludingcataloguemustcomeasadisappointment.Inplaceofaclearcriterionthatcanputanendtodisputes,itappearsthatthesefuzzyvirtuesoftheoreticalsystemswillbehardtoidentifyandthatwearefatedtocontinuethewranglesofthepast.Hempel’sreplywouldbetwofold.First,hewouldpointoutthattheweaponsforwhichpositivistsandoperationalistsyearnaresimplynottobehad.Second,hewouldinsistthathislistofmodesofappraisalmaybetheendofareflectiononthefateofpositivism,butthatitisonlythebeginningofaseriousphilosophyofscience.Thetaskforlogicalempiri-cismistosay,aspreciselyaspossible,whatkindsoflinguisticconnectionsarepresentinvirtuoustheoreticalsystems,whatmakesforexplanatoryandpredictivepower,whatcountsassimplicity,andhowempiricalstatementsareconfirmedbytheresultsofexperientialtests.ForHempel,then,therearefourmainproblemsofthephilosophyofscience.Thelastproblem,theissueofempiricalconfirmation,isalsocentraltothetheoryofknowledge,and,sinceontologicalquestionsdependonobtainingaclearviewofthe149\nPHILIPKITCHERkindsofentitiestowhichthesciencescommitus,clarificationoftheoreticalstructureispertinenttodiscussionsinmetaphysics.Throughouthiscareer,Hempelmademajorcontributionstothreeofthefourproblemshehighlighted:heofferedatheoryofquali-tativeconfirmation,providedmagisterialdiscussionofthefruitfuluseoftheoreticalvocabulary,anddelineatedanaccountofscientificexplanationagainstwhichallsub-sequenttreatmentsofthistopicmustbemeasured.Otherlogicalempiricists,particu-larlyReichenbach(1938)andNelsonGoodman(1949),tookuptheproblemofunderstandingsimplicity,althoughthistopicneverachievedthesameprominencewithinlogicalempiricismastheotherissues.Aphilosopher’slifeworkrarelyconformstoaneatstructure,andHempel’sisnoexception.Duringthecourseofhiscareer,hewroteimportantessaysonthecharacterofmathematicsandtherelationsbetweenmathematicsandthenaturalsciences(1945a,1945b).Healsoretainedastronginterestinthepropertiesofsystemsofclas-sification,fromhisearlyattempttoapplylogicalnotionsintaxonomy(HempelandOppenheim1936)tovariousattemptstoevaluateproposedclassificatoryschemesinpsychiatryandinthesocialsciences(1965:chs6and7).Despitetheundeniableinflu-encethatthesestudieshavehad,Hempel’sattacksontheproblemsofconfirmation,theory-structure,andexplanationare,Ibelieve,hismostenduringaccomplishments.Thefollowingsectionswillconsidertheminorderofascendingimportance.QualitativeconfirmationSupposethathisahypothesisinwhosetruth-valueweareinterested,andthateisastatementthatreportstheresultofsomeempiricaltest.Thegeneralproblemofcon-firmationistounderstandthenatureoftherelationofevidentialsupportbetweenhande.Thisgeneralproblemencompassesseveralspecificquestions:wemightaskthedegreetowhichewouldsupporth(thequantitativeproblem);alternatively,wemightinquireaftertheconditionsunderwhichewouldsupporthatall(thequalitativeproblem);anintermediatequestionprobestheconditionsunderwhichesupportshmorethane¢supportsh¢(thecomparativeproblem;notethatemightbethesamease¢orhidenticalwithh¢).Inhisextensiveinvestigationsofinductivelogic,Carnapfocusedonthequantitativeproblem,consideringformallanguagesadequatefortheformula-tionoffragmentsofscience,andattemptingtodefine,forawideclassofstatementshandewithintheselanguages,thedegreetowhichewouldconfirmh.Bycontrast,Hempeltookthequalitativeproblemtobemorefundamental,andendeavoredtospecifyconditionsunderwhichsingularstatements(conceivedasascribingpropertiestoobjects)wouldconfirm,disconfirm,orbeneutralto,ahypothesis(characteristicallythoughtofasalawlikegeneralization).Hempel’streatmentisnoteworthynotjustforhispositiveproposalbutforhisdisclosureofinterestingdifficultieswithapparentlyplausibleideas.Supposethatthehypothesisofinterestisthegeneralizationthatallravensareblack,formalizedas(x)(Rx…Bx).Itisverynaturaltobelievethatthegeneralizationissupportedbyobservingblackravens.Sowemightarriveatthegeneralproposalofconfirmationbyinstances:thehypothesis(x)(Rx…Bx)isconfirmedbyanysentenceRa&Ba.AsHempelpointsout,themannerinwhichweformulateahypothesisshouldmakenodifferencetotheclassofstatementsthatconfirmit.Thus,ifhandh¢arelogicallyequivalentany150\nCARLG.HEMPELethatconfirmshshouldconfirmh¢,andconversely.Byelementarylogic,(x)(Rx…Bx)islogicallyequivalentto(x)(-Bx…-Rx).Thethesisofinstanceconfirmationnowtellsusthatthelatterisconfirmedby-Ba&-Ra,which,bytheprincipleaboutlogicalequiv-alence,mustalsoconfirm(x)(Rx…Bx).Revertingtoourinterpretationofthenonlog-icalvocabulary,wediscoverthatthegeneralization“Allravensareblack”isconfirmedbystatementsthattellusthataparticularobjectisneitherblack,noraraven.Apparently,learningthattherightmostshoeinmyclosetiswhitewouldsupportanornithologicalgeneralization!The“paradoxoftheravens”hasinspiredalargesubsequentliterature.Hempel’sowndiagnosiswasthatthereisnogenuineparadox,andthatanysenseofsurprisestemsfrom“misguidedintuitions”(1965:20).Whetherornotthisisso,thereisnodoubtthatHempel’sfurtherinvestigationsdisclosesevereproblemsinnaturalconceptions.Manyphilosophers,andscientistsreflectingonmethodologicalissues,haveacceptedthefundamentalideaofhypothetico-deductivism,towitthathypothe-sesareconfirmedwhentheirconsequencesarefoundtobecorrect.Themoststraight-forwardwaytoformulatethatideaisastheConverseConsequenceCondition:ifeisaconsequenceofh,theneconfirmsh.Unfortunately,thatcondition,coupledwitharequirementthatishardtoresist,generatestheconclusionthatanystatementwillconfirmanyhypothesis.ThefurtherrequirementistheEntailmentCondition:evidencestatementsconfirmthelogicalconsequencesofthehypothesestheyconfirm.Givenanystatemente,eisaconsequenceofh&e(whateverhmaybe);sobytheConverseConsequenceConditioneconfirmsh&e;hisaconsequenceofh&e;hencebytheEntailmentConditioneconfirmsh.Hempel’sownaccountofqualitativeconfirmationavoidedthisdifficultybyaban-doningtheConverseConsequenceCondition.Instead,heproposedthatdirectconfirmationresultswhenanevidencestatemententailsarestrictedversionofthehypothesis,effectivelywhatthehypothesiswouldsayiftherejustexistedtheindivid-ualsmentionedintheevidencestatement.Moreexactly,supposethattheevidencestatementecontainsthenamesofexactlytheindividualsa1,...,an;thenedirectlycon-firms(x)(Rx…Bx)justincaseeentailseachofthestatementsRa1…Ba1,...,Ran…Ban.Thegeneralnotionofconfirmationisobtainedbyproposingthateconfirmshjustincasehisentailedbyaclassofsentenceseachmemberofwhichisdirectlyconfirmedbye.Hempel’sapproachisunabletoaccountfortheconfirmationofsentencescontain-ingtheoreticalvocabularybystatementsformulatedinmorebasicterms.ClarkGlymourattemptedtoextendtheHempelianapproachtoofferanaccountofqualita-tiveconfirmation(orofrelevantevidence)thatwouldaddressthisdifficulty(Glymour1980),andhisproposalhasgivenrisetoextensivesubsequentdiscussion.Hempelhimselfbecameconvincedthatthegeneralapproachcouldnotsucceed,onthegroundsthatNelsonGoodman’s“newriddleofinduction”showedtheinadequacyofanypurelysyntacticalanalysisofqualitativeconfirmation(seeGOODMAN).Arguingthatnotalluniversalgeneralizationsaresupportedbytheirinstances,Goodman(1955)exposedthedifficultiesofdistinguishingthosegeneralizationsthatcanbeconfirmedinthiswayfromthosethatcannot.Ironically,muchcontemporarythinkingaboutconfirmationdivergesfromHempel’streatmentataveryearlystage.Themostinfluentialmorerecentproposalis151\nPHILIPKITCHERBayesianism,apositionthatdescendsfromCarnap’sinvestigationsininductivelogicandthatattemptstounderstandhowdegreesofconfirmationofhypothesesadjustinthelightofevidence.ForBayesians,theproblemofquantitativeconfirmationisprimaryandthesolutiontothequalitativeproblemisgeneratedinatrivialfashionfromasolutiontothequantitativeproblem.Tosolvethelatter,weneedtobeabletoassessthevalueoftheprobabilityofthehypothesisgiventheevidence,Pr(h|e);assumingthatthatcanbedone,wecansaythateconfirmsh(orconfirmshrelativetobackgroundinformationB)justincasePr(h|e)>Pr(h)(orPr(h|e&B)>Pr(h|B)).Hempel’sdiscussionsofconfirmationremainofinterestnotsomuchbecauseofhispositiveproposalaboutqualitativeconfirmationasforhiscarefulexposureofdifficultiesinintuitivelyattrac-tivewaysofthinkingaboutconfirmation,andforhisrecognitionofconstraintsonanyadequatesolution.TheoriesLogicalempiricismbeganwiththeconvictionthatthetoolsoflogicdevelopedbyFrege,Russell,andtheirsuccessorscouldbeusedtomakeclearandexplicitthestructureofscientifictheories(seeFREGEandRUSSELL),and,indeed,eveninthe1920s,Reichenbachhadofferedanaxiomatizationofthespecialtheoryofrelativity,intendedtoexhibitwhichpartsofthetheorywereconventionalstipulationsandwhichmadesubstantiveempiricalclaims.Almostunselfconsciously,thelogicalempiriciststookoverthelogi-cian’sconceptionofatheoryasadeductivelyclosedsetofsentencesinsomesuitableformallanguage.Recognizingthatanimportant–and,fromtheempiricistviewpoint,problematic–featureofthetheoriesinphysicsandchemistrythatmostimpressedthemwasthepresenceofspecialvocabularythatresistsexplicitdefinitioninobserva-tionalterms,logicalempiricistsformulatedadistinctiveviewaboutscientifictheories.Ascientifictheoryisadeductivelyclosedsetofstatementsinafirst-orderlanguagewhosenonlogicalvocabularydividesintotwosubsets,thebasicvocabulary(oftenunderstoodascontainingthosetermswhoseapplicationcanbemadeonthebasisofmoreorlessdirectobservation)andthetheoreticalvocabulary(theremainder);thestatementswhoseessentialnonlogicalvocabularycontainsonlytheoreticaltermsarethetheoreticalpostulatesofthetheory,whilestatementswhoseessentialnonlogicalvocabularycontainsboththeoreticalandbasictermsarethecorrespondencerules(avarietyofotherdesignationsforthislastclassofstatementsappearsinlogicalempiri-cistwritings,but“correspondencerules”isthemostpopularlocution);thefunctionofthecorrespondencerulesistoprovidethetheorieswithempiricalcontent,andthey(orasubsetofthem)areoftenconceivedasprovidinganinterpretation(orpartialinter-pretation)ofthetheoreticalvocabulary.ThisconceptionofscientifictheoriesbecamefullyexplicitinthewritingsofCarnap(1956),Nagel(1962),andHempel(1958,1965:ch.8),and,perhapsasaresidueoftheconcernsaboutcognitivesignificance,eachoftheseauthorssoughtwaystochar-acterizethosetheoreticalcontextsinwhichtheintroductionoftheoreticalvocabularyservedimportantscientificpurposes(seeCARNAP).Hempel’sdiscussionsrevolvedaroundafamilyofquestions.Towhatextentcanthecorrespondencerulesbeviewedasfunctioningasdefinitions?Isitproblematictoconcedethatthecorrespondencerulesonlyofferpartialinterpretationsofthetheoreticalvocabulary,orshouldthisbeseen152\nCARLG.HEMPELasasymptomoftheopenendednessofscientificresearch?Canweeliminatethetheo-reticalvocabularywithoutanyscientificloss?Isitreasonabletotreatthetheoreticaltermsascomponentsofaformalapparatusformakingexperientialpredictions,orshouldwesupposethatthosetermsrefertoentitiesandpropertiesthatunderlietheobservablephenomena?Hempel’sdiscussionofthesequestionsembodiesacautiouslyrealisticattitude.Hedoesnotthinkthatwecanprovideafullexplicitdefinitionoftheoreticalvocabularyinobservationalterms,notevenforthoserelativelylow-levelpartsofsciencethatmakeuseofdispositionalconcepts.Althoughhebelievesthatcorrespondencerulesarevehi-clesofpartialinterpretation,hesuggeststhatwecannotneatlyseparatethepartsofatheorythatfunctiontopindownthemeaningsofourtermsfromthosethatmakegenuinelyempiricalclaims,apointinwhichheconcurswithW.V.Quine’scelebratedcritiqueoftheanalytic/syntheticdistinction(Quine1953)(seeQUINE).Partialinter-pretation,Hempelbelieves,hasheuristicadvantages,allowingustointroducenewcor-respondencerulesasweextendthetheorytocopewithpreviouslyuntreatedphenomena.Further,totheextentthattheoreticalvocabularyisineliminable,heholdsthattheevidenceleadingustoadoptthetheoryoughttoinclineustoacceptitspos-tulatesastrueandthustotreatitstheoreticalvocabularyasreferringtoentitiesbeyondthereachofordinaryobservation.Thecrucialissuethusturnsouttobewhetherornotthereisagenerallyavailablemethodforeliminatingtheoreticalvocabulary.Hempelapproachestheproblembyformulatingadilemma.Startingfromthepremisethatthefunctionofatheoryisto“establishdefiniteconnectionsamongobservablephenomena”(1965:186),hesug-geststhatwhensuchconnectionsareestablishedwedonotneedanydetourthroughatheory,sincethetheorywillimplyaconditionalstatement,couchedinthebasicvocabulary,thatassertstheconnection.So,ifthetheoreticaltermsandprinciplesservetheirpurpose,theycan,inprinciple,beeliminated,andarethusunnecessary.If,ontheotherhand,thetheoreticaltermsandprinciplesdonotestablishtheintendedconnec-tions,theydonotservetheirpurpose,andareconsequentlyunnecessary.Thisis“thetheoretician’sdilemma”(1965:ch.8).Behindthedilemmastandsaviewoftheuseoftheoriesinscientificpractice.Itisasthoughthescientistfeedssomedescriptionofobservablesintothetheoreticalmachin-ery,thegearsturn,andtheoutputisanotherstatementaboutobservables.Thepointmaybetopredictsomething(theoutputstatementisonewedidn’tknowbefore)oritmaybetoexplainsomething(wealreadyknewtheoutputstatementbutdidn’tseewhyitwastrue).Ineithercase,thetheoreticalstatementsfunctiontolicenseaninferencefromtheinputtotheoutput,andthusmustsupporttheconditionalstatement,“Ifinputthenoutput.”Whythencanwenotmanagewiththesetofallsuchconditionalstatementscorrespondingtothetransitionsthatthetheorywouldlicense?Afirstresponseisthattheresultantsetwouldbeextremelyunwieldy,thatthetheoryasactuallypresentedprovidesaconcisewayofrepresentingadisparateclassofcon-sequences.AsHempelandhiscolleaguesclearlysaw,however,aresultduetothelogicianWilliamCraigconstructsarecursiveprocedureforgeneratingtheclassofcon-sequences,withoutsteppingoutsidethebasicvocabulary.(ItisinterestingtoreflectthatthesignificanceattributedtoCraig’stheoremrevealedtheholdthatthelogician’scon-ceptionoftheoriesasrecursivelyaxiomatizablecontinuedtoexertonlogicalempiricist153\nPHILIPKITCHERdiscussionsofscientifictheories.)Intheend,Hempel’sresponsetothisandtokindredsuggestionsforeliminatingtheoreticalvocabularydrawsontheproposalthatthetaskofascientifictheoryisnotonlytoachievedeductivesystematizationofobservablephe-nomenabutalsotoprovideinductivesystematization,andwehavenoreasontobelievethatanyoftheeliminationprocedureswillsatisfythisfurtherconstraint(1965:214–15;notethatinthecaseoftheCraigiansurrogate,Hempelisabletoarguethatthesubstitutewillnotachieveanyinductivesystematization).Itisatfirstsightironicthatthephilosopherwhocontributedmosttoourunder-standingofscientificexplanationoverlookedarelativelyobviouspointaboutproposalsforeliminatingtheoreticalterms:eventhoughtheymightbeabletomimicthepredic-tivesuccessesofgenuinetheories,itseemsthattheywouldincursevereexplanatorylosses.Whateversetofbruteempiricalruleswemightdeviseforpredictingtheout-comesofbringingsubstancestogetherinvariousproportionswouldfailtodelivertheexplanatorybenefitsweobtainfromembeddingempiricalgeneralizationswithinthetheoreticaltreatmentofshell-fillingandofionicandcovalentbonds.Ondeeperreflec-tion,however,wecanseethatappealtoexplanatorypowerwouldhaveledHempelintouncomfortablequestionsaboutthesufficiencyofhispreferredaccountofscientificexplanation.Forthereiseveryreasontothinkthatsomeoftheproceduresforelimi-natingtheoreticaltermswouldnotjustdeliversingularconditionalstatementsbutgeneralizationsfromwhichsuchsingularconditionalscouldbederived;becausethegeneralizationswouldserveastheneeded“coveringlaws”inHempel’sschemataforexplanation,bythestandardsoftheHempelianaccountofexplanationtheexplana-torylosswouldbeindiscernible.Duringthe1960sand1970s,theaccountoftheoriesfavoredbyHempel,Nagel,Reichenbach,andotherlogicalempiricistsacquiredthename“thereceivedview,”andlikemostdoctrinessodesignatedcameundervigorousattack(seeSuppe1970forthor-oughanalysis).Oneprincipaldifficulty,recognizedbyHilaryPutnam,wasthattheaccountconflatedtwodistinctions,thedistinctionbetweenobservationalandtheoreti-caltermsandthatbetweenobservableandunobservablethings;asPutnamnoted,sometheoreticaltermsnameobservablethings(“oscilloscope”)andunobservablethingscanbepickedoutusingobservationalterms(“peopletoolittletosee”).Putnam’sobserva-tions,andthediscussionsofreferencethatheandothersinitiated(Kripke1971,Putnam1973)undercuttheoldconcernthatscientistsaresimplyunabletospecifythereferentsoftheirtheoreticalvocabularies.Adifferentlineofobjectionattackedtheideathatatheoryisalinguisticitem.Severalauthors(includingSuppes1967,vanFraassen1980)drewinspirationfrommodeltheoryratherthanfromthesyntaxoflogicalsystems,proposingthattheoriesaretobeidentifiedwithfamiliesofmodels.Theircritiquesinitiatedadebate,asyetunresolved,aboutthecorrectanalysisofthenotionofascientifictheory.Itisperhapsworthrecallingthatboththenewer“semanticconceptionoftheories”andtheolder“syntacticaccount”(or“receivedview”)arephilosophicalreconstructionsoftheprac-ticesofscientists,andthatthestandardsofadequacyforareconstructiondependonwhatpurposes–philosophicalorscientific–oneintendstoachieve.Thequestion“Whatistherealstructureofascientifictheory?”maysimplybeabadquestion,and,dependingonouraims,wemaydrawononeoranotheroftheproposedaccounts(oronsomethingcompletelydifferent).Totheextentthatthesyntacticconceptioncon-154\nCARLG.HEMPELtinuestobevaluableinsuchenterprises,wecanexpectthatHempel’slucidandcarefuldelineationsofpossibilitiesandconstraintswillremainpertinent.ExplanationIntheearlydecadesofthetwentiethcentury,thethoughtthatoneoftheaimsofthesciencesistoprovideexplanations,traditionallypopular,sufferedatemporaryeclipse.Thinkingthatappealstotheexplanatorypowerofatheoryreflectedpurelysubjectivejudgments,scholarswritingaboutsciencetendedtoconcentrateonthecriterionofpredictivesuccess(see,foroneamongmanyexamples,Pearson).Hempelplayedtheleadingroleinrestoringtherespectabilityoftheconceptofscientificexplanation.Fromhisearliestdiscussionsofthetopic,heinsistsontheobjectivecharacterofscientificexplanation(see1965:234(originallywrittenin1942);also1966:ch.5).Takingupathemealreadysoundedbyearlierempiricists(forexampleJohnStuartMill),andperhapsasoldasAristotle,hesuggeststhatexplainingafact,state,oreventconsistsinshowingwhythatfact,event,orstatecouldhavebeenexpectedtooccur,giventhelawsofnature.Thekeytoexplanationisnomicexpectability.Hempelproposedthatanexplanationisanargumentwhoseconclusionisastate-mentdescribingthephenomenontobeexplained(thisstatementistheexplanandum)andwhosepremises(theexplanans)includeatleastonelawofnature.Althoughhisearlywritingsconcentratedoncasesinwhichtheargumentisdeductive,hewasexplicit,fromthebeginning,thatsomeexplanationsarenon-deductivearguments.Healsotookconsiderablepainstopointoutthat,asactuallygiven,explanationsmaynottaketheidealformhespecified.So,forexample,historiansdevelopexplanatorynarra-tivesthatarefarfromcompletearguments,andyet,Hempelcontended,theexplana-toryforceoftheirworkderivesfromthepossibilityofrecognizinggenerallawsofnature,whichincombinationwiththeclaimstheyadvancewouldyieldacompellingargumentfortheexplanandum.TheDeductive-Nomological(D-N)modelofexplanationcanbeencapsulatedinaschema:deductiveexplanatoryargumentstaketheformC1,...,CmL1,...,LnEwheretheCiarestatementsreportingparticularfacts,theLjarelawsofnature,whosepresenceisessentialtothevalidityoftheargument,andEistheexplanandum.(Hempel’smodeldoesnotrequirethattherebeanyCs,althoughtheremustbeatleastoneL;hispresentationssometimesidentifyEasastatementofparticularfact,butheallowsforexplanationsofthisformwhoseconclusionsarelaws,includingprobabilis-ticlaws.)Tobeagenuineexplanation,thepremisesofanargumentfittingtheschemamustallbetrue.Ifoneormoreofthepremisesisnottrue,thentheargumentcountsasapotentialexplanation.Inthe1940s,HempelhopedtoarticulatetheD-Nmodelmoreprecisely,andhepro-posedaformalexplicationofthenotionoflawandofdeductiveexplanation(HempelandOppenheim1948;seeHempel1965:ch.10).Unfortunately,thisattemptprovedvulnerabletotrivializingcounterexamples,and,inanyevent,Goodman’sexplorations155\nPHILIPKITCHERoflaws,counterfactuals,andinduction,convincedHempelthatnoformalaccountofscientificlawscouldbegiven.Thus,throughoutthe1950sand1960s,hisworkonscientificexplanationfocusedonshowinghowhispreferredapproachtoexplanationilluminatedaspectsofthenaturalandsocialsciencesandhowitcouldbeextendedtoincludenon-deductivearguments.Thelattertaskwascomplicatedbyanimportantdisanalogybetweendeductiveandinductivearguments.Addingextrapremisestoadeductivelyvalidargumentpreservesvalidity,buttheincorporationofnewinformationintoanargumentthatisinductivelystrongmaynotonlyunderminetheargumentbutevensupportacontraryconclusion.So,totakeoneofHempel’sownexamples,toclaimthatJonesissufferingfromastrep-tococcalinfectionandthatheisbeingtreatedwithpenicillin,togetherwiththeprob-abilisticlawthat99percentofthosetreatedwithpenicillinrecoverfromsuchinfectionsconfershighprobabilityontheconclusionthatJoneswillrecover,butifwenowlearnthatthisparticularstreptococcalinfectionispenicillin-resistantthenwehavestrongreasonsforthinkingthatJoneswillnotrecover.Hempel’smodelofInductive-StatisticalExplanation(I-S)proposedthatI-Sexplana-tionsareargumentswithtruepremisesoftheform:Pr(B|A)=rAc[r]BcHerethedoublelineindicatesthatthepremisesbestowontheconclusiontheprob-abilityr,whichissupposedtobeclosetoone.Toblocktheproblemofthe“ambiguityofstatisticalexplanation,”Hempelimposesthe“requirementofmaximalspecificity.”Ifsistheconjunctionofthepremisesoftheexplanation,andifkisastatementlogi-callyequivalenttotheset,K,ofacceptedsentences,thenifs&kimpliesthatcbelongstoasubsetA*ofA,thens&kmustalsoimplythatPr(B|A*)=r*wherer*=runlesstheconditionalprobabilityofBonA*issimplyamatterofprobabilitytheory(as,forexample,whenA*isthenullset).Thisintricateconditionisintendedtorequirethatwealwaysemploythemostspecificprobabilisticinformationwehave,and,asHempelexplicitlynoted,itintroducesanunwelcomerelativizationintotheaccountofexpla-nation,for,unlikeD-Nexplanations,I-Sargumentsonlyqualifyasexplanationsrela-tivetoaparticularstate,K,ofourknowledge.TheCoveringLawModelofExplanation,comprisingtheD-NandI-Smodels,wasenor-mouslyinfluential,notonlyrestoringtherespectabilityoftheconceptofexplanationbutalsosparkingmethodologicaldiscussionsinthesocialsciences.Themany-sidedcharacterofHempel’sluciddiscussions,especiallyinthetitleessayofAspectsofScientificExplanation,providesamodelforphilosophicalexplorationofanimportantmetascientificconcept.Nonetheless,forallthesubtletyofhistreatment,Hempel’saccountisnolongerwidelyacceptedamongphilosophersofscience(althoughitcon-tinuestobeadoptedinotherphilosophicaldebatesandinthemethodologicalreflec-tionsofnaturalandsocialscientists).SomeofthedifficultiesstemfromproblemswithwhichHempelstruggled.Theintri-caterequirementofmaximumspecificityprovedinadequatetosalvagethenotionof156\nCARLG.HEMPELI-Sexplanation,yieldingtheunwelcomeconsequencethatbonafideinductiveexplana-tionsturnouttobetacitlydeductive(Coffa1974).Anotherdifficultyofthemodelofprobabilisticexplanationliesinthefactthatwecanapparentlyexplaineventsthatareunlikelytooccur:eventhoughitmaybeimprobablethatanatomicnucleuswillundergoaparticularsequenceofdecay,wecanstill,itseems,usequantumphysicstoexplaintherareoccurrencesthatdotakethatpath.Evenintherealmofpurelydeduc-tiveexplanation,thereareformidablechallenges.AsHempelhimselfnoted,aderiva-tionofBoyle’sLawfromtheconjunctionofthislawwithKepler’slawswouldsatisfytheD-Nschema,eventhoughanysuchderivationisexplanatorilyworthless.Howdowedistinguishsuchargumentsfromtheexplanatoryderivationsoflaws,forexamplethederivationofKepler’slawswithinNewtoniangravitationaltheory?Perhapsthemostsevereproblemscamefromaclusterofexamplesthatshowedhowfamiliarasymmetriesthatoccurinthecontextofcausaljudgmentsalsoaffectourassessmentsofexplanatorypower.Supposethataflagpolecastsashadowofaparticu-larlength.Usingthelawofrectilinearpropagationoflight,togetherwithfactsabouttheheightofthepoleandtheelevationofthesun,itispossibletoderivethelengthoftheshadow,aderivationthatfitstheD-Nmodel.Sofar,sogood,sincethatparticularderivationseemsgenuinelyexplanatory.Thetroubleisthatwecanalsoworkintheoppositedirection.Giventhepropagationlaw,theelevationofthesunandthelengthoftheshadow,wecanderivetheheightofthepole,andthisderivationfitstheD-Nschemaequallywell(Bromberger1966).Anaturalresponsetoexampleslikethisistodeclarethatopaqueobjectsproduce(orcast)shadowsandthatshadowsdonotproducetheassociatedobjects,sothatthereisacausalasymmetryunrepresentedintheHempelianschema.Hempel’sscatteredremarksabouttheconnectionbetweenexplanationandcausa-tionpresentaclearpictureofhisposition.InfluencedbyHumeanworriesaboutthenotionofcausation,heholdsthatourunderstandingofcausalrelationsisgroundedinourabilitytosubsumephenomenaunderlawlikeregularities.Theconceptofexpla-nationispriortothatofcausation,inthataclaimthatccausedeisalwaysderivativefromthethoughtthattheoccurrenceofewouldbeproperlyexplainedbyanargumentinwhichadescriptionofcfiguredamongthepremises,anargumentsatisfyingthecovering-lawmodel.Hence,Hempelcannotappealtocausalasymmetriestoreformu-latehisaccountofexplanation,and,onthefewoccasionsonwhichheconfrontsexam-plesthatembodysuchasymmetries,hearguesstrenuouslythatourintuitiveresponsestothesecasesshouldnotbetrusted(forexample,1965:352–3).Withthearticulationofafamilyofinstanceslikethatoftheshadow-castingflagpole,attemptstodenydif-ferencesinexplanatoryworthcametoappearevermoredesperate,andmostphiloso-phersofscience(includingHempelhimself)haveconcludedthattheproblemofexplanatoryasymmetrycannotbedismissedasillusory.ForaboutaquarterofacenturyHempel’saccountofscientificexplanationalmostachievedphilosophicalconsensus.Sinceitsuccumbedtoahostofproblemsandcriti-cismsnosuccessorapproachhasgarneredsimilarsupport.Inspiredbytheproblemofasymmetry,severalphilosophershaveofferedaccountsthatinvoketheconceptofcau-sation(see,forexample,Humphreys1989,andSalmon1984,1998).OthershavetriedtopreservethemainfeaturesofHempel’saccountbydevelopinganideathatreceived157\nPHILIPKITCHERpassingattentioninhisownwritings,andsuggestingthatexplanationconsistsintheunificationofthephenomena(Friedman1974,Kitcher1981).Yetothershavecon-tendedthatexplanationisanactivitywhosecrucialpropertiesvarywithcontext(Achinstein1983,vanFraassen1980:ch.5).Alltheexistingaccountsfacemajorobstacles(oftengleefullynotedbythepartisansofrivalaccounts).Ifthereisacon-sensus,itscentraltendencyisthat,whileHempel’scovering-lawmodelisinadequate,itisexemplaryindemonstratingtherange,rigor,andclaritythatanysatisfactorytheoryofexplanationshouldstrivefor.Hempel’slegaciesInthepastdecades,logicalempiricismhasbeencriticizedforvariousshortcomings:neglectofthehistoricaldevelopmentofscience(Kuhn1962/1970),overemphasisonthesearchforlawlikeregularitiesinnature(Cartwright1983,1999),andfailuretoappreciatetheautonomyofexperimentalpractice(Galison1987,Hacking1983).Atthesametime,manyphilosophershaveproposedthatHempelandhisco-workersadoptedanunnecessarilyrestrictiveviewoftheformalresourcesonwhichaccountsofconfirmation,theories,andexplanationmightdraw.Despitethesecomplaints,phi-losophyofsciencecontinuestopursuetheagendathatHempelsolucidlyarticulated,and,ifthesetofquestionshasbeenenlargedandtheHempeliananswersarenolongerwidelyaccepted,itwouldbefoolhardytotackletheseproblemswithoutthoroughawarenessofHempel’smanyinsights.NoessayonCarl(“Peter”)Hempelwouldbecompletewithoutsomerecognitionofhisextraordinarypedagogicalinfluence.Notonlywashetheauthorofoneofthegreatintroductionstoanyfieldofphilosophy(Hempel1966),but,throughlecturesandsemi-nars,hewasaninspirationtogenerationsofundergraduates,graduatestudents,andyoungerphilosophers.Thosewhoknewhimsaw,againandagain,ararecombinationofhighscholarlyintegrityandpersonalkindness,acuteintelligenceandgentleness,andhisdailyactionsremindedthosearoundhimthatphilosophybeganwiththedesireforwisdomandforunderstandingthegood.Inhislife,aswellasinhiswork,Hempelwasatruephilosopher.BibliographyWorksbyHempel1936(withOppenheim,P.):DerTypusbegriffimLichtederneuenLogik,Leiden:Sitjhoff.1945a:“GeometryandEmpiricalScience,”AmericanMathematicalMonthly52,pp.7–17.1945b:“OntheNatureofMathematicalTruth,”AmericanMathematicalMonthly52,pp.543–56.1948(withOppenheim,P.):“StudiesintheLogicofConfirmation,”PhilosophyofScience15,pp.135–75.(ReprintedinHempel1965,ch.10.)1958:“TheTheoretician’sDilemma,”inMinnesotaStudiesinthePhilosophyofScienceII,ed.H.Feigl,M.Scriven,andG.Maxwell,Minneapolis:UniversityofMinnesotaPress,pp.37–98.(ReprintedinHempel1965,ch.8.)1965:AspectsofScientificExplanation,NewYork:FreePress.1966:PhilosophyofNaturalScience,EnglewoodCliffs,NJ:Prentice-Hall.158\nCARLG.HEMPELWorksbyotherauthorsAchinstein,P.(1983)TheNatureofExplanation,NewYork:OxfordUniversityPress.Bromberger,S.(1966)“Why-Questions,”inMindandCosmos,ed.R.Colodny,Pittsburgh:UniversityofPittsburghPress.Carnap,R.(1956)“TheMethodologicalCharacterofTheoreticalConcepts,”inTheFoundationsofScienceandtheConceptsofPsychologyandPsychoanalysis,ed.H.FeiglandM.Scriven,Minneapolis:UniversityofMinnesotaPress,pp.38–76.Cartwright,N.(1983)HowtheLawsofPhysicsLie,Oxford:OxfordUniversityPress.——(1999)TheDappledWorld,Cambridge:CambridgeUniversityPress.Coffa,J.A.(1974)“Hempel’sAmbiguity,”Synthese28,pp.141–63.Friedman,M.(1974)“ExplanationandScientificUnderstanding,”JournalofPhilosophy71,pp.5–19.Galison,P.(1987)HowExperimentsEnd,Chicago:UniversityofChicagoPress.Glymour,C.(1980)TheoryandEvidence,Princeton,NJ:PrincetonUniversityPress.Goodman,N.(1949)“TheLogicalSimplicityofPredicates,”JournalofSymbolicLogic14.——(1955)Fact,Fiction,andForecast,Indianapolis:Bobbs-Merrill.Hacking,I.(1983)RepresentingandIntervening,Cambridge:CambridgeUniversityPress.Humphreys,P.(1989)TheChancesofExplanation,Princeton,NJ:PrincetonUniversityPress.Kitcher,P.(1981)“ExplanatoryUnification,”PhilosophyofScience48,pp.507–31.Kripke,S.(1971)“NamingandNecessity,”inSemanticsofNaturalLanguages,ed.D.DavidsonandG.Harman,Dordrecht:Reidel.Kuhn,T.S.(1970)TheStructureofScientificRevolutions,Chicago:UniversityofChicagoPress.(Firstpublished1962.)Nagel,E.(1962)TheStructureofScience,NewYork:HarcourtBrace.Pearson,K.(1911)TheGrammarofScience,3rdedn.,London:A.&C.Black.Putnam,H.(1973)“MeaningandReference,”JournalofPhilosophy70,pp.699–711.Quine,W.V.(1953)FromaLogicalPointofView,Cambridge,MA:HarvardUniversityPress.Reichenbach,H.(1938)ExperienceandPrediction,Chicago:UniversityofChicagoPress.Salmon,W.(1984)ScientificExplanationandtheCausalStructureoftheWorld,Princeton,NJ:PrincetonUniversityPress.——(1998)CausalityandExplanation,NewYork:OxfordUniversityPress.Suppe,F.(ed.)(1970)TheStructureofScientificTheories,Urbana,IL:UniversityofIllinoisPress.Suppes,P.(1967)“WhatisaScientificTheory?,”inPhilosophyofScienceToday,ed.A.DantoandS.Morgenbesser,NewYork:BasicBooks.vanFraassen,B.(1980)TheScientificImage,Oxford:OxfordUniversityPress.159\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200112NelsonGoodman(1906–1998)ISRAELSCHEFFLERNelsonGoodman,distinguishedAmericanphilosopher,wasbornonAugust7,1906inSomerville,Massachusetts,anddiedinNeedham,MassachusettsonNovember25,1998.HereceivedtheBachelorofSciencedegreefromHarvardUniversityin1928andthePh.D.fromHarvardin1941.From1929to1940heoperatedanartgalleryinBoston;throughouthislife,hewasacollectorofancientandmodernart.From1942to1945heservedintheUnitedStatesArmy.Thereafter,hetaughtforoneyearatTuftsUniversitybeforehisappointmenttothefacultyoftheUniversityofPennsylvania,whereheservedasassociateprofessorfrom1946to1951,andthenasprofessorfrom1951to1964.From1964to1967hewastheHarryAustrynWolfsonProfessorofPhilosophyatBrandeisUniversity.From1968to1977,hewasprofessorofphilosophyatHarvardUniversity.Goodman’scontributionstophilosophyarewide-ranging,penetrating,andfunda-mental.Theareasinwhichheworkedincludeepistemology,philosophyofscience,philosophyoflanguage,analysisofsimplicity,theoryofsymbols,aesthetics,andmeta-physics.Hisworkischaracterizedbyunusualoriginality,typicallyrejectingconven-tionalapproachesinordertoreconceivetheproblemstobeaddressedandthenproposingprovocativesolutionstothem.Thus,forexample,herecaststhetraditionalproblemofinductionsoastorequirecodification,ratherthanjustificationofinductivepractice,thereafterdoomingtheprospectsofpurelysyntacticorsemanticapproachestosuchcodificationandofferinganewpragmatictreatmentbasedon“entrenchment.”Totakeanotherexample,hereorientsaestheticsasadivisionofepistemology,con-cernedprimarilywithunderstandingratherthanevaluatingworksofart,aprojectthatleadshimtoformulateacomprehensivenewtheoryofreferentialfunctionsembrac-ingtheliterary,pictorial,andotherartsaswellasthesciences.Finally,rejectingbothphysicalismandphenomenalism,bothrealismandidealism,heemphasizesthediver-sityofequallyadequateconflictingconceptualizationsforanysubjectmatter,thuschampioningwhathecalls“irrealism”;thedoctrinethattherearemanyworldsifanyandthatworldsaremade,notfound.Goodman’streatmentsareanalyticallysubtleaswellasinventive.Hiswritingisterseandtelling,makingbrilliantusebothoflogicandofmetaphor.Tosummarizehisworkinthisbriefspaceisclearlyimpossible.Instead,weshallgiveaselectiveaccountofhismaincontributionsinbasicareasofhisthought.160\nNELSONGOODMANLikenessofmeaningGoodman’sdiscussionswithW.V.QuineandMortonWhiteinthelate1940sledtoawidespreaddiscreditingoftheanalytic/syntheticdistinction.Goodman’spaper,“OnLikenessofMeaning”reflectssomeofhiscontributiontothesediscussions.“Underwhatconditions,”heasks,“dotwonamesorpredicatesinanordinarylanguagehavethesamemeaning?”Heconsidersandrejectssuchanswerstothequestionasthattheystandforthesameessenceorthesameimageoridea,orthatnothingcanbeconceivedthatsatisfiestheonebutnottheother,orthatitisimpossiblethatsomethingsatisfiestheonebutnottheother.Eschewingallreferencetoessences,images,ideas,conceiv-ability,andpossibility,heaskswhethertwopredicateshavethesamemeaningifandonlyiftheyarecoextensive,thatis,applytoexactlythesamethings.Theanswerisno,sincethereareclearcaseswherewordswiththesameextension(orequallylackinginit)differinmeaning,forexample“centaur”and“unicorn.”Extensionalidentityisindeedanecessary,butnotasufficientconditionforsamenessofmeaning.Here,Goodmanproposesthatitisnotonlytheextensionsoftheoriginaltwowordsthemselvesthatweneedtoconsider(so-calledprimaryextensions)butalsotheexten-sionsoftheirparallelcompounds(so-calledsecondaryextensions).Apairofparallelcompoundsisformedbymakinganidenticaladditiontoeachofthetwowordsunderconsideration;thus,addingtheword“picture”to“centaur”andto“unicorn,”wehavetheparallelpair“centaur-picture”and“unicorn-picture.”Now,althoughthereareneithercentaursnorunicorns,therecertainlyarecentaur-picturesandunicorn-picturesand,moreover,theyaredifferent.Althoughtheoriginalwordshavethesameextension,theparallelcompoundsdifferinextension.Goodman’sideais,then,thatthedifferenceinmeaningbetweentwowordsisamattereitheroftheirowndifferenceinextensionorofthatofanyoftheirparallelcompounds.Ingeneral,termshavethesamemeaningifandonlyiftheyhavethesameprimaryandsecondaryextensions.Thisideaisgeneralizedtocovercasesinwhichtheadditionof“picture”yieldsacom-poundwithnullextension.Forexample,“acrid-odor-picture”and“pungent-odor-picture”havethesame(null)extension,neitherapplyingtoanything.Compoundscan,however,beformedbyotheradditions,andGoodmansuggeststhat“description”con-stitutesasuffixcapableofyieldingallthewanteddistinctionsforeverypairofwordsPandQ.Foranyinscriptionoftheform“aPthatisnotaQ”isathingdenotedbythecompound“P-description”butnotbytheparallel“Q-description.”Andanyinscriptionof“aQthatisnotaP”belongstotheextensionof“Q-description”butnottothatof“P-description.”Thus,“pungent-odor-description”and“acrid-odor-description”differextensionallysincethefirst,butnotthesecond,appliestoanyinscriptionoftheform“apungentodorthatisnotanacridodor,”andviceversa.Thus,evenifallpungentodorsareacridandacridodorspungent,theterms“pungentodor”and“acridodor”differinmeaning.Itfollowsfromthisproposal,infact,thatnotwodifferentwordshavethesamemeaning.Totheobjectionthatthecompoundsusedinderivingthisradicalconclusionare“trivial,”Goodmanrepliesthatwhenasingleformofcompoundindeedhasadifferentextensionforeveryterm,“thefactthatithasdifferentextensionsfortwogiventermsisofnostrikingorspecialinterest.Letus,then,simplyexcludeeverycompoundforwhichthecorrespondingcompoundsofeverytwotermshavedifferingextensions...161\nISRAELSCHEFFLERinsteadofsayingthateverytwotermsdifferinmeaningbutthatsomemaynotdifferininterestingways,wesaythattwotermsdifferinmeaningonlyiftheydifferincertaininterestingorpeculiarways”(1972:237–8).ThenewriddleofinductionThestartingpointformoderndiscussionsofinductionisHume’sdenialofnecessaryconnectionsofmattersoffact.Effectscannotbesimplydeducedfromtheircauses,norcanpredictionsbelogicallydemonstratedonthebasisofavailableevidencegarneredfrompastexperience.Whatthencanbetherationaljustificationofthepredictionsuponwhichwebaseallouractions?Humeanswersthatwhilethereisindeednodeduc-tivejustification,thereisamentalhabitwhichunderliestheexpectationthatphe-nomenauniformlyconjoinedinourpastexperiencewillbesoconjoinedinthefuture.Ineffect,heofferstheuniformpastconjunctionofeventsasamarkofthoseinduc-tionswefindcompellinginmakingourpredictions.Thisideaisalsorepresentedbyamodernversion,whichhasfoundwidefavoramongscientistsaswellasphilosophers.Accordingtothisversion,predictionsaremadeinconformitywithgeneralizationsthathaveregularlyworkedinthepast.Suchcongruencewithpastexperienceisofcoursenoguaranteeoffuturesuccess,butitseemstosingleoutthosepredictionsweadoptatanygiventime.Lackingsuchguarantee,however,whatjustificationcantherebeforadoptingthesepredictions?Goodmanarguesthatthejustificationofinduction,likethatofdeduction,isonlyamatterofcodifyingourparticularsanctionedinferences,andcoordinatingthemwiththegoverningrulesofourpractice,thusbringingthemintoagreementwithoneanother.“Aruleisamendedifityieldsaninferenceweareunwillingtoaccept;aninfer-enceisrejectedifitviolatesaruleweareunwillingtoamend.Theprocessofjustifica-tionisthedelicateoneofmakingmutualadjustmentsbetweenrulesandacceptedinferences;andintheagreementachievedliestheonlyjustificationneededforeither”(1983:64).Howpersuasive,then,incodifyingourpracticeistheprevalentregularitydoctrine,theviewthatoursanctionedpredictionsarethoseinconformitywithgeneralizationsthathaveregularlyworkedsofar?HereGoodmanintroducesthenotoriouscaseofthegreenemeralds:supposeallemeraldsexaminedbeforethepresenttimetaregreen.Wepredictthenextemeraldtobeexaminedwillbegreen,sincesuchpredictionconformstothegeneralizationthatallemeraldsaregreen,ageneralizationuniformlyconfirmedbyallourpastevidence.Considernowthepredicate“grue,”applicabletoeverythingexaminedbeforetifandonlyifgreenbuttoeverythingelseifandonlyifblue.Then,allemeraldsexaminedbeforetarenotonlygreen,butalsogrue.Hence,thegeneral-izationthatallemeraldswhateveraregrueissupportedbynolessevidencethanthegeneralizationthattheyareallgreen.Thepredictionthatthenextemeraldtobeexam-inedwillbeblueisthus,bytheregularitytheory,asconfirmedasthepredictionthatitwillbegreen.Thistheory,whetherinHumeorinmodernscientificdress,thusfailsutterlytoseparatetheproperlyconfirmed“green”predictionfromthebogus“grue”one.“Regularityingreennessconfirmsthepredictionoffurthercases;regularityingruenessdoesnot.Tosaythatvalidpredictionsarethosebasedonpastregularities,withoutbeingabletosaywhichregularities,isthusquitepointless.Regularitiesare162\nNELSONGOODMANwhereyoufindthem,andyoucanfindthemanywhere”(1983:82).Iftheoldproblemofjustifyingourinductivepracticehasindeedbeensupplantedby“thenewriddleofinduction”askingforaprincipleddistinctionbetweenvalidandinvalidpredictions(ormoregenerally,“projections”),westillhavealongwaytogo.Goodman’ssolutiondependsonutilizingknowledgetypicallynotusedinattemptstointerpretinduction.Inparticular,hepresumessomeknowledgeofpastprojections,thatis,ofhypothesesthathavebeenactuallyprojectedinthepast,“adoptedaftersomeof[their]instanceshavebeenexaminedanddeterminedtobetrue,andbeforetheresthavebeenexamined”(1983:87).Now,whenweconsulttherecordofpastprojections,wefindthat“green”hasclearlybeenprojectedmuchmoreoftenthan“grue”;itismuchbetterentrenchedthanthelatter.Theentrenchmentofapredicateflowsfromtheactualpastprojectionsofitandofallothercoextensivepredicates.Whileinasenseitisthustheclassthatisentrenched,itbecomessoonlythroughtheprojectionoftermsthatdetermineit.Goodman,uponthisbasis,elaboratesasubtlegeneraltheoryofprojectionwhichwouldtakeustoofarafieldtocharacterizehere.Butthemainpointtobenotedisthathehopestohavegonebeyondtheregularitytheorybyappealingtoregularitiesinourlinguistichabits.“LikeHume,weareappealingheretopastrecurrences,buttorecur-rencesintheexplicituseoftermsaswellastorecurrentfeaturesofwhatisobserved.SomewhatlikeKant,wearesayingthatinductivevaliditydependsnotonlyuponwhatispresentedbutalsouponhowitisorganized;buttheorganizationwepointtoiseffectedbytheuseoflanguageandisnotattributedtoanythinginevitableorimmutableinthenatureofhumancognition.Tospeakveryloosely,Imightsaythatinanswertothequestionwhatdistinguishesthoserecurrentfeaturesofexperiencethatunderlievalidprojectionsfromthosethatdonot,Iamsuggestingthattheformerarethosefeaturesforwhichwehaveadoptedpredicatesthatwehavehabituallyprojected”(1983:97).ConstructionalismandnominalismInhisapproachtophilosophy,Goodmanwasgreatlyinfluencedbytheconstructional-ismexhibitedinRudolfCarnap’sDerlogischeAufbauderWelt,wherebyCarnapstrovetoreduceallconceptstotheimmediatelygiven,withthehelpofthenewmathematicallogic(seeCARNAP).Indeed,inhismonumentalTheStructureofAppearance,GoodmanincludesadetailedexpositionandcritiqueofCarnap’ssystembeforedevelopinganalternativephenomenalisticsystemtakingqualiaasatomsratherthanCarnap’selementarerlebnisse.Inadditiontoofferingacuteanalysesofqualiaandtheirconcretion,ofsize,shape,order,measure,andtime,Goodmandevotesthefirstpartofhisbooktothetheoryofconstructionalsystems.Herehearguesthatnotonlyisintensionalidentitytoostrongasarequirementforconstructionaldefinition;butevenextensionalidentityistoostrong,sinceitprecludesalternativeadequatesystematizations,e.g.takingpointsascertainclassesofvolumesortakingpointsascertainpairsoflines.Heproposesinsteadacriterionof“extensionalisomorphism,”whichheexplainsasallowingagiventermtobedefinablealternatively“byanyofseveralothersthatarenotextensionallyiden-ticalwithoneanother”(1977:17).Thecriterion,whichmustbeappliedtothewhole163\nISRAELSCHEFFLERsetofdefinitionsofthesystem,neverthelessprovidesforthetruth-valuepreservingcharacterofthetranslationsofallsentenceswecareabout.Inhisconstructionalism,Goodmanisconcernedbothwithsystemandwithsim-plicity.Indeed,thesetwoconcernsare,asheargues,identical.“Thepurposeofcon-structingasystemistointerrelateitspredicates.Thesamepurposeisservedbyreducingtoaminimumthebasisrequired.Everydefinitionatoncebothincreasesthecoherenceofthesystemanddiminishesthenumberofpredicatesthatneedtobetakenasprimitive.Thusthemotiveforseekingeconomyisnotmereconcernforsuperficialneatness.Toeconomizeandtosystematizearethesame”(1977:48).Goodman’sattitudetowardconstructionalsystemsissuesinapassionatedefenseoftheAufbauagainstthechargethatitisabstract,static,andbloodless,amerecaricatureofexperience.Onthecontrary,suchachargeisineffectanattackagainstphilosophyingeneral,forallphilosophyinvolvesconceptualization,abstraction,andsystematization,theefforttomapexperience,nottoduplicateit.“Amapisschematic,selective,conven-tional,condensed,anduniform...Themapnotonlysummarizes,clarifies,andsys-tematizes,itoftendisclosesfactswecouldhardlylearnimmediatelyfromourexplorations.”Goodmanemphasizesthefactthatdifferentmapsareusefulfordifferentpurposes.“Letnoonesupposethatifamapmadeaccordingtooneschemeofprojec-tionisaccuratethenmapsmadeaccordingtoalternativeschemesarewrong”(1972:15–16).ThispointisworthremarkingsinceGoodman’sphenomenalisticbasisinTheStructureofAppearanceisnot,ashasoftenbeenmistakenlysupposed,amatterofphilosophicalconvictionprecludingother,forexample,physicalisticbases;heinsiststhatwhileaconstructionalsystemmaybeadequateandilluminating,itcanbynomeansclaimamonopolyofwisdom.Andheseesnovirtueinclaimsofepistemologicalprioritymadeonbehalfofeitherphenomenalisticorphysicalisticsystems.InconstructingsystemsGoodmaninsiststhatthelogicemployedisnotmereneutralmachinery.Interpreteduseofthecalculusofclasses–employingvariablestakingclassesasvalues–commitsthesystemnotonlytotheindividualsexplicitlyacknowl-edged,butalsotoclassesofclasses,etc.ofthese,withoutlimit.“Thenominalisticallymindedphilosopherlikemyself,”hedeclares,“willnotwillinglyuseapparatusthatpeopleshisworldwithahostofethereal,platonic,pseudo-entities.Asaresult,hewillsofarashecanavoidalluseofthecalculusofclasses,andeveryotherreferencetonon-individuals,inconstructingasystem”(1977:26).Goodman’sprincipleis“enti-tiesdifferonlyiftheircontentatleastpartiallydiffers.”Heoffersthefollowingexample:“Aclass(e.g.thatofthecountiesofUtah)isdifferentneitherfromthesingleindividual(thewholestateofUtah)thatexactlycontainsitsmembersnorfromanyotherclass(e.g.thatofacresofUtah)whosemembersexactlyexhaustthissamewhole”(1977:26).ApioneeringpaperbyGoodmanandW.V.Quineis“StepsTowardaConstructiveNominalism”(1947)inwhichtheauthors,startingfromarenunciationofabstractentities,proceedtoofferingeniousnominalistictranslationsofavarietyofnon-nominalisticstatements(seeQUINE).Theysoon,however,concedetheirinabilitythustotranslateallofmathematicsandtheythereforesuggestadifferentwayofsavingmathematicsnominalistically:todeviseanominalisticsyntaxcapableofdescribingthesyntacticrulesbywhichmathematicalinscriptions,asconcreteuninterpretedmarks,aremanipulated.164\nNELSONGOODMANInthisenterprise,theysucceedindefiningasuitablenominalisticsyntaxlanguagefordescribingtheobjectlanguageofmathematics,withitsingredientnotionsof“axiom,”“rule,”“proof,”and“theorem.”Inconclusion,theydeclare“Ourpositionisthattheformulasofplatonisticmathematicsare,likethebeadsofanabacus,conve-nientcomputationalaidswhichneedinvolvenoquestionoftruth.Whatismeaning-fulandtrueinthecaseofplatonisticmathematicsasinthecaseoftheabacusisnottheapparatusitself,butonlythedescriptionofit:therulesbywhichitisconstructedandrun.Theseruleswedounderstand,inthestrictsensethatwecanexpresstheminpurelynominalisticlanguage”(1972:198).Furtherdiscussionofnominalism,includingthelaterdifferinginterpretationsofGoodmanandQuine,aswellaselaborationsandanswerstovariousobjections,canbefoundinGoodman’spaper,“AWorldofIndividuals”(1972:155–72).Here,hedeclaresthathisnominalismisspecifically“therefusaltorecognizeclasses”(p.156),requiringthatallentitiesadmittedtoasystem,nomatterwhattheyare,betreatedasindividuals.TheoryofsymbolsInLanguagesofArt,Goodmandevelopsatheoryofsymbolsthatencompassesbutrangesfarbeyondwhatistypicallyconsideredaesthetics,whilegivingonlyminorattentiontoartisticvalueandcriticism.Hisprimaryconcernistodevelopasystematicapproachtothefunctioningofsymbolsnotonlyintheartsbutalsointhesciencesandinordinarycontextsaswell,forexample,intheeverydayuseoflabelsandsamples.Andintherealmoftheartsproper,hediscussesmusicaswellaspainting,literaryartsaswellasdanceandarchitecture.Hiskeyemphasisisoncognitivefunctionandtothisendhedevisesageneraltheorywhichheemploysincharacterizingnotations,sketches,scripts,andpaintings.Inrelateddiscussions,hisbookoffersanaccountofforgeryofworksofart,atheoryofmetaphorandacategorizationoffiguresofspeech,atreat-mentoffictionalexpressions,andacomprehensiveinterpretationofreferenceasincludingnotonlydenotation,butalsoexemplificationandexpression.Aswatchofclothinatailor’sshopistypicallyusedtoexemplifycertainofitsprop-erties,i.e.itscolorandtexture,butnotitssizeorshape.Toservethus,itmustpossesstheseproperties,butalsorefertothem.Exemplificationisasymbolicfunctionbywhichthesamplestandsforapropertyitpossesses.Expressionimpliesmetaphoricalexempli-fication:ifapictureexpressessadness,itisitselfmetaphoricallysadandalsoreferstosadness.Withthesetwonotionsathand,Goodmanisabletotreatworksofartnotmerelyasobjectsofreferencebut–evenwherenonrepresentational–asreferringsymbolsintheirownright.Andtheingredientideaofmetaphorvastlyexpandstheresourcesavailablefortreatingthewideexpressiverangesofworksofart.Goodman’streatmentofmetaphorhasbeenwidelynoted.“Wherethereismetaphor,”hewrites,“thereisconflict:thepictureissadratherthangayeventhoughitisinsentientandhenceneithersadnorgay.Applicationofatermismetaphoricalonlyiftosomeextentcontra-indicated”(1976:69).Alongwithsuchcontra-indication,theremustalsobeattraction,oraptness.Themetaphoricaluseof“sad”impliesthattherearetworangesfortheterm“sad,”butthatthesetworangesdonotsimplycompriseanambiguity.Inmereambiguity,theseparateusesofthetermare165\nISRAELSCHEFFLERindependent.“Inmetaphor,ontheotherhand,atermwithanextensionestablishedbyhabitisappliedelsewhereundertheinfluenceofthathabit;thereisbothdeparturefromanddeferencetoprecedent.Whenoneuseofatermprecedesandinformsanother,thesecondisthemetaphoricalone”(p.71).Thekeytechnicaldiscussioninthebookpresentsatheoryofnotation,forexample,amusicalscore.Themainfunctionofsuchanotation,saysGoodman,istoprovideauthoritativeidentificationofaworkfromperformancetoperformance,whateverotherusesitmayhave.“Whatisrequiredisthatallandonlyperformancesthatcomplywiththescorebeperformancesofthework”(1976:128).Withthisprincipleasaguide,Goodmandevelopsfiverequirementsforanotationalsystem:unambiguityandsyntacticandsemanticdisjointnessanddifferentiation.Togiveapreciseaccountoftheserequirementswouldexceedthelimitsofthisspace,butabriefsummationoftheirimportisstatedbyGoodmanasfollows:“Asystemisnotational,then,ifandonlyifallobjectscomplyingwithinscriptionsofagivenchar-acterbelongtothesamecomplianceclassandwecan,theoretically,determinethateachmarkbelongsto,andeachobjectcomplieswithinscriptionsof,atmostonepar-ticularcharacter”(1976:156).Usingthisnotionofnotation,Goodmanisablethentocharacterizeotherthingsthan,forexample,scores.Ascript,forexample“isacharac-terinanotationalschemeandinalanguagebut,unlikeascore,isnotinanotationalsystem.Thesyntacticbutnotallthesemanticrequirementsaremet...[itis]acharacterinalanguagethatiseitherambiguousorlackssemanticdisjointnessordifferentiation”(pp.199,201).Ratherthanofferingadefinitionoftheaesthetic,Goodmansuggestsfoursymptomsthattendtobepresentinaestheticexperience:syntacticdensity,semanticdensity,syn-tacticrepleteness,andexemplificationalcharacter.Thefirsttwocallformaximumsen-sitivityofdiscrimination,thethirdcallsforsucheffortalongmanydimensions,whilethefourthisshownbyconcernwithpropertiesexemplifiedbyasymbol,notmerelywiththingsthesymboldenotes.Largelyasaresultofhisanalysis,Goodmanempha-sizestheaffinitiesbetweenartandscience,despitetheirdifferences.Thedifferencebetweenthemisnotthatbetweenfeelingandfactortruthandbeauty“butratheradifferenceindominationofcertainspecificcharacteristicsofsymbols”(1976:264).IrrealismInGoodman’sWaysofWorldmaking,hespeaksofversionsasincludingdepictionsaswellasdescriptions,worksofartaswellasworksofscience.Andheinsiststhatthereareconflictingrightworld-versions,ratherthanasingleworldunderlyingtheright-nessofall.HeconsidershisviewonthesemattersasbelonginginthatmainstreamofmodernphilosophythatbeganwhenKantexchangedthestructureoftheworldforthestructureofthemind,continuedwhenC.I.Lewisexchangedthestructureofthemindforthestructureofconcepts,andthatnowproceedstoexchangethestructureofconceptsforthestructureoftheseveralsymbolsystemsofthesciences,philosophy,thearts,perception,andeverydaydiscourse.Themovementisfromuniquetruthandaworldfixedandfoundtoadiversityofrightandevenconflictingversionsorworldsinthemaking.166\nNELSONGOODMANTheviewthatemerges,hesays,canperhapsbedescribedas“aradicalrelativismunderrigorousrestraints,thateventuatesinsomethingakintoirrealism”(1978:x).Goodmantakeshisadversarytobe“themonopolisticmaterialistorphysicalistwhomaintainsthatonesystem,physics,ispreeminentandall-inclusive,suchthateveryotherversionmusteventuallybereducedtoitorrejectedasfalseormeaningless”(1978:4).Wereallversionsreducibletoasingleone,thatonemightbeplausiblycon-sideredtheonlytruthabouttheoneworld.Butsuchreducibilityisachimera;theclaimofreducibilitytophysicsis“nebuloussincephysicsitselfisfragmentaryandunstableandthekindandconsequencesofreductionenvisagedarevague...Thepluralist’sacceptanceofversionsotherthanphysicsimpliesnorelaxationofrigorbutarecogni-tionthatstandardsdifferentfromyetnolessexactingthanthoseappliedinscienceareappropriateforappraisingwhatisconveyedinperceptualorpictorialorliteraryversions”(p.5).Doesn’ttherightnessofthesevariousversionsimplyanunderlyingworldthatmakesversionsright?No,saysGoodman,wemightbettersaythat“theworld”dependsonrightness.Wecannottestaversionbycomparingitwithaworldundescribed...allwelearnabouttheworldiscontainedinrightversionsofit;andwhiletheunderlyingworld,bereftofthese,neednotbedeniedtothosewholoveit,itisperhapsonthewholeaworldwelllost...Formanypurposes...justversionscanbetreatedasourworlds.(1978:4)Worldsaremadethroughthemakingofversions,butversionscannotbemadeanywaywelike.Goodmandeniesthathisacceptanceofmanyrightworld-versionsimplies“thatanythinggoes,thattallstoriesareasgoodasshortones,thattruthsarenolongerdistinguishedfromfalsehoods.”Althoughitistrue,hesays,that“wemakeworldsbymakingversions,”wecannotdosoatrandomorbywhim(1978:94).“Ofcourse,”hesays,wewanttodistinguishbetweenversionsthatdoandthosethatdonotrefer,andtotalkaboutthethingsandworlds,ifany,referredto;butthesethingsandworldsandeventhestufftheyaremadeof–matter,anti-matter,mind,energy,orwhatnot–arefashionedalongwiththethingsandworldsthemselves.(p.94)Themakingofworldsisbroughtaboutbythemakingofversionsandthemultipleworldsthusmade“arejusttheactualworlds...answeringtotrueorrightversions.Worldspossibleorimpossiblesupposedlyansweringtofalseversions,”saysGoodman,“havenoplaceinmyphilosophy”(1978:94).Goodman’srelativismontheissueofworldmakingisfamiliarfromhisearlierworkonconstructionalism,where,aswehaveseen,heinsistsonhonoringtheconflictingdefinitionsthatcanbeofferedfortheverysameterms.Thecriterionofextensionaliso-morphism,aswehavenotedabove,isamorerelaxedstandardthaneithersynonymyoridentity,butitnonethelessimposesclearanddefiniterestraints.Similarly,histheoryofworldmaking,paradoxicalasitmayseem,recognizesavastarrayofconflictingversions,ratherthanasingleunderlyingworld,yetinsiststhattherearecleardistinctionstobedrawnbetweenrightandwrongversions.Goodman’s167\nISRAELSCHEFFLERrelativismmustthereforebesharplydistinguishedfromnihilism,subjectivism,andculturalrelativism.Hishospitalitytovariantandopposedconceptualizationsisalliedwithadedicationtothehigheststandardsoflogicalrigor.BibliographyWorksbyGoodman1947(withQuine,W.V.):“StepsTowardaConstructiveNominalism,”JournalofSymbolicLogic12,pp.105–22.(ReprintedinGoodman1972,pp.173–93.)1949:“OnLikenessofMeaning,”Analysis10,pp.1–7.(ReprintedinGoodman1972,pp.221–30.)1953:“OnSomeDifferencesaboutMeaning,”Analysis13,pp.90–6.(ReprintedinGoodman1972,pp.231–8.)1963:“TheSignificanceofDerlogischeAufbauderWelt,”inThePhilosophyofRudolfCarnap,ed.P.A.Schilpp,LaSalle,IL:OpenCourtandLondon:CambridgeUniversityPress.(ReprintedinGoodman1972,pp.5–23.)1972:ProblemsandProjects,Indianapolis:Bobbs-Merrill.1976:LanguagesofArt,2ndedn.,Indianapolis:Hackett.(FirstpublishedIndianapolis:Bobbs-Merrill,1968.)1977:TheStructureofAppearance,3rdedn.,Dordrecht:Reidel.(FirstpublishedCambridge,MA:HarvardUniversityPress,1951.)1978:WaysofWorldmaking,Indianapolis:Hackett.1983:Fact,FictionandForecast,4thedn.,Cambridge,MA:HarvardUniversityPress.(Firstpub-lished1955byHarvardUniversityPress.)1984:OfMindandOtherMatters,Cambridge,MA:HarvardUniversityPress.1988(withElgin,CatherineZ.):ReconceptionsinPhilosophyandOtherArtsandSciences,Indianapolis:Hackett.WorksbyotherauthorsElgin,C.Z.(1983)WithReferencetoReference,Indianapolis:Hackett.——(ed.)(1997)ThePhilosophyofNelsonGoodman,4vols,NewYorkandLondon:GarlandPublishing.(ContainsarticlesbyvariousauthorsonGoodman’sconstructionalism,theoryofinduction,philosophyofart,andtheoryofsymbols.)Howard,V.A.(1982)Artistry,Indianapolis:Hackett.McCormick,P.J.(ed.)(1996)Starmaking,Cambridge,MA:MITPress.(Containscriticaldiscus-sionsofGoodman’sirrealism,byHempel,Putnam,Scheffler,andGoodman.)Scheffler,I.(1963)TheAnatomyofInquiry,NewYork:AlfredA.Knopf.——(1997)SymbolicWorlds,Art,Science,Language,Ritual,Cambridge:CambridgeUniversityPress.Stalker,D.(1994)Grue:TheNewRiddleofInduction,Chicago:OpenCourt.168\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200113H.L.A.Hart(1907–1992)SCOTTSHAPIROHerbertLionelAdolphusHartwasbornofJewishparentsinYorkshire,EnglandandwaseducatedatNewCollege,Oxford.AftergraduatingwithaFirstinGreats,HartwascalledtotheBarasaChancerybarristerinLondon.Hespentthenexteightyearsbuild-ingasuccessfullegalpractice,specializingfirstinpropertyconveyancing,trustdraft-ing,andtaxplanning,andthenmovingontocourtworkandadvising.Althoughhisinterestsquicklyturnedfromlawtophilosophy,Hartcontinuedtopracticeand,infact,duringthisperiodhedeclinedaninvitationtoteachphilosophyinOxford.Hislegalcareer,however,wascutshortbyWorldWarII.WhileworkingwithBritishIntelligence,HartmetGilbertRyleandStuartHampshire,fromwhomhelearnedofthenewtrendsinphilosophy.Whenthewarended,HartlefthislawpracticeandreturnedtoOxford.In1952,HartwaselectedtotheChairofJurisprudence,asomewhatsurprisingappointmentgiventhathedidnothaveadegreeineitherlaworphilosophyandhadpublishedlittlebythatpoint.Heoccupiedthatchair,however,withgreatdistinction,publishingseveralseminalworksinlegaltheory,includinghismasterpiece,TheConceptofLaw,in1961.Hartisperhapsbestknownforhisvigorousandsophisticateddefenseofthedoctrineknownaslegalpositivism.Initsbroadestsense,legalpositivismisatheoryaboutthenatureoflawthatdeniesanynecessaryconnectionbetweenlegalityandmorality.Nostipulationismadethat,inordertocountaslaw,anormmustpossessanymoralattributes.Legalpositivists,therefore,believethatitispossibleforalegalsystemtorecognizearuleaslegallyvalidevenifithappenstobeunjust.ThisanalyticseparationbetweenthelegalandthemoralwascapturedbyJohnAustinwhenhesaid:“Theexistenceoflawisonething;itsmeritordemeritisanother”(1954:184–5).Inanefforttocleanseanalyticjurisprudenceofitsmoralcontent,everylegalposi-tivistbeforeHartthoughtitnecessarytorecastthebasiclegalconceptsofobligation,rule,validity,andauthorityintermsofsanctions.Austin,forexample,believedthatlegalrulesarenothingmorethanordersbackedbythreatsofsanctionsissuedbythesover-eign.Sovereignty,inturn,wasunderstoodintermsofcoercivepower,thesovereignbeingtheoneinagroupwhohasthepowertoelicithabitualobediencefromeveryoneandwhohabituallyobeysnoone.Someoneisunderalegalobligationtoact,onthisview,iftheyarelikelytobesanctionedforfailingtoact.169\nSCOTTSHAPIROHartwasfirmlycommittedtotheanalyticseparationoflawandmorality,butthoughtthatthesesanction-centeredtheoriesdistortedandconcealedthevariouswaysinwhichthelawguidesconduct.Forexample,Hartpointedoutthattherearemanylegalrulesthatlacksanctions,inthesensethatnopenaltiesareimposedasaresultofnon-conformitywiththem.Ifapersondraftsawillbutfailstohaveitsignedbytwowitnesses,thatpersonisnotsanctionedfortheinadequateattentionpaidtothetesta-mentaryrules.Hehassimplyfailedtoformavalidwillandhisactionslacklegaleffect.Sanction-centeredtheoriesfallshort,accordingtoHart,becausetheytreatalllegalrulesasiftheirsolefunctionistodiscourageundesirablebehavior.Theirparadigmisthecriminallaw,wheretherulesimposedutiestoactorforbearfromcertainbehaviorandspecifysanctionsintheeventofdisobedience.However,notonlydotherulesrelatedtovalidwillorcontractformationlacksanctions,but,asHartobserved,itdoesnotevenmakesensetospeakofobeyingordisobeyingthem.Theserulesdonotimposeduties;theyinsteadconferpowers.Theirfunctionisnottodiscouragepeoplefromactinginwaysthattheyotherwisemightwish,buttogivethemfacilitiesforrealizingtheirwishes.Theeffacementofpower-conferringrulesisespeciallyproblematicwithrespecttothoserulesthatconferlegalpowersonpublicofficials.Withoutsuchrules,Hartnoted,sanction-centeredtheoriescannotaccountfortheself-regulatingnatureoflegalinsti-tutions:itisadefiningfeatureoflaw,asopposedtopre-legalsocialsystems,thatitsoffi-cialsareempoweredtochangetherulesandtoresolvethedisputesthatmayariseunderthem.Hartalsobelievedthatthesetheoriesgiveamisleadingpictureofthenatureofthelaw’snormativity.Onthesanction-centeredapproach,theonlyreasonsforactionthatthelawprovidesarethreatsofsanctions.ThisignoreswhatHartcalledtheinternalpointofview,whichistheperspectiveofthosewhotreattherulesasstandardsofacceptableconduct.Ineverylegalsystem,Hartclaimed,somemembersofthegrouptreattherulesnotjustasthreats,orpredictionsofwhatcourtswilldo,butasguidestotheirconductandstandardsfortheevaluationofothers–asnormsthatobligateandempower,notmerelyoblige.Byemphasizingtheinternalpointofview,Hartwasnotsimplycriticizingfellowlegalpositivistsforneglectinganobviousfact,i.e.thatatleastsomepeopleinsomecir-cumstancesaremotivatedbythelawqualaw,insteadofsanctions.Rather,Hartwasalsomountingamethodologicaloffensiveagainstthecrudescientismofsomeofhiscontemporaries.Forexample,AlfRoss,theScandinavianLegalRealist,basedhislegalpositivismonhiscommitmenttologicalpositivismandbelievedthat,ifjurisprudenceistohaveempiricalcontent,legalconceptsmustbeoperationalizedinpurelybehav-ioristicterms.Bycontrast,Hartbelievedthattheoriesoflawmustmakeessentialreferencetotheattitudesoflegalactorsbecausethelawisasocialpractice.Inordertoanalyzethepractice,itisnotenoughtorecordregularitiesofbehavior;onemustunderstandhowtheparticipantsunderstandit.Hart’sintroductionoftheinternalpointofviewthusinauguratedthehermeneuticturninjurisprudence,wherethelawisstudiedfromtheinside,thatis,fromtheperspectiveofthosewholiveunder,anddirectlyexperience,thelaw.Byengaginginthishermeneuticenterprise,Hartwasnot,however,givinguponanaturalisticapproachtolegaltheory.Indeed,Hartbelievedthattheinternalpointof170\nH.L.A.HARTviewallowedthelegalpositivisttoanchorrulesinsocialfacts.AccordingtoHart,asocialruleinacommunityexistswheneverasufficientnumberofpeopleengageinapracticefromtheinternalpointofview.Thisinternalaspectofrulesismanifestedexter-nallyinconformingbehavior,aswellascriticismsthatattenddeviationsfromtheprac-ticeandtheuseofnormativelanguagesuchasought,must,andobligationtoexpresssuchdisapprobation.Theexistenceofarule,therefore,isfirmlyrootedinthenaturalworld,thatis,inregularitiesofbehaviormotivatedbytheappropriatecriticalattitude.Hart’stheoryofsocialrulesformsthefoundationofhisapproachtolaw.AccordingtoHart,attherootofeverylegalsystemisasocialruleofaspecialsort,whichhecalledtheruleofrecognition.Thisruleimposesadutyoncourtstoapplyrulesthatbearcertaincharacteristics.IntheAmericansystem,forexample,theruleofrecognitionrequiresjudgestoapplytherulesdulyenactedbyCongress.Theruleofrecognition,therefore,setsoutthecriteriaoflegalvalidity,thatis,thosecriteriathatarulemustpossessinordertobelaw.TheruleofrecognitioniswhatHartcalledasecondaryrule:itisaruleaboutotherrules.Itisalsoanultimaterule:itexistsbecauseitisacceptedbyjudgesfromtheinter-nalpointofview,notinvirtueofitsvalidationbyanotherrule.Theprimaryrules,bycontrast,owetheirexistencetotheruleofrecognition,andnottoanyguidancethattheymightengender.Inadditiontotheruleofrecognition,Hartarguedthateverylegalsystemcontainstwoothersecondaryrules.Theruleofchangeconfersthepoweronlegislativebodiestomodifytheprimaryrules,whereastheruleofadjudicationconfersthepoweroncourtstoadjudicatewhethertheprimaryruleshavebeenfollowedorviolated.Byunderstandingthelawastheunionofprimaryandsecondaryrules,Hartintro-ducedwhatmightbecalledarule-centeredtheoryofthelaw.Onthismodel,thelawguidesconductnotbyissuingnakedthreats,butbyprovidingrulesthatimposedutiesandconferpowers.Thebasiclegalconceptsarealsounderstoodintermsofrules,notsanctions.Aruleisvalidinalegalsystemwhentherulebearsthosecharacteristicssetoutinthatsystem’sruleofrecognition.Anactislegallyobligatory,inturn,whenitisrequiredbyalegallyvalidrule.Apersonhassupremelegalauthoritywhenthesec-ondaryrulesofthesystemconferlegalpoweronthatpersonandnootherhasbeenconferredagreaterpower.Eventheconceptofasanctionisrenderedintermsofrules,forasanctionisnotsimplyacostimposedbythelaw,but,unlikeatax,isapenaltyexactedbecausearulehasbeenviolated.Hartdidnotthinkthatprivilegingtheconceptofarulecompromisedtheanalyticseparationoflawandmorals.Inhismodel,aprimarylegalruleexistsjustincaseitisvalidatedbythatsystem’sruleofrecognition.Thereisnodemandthatthecriteriaoflegalvaliditysetoutmakereferencetotherule’smoralproperties.Itispossible,andregrettablyoftenthecase,thatalegalruleexistseventhough,fromamoralpointofview,itshouldnot.Andwhileitistruethattheruleofrecognitionwouldnotexistunlessjudgesacceptitfromtheinternalpointofview,thisdoesnotmeanthattheyjudgeitmorallyacceptableorthatitismorallyacceptableforthemtotreatitinthisway.DespiteHart’sinsistencethatlawbeseenasasystemofrules,hedidnotthinkthatjudgesarealwaysguidedbytheseruleswhentheydecidecases.Inhisview,courtsarenotsimplythepassiveservantsofthelegislatureoroftradition,restrictedtoapplying171\nSCOTTSHAPIROtheruleslaiddowninadvance,butareactiveplayersinthecreationanddevelopmentofthelaw.Judgesdonotalwaysfindthelaw;theysometimesmakeitaswell.Hart,however,wasnotdisturbedbythefactofjudiciallegislation.Hethoughtthatjudgesshouldbegivenfreereintodecidesomecases,asitenablesthemtofashionsensiblesolutionstounforeseenproblems.Moreover,giventheinherentlimitationofnaturallanguages,hebelievedthatjudiciallegislationwasunavoidable.AccordingtoHart,allgeneraltermsinnaturallanguage(e.g.vehicle)containacoreofsettledmeaning(e.g.car)andapenumbrawherethereferenceclassisill-defined(e.g.tractor).Whenacasefallsintothecoreofageneraltermoftherule,theruleappliesandthejudgeislegallyobligatedtoapplytherule.However,wheninthepenumbra,thelawrunsoutandthejudgemustexercisehisdiscretion.Bynecessity,thejudgecannotfindthelaw,becausethereisnolawtofind,andhencemustmakenewlaw.Althoughsoundingsensibleenough,Hart’srecognition,andsympatheticaccept-ance,ofjudiciallegislationhasbeenattackedbyhischiefcritic,andsuccessortotheChairinJurisprudence,RonaldDworkin.InDworkin’sview,theroleofajudgeistovindicatethelegalrightsofthepartiesandthiscanonlybeaccomplishedifthelawcompletelyregulatesthejudge’sbehaviorineverycase.DworkinfaultedHartforcountingaslawonlyrulesthathavesocialpedigrees,suchaslegislationorcustom,andignoringthemassofimplicitlawrepresentedbymoralprinciplesthatjustifythepedigreedrulesandthatdeterminethelegallycorrectanswerwhentheserulesrunout.Byarguingthat,ineverycase,thereisarightanswer,Dworkinwasnotonlychal-lengingHart’stheoryofadjudicationbutalsohisclaimthatlawandmoralitywereconceptuallydistinct.Forifthelegallycorrectanswerisdeterminedinpartbynormswhoseonlyclaimtolegalvalidityistheirmoralvalidity,thenitwouldseemthatmoral-itywouldbeadeterminateoflegality,contrarytolegalpositivisticstrictures.InthePostscripttothesecondeditionofTheConceptofLaw,publishedposthu-mously,HartagreedwithDworkinthatjudgesareoftenlegallyobligatedtoapplymoralprinciplesthatlackpedigrees,andthatwhenjudgesactonthem,theyareapplyingexistinglaw.However,Hartbelievedthatsuchapositionwasconsistentwithlegalpos-itivism,forhesawnoreasonwhytheruleofrecognitioncouldnotvalidateanormbasedonitsmoralproperties.Legalpositivists,accordingtoHart,onlyclaimthataruleofrecognitionneednotvalidateanormonthebasisofitsmoralcontent,notthatitcannot.Evenwhentheruleofrecognitiondidvalidateprinciplesonthebasisoftheirmoralcontent,Hartdoubtedthattheseprincipleswouldindicateauniqueresultineverycase,thusleavingampleroomfortheexerciseofjudicialdiscretion.Inseparatinglawfrommorals,Hartdidnotmeantoprecludemoralcriticismofthelaw.Quitethecontrary,Hartwasavocalandinfluentialcriticofmanyaspectsofthecriminallaw,especiallytheprohibitionsonso-calledprivatevices.InLaw,LibertyandMorality,Hartattackedthedoctrineknownaslegalmoralism,thebeliefthatsocietyhastherighttousethecriminallawtoenforceitsmoralcode.LordDevlinhadarguedthatsocialcohesionispossibleonlywhenacommoncodeofmoralityisrespectedbyall,andthefloutingofthatcode,eveninprivate,threatenssuchcohesionand,inturn,society’sveryexistence.HartnotedthatDevlinfailedtoproduceanyevidencesup-portinghiscausalclaims,anddoubtedwhetheranycouldbemustered.Moreimpor-tantly,hearguedthatasocietythatcriminalizesbehaviorthatthemajorityfinds172\nH.L.A.HARToffensiveisnotasocietythatrespectsliberty.Torespectliberty,asocietymustprotecttherightofindividualstochoosetheirownlifestyle,evenwhenitdoesnotapproveofthelifestyletheyendupchoosing.Thelibertytoactonlyinwaysthatotherslikeis,asHartpointedout,libertyinnameonly.Incontrasttomostofhiscontemporaries,Harteschewedgrandmoraltheoriesinfavorofamorecommonsenseapproachtonormativeanalysis,whichborrowedelementsfromboththeUtilitarianandKantiantraditions.Forexample,Hartthoughtthatthejustifyingaimofpunishmentisthedeterrenceofcrime.Yet,healsobelievedthatthispursuitmustyieldtothedemandsofjustice,sothatitiswrongtopunishpeopleforcrimestheydidnotcommitorcouldnothavehelpedcommitting.Hewasthuscriticaloftheattemptstoincreasetheefficiencyofthecriminallawbyeliminatingmanyofthetraditionallyrecognizedexcuses,suchastherestrictionsontheuseoftheinsanitydefenseandtheintroductionofcrimesofstrictliabilityandnegligence.AlthoughHartrecognizedthattheavailabilityofexcusesmightallowsometofeignincapacityormistakeandthusevaderesponsibility,heneverthelessthoughtthatthecostsareslightcomparedtothebenefits.Notonlyisitfundamentallyunfairtopunishthosewhocouldnothavehelpeddoingwhattheydid,but,asHartpointedout,asystemofexcusesplacesindividualsincontroloftheirdestinies.Forwhenthelawonlypun-ishespeopleforactionstheycanavoid,peoplecanavoidbeingpunished.Aslongasindividualsneverchoosetobreaktherules,thelawwillletthemgoabouttheirlives.Asaresult,individualsneednotfearthattheywillunwittinglybringthewrathofthelawdownuponthemselves;theycanrelyonthefactthatthelawwillexcusebehaviorthatwasnot,insomesuitablesense,aproductofchoice.Itisamistake,Hartconcluded,tothinkthatreducingcrimebyeliminatingexcuseswillleadtoanincreaseinsecurity.Whenexcusesareunacceptable,peopleareunabletopredicttheconsequencesoftheiractions.Aworldthatisunknowableandun-controllableisaworldinwhichnooneissecure.BibliographyWorksbyHart1955:“ArethereAnyNaturalRights?,”PhilosophicalReview64,pp.175–91.(ReprintedinPoliticalPhilosophy,ed.A.Quinton,Oxford:OxfordUniversityPress,1967,pp.53–66.)1958:“PositivismandtheSeparationofLawandMorals,”HarvardLawReview71,pp.593–629.1959(withHonore,A.M.):CausationintheLaw,Oxford:ClarendonPress.1961:TheConceptofLaw,1stedn.,Oxford:ClarendonPress.(The2ndedn.,1994,includesa“Postscript,”whichisareplytocritics.)1963:Law,LibertyandMorality,London:OxfordUniversityPress.1968:PunishmentandResponsibility,EssaysinthePhilosophyofLaw,Oxford:ClarendonPress.1982:EssaysonBentham,Oxford:ClarendonPress.1983:EssaysinJurisprudenceandPhilosophy,Oxford:ClarendonPress.WorksbyotherauthorsAustin,J.(1954)TheProvinceofJurisprudenceDetermined,London:WeidenfeldandNicolson,pp.184–5.173\nSCOTTSHAPIRODworkin,R.(1977)“TheModelofRulesI”and“TheModelofRulesII,”inTakingRightsSeriously,Cambridge,MA:HarvardUniversityPress.Fuller,L.(1958)“PositivismandFidelitytoLaw:AReplytoProfessorHart,”HarvardLawReview71,p.630.MacCormick,N.(1981)H.L.A.Hart,Stanford:StanfordUniversityPress.174\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200114C.L.Stevenson(1908–1979)JAMESDREIERStevenson’smajorcontributiontophilosophywashisdevelopmentofemotivism,atheoryofethicallanguageaccordingtowhichmoraljudgmentsdonotstateanysortoffact,butratherexpressthemoralemotionsofthespeakerandattempttoinfluenceothers.Stevenson’semotivetheoryofethicallanguageStevensonalwaysstressedthathisworkdidnotincludeanysubstantivemoraljudg-ments,butrathercomprised“analyticethics,”orwhatisnowcommonlycalled“metaethics,”thebranchofmoraltheorythatisaboutethicsandethicallanguage.Whatdowemeanwhenwesaythatsomethingisgoodorbad,orrightorwrong?Onthefaceofit,wearedescribing,attributingtothethingsomeproperty,goodnessorbadness,orrightnessorwrongness.Whatcouldthesepropertiesbe?Howdowefindoutaboutthem?Muchofphilosophicalmoraltheoryexploresvariousanswerstothesequestions.Stevensonthoughtthatquestionsaboutthenatureofmoralpropertiesweremisplaced.Ourmoraljudgmentsdonot,atleastprimarily,describeatall.Utteringmoralsentenceshasadifferentfunction:toexpressemotions,andtoinfluenceorinviteotherstosharethem.AllofhismaincontributionsappearedinEthicsandLanguage,1944,andacollectionofpapers,FactsandValues,1963.Distinguishbetweenexpressingacertainstateofmindandsayingthatoneisinit.IfIsay,“AnnArborisinMichigan,”IexpressmybeliefthatAnnArborisinMichigan,butIdonotsaythatIbelievesuchathing.ForwhatmakeswhatIsaidtrue?NotthatIreallydobelievethatAnnArborisinMichigan;onlythefactthatAnnArborreallyisinMichigan.Stevenson’stheoryofethicallanguage,inanutshell,wasthatwhenIsay,“Inequalityisbad,”Ihaveexpressedacertainnegativemoralattitudetowardinequality,thoughIhavenotsaidthatIhaveit.ItshouldbeclearwhyStevensonstressedthathistheorywas“analytic”ormetaethical,anddidnotcontainanysub-stantivemoraljudgments.Forbyclaimingthatmoraljudgmentsservetoexpressemo-tions,hehadnotexpressedhisownmoralemotionsatall.Besidesexpressingthespeaker’sattitude,Stevensonsaid,moralstatementsalso“createaninfluence,”theyinvitetheaudiencetoshareintheemotionexpressed.Thus,“xisgood”isakinto“Letusapproveofx.”Moralexhortation,afterall,iscommonly175\nJAMESDREIERusedtotrytopersuadetheaudiencetosharethespeaker’ssuggestions,andmoraljudg-mentisoftenacalltoaction.Furthermore,incontext,ethicalstatementscancometohavesomesecondarydescriptivecontent;inVictorianEngland,forexample,callingawoman“virtuous”impliedthatshewaschaste.SoaVictorianmoralistcouldmanagetodescribeawoman,andnotmerelytoevaluateher(expresshisemotionalattitudetowardherandinviteotherstoshareit),bycallingher“virtuous.”SomeadvantagesofemotivismStevenson’stheorywasenormouslyinfluentialinthemiddleofthetwentiethcentury.TakingitscuefromAyer’sshortremarksonethicsinLanguage,TruthandLogic,Stevenson’stheoryaddedsophisticationandsubtlety(seeAYER).In“TheEmotiveMeaningofEthicalTerms,”Stevensonsetsoutsomecriteriaforasuccessfulanalysisofmoralterms,explainingthatwhathecallstraditional“interesttheories”ofethicaltermsfailoneormorecriteria.TheseinteresttheoriesincludetheviewsofHobbes,whomStevensonunderstoodtohavedefined“good”tomean“desiredbyme,”andHume,whomheinterpretedasdefiningittomean“desiredbymycommunity.”Inthefirstplace,wemustbeablesensiblytodisagreeaboutwhethersomethingis“good.”ThisconditionrulesoutHobbes’sdefinition.Forconsiderthefollowingargument:“Thisisgood.”“Thatisn’tso;it’snotgood.”AstranslatedbyHobbes,thisbecomes:“Idesirethis.”“Thatisn’tso,forIdon’t...”Inthesecondplace,“goodness”musthave,sotospeak,amagnetism.Apersonwhorec-ognizesXtobe“good”mustipsofactoacquireastrongertendencytoactinitsfavourthanheotherwisewouldhavehad.ThisrulesouttheHumeantypeofdefinition.Foraccord-ingtoHume,torecognizethatsomethingis“good”issimplytorecognizethatthemajor-ityapproveofit.Clearly,amanmayseethatthemajorityapproveofXwithouthaving,himself,astrongertendencytofavourit...Inthethirdplace,the“goodness”ofanythingmustnotbeverifiablesolelybyuseofthescientificmethod.“Ethicsmustnotbepsychology.”Thisrestrictionrulesoutallofthetraditionalinteresttheories,withoutexception.Emotivismappearstobewellpreparedtosatisfythesecriteria.First,peoplecangen-uinelydisagreewhenonestatesthatXisgoodandtheotherstatesthatXisnotgood;theyaredisagreeinginattitude,asStevensonputsit,notinfactualbelief,butthisisgenuinedisagreementjustasplainlyaswemaydisagreewhenIsuggestthatwegotothemoviestonightandyousuggestthatwegohaveafewdrinksinstead.Second,andperhapsmostsignificantly,ifthejudgmentthatXisgoodisanexpressionoffavorableattitudetowardX,thenitisclearwhyanyonemakingsuchajudgmentwillhaveaten-dencytoactinfavorofX.Finally,whileaddingtoourknowledgebyscientificinvesti-gationmaysometimesresolvecertainethicalissues,therecanbedeeperdisagreementsthatareleftuntouchedbyscientificmethods.Ouremotionalattitudesmaydifferinthefaceofconvergingempiricalknowledge.Afurtherattractionofemotivismisthatitdissolvesknotty-lookingmetaphysicalproblemsofmetaethics.Considerthequestionofwhethermoralpropertiesarenaturalpropertiesorsomeotherspecialsort.G.E.Moorefamouslyarguedthatmoralproper-176\nC.L.STEVENSONtiescouldnotbenaturalproperties,andlaterJohnMackiearguedforskepticismabouttheexistenceofmoralpropertiesonthegroundsthattheycouldnotbenaturalones,andthemetaphysicsandepistemologyofnon-naturalpropertiesistoospooky(or“queer,”asMackiesaid)forsoberphilosophy.Somemetaethicistshavetriedtoshowhowmoralpropertiescouldbepartofthenaturalworldafterall,butitisdifficulttoexplainjusthowourlinguistichabitsandpracticescoulddeterminejustwhichnaturalpropertymoralwrongnesscouldbe,giventhewidediversityanddisagreementinmoralvaluesamongdifferentpeopleandculturesatdifferenttimes.Emotivismresolvestheissuebydenyingthatmoralpredicates,like“wrong”and“good,”servetopickoutprop-ertiesatall.Theyserveasmarkersofmoodoremotioninstead.SothemetaphysicsofallegedmoralpropertiesisavoidedifweadoptStevenson’sview.SomedifficultiesforemotivismEmotivismisnotwithoutitsdifficulties,andthemainoneswereleveledatStevensonsoonafterhebegantopublishhisviews.OnecriticismwasofferedbyBrandBlanshardinapapercalled“TheNewSubjectivisminEthics.”Blanshardcomplainedthatemo-tivismhasanobviouslyfalseimplication.WhenIseearabbitwithitsfootcaughtinatrap,Imightsay(orthink)“That’sabadthing.”Iwouldthen,plausibly,beexpressingmynegativeemotiontowardthepainoftherabbit.ButsupposeIthencontemplatethesituationinwhichImyselfbecomeveryjadedandceasetocareaboutthesufferingofsentientanimals.DoI(now,actually)say,“Well,inthatcase,thesufferingoftherabbitwouldnotbeabadthingatall”?No,ofcoursenot.ButemotivismimpliesthatthisishowIshouldthink.Soemotivismisfalse.Thiscriticismisinstructive,thoughitisnotcorrect.ItillustratestwoimportantpointsaboutStevenson’stheory.First,thefactthatwewouldordinarilysayonethingoranotherisveryimportant,accordingtoStevenson’sapproach.Hewouldneverhavereplied,“Wemightnotsaysuchathinginthatcase,wemightsteadfastlydenyit,butwewouldbemistaken.”Histheorywassupposedtoaccountforourordinaryjudg-ments,andnottoreformthosejudgments.SoitisimportantwhetherBlanshard’sexamplereallydoesshowthatemotivismsometimescontradictsourordinaryethicaljudgments.However,thecriticismisunsound,becauseStevenson’sapproachdoesnot,infact,implythatwedoorshouldjudgethatthesufferingoftherabbitwouldnotbeatallbadifwewerejadedanduncaring.Tothinkthatitdoesimplysuchathingistomistakeemotivismforapoorrelation,subjectivism.Thesubjectivistthinksthat“bad”means(somethinglike)“apttocauseanegativeemotioninme.”Sotocallsomethingbad,accordingtosubjectivism,istosaythatitcausesanegativeemotioninoneself.ButStevensontookgreatpainstodistinguishhisownviewfromsubjectivism,andhegaveverysimilarexamplestoshowwhysubjectivismisincorrect.Accordingtoemotivism,remember,callingsomethingbadisnotsayingthatitdoesordoesn’tdoanything–thatwouldbetodescribethething.Ethicallanguagedoesnot(primarily)describeathingoranemotionorthespeaker,itexpressestheemotionofthespeaker.WhenIcontem-platethesituationinwhichIheartlesslyfeelnosorrowovertherabbit’ssuffering,I(rightnow,actually)feelratherbadaboutthat,andifIweretoexpressmyemotionIwouldsay,“Thatwouldbeabadthing.”177\nJAMESDREIERProbablythemostinfluentialcriticismofStevenson,thecriticismthatlateremo-tivists(andfellownon-descriptivists,seebelow)havebeenmostconcernedtoaddress,wasaproblemnoticedbyPeterGeachandJohnSearle.Itissometimescalledthe“embeddingproblem.”Toputitsuccinctly,theproblemisthatevenifemotivismreallydoestelluswhatsomebodydoeswhensheassertsasimplemoralsentencelike“Itiswrongtokickcats,”itdoesnotseemtotelluswhatsuchasentencemeans.Forthereismoretothemeaningofasentencethanthefactsaboutwhatisaccomplishedorexpressedbyanassertionofit,sincewecanusesentenceswithoutassertingthem,inunassertedorembeddedcontexts.Therearemanykindsofunassertedcontexts.Hereareafewexamples;noticethatinnocasewouldsomeonesincerelyutteringtheentiresentencebeassertingthatitiswrongtokickcats.Ifitiswrongtokickcats,thenitiswrongtokickTibbles.Eitheritiswrongtokickcats,orthereisnothingwrongwithkickingpeople.Iwonderwhetheritiswrongtokickcats.Doyoumeantosaythatitiswrongtokickcats?Manyotherkindsofexamplescouldbegiven,buttheideaisclearenough.CriticsofemotivismpointoutthatwhatStevensonsaidabouttheemotivemeaningofethicaltermsdoesnotseemtoexplainhowasentencelike“Itiswrongtokickcats”embedsintothesecomplexcontexts.What,thatistosay,doesthesentencecontributetothecomplexwhole?Onethingisclearenough:someoneutteringanyofthefourexamplesentencesabovecouldnotbesaidtobeexpressinganegativeemotiontowardkickingcats.Sosomethingmoremustbesaid.Stevensonhimselfneverseemstohavetakenthisproblemtoheart,soheneversaidmuchofanythingbywayofreply.Butsomelaternon-descriptivistshavesaidmore(seebelow).Theembeddingproblemmayappeartobeakindoftechnicality,andperhapsitis,thoughmanyphilosophershavetakenitveryseriously.ThefinalcriticismIwillmentionseemstocutdeeperintothespiritofemotivism.Stevensonsaid,andemotivismgainsmuchofitsplausibilityfromthisidea,thatapersonwhosincerelyassertsorbelievesamoraljudgmentmustnecessarilyfeelsomesortofemotionaltug,sothatwhoeverjudgessomethinggoodmustbeemotivelyinfavorit,andwhoeverjudgessomethingbadmustbeagainstitorinclinedtoavoid,orwouldtrytoeliminateit.Butwemaywonderwhetherthisclaimistrue.Isn’titpossibletojudgesincerelythatsomethingisgood,butfeelnosympathyorother“pro-attitude”towarditwhatsoever?Thereisnouncontroversialanswer.Somefinditobviousthatsuchathingispossible,whileothersareatleastatfirstinclinedtowonderwhatthequestionercouldpossiblyhaveinmind.ButifwetellabackgroundstoryitstartstoseemveryplausiblethatStevensonmayhaveoverstatedtheconnectionbetweenmoraljudgmentandemotion.Surelyitisimaginablethatsomeonecouldbeaself-avowedandsincereamoralist.Suchapersonwouldhavenointerestatallinmoralvaluesorrules,andmightevenbeperfectlyforthrightinadmittingso.Yetamoralistscouldsurelylearntorecog-nizewhichthingsaregoodandbad,evenifthenormalconcernwithsuchthingsmightseemquaintormisguidedtothem.Sotheycouldwithperfectsincerityandunder-standingmanagetojudgethatgivingtocharityismorallygood,orthatbreakingpromisesisbad,andtheycouldmakethosejudgmentswithoutanyemotionor178\nC.L.STEVENSONmotivationortendencytopromotethe“good”thingsordiscouragethe“bad”ones.Allofthisseemspossible.Doesn’titshowthatemotivismisamistakentheory?Perhapsnot.Intheparagraphabove,thewords“good”and“bad”areinquotationmarks.ItisplausiblethatamoralistsusetheseandothermoralwordsinwhatR.M.Harecalledthe“invertedcommassense,”reallymentioningthemratherthanusingthem.Amoralistscannotsay(sincerely,atleast)thatcharityisgood,soinsteadtheysaythatcharityiswhatmostfolkcall“good.”We(moralistsoramoralists)cancertainlymentionemotivewordswithoutexpressingtheiremotivemeanings.SomerelatedtheoriesHare’stheoryisnotemotivist,thoughitisacloseally.AccordingtoHare,themainfunctionservedbymoraljudgmentsisprescription(seeHARE).SoHareagreeswithStevensonthatwedonotfundamentallydescribethingswhenwecallthemgoodorbad,andheevenagreesthatmoraljudgmentscouldbecalled“expressionsofemotion,”sinceprescriptionsareexpressions,inasense.ButHarecautionsagainsttakingStevenson’sideatooliterally.InTheLanguageofMorals,hewrites:Wespeakofexpressingstatements,opinions,beliefs,mathematicalrelations,andsoon;andifitisinoneofthesesensesthatthewordisused,thetheory,thoughittellsuslittle,isharmlessenough.Butunfortunatelyitisalsousedinwayswhichareunlikethese;andAyer’suse(inspeakingofmoraljudgements)oftheword“evince”asitsroughsynonymwasdangerous.Artistsandcomposersandpoetsaresaidtoexpresstheirownandourfeelings;oathsaresaidtoexpressanger;anddancingonthetablemayexpressjoy.Thustosaythatimperatives[ormoraljudgments]expresswishesmayleadtheunwarytosupposethatwhathappenswhenweuseone,isthis:wehavewellingupinsideusakindoflonging,towhich,whenthepressuregetstoogreatforustobear,wegiveventbysayinganimperative[ormoral]sentence.Suchaninterpretation,whenappliedtosuchsentencesas“Supplyandfittodoormortisedeadlatchandplasticknobfurniture”,isunplausible.Inthe1980sand1990sSimonBlackburnandAllanGibbarddevelopedversionsofemotivism(orinGibbard’smoregeneralterminology,“expressivism”)groundedinthesamerootideasasStevenson’stheory.Thesetheoriesaremoresophisticatedinvariousways(inparticulartheymakegoodheadwayintotheembeddingproblemmentionedabove),andtheyhavetosomeextentsupplantedStevenson’semotivism,thoughasinheritors,notasrefuters.BibliographyWorksbyStevenson1944:EthicsandLanguage,NewHaven,CT:YaleUniversityPress.1963:FactsandValues:StudiesinEthicalAnalysis,NewHaven,CT:YaleUniversityPress.(Acol-lectionofpapers.EssayTwoisespeciallyusefulasanintroduction.EssayElevencontainsthemostmatureversionofthetheory.)WorksbyotherauthorsAyer,A.J.(1936)Language,TruthandLogic,NewYork:OxfordUniversityPress.(Chapter6con-tainsthegermofemotivism,whichStevensondevelopedintoafullandsophisticatedtheory.)179\nJAMESDREIERBlackburn,S.(1984)SpreadingtheWord,Oxford:OxfordUniversityPress.(Chapter6especiallydevelopsavariationonemotivismdesignedtoaddressprominentobjections.)Blanshard,B.(1949)“TheNewSubjectivisminEthics,”PhilosophyandPhenomenologicalResearch9,pp.504–11.Geach,P.T.(1960)“Ascriptivism,”PhilosophicalReview69,pp.221–5.(Aninfluential,butsomewhattechnical,objectiontoemotivism.)Gibbard,A.(1990)WiseChoices,AptFeelings,Cambridge,MA:HarvardUniversityPress.Goldman,A.andKim,J.(eds.)(1978)ValuesandMorals:EssaysinHonorofWilliamFrankena,CharlesStevenson,andRichardBrandt,DordrechtandBoston:Reidel.(Acollectionofcriticalessays;thoseonStevenson’sworkarequiteaccessible.ContainsacomprehensivebibliographyofStevenson’swriting.)Hare,R.M.(1952)TheLanguageofMorals,Oxford:OxfordUniversityPress.(Aworkcon-temporarywithStevenson’s,withimportantsimilaritiesandcontrasts.)Searle,J.(1969)SpeechActs,Cambridge:CambridgeUniversityPress.180\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200115W.V.Quine(1908–2000)PETERHYLTONWillardVanOrmanQuinewasbornonJune25,1908.HewasgraduatedfromOberlinCollegewithadegreeinmathematics,summacumlaude,in1930;hisseniorhonorsthesiswasaproofwithinthesystemofWhiteheadandRussell’sPrincipiaMathematica(1910–13),whichhestudiedlargelywithoutaidfromhisteachers.WhiteheadwasinthePhilosophyDepartmentatHarvard,soitwastherethatQuinewenttodograduatework,althoughWhiteheadwasnolongerteachinglogicandQuinehaddonelittleundergraduateworkinphilosophy.Nonetheless,hecompletedaPh.D.intwoyears,graduatingin1932.HisdissertationgeneralizedPrincipia’streatmentofclassessothatitincludeddyadicrelations,insteadoftreatingthelatterseparately.Alongtheway,Quineclarifiedandreformulatedthebasisofthesystem–apointtowhichweshallreturn.Quinespenttheacademicyear1932–3inEuropeonaSheldonFellowship.HespentfivemonthsinVienna,attendingsomemeetingsoftheViennaCircle.Moreimportant,perhaps,wasashorterstayinPrague,wherehehadextensiveconversationswithRudolfCarnap,thencompletingTheLogicalSyntaxofLanguage(Carnap1934).WhileinEuropehewaselectedasoneofthefirstgroupofJuniorFellowsinHarvard’snewlyformedSocietyofFellows:athree-yearfellowship,withoutteachingobligations.Hespentmostoftheseyearsworkingonlogicandsettheory,thoughsomeofitonotheraspectsofphilosophy.Underthislatterhead,hegavethreelecturesonCarnap,essen-tiallyexpounding,inastronglyapprovingfashion,whathetooktobethedoctrinesofTheLogicalSyntaxofLanguage.(ThetextoftheselecturesisnowpublishedinCreath1990.)In1936QuinebecameFacultyInstructoratHarvard.ExceptforserviceintheUSNavyduringWorldWarII,heheldfacultypositionstherefromthattimeuntilhisretirementin1978.Heremainedphilosophicallyactiveandengagedfortwentyyearsafterretirement,continuingtowriteandpublishintohisnineties.ItwillbehelpfultoputQuine’sworkinthecontextofwhatIshallcalltwentieth-centuryscientificphilosophy,amovementwithinthebroaderstreamoftwentieth-centuryanalyticphilosophy.Keyfiguresintwentieth-centuryscientificphilosophy(otherthanQuine)includeBertrandRussellandRudolfCarnap,aswellasothersoftenidentifiedaslogicalpositivistsorlogicalempiricists;FregeandWittgensteinalsomadecrucialcontributionstothemovement.181\nPETERHYLTONLetustrybrieflytocharacterizethismovementbyaimsanddoctrinesratherthanbyitsparticipants.Perhapsmostnotableistheemphasisonknowledge,anditsobjects,ratherthanonethicsorpoliticsoraestheticsorhistoryorthehumancondition,astheprimaryconcernofphilosophy;anemphasis,onemightsay,ontheTrueratherthanontheGoodortheBeautiful.Thisemphasisisequallyanemphasisonscience,espe-ciallyonthenaturalsciences.Itischaracteristicofscientificphilosophytotakethenaturalsciencesasparadigmaticofallknowledge.PartofthisviewisthedoctrinethattheViennaCirclecalled“theUnityofScience.”Thepointhereistheunityofallrealknowledge,fortheGermanwordWissenschaftisbroaderinitsscopethanmostcurrentusesofitsinvariableEnglishtranslation,“science.”(Quine,however,makesitexplicitthatheusestheword“science”broadly;seeQuine1995:49.)Accordingtothisview,therearenofundamentaldivisionsofaimormethodamongthevariousbranchesofknowledge.Alongwiththis,thereisasuspicion,orworse,oftheclaimsofmetaphysics,andofanyclaimsneitheranswerableinstraightforwardfashiontothefindingsofempiricalsciencenorprovablebylogicandmathematics.Thisbriefsketchatonceraisesquestionsabouttherebeinganyroleforphilosophy,eventhatofthescientificphilosophersthemselves.Acentralideaherewasthatphi-losophyisnottoaddtoourknowledgebutis,rather,toanalyzetheknowledgethatthesciencesgiveus,andtherebytogiveusgreaterclarityaboutthatknowledgeanditsbasis.Thetoolofthisanalysiswas,aboveall,logic:thelogicofFregeandRussell(seeFREGEandRUSSELL).Thislogicheldoutanidealofclarity;oneaspectofthephiloso-pher’staskwastoimposeasimilarclarityuponothersubjects.Thislineofthoughtsug-gestsanassimilationofphilosophytologic,butdoesnotbyitselfaccountforthepossibilityofeitherofthesesubjects.Russellsoughttodothisbypostulatinganaprioriinsight,whichmightstrikesomeasalargeconcessiontometaphysics,inthepejora-tivesense.Carnap,drawingontheearlyworkofWittgenstein,heldthatlogicisana-lytic,emptyofcontent,andhencenotgenuineknowledgeatall.Philosophytoomakesnoclaimsabouttheworld.Itsanalysesoflanguagesimplymakeexplicitwhatisalreadythere;itrecommendsacertainkindoflanguageforthisorthatscientificpurpose,butarecommendationisnotaclaim,andispresumablynotinneedofthesamesortofjustification.ThisemphasisonlanguageisconnectedwithCarnap’sviewthatanalytictruthsaretrueinvirtueoflanguage,truebyvirtueofthemeaningsofthewordsmakingthemup.Giventheimportanceofanalyticityasaccountingforlogic,formath-ematics,andforphilosophyitself,thisthrowsanenormousexplanatoryburdenonthenotionoflanguage.Wehavejustverybrieflysketchedthetraditionoftwentieth-centuryscientificphiloso-phy.Quine’spositionrelativetothistraditionisambivalent.Ontheonehand,heisitsgreatestexponentinthelastfortyyearsofthecentury.Ontheotherhandherevo-lutionizesit,insuchawaythatonemightsaythatherejectsthetraditionratherthancontinuingit.BothRussellandCarnapattributedgreatimportancetothenaturalsciencesbutneverthelessheldthatlogic,mathematics,andphilosophyitself,allhaveastatusthatisquitedifferentfromthatof,say,physicsorchemistry,orhistoryorsociology.Theformerareindependentofobservation,andthusapriori–howeverexactlythatideaistobeunderstood–whilethelatterareaposteriori,empiricalandultimatelyanswerabletoobservationandsensoryexperience.Quinerejectsthe182\nW.V.QUINEideathatthereisafundamentalepistemologicaldistinctionhere.Thisrejection–whichQuinehimselfsometimesspeaksofashisnaturalism–isfundamentalforhisphiloso-phy.Weneedtoseewhyherejectstheapriori,andhowhecangetbywithoutit;weshallthentobegintoshowhowhisgeneralapproachtophilosophyflows,inlargepart,fromthisstep.AnalyticityandtheaprioriLikeCarnap,Quinerejectsanyideaoftheaprioriasbasedonpureintuition,oronpurereason;suchanidearunscountertohisscientificandempiricistpredilections.Carnapappealedtotheideaofanalyticityasanalternative(seeCARNAP).Quine,famously,alsorejectsCarnap’suseofthisideaandwithitanysignificantideaoftheaprioriorofnecessity.WeshalldiscussQuine’sargumentsagainstCarnap’snotionofanalyticity,oragainsttheideathatthereisaseriousandsignificantdistinctiontobemadebetweentheanalyticandthesynthetic.Quine’srejectionofthedistinction,however,isonlyhalfthestory.Theotherhalfistoshowhowhecanmakesenseoftheapparentlyaprioristatusoflogicandmathematicswithoutit–or,better,perhaps,howhecanaccountforthosefactswhichhaveledphilosopherstothinkthatlogicandmathematicsmustbeapriori.UnderstandingQuine’sattackonCarnap’snotionofanalyticityiscomplicated,partlybecauseitwasforatime,Ithink,notentirelycleareventoQuinehimselfexactlywhatheisattackingandhow.QuinethinksofCarnapiananalyticityastruthinvirtueofmeaning,andsoalsothinksthatifwehadaclearunderstandingofthenotionofmeaning–moreprecisely,ofsynonymy,orsamenessofmeaning–thenwewouldhavegonealongwaytowardsmakingclearsenseofanotionofanalyticity.SoforalongtimeQuine’sattackonanalyticityseemedtobepartandparcelofanattackonthenotionofmeaning,asunclearorundefined.AndcertainlyQuineisskepticalastohowfarwecanmakeclearsense–whichforhimmeanssenseinscientific,especiallybehav-ioral,terms–oftheideaofsynonymy.(Heisnot,however,skepticalofthenotionofmeaningfulness.Seehisessay“TheProblemofMeaninginLinguistics,”inQuine1961,theburdenofwhichispreciselythatthenotionofmeaningfulnessisnotafflictedwiththesamesortsofproblemsasthenotionofsamenessofmeaning,butlendsitselftoaready,ifsomewhatrough,understandinginbehavioralterms.)Hisviewofsynonymyisnotthatthereisnosenseatalltobemadeofitanywhere,eventhoughhehasreasontothinkthatwemaynotbeabletomakecompletesenseofiteverywhere.SoweseeQuine,asearlyas“CarnapandLogicalTruth”(1963,writtenin1954)acceptingthattheremaybealimitednotionofanalyticitytobehad.Hiswillingnesstoacceptsomenotionofanalyticitybecomesmoremarkedastimegoesby.InRootsofReference(1974)heproposesatentativedefinitionofsynonymy,andwithitanunderstandingofana-lyticity;in“TwoDogmasinRetrospect”(1991)wefindQuinearguingthat(first-order)logicisanalytic.Howarewetounderstandthissituation?Quinecontinuestorejecttheideaofanotionofanalyticitythatwouldplayany-thinglikethecentralphilosophicalrolethatCarnapallottedit.Inordertoplaythatrole,anotionofanalyticitywouldhavetomeettworequirements.First,itwouldhavetohavetherightscope:thetruthsoflogicandmathematics,atleast,mustcomeoutasanalytic.Second,itmustalso,atleastinQuine’sview,markasignificantepistemo-183\nPETERHYLTONlogicaldistinction:analytictruthsneednojustification(orelsewhatcountsas“justifi-cation”forthemiswhollydifferentinkindfromthejustificationofsynthetictruths).WhileQuineacceptsanotionofanalyticity,itisnotonethatsatisfieseitheroftheserequirements.Letusbeginwiththequestionofscopeanddefinition.Quine,aswesaw,takesanalyticitytobetruthinvirtueofmeaning.Buthowarewetounderstandtheideaofmeaning,asitoccurshere?ForQuine,theonlythingthatcouldberelevanttothemeaningofawordorasentenceinagivenlanguageishowitisusedbyspeakersofthatlanguage.Thisisanimportantpoint.Quinehasbeenaccusedofbeingundulybehavioristic,especiallyaboutlanguage.Certainlyhehasageneralbiasinfavourofabehavioristapproachtothemind.Heclaims,however,thathisinsis-tenceonapproachinglanguage-usebehaviorallyisnotmerelytheresultofprejudice.Indeedheoffersanargumentforsomeformofbehaviorisminthiscontext.Thepassageisworthquotingatsomelength:Inpsychologyonemayormaynotbeabehavioristbutinlinguisticsonehasnochoice.Eachofuslearnshislanguagebyobservingotherpeople’sverbalbehaviorandhavinghisownfalteringattemptsobservedandreinforcedorcorrectedbyothers.Wedependstrictlyonovertbehaviorinobservablesituations.Aslongasourcommandofourlanguagefitsallexternalcheckpoints...solongalliswell.Ourmentallifebetweencheckpointsisindif-ferenttoourratingasamasterofthelanguage.Thereisnothinginlinguisticmeaningbeyondwhatistobegleanedfromovertbehaviorinobservablecircumstances.(Quine1990:37–8)ForQuinetherecanbenomoretomeaningthanisimplicitintheactualusethatismadeofthelanguage.Quine’sinterestisexclusivelyinknowledge,andtheaspectoftheuseoflanguagethatprimarilyconcernshimisouracceptingornotacceptingsentences.Thushesays,earlyon:“inpointofmeaning...awordmaybesaidtobedeterminedtowhateverextentthetruthorfalsehoodofitscontextsisdetermined”(1936:89).Butthenthequestionis:whichofthecontextsofawordmustbesodeterminedinordertodeter-mineitsmeaning?Withoutsomereasontodiscriminate,wehavenoreasontotreatonecontextasmoredefinitiveofaword’smeaningthananyother.Butthennotruesentenceinwhichthewordappearswouldhaveanybetterclaimtobeanalyticthananyothersuchsentence;clearlynousefulanalytic/syntheticdistinctioncanbeerectedonthatbasis.Whatsortofthingmightgiveusreasontodiscriminateamongcontexts?Ifmasteryofsomesmallsubsetofaword’susesgaveonemasteryofitsuseasawhole,thentherewouldbereasontosaythatthoseuses,thosecontexts,constituteditsmeaning.Andclearlythishappensinsomecases.Achildwhootherwisehasafairdegreeoflinguis-ticsophisticationbutdoesnotknowtheword“bachelor”canbegivenamasteryofthatwordallatonce,atasinglestroke,bybeingtoldthatbachelorsareunmarriedmen.Thisfactgivesuseveryreasontosaythat“bachelor”means“unmarriedman,”andthatthesentence“Allbachelorsareunmarried”isanalytic–whichQuine,atleastinhislaterwork,certainlyaccepts(seeQuine1991:270).Alongtheselines,heproposesadefinitionofanalyticity:“asentenceisanalyticifeverybodylearnsthatitistruebylearningitswords”(1974:79).Hearguesthatfirst-orderlogicisanalyticbythissort184\nW.V.QUINEofdefinition,butthatotheranalytictruthswillallbetrivial.Inparticular,thereisnoprospectofarguingonthissortofbasisthatmathematicsisanalytic;apartfromotherconsiderations,Gödel’sincompletenesstheoremwouldbeaninsurmountablebarriertosuchanargument(seeTARSKI,CHURCH,GÖDEL).Wehaveyettodiscussthequestionoftheepistemologicalsignificanceofthenotionofanalyticity.This,Quinecametosee,isthecrucialquestion;inthe1980shewrote:“Inowperceivethatthephilosophicallyimportantquestionaboutanalyticityandthelinguisticdoctrineoflogicaltruthisnothowtoexplicatethem;itisthequestionoftheirrelevancetoepistemology”(HahnandSchilpp1998:207).Whyshouldanyonethinkthatshowingasentencetobeanalyticforagivenlanguage–learnedinthecourseoflearningthelanguage–showsanythingaboutitsepistemologicalstatus?Whymightonethinkthatitshowsthatforthatsentencenojustificationisrequired,orthatthequestionofjustificationissomehowmisplaced?Well,clearlyitmightbethoughttoshowthatgiventhatwearespeakingthatlanguagethequestionofthejustificationofthatparticularsentencedoesnotarise.Butwhydoesthequestionnotsimplybecomeoneofthejustificationforspeakingthatlanguage?Wearepresumablyoperatingherewithverytightidentity-criteriaforlanguages,sothatshiftingthemeaningoftheoneword“bachelor”wouldmeanthatwewerespeakingadifferentlanguage(ifthisseemsexces-sivelyodd,wemightspeakintermsofidiolectsratherthanlanguages;butthepointisthesame).Andgiventhatconceptionofalanguage,itisnotobviouslyabsurdtoaskforthejustificationforspeakingagivenlanguage(itisnolongerenoughtosay,“itistheoneIwasbroughtupwithandfeelmostathomein,”forthisqualitywouldsurviveminorshifts).ForCarnap,thechoiceofalanguageisinepistemologicallyimportantwaysunlikethechoiceofatheorywithinalanguage.Theformerisnotamatterofcorrectness,ofrightorwrong;itisapracticalmatterhavingtodowithpragmaticfactorssuchasthesimplicityofagivenlanguageanditsconvenienceforthisorthatgoal.Thisideaissuesinwhathecallsthe“PrincipleofTolerance”:sincechoiceoflanguage,unlikethechoiceofatheorywithinalanguage,isnotamatterofcorrectnessorincorrectness,weshouldbetolerant,andallowpeopletoworkwithwhateverlanguagetheychoose.Withinagivenlanguage,justificationismoreorlessrule-governed,governedbytherulesofthatparticularlanguage;justification,likeothersignificantphilosophicalnotionsinCarnap’sview,isthuslanguage-relative.Butthechoiceofalanguageitselfisnotsome-thingthatcanbejustifiedinthesamesortofway,sincewithoutalanguagewehavenorulesofjustificationtowhichtoappeal.Carnapemphasizesthedistinctionbetweenthejustificationofchoiceofatheorywithinalanguageandthejustification(orthelackofneedforjustification)ofchoiceoflanguage.Theideathatattributinganalyticitytoasentencehasepistemologicalsig-nificancedependsuponthisdistinction.Sayingofasentencethatitisanalyticwouldmeanthatitisinsomesenseintegraltothelanguagethatwecurrentlyspeak.Soifweceasedtoacceptthatsentencewewouldhavemodifiedthelanguage.Butthatwouldleaveopenthepossibilitythatwemighthaveevidencewhichwouldjustifythatmodi-ficationofthelanguage.Thusitwouldseemthatevidencemightbearonananalyticsentenceinthesamesortofwayinwhichitbearsonasyntheticsentence,unlessthenotionsofevidenceandjustificationinplayareofdifferentkindsinthetwocases.Carnap,ofcourse,holdstherethereisjustsuchadifferenceinkind.Heclaimsthat185\nPETERHYLTONtherearequitedifferentconceptionsofevidenceandjustificationatwork.Withinthelanguage,justificationisarule-governedprocedure,andamatterofrightorwrong;whenthelanguageitselfisbeingchosen,however,therearenorulestowhichtoappeal,andthechoiceispurpose-relativeandtosomeextentarbitrary.Quineattacksthisdistinctionfrombothsides.Hedeniesthat(internal)justificationistoanysignificantextentarule-governedprocedure.Hesays,forexample:Iamimpressed...apartfromprefabricatedexamplesofblackandwhiteballsinanurn,withhowbafflingtheproblemhasalwaysbeenofarrivingatanyexplicittheoryoftheempiricalconfirmationofasyntheticstatement.(1961:41–2)Healsoinsiststhatallourcognitivechoices,includingthechoiceofalanguageforknowledge,aredirectedtowardsthesameend:achievingthemostsuccessfultheory,whereacrucialtestofsuccessisthegenerationoftruepredictions.Vaguervirtues,suchassimplicityandfruitfulnessarealsorelevant.ThesearethesortsofthingsthatCarnapcountedas“pragmaticfactors,”applicabletoquestionsoflanguage-choice.QuineclaimsthattheyareapplicablealsotowhatCarnapwouldcountasempiricalbeliefs.Theymaynotinanyveryobviouswaybeapplicabletothequestionwhetherthereisnowadeskinfrontofme,butcertainlytheyaretomoreorlessabstractclaimsoftheoreticalphysics.ForQuinethereisacontinuumhere,withnosharpbreakstobehad.Quinethusholdsthatevenwherewehaveasignificanttruthwhichisanalytic,thisstatussimplydoesnotmatterepistemologically:“Momentumisproportionaltovelocity”countsasanalytic.Butdowecare?Einstein’srelativitytheorydeniestheproportionalitylaw,complicatingitwithaformulainvolvingthespeedoflight.ButinsteadofaccusingEinsteinofacontradictioninterms,wesimplystandcorrected.(Quine2000)NowCarnapmightagreethatwe“standcorrected”becauseweacceptthatEinsteinhasshownusthatanon-Newtonianlanguageworksbetterformakingsomepredic-tions,buthewouldinsistthatthisisadifferentsenseofcorrectionfromthatinwhichwearecorrectedwhenwechangeourmindaboutabeliefthatdoesnotinvolveachangeoflanguage.ButthisispreciselywhatQuinedenies,aswehaveseen.OneissuewhicharoseabovewasQuine’sviewofthenatureofjustification,therela-tionbetweentheevidencewehaveandthebeliefsthatweholdonthebasisofit.Thepointtherewasthatthereisnot,ingeneral,asimplerelationshipbetweenasentence,ontheonehand,andanobservationorgroupofobservationsthatjustifyit,ontheotherhand.Justificationisnot,ingeneral,asimpleandrule-governedmatter.Ofcoursetherearesentences,suchas“thereisadeskinfrontofmenow,”whichdoseemtohaveaverystraightforwardrelationtoobservations.Whatmakesthatcasestraightforwardisthatithardlymatterswhatelseapersonbelieves:giventherightobservations,almostanyonewillacceptthatthereiscurrentlyadeskinfrontofthem.Thejustificationrela-tionhereholdsbetweenobservationsandtheindividualsentencebelieved,whateverone’sotherbeliefsmaybe.InQuine’sview,however,thisisapoorparadigmtouseforknowledgeasawhole.Ingeneral,justificationisholistic,meaningthatitdoesnotapply186\nW.V.QUINEtosentencestakenindividually,inisolationfromothers,butrathertolargerorsmallerchunksoftheory,madeup,insomecases,ofalargenumberofsentences.Manyofthesentencesweaccept–mostobviouslythemoreabstractandtheoreticalones–haverelationstoobservationsonlyifwetacitlyassumemanyothersentences.Theseothersentences,backgroundassumptions,arerequiredifthesentenceinwhichweareinter-estedistohaveanyobservationalconsequencesatall.Fromthepointofviewoftheworkingscientist,thebackgroundassumptionsmaybeconfidentlyaccepted,andonlytheindividualsentenceupfortesting.Fromamoreabstractpointofview,suchasQuine’s,however,whatistestedbyobservationisnottheindividualsentencealone,butratherthewholesetofsentencesthatimpliestheobservationalconsequences.Fromasufficientlyabstractpointofview,indeed,itisalwaysthewholeofourknowledgethatistested.Anytestofasentencepresupposestruthsoflogicamongitsbackgroundassumptions.Logic,however,isusedeverywhereinoursystemofbeliefs,soinaratherPickwickiansenseitisthatsystemasawholethatisatstake.(Quinecallsthisextremeholism“legalistic”(1991:268).Itakethistomeanthatitholdsfromaveryabstractpointofview,butnotthatitisunimportant.)HolismisnotnewwithQuine.When“TwoDogmas”wasreprinted,QuineaddedafootnotetoDuhem,andthereisaclearstatementoftheviewinCarnap’sLogicalSyntaxofLanguage,withreferencestoPoincaréaswellastoDuhem(Carnap1937:318).Quine’susesofthedoctrine,however,arenovel.Oneusewehaveseen:itisthebasisoftheclaimthatjustification(withinalanguage)isnotthesortofrule-governedpro-cedurethatCarnapsometimessuggests,andsoisnotdifferentinkindfromthesortofjustificationthatappliestothechoiceofonelanguageratherthananother.Thisclaim,inturn,iscrucialforQuine’sviewofanalyticityandtheapriori,discussedinthepre-viousfewpages.AsecondusethatQuinemakesofholismalsorelatestothequestionoftheapriori,butinaquitedifferentway.Heattemptstoundercuttheideathattheremustbeaprioriknowledgebyinvokingholismtoexplainthephenomenawhichledsomephilosopherstoinvoketheideaoftheapriori,buttodosowithoutinvokingthatidea.(Ifthosephe-nomenacanindeedbeexplainedwithouttheapriori,thentheirexistencenolongerconstitutesareasontoaccepttheapriori.)Thereisnodoubtthatthetheoremsofmath-ematicsandoflogicarenotdiscoveredbyexperimentor,atleastinanyordinarysense,justifiedbyobservation.Andtheirfalsehoodseemscompletelyinconceivable.HowcanthesefactsbeexplainedinaccordancewithQuine’sviews?Holismprovidestheanswer.Logicfigureseverywhereinoursystemofbeliefs;mathematicsisusedinmanybranchesofknowledge.Nooneobservationorexperimentbearsonthem,butthesuccessofoursystemofbeliefsasawholeinpredictingexperienceprovidesjustifica-tion;thisjustificationisexceedinglyindirect.ForQuinethatonlyputsitatoneendof2acontinuumwhichwealreadyhavereasontoaccept,for“e=mc”isjustifiedverymuchlessdirectlythanis“thereisadeskinfrontofmenow.”Equallytheunimagin-abilityofthefalsityoflogicbecomesintelligible.Giventheubiquityoflogic,changingitmeansmakingchangeseverywhereinoursystemofknowledge.Itisnottobewonderedatifthisishardtoconceive.OurdiscussionsofarhasbeenfocusedonQuine’srejectionoftheapriori.Heisthusleftwithnokindofknowledgeotherthantheordinaryknowledgeofcommonsense187\nPETERHYLTONand(better)science.Philosophytoo,sinceitclaimstoyieldknowledge,mustbeofthissamegeneralsort.ThisisthedoctrinethatQuinecallsnaturalism;itisabsolutelyfun-damentaltohisthought.Therestofthisessaywilltakenaturalismasitsstartingpoint,andinvestigateQuine’sphilosophyasanunfoldingofthatdoctrine.Ishallbeginwithtopicshavingtodowithepistemology,andthenmovetotopicswhosefocusisontologyandmetaphysics(oritsQuineananalogue).Thereis,however,onefurtherissuewhichweshouldbrieflymentionhere.Quinehasarguedforthepossibilitythattwotranslatorsmightcomeupwithdiffer-enttranslationsofsomesentences–notmerelystylisticallydifferent,but“notequiva-lentinanyplausiblesenseofequivalence,howeverloose”(Quine1960:27).Thisisthecontroversialdoctrineknownastheindeterminacyoftranslation.Wheresentencesarenotassertedordeniedonanyverydirectobservationalbasis,theevidence(behavioralevidence,ofcourse)foronetranslationoveranotherismediatedbyothersentences;alternativetranslationsofallofthemmightcancelout,leavingeachoftwooverallschemesoftranslationasequallyjustified.InQuine’sviewtherewould,insuchacase,benorightandwrong,nofactofthematter:itwouldnotbeourknowledgewhichwaslacking,butratherthattherewasnofacttobeknown.Thisideaprovokedanenormousamountofdiscussioninthe1960sandearly1970s,andsomecommentatorshaveeventhoughtthatitiswhatreallyunderliesQuine’sobjectionstoanalyticity.Inmyview,however,itisnotofgreatimportancetohisthought,exceptasdramatizingtheideathatmeaningmustultimatelybeanswerabletobehavior(seeHylton1990).NoteinthisconnectionthatQuinenowspeaksofindeterminacyas“aconjecture”(seeHahnandSchilpp1986:728).KnowledgeandtherealmofthecognitiveHowshouldweconceiveofknowledge,ifwearetotakeascientificapproachtoit?FundamentaltoQuine’sthoughtistheideathatknowledgeistobeunderstoodasabiologicalphenomenon.Humanknowledgeisthoughtofasaconditionofthehumananimal.Itoriginatesinthestruggleofonespeciesofprimatetosurvive.TheopeningsentencesofFromStimulustoSciencereadlikethis:“Weandotheranimalsnoticewhatgoesonaroundus.Thishelpsusbysuggestingwhatwemightexpectandevenprevent,andthusfosterssurvival”(Quine1995:1).ThisiswhatonemightcallaDarwinianconceptionofknowledge:knowledgeasanadaptivemechanism,fosteringthesurvivalofthespecies.Inthisviewweseeadecisiverejectionofthelong-standingphilosophicaltraditionthatsharplydistinguishesbetweenrealknowledgeandmerebelieforopinion,betweenscientiaanddoxa.Thattraditiontendstoassimilaterealknowledge–scientia–toknowl-edgeofapropositionofmathematics,whenitisknownonthebasisofathoroughlyunderstoodproof.Realknowledgeisaccordinglythoughtofasinfallibleandknownwithcertainty.Noneoftheseideasfitswiththeideaofknowledgeasabiologicalphe-nomenon,aswhathelpsthehumananimaltogetbyinitsdealingswiththeworld.Somephilosophershaveheldthatthoseideasareimplicit,moreorless,intheword“knowledge.”Quine’sresponseissimplytoabandonthatwordfor“scientificandphilo-sophicalpurposes”(1987:109).Hecontinuestouseitinformally(asshallIinexpoundinghisviews),butwithouttheweightthatitmaybethoughttocarryin188\nW.V.QUINEgeneraluse.Hemakesnosharpdistinctionbetweenourknowledge,andthatwhichwe,orexpertsamongus,acceptuponreflection.Quine’sconceptionofcognition,then,isfallibilistthroughandthrough:nopartofoursystemofbeliefscanbecountedaswhollyimmunefromrevision,thoughsomepartsarenodoubtfarmoresecurethanothers.AsecondverygeneralpointaboutQuine’sconceptionofknowledgeisthathetakesittobelinguisticorverbal;atleastfor“scientificandphilosophicalpurposes,”hethinksofoursystemofbeliefsasbeingembodiedinsentences:Whatsortofthingisascientifictheory?Itisanidea,onemightnaturallysay,oracomplexofideas.Butthemostpracticalwayofcomingtogripswithideas,andusuallytheonlyway,isbywayofthewordsthatexpressthem.Whattolookforinthewayoftheories,then,arethesentencesthatexpressthem.(Quine1981:24)MuchofQuine’sinterestinlanguageandinitsanalysisarisesfromthefactthatourknowledgeisembodiedinlanguage.Inoneway,aswesaw,theideaofknowledge,asQuineemploysit,marksnoverysharpdistinction.Unlikesomephilosophers,hedoesnotusethetermasanhonorific,connotingsomeparticularhighdegreeofjustificationorofcertainty.Hedoes,however,makeasharpdistinctionbetweentherealmofthecognitiveandtherestofhumanactivity.Tocallahumanactivitycognitiveis,roughly,tosaythatitisanswerableto,ifnotexclusivelyaimingat,predictionsofsensoryexperience.Itisperhapsapresuppo-sitionherethatcognitiveactivityandcognitivelanguagecanbepeeledofffromthechaoticmassofhumanactivityandlanguagegenerally–oratleastthatwecanabstractwithoutdistortion,andtalkofthecognitivewhileignoringtherest.Thepredictionofexperienceisapracticalmatter;atthelimit,asweindicated,survivalisatstake.ThisisthesenseinwhichQuine’sconceptionofknowledgeisDarwinian.ThusfarQuineiswiththepragmatists.Ontheotherhand,tyingtheconceptofknowledgetothepredictionofsense-experienceenablesQuinetomakecleardistinctions,andtoerectbarriers,inplaceswherethepragmatistswouldnot.Activitieswhichpredictexperiencearecognitive;others,thoughtheymaycontributetohumanflourishinginotherways,arenot.Thejustificationforthisdistinctionispresumably(forQuineisnotexplicithere)thatsense-experienceisouronlywayoffindingoutabouttheworld(weshalldiscussthisideashortly).InspiteofQuine’spractical,Darwinianviewofknowledge,thereisthusasenseinwhichhisviewmakesacleardistinctionbetweenthetheoreticalandthepractical.Theoreticalsuccessissuccessinpredictionofsensoryexperience.EvidenceAstraditionallyconceivedbyphilosophers,thisnotionincludestwostrands.Ontheonehand,evidenceisthoughtofasconsistingoftheepistemologicallymostfundamentalitemsofourknowledge:evidenceisthatwhichis,sotospeak,firstintheorderofknowledge(thatwhichisevident),andsoalsothatfromwhichotheritemsofknowl-edgemustbeinferred.Ontheotherhand,evidenceisalsotoconsistofimmediatelygivendata,devoidofanyconceptualimpositionsofourown;sincenointerpretationis189\nPETERHYLTONinvolved,thereisnoroomfordoubt.Thesetwostrandsareintension.Whatisliterallyfirstintheorderofknowledgeseemtobefactsaboutotherpeopleandordinaryphysi-calobjects.Yetsentencesrecordingsuchfactsdonotseemsimplytorecordrawdata.Theyinvolveconceptualization,andare(notoriously)opentodoubt.Quine’sresponsetothesedifficultiesistoabandonthetraditionalconceptionofevi-dencecompletely,infavourofaphysicalisticalternative.Hespeaksnotof“thegiven”butofthestimulationofoursensoryreceptors,andofobservationsentences:roughly,sentencesthatanyspeakerofthelanguageisdisposedtoacceptorrejectsimplyonthebasisofcurrentstimulation.Thushesays:ourimmediateinputfromtheexternalworld[is]thetriggeringofoursensoryreceptors.Ihavecutthroughallthis[i.e.thedifficultiesofanalyzingthenotionsofobservationandexperience]bysettlingforthetriggeringorstimulationitselfandhencespeaking,oddlyperhaps,ofthepredictionofstimulation...Observationdropsoutasatechnicalnotion.Sodoesevidence,ifthatwasobservation.Wecandealwiththequestionofevidenceforsciencewithoutthehelpof“evidence”asatechnicalterm.Wecanmakedoinsteadwiththenotionofobservationsentence.(1990:2)Whatisatstakeinphilosophicaltalkofevidenceis,fromQuine’spointofview,theissueofhowwefindoutabouttheworld.Herewehaveacrucialexampleofhismethod,oftheideathatthestudyofknowledgeistobenaturalized,andasfaraspossibleputonaphysicalisticbasis.Forhim,theissueistobetakenasascientificquestion:howdowecomebyinformationabouttheworld?Andtheansweristhatwedosobytheimpactofvariousformsofenergyonoursensorysurfaces.Physicswilltelluswhatformsofenergythereare;physiologyandpsychologywillsaywhichformsofenergyhumanbeingscandetect,i.e.towhichformshumanbeingsrespond.Thecentralfacthereisthatitisonlythroughstimulationofournerve-endingsbyenergyimpingingonoursensorysurfacesthatwehumanbeingsknowanythingatallabouttheworld.ThisisfundamentaltoQuine’sepistemologyanditis,heemphasizes,“afindingofnaturalscienceitself”(1990:19).Quine’suseofthenotionofstimulationsisthussymptomaticofhisgeneralshiftinperspective.Thequestionisnot:whatisgiventome,attheoutsetofmycognitiveendeavors?Butrather:howdowehumansgainknowledgeoftheworld?Thisisonhisviewastraightforwardscientificandcausalquestion.Theanswerrefersustostimula-tionsofoursensorysurfaces.Quinethusabandons,orgreatlymodifies,thetraditionalconceptofsensoryevidence.Insodoingheshiftsthequestiontowhichtheoriginalconceptionwasananswer.Thetraditionalphilosopher’sdemandsarenotmetbyQuine’sview.Inthemoststraightforwardsense,theoccurrenceofstimulationsisinde-pendentofandpriortotheory.Ourknowledgeofsuchmatters,however,isclearlynotindependentoftheory.Sothereisherenoprospectofwhatthetraditionalphilosophersought:supportofthetheorywhichiswhollyindependentofthetheory.Totheextentthatthisstrikesusaproblemoraparadox,tothatextentwehavenotacceptedQuineannaturalism.ThepointwehavejustnotedlinksQuine’srejectionofthesensorygivenwithhisrejectionofthetraditionalconceptionoftheapriori,orofanyanalogue.Thegiven190\nW.V.QUINEandtheaprioriwereeachconceived,bysomephilosophers,asakindofextra-theoreticalknowledge,knowledgesomehowfreeofthevicissitudesaffectingtheordinaryknowledgeofcommonsenseandscience.Quinedeniesthatthereisanysuchkindofknowledge.Withinwhatwetakeourselvestoknowthereis,nodoubt,betterandworse;thereis,however,noknowledgeofawhollydifferentandsuperiorkind.Quineisconsistenthere.Hedoesnottakeeventhemostfundamentalpointsofhisownphilosophytobeapriori–andthisincludesthedoctrinethatweknowabouttheexter-nalworldthroughimpactsonoursensorysurfaces.Whileitisextremelyunlikely,thereareimaginablecircumstancesunderwhichwemightdropourpresentideaofsensoryevidenceentirely.Ifsufficientconfusionresultedfromourfollowingtheevidenceofoursenses,aswenowunderstandthem,andsufficientsuccessonthepartofthosewhoclaimtohearvoicesintheirheads,say,ourestimateoftheroleofstimulationofoursensorysurfacesmightchange.(Thispointisexplicit;seeQuine1990:20f.)ForQuine,aswehavesaid,thecentralfactaboutknowledgeisthatitisonlythroughstimulationofournerve-endingsthatweknowanythingatallabouttheworld.Suchstimulationsprovidetheonlyempiricalconstraintonoursystemofbeliefs,theonlyexternalcriterionofsuccess.(ByspeakingofanexternalcriterionImeantoleaveroomforwhatonemightthinkofasinternalfactors:suchastheoverallsimplicityofthesystem.)Thisfactsuggeststhattheremaybeempiricalslackbetweenevidence,eventhetotalityofallpossibleevidence,andtheory.Wecanfocusthisideabyaskingwhethertheremightbetwosystemsofbelief,differentfromoneanotherbuteachfullysuccessfulat“predictingstimulations.”Quine’sanswer,thoughsomewhatqualifiedandcomplicated,isthatnothingrulesthisout.Thisisthedoctrineknownastheunder-determinationoftheorybyevidence.Evenif,perimpossibile,alltheevidencewerein,stillanygiventheorybasedonthatevidencemightnotbeuniquelyjustified.Ifwehadatheorythatexplainedandpredictedtheevidencesatisfactorily,thatwouldofcoursejustifyit;sinceitwouldatleastinprinciplebepossibleforanothertheorytodoaswell,however,thejustificationwouldnotbeunique.Therelationofevidencetoknowledge:observationsentencesSofarwehavesaidsomethingaboutevidenceandsomethingaboutknowledge,asQuineconceivesthem,butnothingexplicitabouttherelationbetweenthetwo.Clearlythereisagulfbetweenthesensorystimulationsthatareouronlysourceofinforma-tionabouttheworld,andthemassofsentencesinwhichourbeliefsabouttheworldareembodied.Howisthisgulfbridged?Forthesentencesmostdirectlytiedtosensoryevidence,observationsentences,thereisafairlyclearaccount.Theothersentencesofourtheoryoftheworldgettheirrelationtoevidenceviatheirrelationtoobservationsentences,andheretheaccountismuchsketchier.(Afullaccountis,Quinethinks,notyetavailable,andmayneverbe.)Discussingobservationsentenceswillatleastgiveusanideaofhowthegapbetweenevidenceandtheoryisbridged.ThefirstpointtomakeaboutwhatQuinecallsobservationsentencesisthattheyare–byhislights–sentences.Thisdoesnotmeanthattheyareallsentencesinthegrammaticalsense:“Rabbit,”takenasacompleteutterance,isQuine’sownexample.Theideaofasentencehereisthatofapieceoflanguagewhichmaybeusedto191\nPETERHYLTONsaysomething,andthusmaybetrueorfalse.ForQuine,thefundamentalevidentialrelationholdsbetweensensorystimulationsandunitsoflanguageofthatsort.Wecanbringoutthesignificanceofthisideabycontrastingsentences(asweareusingthatword)withreferringexpressions,or“terms,”asQuinesays.Inmostuses,theword“rabbit”isaterm,notasentence:ifIsay“Iseearabbit”theninthatsentenceitisatermwhichfunctionsaspart,butonlypart,ofasentence.Masteryofthewordinthatsortofuse,Quineclaims,requiresmorethanamereabilitytorespondtothepresenceofrabbits.Itrequiresalsothatonecandistinguishthecircumstancesthatlicensetheclaim“There’sonerabbit”fromthoselicensingtheclaim“Therearemanyrabbits”;thatonecansimilarlydistinguish“There’sthesamerabbitagain”from“There’sanotherrabbit”;andsoon.Theabilitytomaketheseandrelateddistinctionsrequiressomeknowledgeofthewaysofphysicalobjectsingeneral,andofrabbitsinparticular.Sentencescontainingtermsarethusnotepistemologicallybasic.Masteryoftheuseofatermalreadyrequiresthatonepossesssomeknowledge,andthentheques-tionoftheevidenceforthatknowledgemustarise.Masteryoftheword“rabbit”asanobservationsentence,bycontrast,requiresonlythecapacitytorespondtoenviron-mentscontainingrabbitsinwaysinwhichonedoesnotrespondtoenvironmentswhichdonotcontainrabbits;thiscapacityrequiresnoauxiliaryknowledge.Thiskindofabilityisallthatispresupposedbythemasteryofanobservationsentence.Itisprimi-tiveandfundamental.Itiswhatunderliesallcognitivelanguage-use,includingthemostsophisticated,butotherformsoflanguagegobeyonditinprinciple.Tothispointwehavesaidlittlemoreaboutobservationsentencesthanthattheyaresentences,thoughthatideahasprovedtobefarfromtrivial.Beyondthat,anobserva-tionsentenceiswhatQuinecallsan“occasionsentence,”i.e.itisonethatistruewhenutteredundersomecircumstancesandfalseunderothers(so“It’sraininginLondon”isanoccasionsentence;“Goldisametal”isnot).TheroughideathatQuineintendstocaptureisthatanobservationsentenceisanoccasionsentenceaboutwhichthereiscommunity-wideagreementunderanygivencircumstances.Alittlemoreprecisely,wecandistinguishtwoconditionswhichanoccasionsentencemustsatisfytocountasanobservationsentence.First,whetheragivenspeakerofthelanguageisdisposedtoassentto,ordissentfrom,anobservationsentenceatagiventimeissimplyamatterofthestimulationsthatthatindividualisundergoingatthattime.Foreachindividual,thesamestimulation-patternwilltypicallyleadtothesameverdicteachtime.Second,anyfullycompetentspeakersofthesamelanguage,inthesamecircumstances,willagreeonanobservationsentence;Quinespeaksof“unhesitatingconcurrencebyallqualifiedwitnesses”(1995:44).(Thereissomevaguenesshere,arisingfromthevague-nessof“inthesamecircumstances,”andof“qualifiedwitness.”)Observationsentencesthusassertthepresenceofsomethingreadilydetectablebythesenses.Whethersuchasentenceiscorrectlyassertableinagivensituationdoesnotdependuponancillaryinformation,unlessitissharedbyallspeakersofthelanguage.Hence“It’scoldhere!”mightqualify,but“That’sQuine!”wouldnot,sincenotallEnglish-speakerswouldrecognizethatphilosopheronsight.Therangeofobservationsentenceswillvarywithourdecisionastoexactlywhoshouldbeincludedamongthefully-functioningspeakersofthelanguage.“That’sred”willpresumablycountifweexcludetheblindandthecolor-blind,butotherwisenot.Observationsentencesaretheepistemologicallymostbasicpartsofourtheoryoftheworld.Theycanbeknownbefore192\nW.V.QUINEanythingelse.(Theythusplayoneoftherolestraditionallyaccordedtothenotionofevidence:theyarefirstintheorderofknowledge.)Theknowledgethattheyembodyissorudimentarythatinalmostallcasesitgoesunspoken,butcanbeelicitedbyraisingthesentenceasaquestionandnotingthesubject’sreaction.Consideredholophrastically,i.e.asunanalyzedwholes,observationsentencesaresimplyresponsestostimulation,andareinonlythemostminimalsenseconceptualortheoretical.Hencesuchsentencescanbemasteredbyachildotherwisequiteinnocentoflanguage.AsQuinesays,“Theirdirectassociationwithcurrentstimulationisessen-tialifthechildistoacquirethemwithoutpriorlanguage”(1990:5).Consideredasmadeupofparts,however,theyconnectwithsophisticatedtheory,forthewordswhichmakethemuprecurinmoretheoreticalcontexts.Thisdualaspectisessentialtothefunctionofobservationsentencesasthestartingpointoflanguageandconceptual-ization.Becausetheypresupposesolittle,observationsentenceswillbethefirstsen-tenceslearnedbyachild(or,indeed,byanadulttryingtofindhisorherunaidedwayinawhollystrangelinguisticcommunity);theirlearningpresupposesnopriorcon-ceptualortheoreticalresources.Becausetheirtermsrecurinhighertheory,learningsuchasentenceisastartonlearningthelanguageasawhole;itisonlythissharingofvocabularywhichunitestheobservationsentenceswiththerestofthelanguage(1990:8).Observationsentences,wesaw,areoccasionsentences:thetruth-valueofsuchasentencewillvaryfromoneoccasionofutterancetoanother.Ourscientifictheories,however,andmostotherseriousknowledge,consistsofstandingsentences:sentencestrueorfalseonce-for-all.Howisthisgapinturntobebridged?Quine’sanswerappealstothenotionofanobservationcategorical.Thisisasentencecompoundedoftwoobser-vationsentences,sayingthatwheneveroneofthemholdstheotherwillalsohold:“Wheneverthereissmoketherewillbefire,”forexample.Thisisastandingsentence,andsomightbeimpliedbyaseriousbranchoforganizedknowledge.Ontheotherhand,bothofitscomponentpartsareobservational(orsowearesupposing).Sowecantellrightoffwhetheragivensituationisoneinwhichthereissmoke,andwhetheritisoneinwhichthereisfire.Hencewecantellrightoffwhetheragivensituationisoneinwhichtheobservationcategoricalisfalsified.Becausethesentenceisineffectageneralizationoverallsituations,itcannot,ofcourse,beverifiedbyasinglesituation,butit–andhencethetheorywhichimpliesit–canbefalsified.Quinereadilyacceptsthisasymmetrybetweenverificationandfalsification,whichfitswithhisgeneralfallibilism.Sinceourtheorieshaveinfinitelymanyobservationalconsequences,theycannotbeconclusivelyverified;inprinciple,however,asingleobservationmayfalsifyatheory.NaturalizedepistemologyandnormativityWehavebeenarticulatingQuine’sgeneralconceptionofknowledge,evidence,andtherelationbetweenthem.Thisconceptionisthoroughlynaturalistic:Quinetreatsknowl-edgeasanaturalphenomenon,tobestudiedbytheproceduresofscience.MostofQuine’sownworkinepistemologyisanarticulationanddefenseofthisverygeneralconception.Hesuggests,however,thatthereisalsoroominepistemologyfordetailedpiecemealworkofamorerecognizablyscientifickind.Thiswouldconsistintracingthe193\nPETERHYLTONconnectionsbetweentheoryandevidenceinapsychologicallyrealisticfashion,toseehowourknowledgeisinfactrelatedtotheevidencethatwehave.Epistemologyofthissortisthusabranchofpsychology.(Ofcoursepsychologyisitselfamongtheitemsofknowledgewhoserelationtoevidenceistobeinvestigatedinthisway;Quinespeakshereof“reciprocalcontainment”(1969:83).)Epistemology,astraditionallythoughtof,isanormativesubject:itaimstotellusnotmerelyaboutwhatisbutalsoaboutwhatoughttobe;itaimstotellusnotonlywhatwedoinfactbelieve,andonwhatevidence,butalsowhichbeliefsarejustifiedonthebasisoftheevidencethatwehave.Quineanepistemology,atleastaccordingtohisaccountofthematter,isdescriptive.Towhatextent,ifany,canthisdescriptivesubjecttakeontheburdenoftraditionalnormativeepistemology?Thisisalargeandcomplexquestion(seeGregory1999,towhichIamindebtedhere).RoughlywemaysaythatQuinehasnoroomfortheverylarge-scalequestionsanddoubtswhichareonekindofstartingpointfortraditionalepistemology.Hehasnosympathyatall,forexample,withglobalskepticism.Theaimofourknowledgeistopredictsense-experience.Atheorythatdoesthatsatisfactorilydoesallthatwecanask.Thereisnofurtherquestionastowhetherittellsusaboutthenatureofreality:whatif...wehaveachievedatheorythatisconformabletoeverypossibleobservation,pastandfuture?Inwhatsensecouldtheworldthenbesaidtodeviatefromwhatthetheoryclaims?Clearlyinnone.(Quine1981:22)Wecannotdivorcetheideaofrealityfromthatsense-experience.(Thispointwillemergefurtherinthenextsection,below.)Thereis,nevertheless,asenseinwhichQuineanepistemologyisnormative.Thecriterionofsuccessforallputativeknowledge,forscienceinQuine’sbroadsense,isthepredictionofsense-experience.Quineseesthisasdefiningthenotionofscience(1990:20).Thisdefinitionisnotarbitrary:ourprimaryaim,inscience,istofindoutabouttheworld,andonethingweknow–awell-establishedpieceofscientificknowledge–isthatitisonlythroughsense-experiencethatwecometoknowabouttheworld.Giventhatourgoalisfixed,therearequestionsofanormativesortaboutthebestwaysinwhichtoachievethataim,andinthisinstrumentalsenseepistemologyisnormative:Naturalizationofepistemologydoesnotjettisonthenormativeandsettlefortheindis-criminatedescriptionofongoingprocedures.Formenormativeepistemologyisabranchofengineering....Thereisnoquestionhereofultimatevalue,asinmorals;itisaques-tionofefficacyforanulteriorend,truthorprediction.Thenormativehere,aselsewhereinengineering,becomesdescriptivewhentheterminalparameterisexpressed.(HahnandSchilpp1998:664–5)RealismAswehaveseen,Quine’snaturalismcanbeidentifiedwiththeviewthatthereisessentiallyonlyonekindofknowledge.Inparticular,thereisnospecialphilosophicalperspectivefromwhichwecanattainknowledgethatisindependentofour194\nW.V.QUINEordinaryscientificorcommonsensicaltheoryoftheworld.Thusonhisaccountwearealwaysinsidethattheory,modifyingit,perhaps,butnotwhollytranscendingit.Thereisnotranscendentalstandpointthatisindependentofourordinaryknowledge,andfromwhichwecanevaluatethatknowledgewithoutpresupposingit.Quine’srealismisanimportantapplication,andillustration,ofthisview.Quine’sworkisfullofremarkswhichmightsuggestthathedoesnottakeourtheories–includingthe“theory”thatiscommonsenseknowledge–tobe(really)true,anddoesnottaketheobjectsthatthosetheoriespresupposetobe(really)real.Hesaysthatourtheoriesfaroutruntheevidencethatwehaveforthem,andthatmorethanonetheoryiscompatiblewiththatevidence.Thedisparitybetweenourevidenceandourknowledgeisarecurrentthemeinhiswork.Hisinsistenceonthisdisparity,togetherwithhisviewthatourknowledgeisjustifiedbyitsefficacyinpredictingandunderstandingthecourseofexperience,mightleadonetosupposethathisisaninstrumentalistorpragmatistview:thatourtheories,evenattheirbest,arenotreallytrue,thattheydonotaimtocorrespondtoanextra-theoreticalworld,butaresimplyusefulinstrumentsforpredicting,understanding,andcontrollingfutureexperience.SomecriticshavetakenthistobeQuine’sview,andhaveseenhisinsistenceonrealismasacontradiction,orasamis-statementofhisactualposition(see,forexample,Lee1986,andSmart1969).Quine,however,insiststhathisviewis“robustrealism”(1990:21),andthatappearancestothecontrarycanbedispelledbytakingnaturalismseriouslyenough.OnQuine’sview,theobjectsthatourtheoriespresupposedoexistinextra-theoreticreality.Itispartofourtheory–thatis,partofthebestunderstandingoftheworldwehave–thatthoseobjects(withafewexceptions,mostofthemstraightforward)arenotdependentonusorourtheorizing.Nowthecriticmayprotest:itmaybepartofourtheorythatourtheoreticalobjectsreallyexist,independentofourtheory–butthat’sjustpartofthetheory.Dotheobjectsreallyexist?Butthisisanattempttoaskaques-tionfromastanceindependentofourtheorizingabouttheworld:thepointofQuine’snaturalismisthatthereisnosuchstance.Theobjectsthatwebelieveinexist,andarereal,intheonlysenseofthoseideasthatweactuallyhave.Moregenerally,theapparentlyskepticalremarksthatQuinemakeswhendiscussingouracquisitionofknowledgedonotaffecthisbeliefinthetruthoftheknowledgethatisthusacquired:hisontologyisinthisrespectinsulatedfromhisepistemology.ThisisaconsequenceofQuine’sversionofnaturalism,andinparticularofthereciprocalcon-tainmentofscience,withitsontologicalclaims,withinepistemology,andviceversa.Letusseehowthisgoes.Weaccept,letussuppose,thebestoveralltheoryoftheworldthatisavailable.Freelydrawingonthistheorywedoepistemology,i.e.weinvestigatethewayinwhichhumanbeings–includingourselves–cometoformulatetheoriesandtoposittheexistenceofobjects;amongthetheoriesthusinvestigatedarethosethatwearedrawingoninthecourseofourinvestigation.Nowthecrucialpointisthattocallabodyofknowledgea“theory,”ortocallanobjecta“posit,”doesnotintheleastimpugnitstruthoritsreality.“Theory”hereisnotmeretheory,contrastedwithrealknowledge,foranybodyofknowledgewillcountasatheoryfromthepointofviewofepistemology;noris“positedobject”contrastedwithrealobject.Thus,asQuinefamouslysays:195\nPETERHYLTONTocallapositapositisnottopatronizeit...Everythingtowhichweconcedeexistenceisapositfromthestandpointofadescriptionofthetheory-buildingprocess,andsimul-taneouslyrealfromthestandpointofthetheorythatisbeingbuilt.Norletuslookdownonthestandpointofthetheoryasmake-believe;forwecanneverdobetterthanoccupythestand-pointofsometheoryorother,thebestwecanmusteratthetime.(1960:22;myemphasis).Thecrucialpointistheoneemphasized:thereisnoalternativetooccupyingsomesub-stantivetheoryoftheworld,andtodothismeansacceptingthattheory,atleastforthemoment,astrue,andacceptingitsobjectsasreal.Ofcoursewemaydevelopourtheoryintoadifferentone,butwecannotoccupysomeneutralphilosophicalvantagepoint;norcanweacceptatheorywhilestillpretendingthatwearenotacceptingitastrue.InQuine’suse,then,theoreticalknowledgeisnotcontrastedwithordinaryknow-ledge.Similarly,theoryforQuineisnotcontrastedwithfact.Allknowledgeistheoreti-cal,inQuine’ssense.Justasthereishereastretching(oradistortion)oftheword“theory,”soalsothereisastretchingoftheword“posit.”Whenwesayof,neutrinos,forexample,thattheyareposits,wewouldgenerallybetakentomeanthatsomepersonorgroupofpeopleconsciouslypositedthem.Quine,however,speaksofphysicalobjectsingeneralasposits,andherethereisnosuchimplication.Noconsciousdecisionwasevertakentopositsuchthings.Asinthecaseof“theory,”Quine’susethusassimilatesideaswhichonemightsupposetobeimportantlydifferent.Metaphysicsandregimentation:logicandextensionalityTothispointwehaveseenQuinereflectingonthenatureofourknowledge,itssourcesandbases,andonitsstatus.Thephilosopher’stask,asheconceivesit,alsoincludesclarifyingourknowledgeandhelpingustoattainaclearviewofjustwhatitcomestoandwhatitreallycommitsusto.ThislatterkindoftaskmaybethoughtofastheQuineanversionoranalogueofmetaphysics.Liketheothers,itaims,inQuine’sview,tocontributetotheoverallscientificenterprise.Itshould,forexample,enableustoavoiduselessormisleadingquestions,helptosuggestfruitfullinesoffurtherinquiry,andexposepotentialproblemsthatmaylurkinscientifictheories.Howisthistasktobeapproached?Themethodhereistoshowhowvariouspartsofourknowledgecouldbereformulatedintheclearestpossibleterms.Wehavereasontotaketheobjectsandcategoriesrevealedbythisreformulationasrealandfundamental.AnextendedpassagefromWordandObjectisworthquotingatlengthonthistopic:Thesamemotivesthatimpelscientiststoseekeversimplerandclearertheoriesadequatetothesubjectmatteroftheirspecialsciencesaremotivesforsimplificationandclarifica-tionofthebroaderframeworksharedbyallthesciences.Heretheobjectiveiscalledphilo-sophical,becauseofthebreadthoftheframeworkconcerned;butthemotivationisthesame.Thequestofasimplest,clearestoverallpatternofcanonicalnotationisnottobedistinguishedfromaquestofultimatecategories,alimningofthemostgeneraltraitsofreality.Norletitberetortedthatsuchconstructionsareconventionalaffairsnotdictatedbyreality;formaynotthesamebesaidofaphysicaltheory.True,suchisthenatureofrealitythatonephysicaltheorywillgetusaroundbetterthananother;butsimilarlyforcanonicalnotation.(1960:161)196\nW.V.QUINEThemesthatweexaminedinthefirstpartofthisessay,havingtodoespeciallywiththerejectionofthePrincipleofToleranceandwiththeremotenessofsomepartsofourtheoryfromexperience,re-emergehere.Simplicity,clarity,andconveniencearenotmerely“pragmatic”virtueswhicharetobedistinguishedfromthecognitiveortheo-reticalvirtueoftruth.Ifwehavealanguagewhichdisplaysthosevirtuestothehighestextent,thenwehavereasontotakethestructureofthatlanguageastellingussome-thingabouttherealworld,justaswetaketheorieswhichdisplaythosevirtuestotellussomethingabouttheworld.Oneaspectofthephilosopher’stask,then,istofindtheclearestandsimplestframe-workinwhichtoformulateourknowledge.InQuine’sview,thisclarityandsimplicityistobefoundinfirst-orderlogic.(By“first-order”logicismeantlogicoftruth-functionsandquantifierswhichbindvariablesinpositionsoccupiedbysingularterms,butnotpositionsoccupiedbypredicates.)Hetakesitthatourknowledgeisatitsclearestwhenitisformulatedinthesyntaxoffirst-orderlogic:asyntaxwhichusesonlytruth-functions,predicates,variables,andquantifiers.Thisisnottosaythatheadvocateslan-guagereform,orthathethinksthatweshouldinfactreformulateallofourknowledgeinthoseterms.Butwhereourconcerniswithgettingclearaboutwhatsomepartofourknowledgereallycommitsusto,wewoulddowelltoconsiderhowitmightbephrasedinlogicalsyntax.Thissyntaxisextraordinarilytransparentandeconomical,whichmakestheorizingaboutitsimple,andyethassurprisingexpressivepower.Itcon-tainsneitherpropernamesnorfunction-symbolsamongitsprimitiveexpressions,forexample,yettheeffectofeachcanbeeasilyachieved.(ThebasictechniquesofdoingsoderivefromRussell’sTheoryofDescriptions(seeRUSSELL).)Onemoraltobedrawnhereisthatwemaybeabletoachievetheeffectofaparticularconstructionwithoutinfacthavingtoaugmentourstockofprimitives.Quineismuchconcernedtotakeadvantageofthiskindofeconomywhereveritispossible,forthesakeoftheclarityandsimplicityoftheoveralltheory.(Inontology,asweshallsee,Quinehasasimilarconcernwitheconomy:toshowhowparticularkindsofobjectwhichappeartobeassumedinwhatwetakeourselvestoknowneednotinfactbetakenforgrantedasprimitive,becausetheireffectscanbeduplicatedbyothermeans.)Onefeatureofthisframeworkwhichhasprovedextremelycontroversialisitsexten-sionality.Alanguageisextensionalwhenanysentenceofitretainsitstruth-valueunderanyone(ormore)ofthreekindsofchanges:(1)anynameinitmaybereplacedbyanyothernameofthesameobject;(2)anypredicateinitmaybereplacedbyaco-extensivepredicate(i.e.onetrueandfalseofexactlythesameobjects);(3)anysentenceembeddedinanothersentencemaybereplacedbyasentenceofthesametruth-value.Itisimportanttonotethatthesethreerequirementswhichalanguagemustsatisfytocountasextensionalinterlockinwaysthatmakeitveryhardforareasonablycom-prehensivelanguagetosatisfyanyoneofthemwithoutsatisfyingallthree.Thispointisimportantbecausethefirstrequirementis,onthefaceofit,farmoreplausiblethanthethird.Itisextremelyplausiblethatifwereallyunderstandwhatagivensentenceisabout,thenwecouldreplacethenameofthatobjectinthesentencewithanothernameofthesameobjectwithoutalteringthetruth-valueoftheoriginal.Ifasentenceisgenuinelyaboutanobject,howwenamethatobjectshouldbeamatterofindiffer-encetothetruthorfalsehoodofthesentence,thoughitmayaffect,forexample,itspoeticquality.(Onemight,indeed,takethisideaasapartialdefinitionofthesomewhat197\nPETERHYLTONvagueideaofaboutness.)Nosuchsuperficialplausibilityattachestothethirdrequire-ment,yetitturnsouttobeveryhardtoseehowonecanacceptthefirstwithoutalsoacceptingthethird.(See,forexample,Quine1995:91–2.)Anylanguagewhichusesthesyntaxoffirst-orderlogic(alongwithanystandardsemantics)is,invirtueofthatfact,anextensionallanguage.YetitwouldbeamistaketothinkQuineacceptsextensionalitysimplyinordertobeabletousethesyntaxoflogic.Onthecontrary:hetakesoneoftheadvantagesofthatsyntaxtobethatitenforcesextensionality,whichheholdstobedesirableforitsownsake:Ifindextensionalitynecessary...thoughnotsufficient,formyfullunderstandingofatheory.Inparticular,itisanaffronttocommonsensetoseeatruesentencegofalsewhenasingularterminitissupplantedbyanotherthatnamesthesamething.Whatistrueofathingistrueofit,surely,underanyname.(1995:90–1)Quine’sinsistenceonextensionalityisaverylong-runningthemeofhisthought,goingbackasfarastheclarificationofPrincipiaMathematica(WhiteheadandRussell1910–13)inhisdoctoraldissertation.(Seethefirstparagraphofthisessay;alsoQuine1991:265–6.)Therequirementofextensionalityhasbeencontroversialbecausetherearelargeareasofdiscoursewhich,atleastiftakenatfacevalue,arenotextensional.ThisforQuineisreasonenoughnottotakesuchdiscourseatfacevalue.Instead,hethinks,weshouldeithertrytoreformulateitsothatitbecomesextensionalorelseexcludeitfromthemorescientificandrespectablepartsofourknowledge,thosepartswhichwetakeasreallytellingusabouttheobjectiveworld.Weshalldiscussthreeexamples;ourdis-cussionofeachwillbeverybrief,althoughthesecondandthirdofthemareissueswhichhavegeneratedmuchcontroversy,andonwhichQuinehas,largelyforthatreason,writtenextensively.Thefirstsortofprima-facieviolationofextensionalityisalmosttrivial;indeeditmightbesaidthattheideathatwereallyhavesuchaviolationhereissimplyamistake.Thisisthecaseofquotation,whichisworthexaminingbecauseitfunctionsassome-thingofaparadigmforQuine.Itistruetosay:“‘Quine’hasonesyllable”;itisfalsetosay“‘TheauthorofWordandObject’hasonesyllable”;yettheonesentencemightseemtobeobtainedfromtheotherbyreplacingasingulartermwithanothersingulartermdesignatingthesameobject,sinceQuine,ofcourse,istheauthorofWordandObject.Herethesolutiontotheapparentpuzzleiseasy.Weshouldnotconstruethesubjectsofthesentencesasreferringtoaperson,butrathertothewords,theexpressionsthem-selves.Itisoftheword“Quine,”notofthephilosopherQuine,thatwesayithasonesyllable.Andthenofcoursethesubstitutionnolongerreplacesanamebyanothernameforthesameobject,sincethetwoexpressionsarenotthesame.Thesecondsortofexampleisfarlesseasytodismiss.Thecaseofindirectdiscourse,whereonereportsthespeechorthethoughtsofanother,alsogivesrisetoprima-faciecasesofnon-extensionality.Othello(supposinghimforthemomenttobearealperson)hasthefalsebeliefthatDesdemonalovesCassio,butdoesnot(presumably)havetheequallyfalsebeliefthatthemoonismadeofgreencheese.Clearly,instatementsoftheform“Abelievesthatp”itisnotonlythetruth-valueofpthatisrelevanttothetruth-valueofthewhole.Similarlyinthecaseofsingularterms.Touseanexampleof198\nW.V.QUINEQuine’s,TommaybelievethatCicerodenouncedCatilinewithoutbelievingthatTullydenouncedCatiline,forhemaynotknowthatCiceroisTully.Oneresponsetosuchcasesistosaythatwhatisbelievedordisbelievedisnotafactaboutaperson(orotherobject)butratheraproposition.(Forthisreason,philosophersoftenspeakofsuchcasesasstatementsof“propositionalattitudes.”)Thenitisclaimedthatwehavetwodistinctpropositions:first,thatCicerodenouncedCatiline,and,second,thatTullydenouncedCatiline.Soconstrued,belief-contextsbecomeextensional,forwecannolongerobtainafalsehoodfromatruthbysubstitutingaco-designativeexpression(sincethisnowmeansanexpressionreferringtothesameproposition).Quinerejectsthisidea,chieflyonthegroundsthatnoclearandpreciseidentity-conditionshavebeengivenforpropositions,whichshouldthereforenotbeacceptedforscientificandphilosophicalpurposes.Thisisofapiecewithhishavingsomedegreeofskepticismaboutmeaning,foronecanthinkofapropositionasbeingthemeaningofadeclarativesentence.(Inthatcasethequestionoftheidentity-conditionsofpropositionsisthesameastheissueofsynonymyfordeclarativesentences.)Quine’streatmentofindirectdiscoursereliesnotonthemeaningsofsentencesbutsimplyonthesentencesthemselves.Formally,heassimilatesthemtocasesofquota-tion,byconstruingbelief(anddoubt,andhope,andsoon)asattitudestowardssen-tences.Perhapsthereissomethingoddorcounterintuitiveaboutspeakingofbelievingornotbelievingasentence,butthiskindofoddityissomethingthatQuineis,aswehaveseen,fullypreparedtoacceptintheinterestsofclarity(asheconceivesit).Andifwedotalkthisway,thenthereisnothingpuzzlinginthinkingthatTom,toreverttoourexample,maybelievetheonesentencewithoutbelievingtheother,fortheyaredis-tinctobjects.Twopointscallforcomment.First,Quinewasforaperiodconvincedthattherearetwokindsofbelief.Thereistheordinarykind,variouslyconstruedasanattitudetowardsasentenceoraproposition.Butthenthereisalso,hethought,anotherkind,derebelief,whichisbeliefgenuinelyaboutanobject,sothatthewayinwhichtheobjectisdescribedisirrelevanttothetruthofthebelief-attribution.Hespentmuchefforttryingtomakesenseofcaseswhichappeartobeofthissort,butsubsequentlycametothink(quitecorrectly,inmyview)thattheappearanceoftwodistinctkindsofbeliefismistaken,andthatthephenomenathathewasattemptingtounderstandareinfactnotreallycasesofadifferentkindofbelief.Second,itwouldbeamistaketothinkofthedifficultiesofmakingclearsenseofbeliefcontextsasmerelyformal.ItisQuine’sview,afterall,thatwewantanextensionallanguagenotforitsownsakealone,butbecauselackofextensionalityisasignoflackofclarity.Hisre-construalofbeliefasanattitudetowardssentencesputstheemphasisinwhathethinks,forindependentreasons,intherightplace.Ifweareconstruingevidenceasausterelyandstrictlyaspossible,theevidenceforabelief-ascriptionissimplyareportofwhatthepersoncon-cernedsaid–theverywordsuttered–anddid.Inreportsofthesesortsthereisnovio-lationofextensionality.Beyondthat,belief-ascriptionreliesonempatheticprojection:theascriberimagineswhatitwouldbeliketobeinthebeliever’ssituation.Here,itmaybethought,weareoutbeyondtherealmofhardfact.Athirdareaofdiscourseinwhichextensionalityfails,atleastonQuine’sview,isthatconcerningmodality.Supposeitsaidthatnineisnecessarilygreaterthanfive(theexample,again,isQuine’s).Replacing“nine”byaco-designativeexpressionwecan199\nPETERHYLTONobtain:thenumberoftheplanetsisnecessarilygreaterthanfive.Quine,however,seesnohopeofmakingsenseofanynotionofnecessityaccordingtowhichthisistrue(cf.MARCUS).Hedistinguishesanotionofnecessityofthissortfromonewhichattachestosentences(andhencecanbeassimilatedtothecaseofquotation)ortreatsitasastatementoperator.Inthesecasesthenecessityissupposedtoholdofthesentence,notoftheobject,sothereisnoriskofthesortofviolationofextensionalityindicatedabove.Thegeneralissueofmodality,however,isnotacrucialonefromaQuineanpointofview,forheseesnoneedtoacceptanynotionsofnecessityorpossibility.Heholdsthatareconstructionofourknowledge,orofthoseaspectsofitwhichwewanttotakewithfullseriousness,astellingusabouttheobjectiveworld,requirenosuchnotions.Wehavebrieflydiscussedthreecasesofprima-facienon-extensionality.Itisworthemphasizing,however,thatthesethreedonotexhaustthematter.Counterfactualcon-ditionals,statementsofcausality,and,iftakenatfacevalue,statementsoftimeandtense,areamongothersuchcases.OntologyanditsrelativityAcrucialpartofthetaskofgettingclearaboutwhatourtheoriescommitusto,onQuine’saccount,isgaugingtheirontologicalcommitments.Bytheontologicalcommit-mentsofatheoryQuinemeanswhatentitiesthattheorysaysthereareintheworld.InQuine’sview,thewaytosettlethisisbyseeingwhatobjectsmustbeintherangeofthetheory’svariablesifitistobetrue.Quine’semphasisonontology,andontherangeofvariablesasthemeasureofontologicalcommitment,isperhapsinparttobeexplainedbyhisearlyworkinset-theory,andhisabidinginterestinthatsubject.Inthatcontext,therangeofthevariablesofthetheoryisanaturalmeasureofitsstrength,andthethreatofparadoxmakesthismatterofvitalconcern.(TheimportanceofthispointwasemphasizedtomebyStephenMenn,towhomIamgrateful.)Quinetakesontologytobeaproductofself-consciousscientificandphilosophicalreflection.Hedoesnotseethephilosopher’sontologicaltaskasthatofcapturinginper-spicuousformtheontologyimplicitinordinarythoughtanddiscourse,forheinsiststhatthereisnosuch“ordinaryontology”:afencedontologyisjustnotimplicitinordinarylanguage.Theideaofaboundarybetweenbeingandnon-beingisaphilosophicalidea,anideaoftechnicalscienceinabroadsense.Scientistsandphilosophersseekacomprehensivesystemoftheworld,andonethatisorientedtoreferenceevenmoresquarelyandutterlythanordinarylanguage.Ontologicalconcernisnotacorrectionofalaythoughtandpractice;itisforeigntothelayculture,thoughanoutgrowthofit.(1981:9)Whenweareconcernedwithontologicalquestions,ormetaphysicalquestionsmoregenerally,wecannotsimplyexamineourbeliefsinthetermsinwhichweareatfirstpronetoexpressthem.Ourbeliefsmust,rather,becastintoastandardnotation,whichwilllettheirpresuppositionsshineforth.Ontology,asQuineinterpretsit,thuspresup-posesregimentation:itisonlyinsofarasweconceiveofourknowledgeascastinregimentednotationthatitmakessensetoraiseontologicalquestions.200\nW.V.QUINETheartificialityofontology,inQuine’sview,isanimportantpoint.Heisoftencriti-cizedfordistortingordinarythought,ordoingviolencetooursupposed“intuitions.”Onhisviewthesecriticismsmisstheirmarkentirely.Morepositively,hetakespartofthetaskofontologytobethatofshowingjustwhatourtheoriesreallycommitusto.Herehehasaninterestinontologicaleconomy.Inmanycasesourtheoriesseemtocommitustoacceptingentitiesofacertainkind,butartfulre-construalofthetheo-riesshowsthatinfactweneednotpresupposeentitiesofthegivensort.Hereartifi-cialityisinevitable.Inthisspirit,Quinetakesthereductionofnumberstosets,andoforderedpairstosets,tobeclearphilosophicalachievements.(Section53ofQuine1960isentitled“TheOrderedPairasPhilosophicalParadigm.”)Ineachcasethereductionshowshowourdiscoursecouldberephrasedsoastoavoidcommitmenttoakindofentitytowhichitappearstobecommitted.Therephrasingisnotmeantasapracticalsubstitutefornormalarithmeticalorset-theoreticlanguage.Norisitclaimedthattheparaphrasedversiongetsatthe“realmeaning”orhiddenstructureoftheoriginal.Itissimplythatthereductionshowsushowwecoulddoeverythingthatweneed,forscientificpurposes,withoutpresupposingthatthereareorderedpairs(ornumbers),andhencethatweneednottakeourselvestobecommittedtotheexistenceofsuchthings.Norisitonlytowardssuchtechnicalsubjectsasarithmeticandset-theorythatQuinetakesthisattitude.Onthecontrary;hetakesthesamevieweverywhere.Heholds,forexample,thatwecaneliminatemindsintermsofbodies,justaswecaneliminatenumbersintermsofsets.Insteadofspeakingofaperson’smindatagivenmoment,wecouldinsteadspeakofhisorherbodyatthatmoment.Mentalisticpredicates,suchas“isthinkingofVienna”persist,butnoexclusivelymentalobjects.Themaneuveristrivial,butnot,inQuine’sview,anytheworseforthat;hetakesitasshowingthatwehavenoontologicalcommitmenttominds,overandabovebodies.Quinethusstressestheartificialityofontology,andtheuseofparaphraseorre-construaltoeliminateapparentontologicalcommitments.Histacticsherepresup-posesomethingimplicitinourearlierdiscussionofknowledge,especiallyofobser-vationsentences.Inhisviewlanguageis,ofcourse,referential–itisaboutthings.Werefertopeopleandotherobjectsinalmosteverythingwesay.Buttherelationofreferenceisnotourfundamentalcognitiverelationtotheworld.Thefundamentalrelation,thewaythatlanguagegetstobeabouttheworldatall,istherelationofobservationsentencestopatternsofstimulation.(Inparticular,therelationofagivenobservationsentencetothepatternsofstimulationonthebasisofwhichspeakersofthelanguageareinclinedtoacceptorrejectthegivensentence.)Thisisnotarelationofreference,forobservationsentencesarenotaboutpatternsofstimulation.Referenceisaderivativerelation.Indeedtheverynotionofanobject,thattowhichwerefer,isinQuine’sviewderivative–thoughitcomessonaturallytousthatitseemsinevitable.Iftherewerealanguageconsistingonlyofobservationsentences,therewouldbenoreasontoattributeanyreferencetothatlanguage.Thevastlygreatercomplexityofourlanguagerequiresinferentiallinksbetweensentences,linkswhichwecangrasponlybyattributingstructuretothesentences.Seeingsentencesasdividedintoterms,someofwhichrefer,ispartofthisprocess:201\nPETERHYLTONReferenceandontologyrecedethustothestatusofmereauxiliaries.Truesentences,observationalandtheoretical,arethealphaandtheomegaofthescientificenterprise.Theyarerelatedbystructure,andobjectsfigureasmerenodesofthestructure.(Quine1990:31)ThederivativecharacterofontologymanifestsitselfmostdramaticallyintheQuineandoctrineknownasontologicalrelativity,ortheinscrutabilityofreference.Quine’sclaimisthatitwouldbepossibletocarryoutalarge-scalere-construalofourknowledge,replacingeveryreferringtermbyanother.Wecould,forexample,replaceeverytermreferringtoaphysicalobjectbyatermreferringtoitsspace-timecomple-ment,thatis,tothewholeofspaceandtimeotherthanthegivenobject.Alongwiththatwewouldre-construeourpredicates,sothatapredicatetrueofagivenobjectwouldnowbetakenastrueofitsspace-timecomplement.Thesere-construalswouldcancelout,leavingthetruth-valueofeachsentencethesameasbefore.Theresultwouldbeasystemofknowledgeexactlylikeourowninitsstructure,includingitsrela-tiontoevidence,butinwhicheachtermreferstoobjectsdifferentfromthosetowhichitsunreconstructedanaloguerefers.Nothing,Quineclaims,preventssuchare-construal.Thepassagequotedinthepreviousparagraphcontinues:“Whatparticularobjectstheremaybeisindifferenttothetruthofobservationsentences,indifferenttothesupporttheylendtotheoreticalsentences,indifferenttothesuccessofthetheoryinitspredictions”(Quine1990:31).Quineclaimsthatthissortofwholesalere-construalofourknowledgeispossible,thatitwouldnotbeinconsistentwithanypartofit.True,Icanandwouldstoutlymain-tainthatmyword“rabbit,”say,referstorabbitsandnottotheirspace-timecomple-ments.Butallmyusesoftheword,includingthoseusesinwhichIinsistthatitisrabbitsthatIamtalkingabout,areopentore-construal.Fromthepossibilityofthissortofre-construal,Quinesometimesinfersthatthereisnofactofthematteraboutreference–nofactofthematterastowhethermyword“rabbit,”sayreallyreferstorabbits,orrathertothespace-timecomplementofalltherabbits.Thethesisofontologicalrela-tivity,however,isperhapsbestthoughtofasarejectionofthisideaof“reallyrefers,”insofarasitoutrunstheordinaryideaofreference.Takingourownlanguageforgrantedwecansaytowhatthewordsofanotherlanguage(or,trivially,ofthatsamelanguage)refer;whilewestaywithinthatlanguageweuseittoreferwithnomorethantheusualdifficultiesorambiguities.Ontologicalrelativityis,Quinesays,“unproblem-aticbuttrivial”(HahnandSchilpp1998:460).Onlywhenwetranslate,ormapourlanguageontoitself,doestheontologicalrelativityemerge.Andhereitshouldperhapsbeseenasareminderofthederivativestatusofreference.Quinehascometogiveincreasingemphasistothedoctrineofontologicalrelativity.Shouldweseeitasconflictingwith,orundermining,hisinsistenceonrealism?Sinceitleavestruth-valuesunaffected,itdoesnotaffecthisclaimstobearealistabouttruth.Thatlatterclaim,aswesaw,wasalsoonethatmightbedoubted.Quine’sclaimthereisperhapsthatheisasmuchofarealistasitmakessensetobe:thattheresimplyisnocoherentsenseofrealismstrongerthanhis.Similarlyhere,perhaps.Quineclaimsthatontologicalrelativityisbeyonddoubt;it“admitsoftrivialproof,”hesays(HahnandSchilpp1998:728).Sothereisinhisviewnochanceofdefendingaversionofrealismthatdeniesit.Ontologicalrelativitymightthusbetakenasshowinguswhatrealism202\nW.V.QUINEcancometo.Ontologysimplyisderivativeupontruth,andgivenanysystemoftruthstheresimplywillbemorethanonewayofconstruingitsontology.Ifthisunderminesourpreviousconceptionofrealismthen,fromQuine’spointofview,somuchtheworseforthatconception.ConclusionIshallnotattempttosummarizewhatisalreadyaverycompressedtreatmentofQuine’sthought.Itisperhapsworthsaying,however,thatIhaveignored,ortreatedverybriefly,anumberofissuesthathaveoccupiedconsiderablespace,bothinQuine’sownwritingsandintheworksofcommentatorsandcritics.Notableexampleshereincludetheindeterminacyoftranslation,Quine’sviewsofindirectdiscourseandofmodality,thequestionofhisphysicalism,andwhatitamountsto,andhissubstantiveviewsonontology.MydecisionsastohowtousethelimitedspaceavailablehereareofcoursebasedonmyviewofwhatismostimportantinQuine’sthought.Butthereader1shouldperhapsknowthatothersmighthavemadethesedecisionsratherdifferently.Note1FortheircommentsonearlierdraftsofthisessayIamindebtedtoBillHart,PeterHacker,GaryKemp,andAlMartinich.BibliographyWorksbyQuine1936:“TruthbyConvention,”inPhilosophicalEssaysforA.N.Whitehead,ed.O.H.Lee,NewYork:Longmans.(ReprintedinQuine,TheWaysofParadox,rev.edn.,Cambridge,MA:HarvardUniversityPress,1976.)1960:WordandObject,Cambridge,MA:MITPress.1961:FromaLogicalPointofView,2nd,rev.,edn.,Cambridge,MA:HarvardUniversityPress.(Firstpublished1953.)1963:“CarnapandLogicalTruth,”inPaulA.Schilpp(ed.)ThePhilosophyofRudolfCarnap,LaSalle,IL:OpenCourt.(ReprintedinQuine,TheWaysofParadox,1976.)1969:OntologicalRelativityandOtherEssays,NewYork:ColumbiaUniversityPress.1974:RootsofReference,LaSalle,IL:OpenCourt.1981:TheoriesandThings,Cambridge,MA:HarvardUniversityPress.1987:Quiddities:AnIntermittentlyPhilosophicalDictionary,Cambridge,MA:HarvardUniversityPress.1990:PursuitofTruth,Cambridge,MA:HarvardUniversityPress.1991:“TwoDogmasinRetrospect,”CanadianJournalofPhilosophy21,pp.265–74.1995:FromStimulustoScience,Cambridge,MA:HarvardUniversityPress.2000:“ResponsetoO’Grady,”TheProceedingsoftheWorldCongressofPhilosophy,Boston,1998,BowlingGreen,OH:PhilosophyDocumentationCenter.WorksbyotherauthorsCarnap,Rudolf(1934)DielogischeSyntaxderSprache;Vienna:SpringerVerlag.(Published,withadditions,asTheLogicalSyntaxofLanguage,trans.AmetheSmeaton,London:RoutledgeandKeganPaul,1937.)203\nPETERHYLTONCreath,Richard(ed.)(1990)DearCarnap,DearVan,BerkeleyandLosAngeles:UniversityofCaliforniaPress.Gibson,Roger(1982)ThePhilosophyofW.V.Quine,Tampa:UniversityPressesofSouthFlorida.Gregory,Paul(1999)“Quine’sNaturalEpistemology,”Ph.D.dissertation,UniversityofIllinois.Hahn,EdwinandSchilpp,P.A.(eds.)(1998)ThePhilosophyofW.V.Quine,expandededition,ChicagoandLaSalle,IL:OpenCourtPress.(Previouslypublished1986.)Hylton,Peter(1990–1)“Translation,Meaning,andSelf-Knowledge,”ProceedingsoftheAristotelianSociety91.Lee,HaroldN.(1998)“DiscourseandEvent:TheLogicianandReality,”inHahnandSchilpp,pp.295–315.Smart,J.J.C.(1969)“Quine’sPhilosophyofScience,”inWordsandObjections:EssaysontheWorkofW.V.Quine,ed.DonaldDavidsonandJaakkoHintikka,Dordrecht:Reidel,pp.3–13.Whitehead,AlfredNorthandRussell,Bertrand(1910–13)PrincipiaMathematica,3vols,Cambridge:CambridgeUniversityPress.2ndedn.1927.204\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200116A.J.Ayer(1910–1989)T.L.S.SPRIGGELanguage,TruthandLogicGeneralcharacterofthebookA.J.Ayerrosetoearlyphilosophicalfamewiththepublicationin1936,whenhewas25yearsold,ofwhatremainedhismostfamous,orinfamous,book,Language,TruthandLogic.TheworkishisownversionofthelogicalpositivismcharacteristicoftheViennaCircle(whosemeetingshehadattendedforthreemonthsin1932–3),hisoutlookbeingclosesttothatoftheirleader,MoritzSchlick.ThebookisalsostronglyinfluencedbytheBritishempiricisttradition,inparticularbyHumeandRussell.ItwassomethingofabombshelltoBritishphilosophersandbecameforthemtheparadigmstatementoflogicalpositivism,threateningtheoutlookofsome,providinganexcitingintellectualliberationforothers.Thebookopenswiththestrikingstatement:Thetraditionaldisputesofphilosophersare,forthemostpart,asunwarrantedastheyareunfruitful.Thesurestwaytoendthemistoestablishbeyondquestionwhatshouldbethepurposeandmethodofphilosophicalenquiry.(1946:33)Sofarasphilosophygoes,Ayer’sconcernistoshowthemeaninglessnessofmeta-physicaltheoriesaboutarealitybeyondtheempirical.Moregenerally,healsoclaimstoshowthatreligiousstatements,asusuallynowintended,aremeaningless,asalsoarestatementsoffundamentalethicalprinciple(exceptasmereexpressionsof1emotion).ToestablishthemeaninglessnessofallsuchstatementsAyerputsforwardtheveri-ficationprinciple.Accordingtothisastatementisfactuallysignificanttoanygivenperson,if,andonlyif,heknowshowtoverifythepropositionwhichitpurportstoexpress–thatis,ifheknowswhatobservationswouldleadhim,undercertainconditions,toacceptthepropositionasbeingtrue,orrejectitasbeingfalse.If,ontheotherhand,theputativepropositionisofsuchacharacterthattheassumptionofitstruth,orfalsehood,isconsistentwithanyassumptionwhatso-everconcerningthenatureofhisfutureexperience,then,asfarasheisconcerned,205\nT.L.S.SPRIGGEitis,ifnotatautology,amerepseudo-proposition.Thesentenceexpressingitmaybeemotionallysignificanttohim;butitisnotliterallysignificant.(1946:35)Thusameaningfulstatementmusteitherbeempiricallyverifiable,orbeatautology,thatis,analyticortruebydefinition.Thepassagejustquotedissupposedtobea“somewhatvague”formulationofaprinciplethatAyerproceedstoexpressmoreprecisely.Butactually,sinceitturnedoutdifficulttofindasatisfactorypreciseformulation,itremainsasgoodaformulationasany.SincetherearetwotypesofmeaningfulstatementforAyer,theempiricallyverifi-ableandtheanalytic,hisaccountofeachofthesewillbeconsideredinturn.EmpiricalstatementsAyerdistinguishesbetweenstrongandweakverification.Astronglyverifiableproposi-tionisonewhichcouldbeconclusivelyestablishedbysenseexperience,aweaklyverifiablepropositionisonewhichcouldbemadeprobablebysenseexperience.Itistoomuchtodemandstrongverifiabilityofameaningfulfactualstatement(itisdoubtfulindeedifanypropositionisstronglyverifiable)andsosomeformofweakverifiabilityistheappropriatecriterion.Ayertriestogiveanexactformulationofthisasfollows.Letuscallapropositionwhichrecordsanactualorpossibleobservationanexperientialproposition.Thenwemaysaythatitisthemarkofagenuinefactualproposition,notthatitshouldbeequivalenttoanexperientialproposition,oranyfinitenumberofexperientialpropositions,butsimplythatsomeexperientialpropositioncanbededucedfromitinconjunctionwithcertainotherpremiseswithoutbeingdeduciblefromtheseotherpremisesalone.(1946:38–9)Thegeneralidea,here,isclearenough.Ameaningfulempiricalstatementmustbeagenuineaidtotheanticipationoftheexperienceswecanexpecttohaveundervariouscircumstances(identifiedintermsoftheotherexperiencesthenavailable),thoughitneednottelluswhatexperiencestoexpectallonitsown.ToillustratehispointAyergivesanexampleoftwoquestionsthatmightberaisedaboutapainting.(1)WasitpaintedbyGoya?(2)IsthepaintingasetofideasinGod’smind?Peoplemaydisagreeintheiranswerstoeachofthesequestions,butinthefirstcasetheyknowwhatkindofempiricalevidencewouldsupporttheirclaimagainstthatoftheiropponents,inthelattertheydonot(1946:40).Laterinthework,especiallyinchapterVIII,“SolutionsofOutstandingPhilosophicalDisputes,”Ayershows,orclaimstoshow,howmetaphysicalquestionsareallmean-inglessinmuchthesamemanner,unlesstheyareunderstood,asisoftenappropriate,asmisleadingwaysofdiscussinghowpropositionsofacertaintypearetobeanalyzed.(Seethesectionbelow,“Whatisthetaskofphilosophy?”)UnfortunatelyAyerlaterdiscoveredthatthetechnicalformulationofthisrevelationofthemeaninglessnessofmetaphysicalquestionswasunsatisfactory.Thisisbecauseanystatementwhatever,callit“P,”canmeettheconditionsimplyinvirtueofthefact2thatitsconjunctionwith“IfP,thenO”(where“O”isaexperientialstatement)entails206\nA.J.AYER“O,”as“IfPthenO”doesnotalone.Thus“Godisannoyed,”whichAyerwouldhatetofindmeaningful,entailsanobservationstatement“Youwillshortlyhearthunder”whenconjoinedwith“IfGodisannoyedwithwhatyousaid,youwillshortlyhearthunder.”Inhis1946introductiontothesecondeditionAyerofferedamorecomplicatedformula-tion,which,however,hehadlatertoconcede,fellfouloftechnicalcriticismsfromAlonzoChurchandC.G.Hempel.(SeeChurch1949andHempel1959.)Ifoneisnottooinfatuatedwithsemi-formalization,however,onecan,surely,sayclearlyenoughwhatAyerwasgettingat,whetheroneacceptsitornot.Surelytherealpointoftheverificationprinciple,sofarasfactual(non-analytic)statementsgo,wasthis.Suchastatementismeaningfultoaparticularindividualifandonlyifitispossibleforeitheritoritsnegationtobeapracticalaidtohiminformingcorrectexpectationsaboutwhatheisliabletoexperienceinthefuture.Ifthereisnosuchpossibilitythenitisfactuallymeaningless,howevermuchhemaysupposehimselfto3understandit.Ifthereisaproblem,here,itisaboutwhat“possible”means,butperhapsitissufficientthattheindividualdoesnotutterlyruleoutitsoccurring.Itistobenoted,however,inthisconnection,thatAyerisanxioustodistinguishpracticalverifiabilityfromverifiabilityinprinciple.Thus,inanintriguinglydatedexample,“Therearemoun-tainsontheothersideofthemoon”wassaidtobeunverifiableinpracticebutverifi-ableinprinciple(anexampletakenfromMoritzSchlick).Animportantquestioniswhethertheverificationprincipleisintendednotonlytotelluswhetherastatementismeaningfulornot,butalsototelluswhatitsmeaningis.Ineffect,ratherthaninactualformulation,Ayertreatsitasdoingsoandsurelythisiscorrect.Forifafactuallymeaningfulstatementmustbeapossibleaidtoknowingwhatexperiencestoexpectundervariouscircumstancesthenitsmeaningmustlieinthetotalityofsuchaidasitiscapableofgiving.Ifthereissomeresidueofpurportedfurthermeaningitwouldseemthatthiscouldbecreamedoffasanunverifiablestate-mentincludedwithinit.Thattheverificationprincipleisintendedtoexhibitthemeaningoffactualproposi-tionsisplainfromAyer’sdeductionsfromitconcerningtheanalysisofawholerangeofordinarystatementsoffact.Thustheverificationprincipleissaidtomakeinevitableaphenomenalistanalysisofstatementsaboutmaterialobjects,sinceitisonly“bytheoccurrenceofcertainsense-contentsthattheexistenceofanymaterialthingcaneverbeintheleastverified”(1946:53).Itisnogoodsomeobjectorsayingthattheexistenceofaphysicalobjectisnotmerelyafactaboutwhatsense-contentsareavailabletous,thoughitisbythisthatitisverified,forultimatelyallthatcanbeverifiedbyfactsaboutsense-contentsarefactsaboutsense-contentsandtheprobabletruthofwhatcanbeinferredfromsuchfactsinductively.(Seethefollowingsection).Ayer,however,likeQuinelater,eschewstalkofmeaningsasentities,substitutingfortalkofmeaningstalkofsynonymy(1946:68)(seeQUINE).AnalyticoraprioristatementsTheothersortofmeaningfulstatementsforAyerwereanalyticstatements.Allgenuinelynecessaryoraprioristatementsareofthistype;thusanythinglikethesyntheticaprioriofKantandothersisrejected.207\nT.L.S.SPRIGGEOneinitialpointworthremarkingisthat,whileAyerisclearthat,ifastatementisempiricallymeaningfulthensoareitscontrariesandcontradictory,hisassertionthattheonlytwotypesofmeaningfulstatementsareempiricalhypothesesandanalyticpropositions,takenstrictly,impliesthatthisisnotsointhecaseofthelatter.Ifso,while“5+3=8”ismeaningful,theproposition“5+3=9”isnotfalse,butmeaningless.Howfarthisisintendedisunclear,sinceAyerdoes,infact,talkoffalsemathematicalstatements.(See1946:86.)Ayer’sdiscussionofanalyticpropositions(in1946:ch.IV)startsoutfromtheproblemwhichaprioritruthissupposedtoposeforempiricism(ofwhichhislogicalpositivismisavowedlyaspecies).Forempiricismcancountenancenoclaimtoknowl-edgethatisnotbaseduponsenseexperience,andeventhenwhatiscalled“knowledge”isalwaysprobablehypothesisratherthanabsolutecertainty.Wheretheempiricistdoesencounterdifficultyisinconnectionwiththetruthsofformallogicandmathematics.Forwhereasascientificgeneralisationisreadilyadmittedtobefallible,thetruthsofmathematicsandlogicappeartoeveryonetobenecessaryandcertain.Butifempiricismiscorrectnopropositionwhichhasafactualcontentcanbenecessaryorcertain.Accordinglytheempiricistmustdealwiththetruthsoflogicandmathematicsinoneofthetwofollowingways:hemustsayeitherthattheyarenotnecessarytruths,inwhichcasehemustaccountfortheuniversalconvictionthattheyare;orhemustsaythattheyhavenofactualcontent,andthenhemustexplainhowapropositionwhichisemptyofallfactualcontentcanbetrueandusefulandsurprising.(1946:72–3)Havingdismissedthefirstalternative,thatofJ.S.Mill,accordingtowhich,forexample2¥3=6issimplysowellconfirmedastatementoffactthatwe(wrongly)thinkthatitcouldnothavebeenotherwise,Ayeroptsfortheviewthatallaprioriandnecessary(theseareidentified)so-calledtruthsarereallyanalytic.RejectingKant’saccountofanalyticityforvariousreasons,Ayerformulateshisownaccount.This,however,issomewhatshifting.Themostdefinitiveformulationwouldseemtobethis:apropositionisanalyticwhenitsvaliditydependssolelyonthedefinitionsofthesymbolsitcontains,andsyntheticwhenitsvalidityisdeterminedbythefactsofexperience.Thus,theproposition“Thereareantswhichhaveestablishedasystemofslavery”isasyntheticproposition.Forwecannottellwhetheritistrueorfalsemerelybyconsideringthedefi-nitionsofthesymbolswhichconstituteit.Wehavetoresorttoactualobservationofthebehaviourofants.Ontheotherhand,theproposition“Eithersomeantsareparasiticornoneare”isananalyticproposition.Foroneneednotresorttoobservationtodiscoverthatthereeitherareorarenotantswhichareparasitic.Ifoneknowswhatisthefunctionofthewords“either,”“or,”and“not,”thenonecanseethatanypropositionoftheform“Eitherpistrueorpisnottrue”isvalid,independentlyofexperience.Accordingly,allsuchpropositionsareanalytic....However,when...wesaythatanalyticpropositionsaredevoidoffactualcontent,andconsequentlythattheysaynothing,wearenotsuggestingthattheyaresenselessinthewaythatmetaphysicalutterancesaresenseless.Foralthoughtheygiveusnoinformationaboutanyempiricalsituation,theydoenlightenusbyillus-tratingthewayinwhichweusesymbols.(1946:78–9)208\nA.J.AYERThereareseveraldifficultieswithpropositionsofthetypeheclassifiesasboth“analytic”and“apriori.”Onhisaccount,arethesepropositionsnot,ineffect,statementsabouthowcertainsymbolsarenormallyused?Butifso,theyseemtobeempirical,sinceitisanempiricalfactthatweusewordsaswedo.Ayertacklesthisquestioninhisintroductiontothesecondedition.(See1946:16–18.)Hisreplyisthat,althoughtheyaresimplytheconsequencesofstickingtoacertainconsistentuseofcertainsymbols,theydonotsomuchstateaspresupposesuchrulesoflanguage.Andthis,thinksAyer,explainswhytheycanbesurprising.Forthereare(doubtlessinfinitely)manyconsequencesofthissortwhichitrequiresconsiderableintellectualpowertograsp.WhatistroublingaboutthisansweristhatwhilethesimplerstatementsthatAyercalls“analytic”maybethoughtofaslittlemorethanremindersofhowweoptionallyusecertainsymbols,thereareinnumerableconsequencesofsuchusewhichfollowtherefrominamannerthatisnotsimilarlyoptional.Comparetherulesofchess.Thereisnoproperanswer,otherthanapresentlyirrelevanthistoricalone,astowhythepiecesmaybemovedjustastheymay,butfromthissetofoptionalrulesuntoldcon-sequencesfollowastohowthegamecanbestbeplayedtowin.Ortotakeacasemoretothepoint,evenifthepropositionthat7is6+1issimplyareminderofthemeaningof“7,”thepropositionthat7isaprimenumberisnotareminderofhowweusethesymbolbutanecessaryconsequencethereof.Itisofthislattersortofnecessitythat4Ayergivesnosatisfactoryaccount.Orsoatleastitseemstosomeofus.Bethatasitmay,thegeneralideaisclear,namelythatallnecessaryoraprioritruthsarereallytheconsequencesofanoptionaluseoflanguageandtellusnothingaboutanythingnon-linguistic.Nordotheyexactlysayanythingaboutlanguage,theysimplyhelpfullyreflectbacktousthecharacterofthelanguageinwhichourknowledgeorbeliefsareexpressed.PerhapsAyer’spositionis,ineffect,thatthefunctionofanalyticstatementsisnotstrictlytosayanythingbuttoserveasakindofverbaldrillwherebywereinforceandimproveourcommandoftherulesandacceptedtransformationswhichgiveverbal5expressionstheirmeaning.Whatistruth?Somuchforlanguageandlogic.Whatoftruth?OnemightsupposethatforAyertruthwouldconsistinbeingareliablepredictorofsenseexperienceinthecaseofempiricalpropositions,andfollowingfromlinguisticrulesinthecaseofanalyticpropositions.Actually,inchapterVAyerputsforwardwhathasbeencalledtheredundancytheoryoftruth.Accordingtothis,thecorrectaccountofwhat“true”meansisthatitsimplyemphasizestheassertionofthepropositionsaidtobetrue.Thustosaythatthepropo-sitionthatdogsbarkistrueissimplytosaythatdogsbark.Similarlytosaythatapropo-sitionisfalseissimplytoassertitsnegation.Tosaythatitisfalsethatcatsbarkissimplytosaythatcatsdon’tbark.Buthowarewetoexplainoccurrencesof“true”wherethepropositionsaidtobetrueisnotformulated,forexample,“Everythinghesaidinthelectureistrue”?Ayer’simpliedansweristhatthismeans“(p)(hesaidthatpimpliesp).”Thereisaproblemwiththisanswer,intowhichweshallnotenter,inthatthevariable“p”occursfirstlyasa209\nT.L.S.SPRIGGEnamevariableandsecondlyasapropositionalvariable;thatis,thefirstoccurrenceof“p”standsinforthenameofaproposition,anditssecondoccurrencefortheactualformulationofaproposition.InthelightofthisAyersaysthatphilosophicalattemptstoanswerthequestion“Whatistruth?”arelargelymisconceived.Whentheyhaveadefinitemeaning,thequestionaskedisreally,“Howarepropositionsvalidated?”Themeaningof“validated”isnotmadeclear,butthequestionseemstomean“Whatarethecriteriaweproperlyuseindecidingwhethertoaffirmordenythem?”(Ayeroftenuses“validate,”asitseemstome,toavoidtalkofjudgingtobetrue.)Theansweristhatanalyticpropositionsarevalidatedbytheirbeingconsequencesofthewayweusewords,whileempiricalpropo-sitionsarevalidatedbythefactthattheyhavebeenfoundsuccessfulasawayofpre-dictingwhatwewillexperienceundervariouscircumstances,andaretherebytakenaslikelytobesimilarlysuccessfulinthefuture.(See1946:99.)Ifweraise“theproblemofinduction”associatedespeciallywithHume,astowhatrightwehavetotakepastexperienceasaguidetothefuture,Ayer’sansweris,roughly,thattobeguidedinthiswayisjustwhatwepresentlymeanbybeing“rational.”Whatisthetaskofphilosophy?Ifmetaphysicsisnonsense,andtobeabandoned,isthereanytypeofphilosophythatismoreintellectuallyrespectable?Ayer’spositiveansweristhatthesortofphilosophythatisaworthwhileactivityis(conceptual)analysis.(See1946:chsIIandIII.)Andinfactthiswaswhatgenuinelygreatphilosophershavealwaysbeenmainlyengagedin.Oftentheyhaveputtheirquestionsintheformof“WhatisX?”e.g.“Whatismatter?”,“Whatistime?”,“Whatistheself?”,andthereisnothingwronginthismodeofexpres-sion,ifitisproperlyunderstood.Thusunderstood,thesequestionsarereallyrequestsfordefinitionsofsomeoftheverygeneralexpressionsinourlanguagewhicheitherpuzzleusorleadtometaphysicalnonsense.Thetraditionaltypeofdefinitionprofessestoexplicatethemeaningofawordbyofferingsomeothermorecomplexverbalexpressionwhichisitsequivalent.Ayer’strivialexampleis“Anoculistisaneyedoctor,”whichtellsusthat“oculist”and“eyedoctor”meanthesame(1946:60).Definitionsofthistypeareforthemostpartoflimitedusetophilosophy,whichisconcernedwithamorefundamentalclarificationofwhatbothexpressionsaremeanttostandfor.Insteadthephilosopherrequiresso-called“definitionsinuse.”(Seepp.60–3.)Adefinitionofthissortisaninstructionfortrans-latingstatementsaboutXorXsintoequivalentstatementswhichhavenowordorexpressionreferring(grammaticallyconsidered)toXorXs.TheseareusefulwhenXsstrikeaphilosopherassomehownotbelongingtothebedrockofreality.Thusaphilosophermighttrytoanswerthequestion“Whatisanation?”byshowinghowspecimenstatementsaboutnationscanbetranslatedintostatementsaboutpeople.Forexample,hemightseekawayoftranslating“BritainandGermanywereatwarfrom1939to1945”intoacomplicatedstatementabouthowpeoplewhosehomeswereononepartoftheearth’ssurfacebehavedtowards,andwereaffectedby,peoplewhosehomeswereonanotherpartoftheearth’ssurface.Ifsuchatranslationofstatementsaboutnationsintostatementsaboutpeopleandlandispossible,thenanationmaybedescribedasa“logicalconstruction”outofpeopleandland,thoughitshouldbe210\nA.J.AYERrealizedthatthisisnotastatementabouthowtwosortsofthingarerelatedbutbetweenlinguisticexpressions,oneofwhichissupposedlypuzzlinginawaywhichtheotherisnot.Eveninsuchratherobviouscasesasthatofnationssuchdefinitionsareusuallygesturedtowardsratherthanactuallyformulated.Thegesturemaybesufficient,however,toshowthatwhatthestatementaboutthenationssaysinasimple,butmis-leading(becauseitsuggeststhatnationsaresomethingoverandabovepersonsandland)way,issomethingwhichinprincipleshouldbesayableinawaythatgivesusnoexcuseforbeingthusmisled.Asforthedifficultyinfindingquitesatisfactoryactualtranslations,this,itmaybesuggested,maybebecausetheywouldhavetobeimpos-siblycomplicated,orbecausetheywouldhavetobepreciseaboutdetailsthatthestate-mentsbeingexplainedleavevague.Soalthoughwariscertainlyamatterofpeopledoingthingsonecannotbepreciseaboutjustwhatpeoplemustdotobeatwar.(Cf.1954:141–3.)Sothetaskofphilosophyistopointtowardsdefinitionsinuseofexpressionsthatareliabletopuzzleusortosuggestthattherearethingsoverandabovethoseweactuallyencounterempirically.PhenomenalismregardingphysicalobjectsCentraltoAyer’stypeoflogicalpositivismisaphenomenalistviewofphysicalorma-terialobjects.Accordingtothis,everystatementaboutthephysicalworldis,inprinci-ple,translatableintoapropositiontotheeffectthatundersuchandsuchconditionssuchandsuchsensationswillorwouldoccur.Thusthepropositionthataphysicalthingexistsalwaysmeans“that,ifcertainconditionswerefulfilled,certainsense-6contents...wouldbeexperienced”(1946:141).“Experiencedbywhom?”onemaywellask.Theanswerwouldseemtobe“bywhom-soeveritiswhoisaffirmingtheproposition”;inshort,whenyouaffirmit,ittellsyouwhatsensationsyoushouldexpectundersuchandsuchcircumstances,whilewhenIaffirmit,ittellsmewhatsensationsIshouldexpectundersuchandsuchcircumstances.Thissuggeststhateachofusgivesourownprivatemeaningtotheproposition.Andinfactthedoctrineofthebook,withoutperhapstheauthorbeingfullyawareofit,isthatwhatcountsasthesamefactualstatementhasadifferentmeaningforeachperson,sinceforeachofthemtheinformationitprovides,ifitistrue,concernsjusttheirownactualandpossibleexperience.Thisisdisguised,somewhat,bythefactthatpersonsarethemselvessupposedtobelogicalconstruc-tionsoutofsensations,thoughnotthesamesortoflogicalconstructionasphysicalobjectsare.Foraperson,sofarashisconsciousmindgoes,is,accordingtoAyer,alogicalcon-structionoutofthosesense-contentsthatoccurinthesamesense-fieldsasdotheorganicsenseimpressionsofthatbody.Amongtheconsequencesthatfollowfromthisisthatanindividual’ssurvivalofbodilydeathisameaninglessidea,insofarastherecanbenoorganicsensationsofhisbodythereafter.HereAyerdifferedfromtheleaderoftheViennaCircle,MoritzSchlick,whothoughtitperfectlymeaningfultosupposethatImightverifymyowndeathbyhavingsenseimpressionsasofseeingmyfuneralfromapointofviewunoccupiedbyahumanbody.(SeeSchlick1949:159–60.)211\nT.L.S.SPRIGGEPropositionsaboutothermindsThisleadsnaturallytotheaccountofourknowledgeofothermindspresentedinLanguage,TruthandLogic,chapterVII.Itislikelytobecharged,hesaysineffect,thatifIacceptthedoctrineofthatbookthenIamcommittedtosolipsism,totheviewthatonlyIexist,asaconsciousindividualwithmyownsenseexperiences.Fortheexistenceofthesense-experiencesofotherpeopleappearstobe,inprinciple,somethingthatIcannotverify,sincemysense-impressionscanonlybeassociatedwiththeorganicsensationsofmyownbody,notwiththoseofanotherperson.ButAyerrejectsthisconclusion,contendingthateachofusmustdefinetheexistenceofotherpersons,includingthesensationsthatgowiththeorganicsensationsoftheirbody,intermsofthebehavioronthebasisofwhichIwouldordinarilyconcludethattheywereconsciousandhadsensations.Thusheheld,atthatstage,thatwhat“I”(whoeverIam)meanwhenIspeakofmyownsensationsisthatsuchandsuchsensationsactuallyoccurtogetherwiththeorganicsensationsofmybody,butthatwhenIspeakofthesensationsofanotherpersonImeanthattheyarebehaving,oraredisposedtobehave,insuchandsuchaway(thisbeingaphysicalfactultimatelyconsistinginfactsaboutmysensationsasofperceivingtheirbodiesmoveandmakenoisesetc.ofsuchandsuchasort).Thisviewaboutwhatwemeanbyspeakingaboutthesensationsofothers,reveals,asAyerlaterhimselfinsisted,apeculiardouble-takeonthewholebusinessofwhatwemeanbywhatwesay(see1956:245–7).TheofficialviewofLanguage,TruthandLogicisthateveryonemeansbyeverythingtheysaysomethingabouttheirownsensations.Butthisisaviewdeliveredastrueofeveryspeaker,byAyer,quaphilosopher,who,thereby,isclearlysupposingthatotherpeoplehavesensationsinthesamesenseashedoes(i.e.inawaynotanalyzablebehavioristically)thoughitispartofthetheorythatthisrealistconceptionofthesensationsofothersis,foreachofus(andthatmustincludeAyerhimself)meaningless(see1946:141).PropositionsaboutthepastAsstrange,orstranger,thanthisviewofthemeaningofassertionsabouttheexperi-encesofothersistheviewthatpropositionsaboutthepastcanonlybemeaningful(becauseotherwiseunverifiableinprinciple)iftheyareequivalenttopredictionsaboutthekindofso-calledhistoricalevidencethatwouldsupportthem.Anoddity(wemayremark)ofthisviewisthat,whileempiricalknowledgeissaidtoconsistinpredictionsaboutfutureexperienceonthebasisofpastexperience,thefactthatthepastexperiencesoccurredisitselfapredictionofthesameessentialkind.TheparadoxicalcharacteroftheseconclusionsisamongthefactorswhichledAyerawayfromtheprecisepositionsofLanguage,TruthandLogic,thoughhestruggledtoremaintruetoatleastthegeneralspiritoftheverificationprinciple.CritiqueofethicsandtheologyOneposition,however,whichAyerneverabandonedwastheemotivetheoryofethicsadvancedinchapterVI(seeSTEVENSON).Accordingtothisethicalconceptsarepseudo-212\nA.J.AYERconcepts,thatis,theylackfactualmeaning.Tosaythatbehaviorofacertainsortiswrongisnottostateanythingaboutitwhichcanbetrueorfalse.Itissimplytoexpressafeeling.True,theremaybeethicalstatementswhichincludeafactualelement.IfIsay“Hedidwrongtokillthecat,”then,inasmuchasthissaysthathekilledthecatitismeaningful,andtrueorfalse.Butcallingitwrongaddsnothingcognitivelymean-ingful.AndifImakeastatementofpureethicalprinciple,suchas“Suicideiswrong”thenthisisjustasthoughIheldupmyhandinhorrorattheideaofpeoplecommit-tingsuicide.Thishasbeencalled(notbyAyer)the“boo-hurrah”theoryofethics.Hegaveitamilderstatementinalateressay(1954:essay10)pointingoutthathewasnotmakinganegativevaluejudgmentaboutmoralthinking,and,infact,Ayerwaspersonallyandpubliclycommittedtostrongliberalprinciples.Asforreligiouspropositions,statementslike“Godexists”or“Godlovesus,”asmostpeoplenowthinkthattheyunderstandthem,theyaremeaningless.Thereisnoobser-vationaltestthatcouldbeusedtodeterminetheirtruthorfalsehood.Godisnot,asthemountainsontheothersideofthemoonwerethen,somethingwhoseexistenceisinfactunverifiable,butinprincipleverifiable.ItisworthnotingthatevenifAyerrevisedhisviewthatlifeafterdeathwasmeaninglesshecouldstillputupagoodcaseforsayingthattheexistenceofGodwouldremainso,forwhateverexperiencesImighthaveinsomeotherworld,noneofthem(soAyercouldeasilyargue)wouldshowthattherewasorwasnotaGodassophisticatedmonotheismdescribesHim.Ofcourse,ifGodisconceivedofasanenormouslypowerfulbeinginhumanform,asdepictedinreligiouspaintings,thematterwouldbedifferent,butareligioussophisticatewillsaythatthatisamereimageofatruthwhichcannotbeexpressedinsensoryterms.AtthatpointAyersaidthathewasnotanatheist,since“Goddoesnotexist”isasmeaninglessas“Godexists.”LaterherelaxedthissomewhatandwaspreparedtocallhimselfanatheistonthegroundsthatnomeaningcanbegiventothepropositionthatGodexiststhatmakesitremotelylikelytobetrue.(SeeAyer1973.)ThefutureofphilosophyOfficiallyinLanguage,TruthandLogicAyerregardedthepositivetaskleftforphilosophy–aftertheeliminationofmetaphysicsandthefinalanalysisofthestatementsofevery-daylife,whichaverificationistapproachlikehishad,atleastalmost,finallyachieved(seepp.152–3)–astheanalysisoftheconceptsofscience.Butthiswashardlyunder-takeninthatwork,andwasneveramainconcernofAyer’s(excepttosomeextentinhistreatmentofprobability).Hisinterestwasalwaysinthosetraditionalquestionsofphilosophyhisradicalanswerstowhichwehavebeendiscussing,andmostofhislaterworkconsistsinattemptstofindmorepersuasiveanswerstotheminthelightofalessextremeformofverificationism.LaterpositionsItisunfairtotakeLanguage,TruthandLogicasbeingAyer’smaincontributiontophilosophy.Thereismuchinsufficientlyadmiredlaterwork.AtpresentAyerisoutoffashionandundervalued.Hewouldhardlyhavegroundforcomplaintatthis,scofferashewasasayoungmanathispredecessors,butgreaterjusticewillbedonetohimoneday.213\nT.L.S.SPRIGGEHeisprobablyrightthathisbestbookwasTheProblemofKnowledge(1956).Thispresentsphilosophicalepistemology(theoryofknowledge)asprimarilyconcernedtoexplainwhatknowledgeisbyanexaminationofskepticaldifficultiesfoundinvariousordinaryclaimstopossessit.IntheprocesshepresentsarevisedviewofsuchmattersaswehaveconsideredinconnectionwithLanguage,TruthandLogic,revisionstoagreatextentalreadypresentedintheessayscollectedinPhilosophicalEssays(1954).Ayercontendsthatforsomeonetoknowthatsomethingisthecase,itisrequired(1)thathefeelssurethatitisso,(2)thatitisso,and(3)thathehastherighttobesurethatitisso.(See1956:34).Skepticshaveraisedallsortsofdoubtsaboutourrightevertobesureofthetruthofanypropositionsconcerningphysicalobjects,otherminds,andpastevents,whetherthesepropositionssimplyassertthattherereallyaresuchthingsorsaysomethingmorespecificaboutparticularcases.Thepatternofskepticismisasfollows.Acontrastisdrawnbetweentheevidenceonwhichsuchpropositionsarebelievedandwhattheyclaimtobethecase.Thusourevi-denceforpropositionsconcerningphysicalobjectsalwaysconsistsinfactsaboutourownsense-data,forpropositionsconcerningothermindsinfactsaboutthebehaviorofotherorganisms,andforpropositionsconcerningpasteventsinapparentmemoriesorrecordsofthem.Therearefourtypesofphilosophicalripostetosuchskepticism.(1)Naiverealism.Thisdeniesthatourevidencefortheproblematicknowledgeclaimsisindirectinthemannertheskepticalleges,ratherdowehaveadirectorimmediateexperienceofphysicalthingsorothermindsorthepast(whicheverisinquestion).(2)Reductionism.Thisanalysestheproblematictruthsintotruthsaboutthethingswhichfeatureintheevidenceforthem.Thusfactsaboutphysicalobjectsarereducedtofactsaboutsense-dataandfactsaboutothermindstothebehaviorofothers.TheinterpretationsofsuchpropositionsinLanguage,TruthandLogicareparadigmcasesofreductionism.(3)Thescientificapproach(orcausalinferencetheory).Theproblematicpropositionsareinferredasthecausesofthethingswhichareourevidenceforthem.Noneoftheseapproachesisaltogethersuccessful.Statementsaboutphysicalobjectscannotbetranslatedintostatementsaboutsense-datanorstatementsaboutotherpeople’sexperiencesintofactsabouttheirbehavior.Naiverealismsimplyduckstheproblem.Andcausalrealismisvulnerabletotheobjectionthatapropercausalinfer-encemustbetoarealitywhichwecouldknowaboutmoredirectly,andthusbeabletocheckthatthecausalrelationbetweenourevidenceandwhatwetakeitasevidenceforreallyholds.(4)Descriptiveanalysis.Thestingoftheproblemasthedescriptiveanalystseesitliesinthefactthatwecomplainthatwehavenotgotawayofknowingaboutthesethingswhenitislogicallyimpossiblethatweshouldhave.Forexample,thecomplaintthatothermindsareclosedtouslosesitsalarmingqualitywhenwerealizethatitisanecessarytruth(analytic)thatourbeliefinanythingmustrestonourownexperience,sothatitisnotthatwelacksomepowerwhichitwould214\nA.J.AYERevenmakesenseofsomeonehaving.Oncethisisrealized,wecanbecontentwithacarefuldescriptionofournormalwayofformingbeliefsaboutthemindsofothers,sincethisisoftenasgoodasitmakessensetowishfor.Similarlyfortheotherproblems.Ayer’sownfavoredstrategyineachcaseisthelast,thatofdescriptiveanalysis.However,itmustbesaidthatinthecaseofthefirstproblem(thatofourknowledgeofphysicalobjects,alwaysaspecialinterestofhis)itisdoubtfulhowfarAyerreallymeanstodistancehimselffromreductionism.Forvarioussomewhattechni-calreasonstherecanbenoactualtranslationofstatementsaboutphysicalobjectsintoif–thenstatementsaboutsense-data,butthesuggestionlingersintheairthatsomehowthereisnothingelseforthemtotellusabout.(SeeAyer1956:131–47and1954:essay8.)Ayer’sapproachtotheproblemofothermindsissomewhatdifferent.(See1956:243–54.)Theobviousaccountofourknowledgeofothermindsisthatitrestsonanargumentfromanalogy(whichbelongstothecausalinferencetypeofapproachlistedabove).Thetroubleisthattheobviouslyrespectablecasesofargumentfromanalogyarewheretheconclusionconcernstheexistenceofsomethingofthesamegeneralkindasthingswehaveencounteredmoredirectly(saythattheremaybelifeonsomedistantplanetsufficientlyanalogoustoourown).Ayerisstilltroubled,ineffect,byaverificationistscrupleaboutanythinginprincipleunobservable.AtonestageAyersuggestedanintriguingsolution.(SeeAyer1954:essay8,also1956:247–9.)WhenIsayofanotherthatheishavingacertainexperience,whatImeancanbeanalyzedintoastatementoftheform“Apersonofsuchandsuchadescription(e.g.presentlystandinginacertainposition,female,capableinphilosophi-calargument,etc.,etc.)ishavingsuchandsuchanexperience.”Nowisit,Ayerasksus,anecessarytruththatitisnotImyselfwhoanswertothatdescription?Ifnot,anditisconceivable,thoughprofoundlycontrarytofact,thatImighthavedoneso,thenitisalsoconceivablethatIcouldhaveverifiedthepropositiondirectly.(Soanother’sexperiencebecomesmoreliketheothersideofthemoon–asitseemedtobethen–than,say,God.)Thereissomethingthatwillseemtomostpeoplerathersuspiciousaboutthissolu-tionofthedifficulty,butIcannotpursuethematterfurther.WhatitdoesshowisthatAyerwasneverquitesurewhetherhecontinuedtowantstatementswhicharemean-ingfulformetobeoneswhichinprincipleImyselfcouldverify,orwhetheritisonlyrequiredthatsomehumanbeing,orlikecreature,couldverifythem.OnallthesetopicsAyerdevelopedhispositionfurther.TheCentralQuestionsofPhilosophy,inparticular,includessomequitenovelsuggestionsonourconstructionofthephysicalworld,which,unfortunately,wecannotconsiderhere.NoristherespaceheretoconsiderthemanyotherphilosophicalproblemsAyerdealtwithinhiswork.Hisdiscussionofprobability,forexample,doeswelltoinsistonsomeeasilyoverlookedfactsaboutit,e.g.thatifstatementsofprobabilitycanonlyasserthowprobablesomethingisrelativetocertainspecifiableevidencethenthereisnowayinwhichwecanassertthatitismoreprobablethatapropositionbasedonmorecom-prehensiveevidencewillbetruethanwillonebasedonlesscomprehensiveevidence.215\nT.L.S.SPRIGGE(See1963:188–98and1972:54–8.)Hislatermorepositiveviewofmetaphysics,asconceptualclarificationwhichmayimproveourwayofunderstandingthings,shouldalsobementioned.(See1967:essay5.)Altogetherhisphilosophy,rightorwrong,hasadmirablequalitiestowhichthisaccount,concentratingasitdoes,forhistoricalreasons,onLanguage,TruthandLogic,maynothavedonejustice.Foronething,hewroteinphilosophicalproseofunrivalledexcellence.Itisthoroughlystraightforwardandextremelylucid.Apassageonceunderstoodstayseasilyinthemindasthebasisforwhatfollowsasonereadson;therewasnoneedforthoseirritatingnumberedpropositionsatwhichonemustbeforeverlookingbackinmuchphilosophywrittentoday.AyerisfornowthelastgreatfigureinthegreattraditionofBritishempiricistphilosophyinthelineofHume.Itsfaults,sofarasitisfaulty(amatternotconsideredhere),arethoseofthatwholetraditionwhichhemay,indeed,havecarriedforwardaswellasiteverwillbeagain.Notes1IshallmostlyfollowtheusagerecommendedintheintroductiontothesecondeditionofLanguage,TruthandLogic,accordingtowhichanysetofsynonymousdeclarative(hesays“indicative”)sentencesisspokenofasexpressingastatement,andthisstatementissaidtobeapropositionifandonlyifitisliterallymeaningful(1946:8).2Itisnotaltogetherclearwhatanexperientialstatementorpropositionis.(Ayergivesnoexampleshere.)Doesitreportapresentexperienceorobservationordoesitpredictoneexpectedtotakeplaceshortly?Ayer’sformulationsuggeststhefirst,inwhichcaseitisratherpointless.YetthatseemstobewhatAyerhasinmind.Inanycase,experientialpropositionsseemveryliketheostensivepropositionswhichAyerhadrejectedintheoriginalfirsteditiontext(1946:91–3).3Somuch,indeed,issaidbyAyerhimself,thoughsomewhatasanaside.“Foritwillbeshownthatallpropositionswhichhavefactualcontentareempiricalhypotheses;andthatthefunctionofanempiricalhypothesisistoprovidearulefortheanticipationofexperience”(1946:41,seealsop.151andpassim).ItisworthnotingthatwhereasthesomewhatsimilarlymindedAmericanpragma-tistsemphasizedtheimportanceofgenuinefactualknowledgeasfacilitatingourcontrolofthings,AyeralmostexclusivelyspeaksofpredictionandIrememberhimoncearguing(inaseminar)againstC.I.Lewis’sclaimthatonlyanagent,withsomecontroloverevents,couldunderstandfactualstatements.(Forarareuseof“control”see1946:50.)4Itseemsthatwemusteitherrecognizeanon-conventionalnecessityhere,oragreewiththeideamanyfindinWittgensteinthatsomehoweachdeductionfromaruleisitselfultimatelyafreedecision,oratleastonenecessaryonlyinthesenseofbeingsociallyenforced.5CompareStevenson1945:68–70.Thequestionwassometimesraisedwhethertheverifica-tionprinciplewassyntheticoranalytic.Ayer’sanswerwasthatitwasanalyticwithreferencetothemeaningwhichgiveswhatwecallfactualstatementstheirpoint.Seeintroduction,1946:15–16.6Itdoesnotfollowthatphysicalthingscannotexistunperceived,sincetheirexistenceconsistsnotintheirbeingperceivedbutinthefactthatifcertainconditionswerefulfilledtheywouldbe.ThusAyerthinkstodistinguishhimselffromidealism.See1946:145.216\nA.J.AYERBibliographyWorksbyAyer1946:Language,TruthandLogic,2ndedn.,London:VictorGollancz.(Firstpublished1936.)1954:PhilosophicalEssays,London:Macmillan.1956:TheProblemofKnowledge,London:Macmillan.1963:TheConceptofaPerson,London:Macmillan.1967:MetaphysicsandCommonSense,London:Macmillan.1972:ProbabilityandEvidence,London:Macmillan.1973:TheCentralQuestionsofPhilosophy,London:WeidenfeldandNicolson.(ReprintedPenguinBooks,1976.)WorksbyotherauthorsChurch,A.(1949)ReviewofLanguage,TruthandLogic,JournalofSymbolicLogic14,pp.52–7.Hempel,C.G.(1953)“TheEmpiricistCriterionofMeaning,”inLogicalPositivism,ed.A.J.Ayer,Glencoe,IL:FreePress,pp.115–16.Schlick,M.(1949)“MeaningandVerification,”inReadingsinPhilosophicalAnalysis,ed.H.FeiglandW.Sellars,NewYork:Appleton-Century-Crofts,p.158.(FirstpublishedinPhilosophicalReview45(1936).)Stevenson,C.L.(1945)EthicsandLanguage,NewHaven,CT:YaleUniversityPress.217\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200117J.L.Austin(1911–1960)JOHNR.SEARLEJohnLangshawAustinreceivedhisuniversityeducationinclassicsatBalliolCollegeOxford.Aftercompletinghisdegreein1933hebecameafellowofAllSoulsCollegeandin1935afellowofMagdalenCollege.DuringtheSecondWorldWar,from1939to1945,heservedasanofficerinBritishintelligence,risingtotherankofLt.Colonel.HeissaidtobelargelyresponsiblefortheextraordinaryaccuracyoftheAlliedintelli-genceatthetimeoftheNormandyinvasion,andhereceivedcitationsfromtheBritish,French,andAmericangovernmentsforhiswarwork.AfterthewarhereturnedtoOxfordandin1952hebecameWhite’sProfessorofMoralPhilosophy,aposthehelduntilhisdeathin1960.WhenAustinwasprofessor,therewereaboutsixtypracticingprofessionalphiloso-phersinOxford,andonlythreeheldtherankofprofessor(theothertwowereGilbertRyleandH.H.Price).AustinwasthemostinfluentialofaverydistinguishedgroupofOxfordphilosophersofthatperiod.DuringthefiftiesmostpeopleinOxfordthoughtitwasthebestuniversityintheworldforthestudyandpracticeofphilosophy,andtherewasnoquestionthatphilosophywasthedominantsubjectintheuniversityatlarge.Itishardforpeopleeducatedinotheruniversities,eveninBritain,toimaginethestatus,prestige,andintellectualcentralityaccordedtophilosophyinOxfordatthattime.TheperiodofAustin’sascendancymatchedcloselymyownstayinOxford,frommyentryasafreshmanin1952until,asalectureratChristChurch,Ileftin1959.Igottoknowhimquitewell,andtheseremarksarebasedinpartonmyownpersonalrecollections.Austin’sinfluencewasnotprimarilyduetohiswriting.Hepublishedonlysevenarticlesinhislifetime,andoftheseonlyone,“OtherMinds,”canbesaidtohavebeentremendouslyinfluentialatthetime,thoughthreeothers,“Truth,”“APleaforExcuses,”and“IfsandCans”receivedagooddealofattention.ThereisasenseinwhichmostofAustin’spublishedworksduringhislifetimewerepopulariza-tions.Thesewerearticlesandlecturestomeetsomeparticularrequestordemand.FourofhisarticleswerepreparedasinvitedcontributionstosymposiaoftheAristotelianSocietyandonewashispubliclecturetotheBritishAcademy.AsixthwashispresidentialaddresstotheAristotelianSociety.Onlyone,“HowtoTalk–SomeSimpleWays”was,sotospeak,unprovoked.ItwaspublishedasaseparatearticlebytheAristotelianSociety.218\nJ.L.AUSTINDuringhislifetimeAustin’sinfluencewasdueprimarilytotwofactors:first,hehadanoriginalconceptionofhowphilosophymightbepracticed;andsecond,hehadaforcefulintellectandpersonalitythatheexhibitedinhisteaching,andaboveallinphilosophicaldiscussions,bothwithstudentsandcolleagues.Hislecturesonspeechactswerepublishedafterhisdeath,andthisworkishisgreatestlegacy,thoughitwaslargelyunknowninhislifetime,excepttopeoplewhohadbeenhisstudents.Nowthathehasbeendeadforseveraldecades,wecanappraisehiscontributionsfromalongerperspective.Itseemstometherearefourdifferentsubjectsthatneedtobediscussed.First,histheoryofspeechacts.Second,hisconceptionofordinarylanguage,andofordinarylanguagephilosophy,andhowitmightbeusedconstructivelytogiveusgreaterphilosophicalinsight.Onthistopictheclassicworkishisarticle“APleaForExcuses.”Third,Austin’sconceptionofordinarylanguagephilosophyandhowitmightbeusedcriticallyintheexaminationoftraditionalphilosophicalissues.Austin’scriti-cismofsense-datatheoriesofperception,inhisposthumouslypublishedbookSenseandSensibilia,isthepurestexpressionofhiscriticaltechnique.Fourth,muchofAustin’sinfluencebothonhiscontemporaries,whenhewasalive,andonthesubsequentworkofhisstudentsandcolleagues,wasduetohisqualitiesofcharacterandintellect.Iconcludethechapterbygivingabriefassessmentofhisprincipalachievements.ThetheoryofspeechactsIbelieveAustin’smostimportantcontributiontothehistoryofphilosophyisinhisoverallphilosophyoflanguageasmanifestedinhistheoryofspeechacts.DuringhislifetimeAustin’smostimportantdiscoverywassupposedtobethatof“performativeutterances,”andcorrespondinglyofperformativeverbsandperforma-tivesentences.IntheperiodinwhichAustinworked,philosophersgenerallysupposedthatthemainfunctionoflanguagewastomaketruthclaims.Therewerevariouswaysofdescribingthese,anditwascommontosay,asthelogicalpositivistsdid,thatallofourcognitivelymeaningfulutterancesdividedintotheanalyticandthesynthetic,anditwascommoninethicalphilosophytoinsistthattherewasadistinctionbetweenthoseutteranceswhichwere“descriptive,”andthosewhichwere“evaluative.”Austinthoughtthatallofthesesimpledistinctionsweremuchtoocrude.Hewasthefirstphilosophertonoticethatthereisanimportantclassofutterancesmadewithindica-tivesentencesthatdonotsetouttobetrueorfalse,becauseintheseutterancesthespeakerisnotdescribingasituation,butratherperforminganaction,andperforminganactionwheretheutteranceofthesentenceconstitutestheperformanceoftheactionnamedbythemainverbofthesentence.SoifIsay“Ipromisetocomeandseeyou,”inappropriatecircumstances,Iamnotdescribingapromise,Iammakingapromise.AccordingtoAustin,theutteranceofthesentenceservestoperformanaction,nottodescribeanything.Thisledhimtomakeadistinctionthathethoughtwouldenableustoseemattersmorecorrectly:thedistinctionbetweenperformativeandconstativeutterances.Therearethreewaysinwhichperformativesdifferfromconstatives.First,performativessuchas“Ipromisetocomeandseeyou”typicallyhaveaspecialverbforperformingtheactioninquestion,andthereisevenaspecialadverb,“hereby,”whichwecaninsertinperformativesentences;forexample,“Iherebypromisetocomeandseeyou”.Constatives,forexample,“Itisraining,”or“Snowiswhite,”donothaveor219\nJOHNR.SEARLEneedaspecialverb.Second,constativescanbetrueorfalse,butperformativesarenottrueorfalse,rathertheyareeitherfelicitouslyorinfelicitouslyperformed.Correspondingtothetrue/falsedimensionforassessingconstativesisthefelicitous/infelicitousdimensionforassessingperformatives.Andthird,theperformativeutter-anceisanaction,adoing,whereastheconstativeisastatementoradescription.However,Austin’spatientresearcheventuallyshowedthatthiswayofmakingthedistinctiondoesnotwork.Itturnsoutthatallofthefeaturesthataresupposedtobespecialtotheperformativearetrueoftheconstativeaswell,andthuswhatwasorigi-nallysupposedtobethespecialcase,performatives,seemstoswallowthegeneralcase,constatives,whichnowturnouttobeperformancesofactionslikeanyotherutterance,andthisledAustintoageneraltheoryofspeechacts.Goingthroughthethreecriteriainorder:first,justasthereareperformativeverbsforpromising,ordering,andapologizing,soalsothereareperformativeverbsforstating,claiming,andotherconstatives.Thus,justasonecanpromisebysaying“Ipromise,”soonecanstatethatitisrainingbysaying“Istatethatitisraining,”andthecriterionthatAustinhadhopedtousetoidentifyperformativeverbs,namelythepossibleoccurrenceoftheadverb“hereby,”asin“Iherebypromisetocome,”alsocharacterizesconstatives,asin“Iherebystatethatitisraining.”Second,theso-calledconstativesalsohaveafelicitous/infelicitousdimensionofassessment,andmanyso-calledperformativescanbeappraisedastrueorfalse.Forexample,ifImakeastatementthatIamnopositiontomake,myutterancewillbeinfe-licitousinexactlythesamesensethatapromisecanbeinfelicitousif,forexample,IamunabletodothethingIpromisedtodo.SupposeIsayrightnow,“Thereareexactlythirty-fivepeopleinthenextroom,”whenIhavenobasiswhateverformakingthatstatement,thenthestatementisinfelicitousinthesamesenseinwhichperformativescanbeinfelicitous.Furthermore,thereclearlyareapparentperformativesthatcanbejudgedastrueorfalse.IfIsay,“Iwarnyouthatthebullisabouttocharge,”whenitisnotthecasethatthebullisabouttocharge,thenIhaveissuedafalsewarning,eventhoughawarningisaperformativeonAustin’soriginaldefinition.Third,makingastatementisjustasmuchperforminganactionasmakingapromise.AttheendofAustin’sdiscussiontheconclusionisobvious:weshouldthinkofeveryutteranceastheperformanceofaspeechact.Thenotionofaperformativeshouldberestrictedtothoseutterancescontainingtheperformativeuseofaperformativeexpression.Thetheoryofspeechactsbeginswiththerejectionoftheperformative/constativedistinction.WithinthetheoryofspeechactsAustinthenmadeadistinctionbetweenthreedifferentlevelsofdescriptionofanutterance:(1)thelevelofthelocutionaryact,whichisdefinedasutteringwordswithacertainmeaning,where“meaning”isexplainedassenseandreference;(2)theleveloftheillocutionaryact,whichisdefinedastheutteranceofwordswithacertainforce,whichAustinbaptizedas“illocutionaryforce”;and(3)theperlocutionaryact,whichisdefinedastheproductionofcertainsortsofeffectsonthehearer.TotakeAustin’sexample,ifIsay“Shoother,”thenifby“shoot”Imeanshoot,andby“her”Irefertoher,thenIwillhaveperformedacertainlocutionaryactofsaying“Shoother.”ButifIutteredthatsentencewiththeforceofanorder,oradvice,orrequest,thenthoseverbswillnametheillocutionaryforceofmyutteranceandhencetheillocutionaryactthatIwasperforminginmakingtheutter-220\nJ.L.AUSTINance.AndifIpersuadethehearertoshoother,persuadingistheproductionofaneffectonahearerofasortthatAustincalledaperlocutionaryact.Thedistinctionbetweentheillocutionaryandtheperlocutionaryseemstomeessen-tialforanytheoryoflanguage,anditisespeciallyimportantforthosetheoriesthattakelanguageasamatteroflinguisticbehavior;because,ofcourse,thelinguisticbehaviorwhichinvolvesproducingeffectsonpeopleintheformofperlocutionaryeffects,needstobedistinguishedfromthelinguisticbehaviorwhichinvolvesperformingspeechacts,regardlessofthesubsequenteffectsonthehearers.ImplicitinAustin’sworkisthecon-ceptionthattheillocutionaryact,nottheperlocutionaryact,isthefundamentaltargetofanalysisinthephilosophyoflanguage.ThathasbeentheassumptiononwhichIandalargenumberofotherresearchershaveproceededinattemptingtocarryonAustin’spioneeringefforts(seeSEARLE).Thedistinctionbetweenthelocutionaryandtheillocutionary,however,doesnotseemtometowork.Thereasonisthatthemeaningofthesentence,whichissupposedtodeterminethelocutionaryact,isalreadysufficienttofixacertainrangeofillocu-tionaryforces.Youcannotdistinguishbetweenmeaningandforce,becauseforceisalreadypartofthemeaningofthesentence.ThereisnowaythatIcanutterthesentence“Itisraining,”orforthatmatter,“Shoother,”withoutperformingsomeillocutionaryactinsofarasitisalocutionaryact.Thereisnodistinctionbetweenthelocutionaryandtheillocutionary,becausethelocutionaryiseoipsoillocutionary.Austinalsogaveusataxonomyoftypesofillocutionaryacts.Histaxonomyincludesthefollowing;first,verdictives.Thesearefindingsoffactorvalueonsomematter.Anexampleofaverdictiveisgivingaverdict.Second,exercitives.Thesearetheexercisingofpowers,rights,andinfluence.Exampleswouldbeappointingandvoting.Third,com-missives.Thesearealwayscasesofcommittingthespeakertoacourseofaction.Thefavoriteexample,ofcourse,ispromising.Fourth,behabitives.Thesehavetodowithsocialbehavior.Examplesareapologizingandcongratulating.Fifth,expositives.Thesemakeplainhowourutterancesfitintothediscourse.Examplesarereplying,arguing,andconceding.Aswiththelocutionary/illocutionarydistinction,itseemedtomethistaxonomyneedsrevisionandextension,becausethereisnoclearcriterionfordistinguishingbetweenthevariouscategories.IandseveralotherphilosophershaveattemptedtocriticizeandimproveonAustin’staxonomy;howeveritisimportanttoemphasizethatthecriticismsandrevisionsofhisviewsaremadewithinaframeworkheinventedandusingtoolshegaveus.IseethemanycriticismsofAustin’sspecificdoctrinesbysubsequentspeechacttheoristsnotasrefutationsbutasfurthercontributionstoadiscussionthathebegan,butdidnotlivetocomplete.ItisimportanttoemphasizethatwhenwereadAustin’smostfamousbook,theworkposthumouslypublished,HowToDoThingsWithWords,wearereadinghislecturenotes.Austinwouldneverhavepublishedthismaterialinthisform.IknowthisforafactbecauseIwantedhimtopublishitsothatIcouldpublishmycriticismsofit,evenwhenIwasastudent.Ionceaskedhim“HowsooncanwehopethatyourWilliamJameslectureswillbepublished?”thusgivinghimanopeningIshouldneverhavedone.Herespondedimmediately,“Youcanhopeitwillbepublishedanytimeyoulike.”Furtherdiscussionrevealedthathedidnotthinktheworkreadyforpublication:“Itistoohalf-baked,”hesaid.221\nJOHNR.SEARLEOrdinarylanguagephilosophy:theconstructivefunctionAustinwasmostfamousduringhislifetime,notforhistheoryofspeechacts,aboutwhichonlythetheoryofperformativeswasgenerallyknown,butratherforhispar-ticularconceptionofphilosophy,andhisstyleofdoingphilosophy.Hewasalwaysanxioustoinsistthathedidnotthinkthathiswastheonlywayofdoingphilosophy,butmerelythatitwasonepossiblewayofdoingonepartofphilosophy.Hethoughtthatthefirststeptobetakeninphilosophywastomakeaverycarefulanalysisoftheordinaryuseofexpressions.TheordinaryexpressionsofanaturallanguagelikeEnglish,hethought,embodiedallthedistinctionsabouttheworldthatpeoplehadfounditnecessaryandusefultomakeinthecourseofmillennia.Hedidnotthinkthatordinarylanguagewasthelastword,buthedidthinkitwasthefirstword.InadebatewithBertrandRussell,whenRussellaskedhimifhethoughttheexaminationofordi-narylanguagewasthebe-allandtheend-allofphilosophy,Austinisreportedtohaveanswered,“Itmaynotbethebe-allandtheend-all,butitcertainlyisthebegin-all.”TheanalysisoftheordinaryuseofexpressionsservedtwophilosophicalpurposesforAustin.Onewasacorrectivepurposeofshowingthatmanyoftheclaimsthatphilosophershadmaderestedsimplyonmistakesabouttheordinaryuseofexpres-sions.Hismostfamousdiscussioninthisregardisprobablyhiscriticismoftheargumentsforthesense-datatheoryofperception,inhislecturesSenseandSensibilia.Thesecondpurposeoftheanalysisoflanguagewasmoreconstructive:hethoughtwecouldlearnagreatdealabouttheworldbyanalyzingtheexpressionsweusetodescribetheworld.Austinthoughtthathismethodofdoingphilosophyallowedfortwofeatureswhichphilosophyisthoughtnottopossess.First,philosophyonhisconceptionisacoopera-tiveenterprise.Itisnotsomethingyoudoaloneinyourstudy,butratheryougetagroupofpeopleandtrytodiscussexamplestoseehowwordsareusedindescribingthoseexamples.Andsecond,philosophysoconstruedallowsforprogress.Itistypicalthatpeoplewhocarryonphilosophicaldiscussionofthistype,attheendoftheday,feeltheyhavemadedefiniteprogressinanalyzingtheapplicationofwordstoconcreteexamples.ItischaracteristicofAustin’sapproachthathecanoftenshowthatwhatseemliketwosynonymsornearsynonymsarereallyquitedifferent.Inafamouscasehetookthetwoexpressions“byaccident”and“bymistake”andshowedthattheyreallyhadquitedifferentmeanings,eventhoughatfirstsightmostEnglishspeakerswouldprob-ablysaytheymeanprettymuchthesamething.Herewashisdemonstration.SupposeIgooutintothefieldtoshootmydonkey.SupposeIseeyourdonkey,whichlooksverymuchlikemydonkey,andIshootyourdonkey.DidIshootyourdonkeybyaccident,orbymistake?SupposeIgooutintothefieldandshootatmydonkey,butjustasIampullingthetrigger,thetwobeastsmove,andmybulletstrikesyourdonkey.DidIshootyourdonkeybyaccident,orbymistake?Ithinktheexamplesareabsolutelyclearinbothcases.TheconstructivesideofAustin’smethodofdoingphilosophyismostpowerfullyexemplifiedbyhisarticle“APleaForExcuses”(togetherwithitsposthumouslypub-lishedcompanionpiece,“ThreeWaysofSpillingInk”).“APleaforExcuses”is,infact,asummaryofanentireseriesofseminarsthatAustingaveduringthe1950s.I222\nJ.L.AUSTINattendedtheseminars,andtherewaseasilyenoughmaterialpresentedtofillanentirebook,butitisperhapstypicalofAustinthatthematerialthatamoreaveragephilosopherwoulduseforacompletebookhecondensedintoasinglearticle.Austin’smethodisillustratedbythefollowing:mostphilosophers,myselfforexample,ifexam-iningtheproblemofaction,wouldbeginbyaskingwhatfactaboutaneventmakesitintoahumanaction.Austinthinksthatthisapproach,ashefrequentlysaidincriticizingmyviews,is“muchtoofast.”Hisownapproachis,sotospeak,tosneakupontheproblemindirectlybyaskingwhatsortsofexcuses,justifications,extenuations,andexplanationsweofferforouractions.“Excuses,”heinsists,isjustatitle,notadescriptionofthewholesubjectmatter.Theresultsoftheanalysisareaseriesofthesesthatheadvancesaboutthecharacterofourconceptualapparatusfordiscussingactions.Manyofthesearequitesurprising.So,forexample,Ithinkmostphilosophersintuitivelywouldsupposethatanyactionisdoneeithervoluntarilyorinvoluntarily.ButAustinpointsoutthatthewholequestionofnegationsandoppositesismuchmorecomplexthanthat.Theoppositesoftheword“voluntarily,”hesays,mightbe“underconstraint”or“duress”or“obligation.”Theoppositeof“involuntarily”mightbe“deliberately”or“onpurpose”orthelike.Austinurgesusnottotakeanythingforgrantedaboutnegationsandopposites.Again,Ithinkmanyphilosopherssupposethatthereisnotmuchdifferencebetweendoingsomethingintentionally,deliberately,andonpurpose.Austinmakesitabundantlyclearthatthesearenotatallthesame.Healsourgesustopaycloseattentiontolegalcasesandpsychologicalstudies;heexaminesonecase,Reginav.Finney,insomedetail,showingthatthelawyersandthejudgemakeseriousmistakes,treatingseveraltermsofexcuseasequivalentwhentheyarenot,andbeingunclearaboutwhatactionexactlyofthedefendantisbeingqualifiedbywhatexpression.Itisimpossibletosummarizethisarticlebecausethearticleisitselfasummaryofaquiteextendedprojectofresearch,andtheinterestoftheresultsisinthespecificdetails.ButthearticlerevealsboththestrengthsandsomeofthelimitationsofAustin’smethod.Ordinarylanguagephilosophy:thecriticalfunctionIbelievethepurestcasewhereonecanobserveAustin,sotospeak“inaction,”isinhisbookSenseandSensibilia.TheactualtextthatwehavebeforeusnowwaspreparedfromnotesofnumerousstudentsbyGeoffreyWarnock,butWarnockdoesanexcellentjobofconveyingtheflavoroftheactuallectures,asIcansayfromhavingattendedthem.IfAustinhadlived,Idoubtthathewouldeverhavepublishedtheselecturesastheystand.Theirresultsarealmostuniformlynegative,andthetoneisoftenmoreharshthanAustinwouldnormallyhaveallowedinpublication.Nonetheless,theyareabeau-tifulexemplificationofhismethodofphilosophicalanalysis.Hesimplydoesacarefulword-by-wordexaminationofaseriesoftraditionalphilosophicalargumentsdesignedtoshowthatweneverperceive“materialobjects,”butonlyperceive“sense-data.”AustintakesAyer’sbookTheFoundationsofEmpiricalKnowledgeashis“stalkinghorse,”andhealsodiscussesargumentsfromPriceandWarnock(seeAYER).Austingoespatientlythroughtheargumentsthataretraditionallycalled“theargumentfromillusion,”whichattempttoprovethatallweeverperceivearesense-data,andheshowsthatwithoutexceptionthearguments,aspresentedbyAyer,arehopelesslymuddled223\nJOHNR.SEARLEandconfused.Ayerassumesthatsuchwordsas“look,”“appear,”and“seems”canbeusedindifferentlyasiftheymeantthesamething,butAustin’spatientanalysisshowsthattheyarereallyquitedifferent.Inthestandardargumentsforsense-dataAustinfindsonlycarelessness,muddle,andconfusion.Thestagesoftheargumentsthathefindsare1Thephilosophersassumethattherearetwoexclusiveclassesofsenseexperience,thoseofmaterialobjects,andthoseofsense-data.2Theyarguethattheremustbenodiscriminabledifferencesinthecharacteroftheperceptionssincewecanconfuseonethingfortheother.3Theyconcludethat,sinceonewouldexpectaconsiderabledifferencefromtwosuchdifferentsortsofentities,theremustbeonlyoneclassthatweareactuallyperceiving,andthatonemustbesense-data.BypatientlyworkingthroughthetextsAustinchallengeseachoftheseclaims.1Thereareallsortsofthingsweperceivethatdonotfiteithercategorycomfortably,suchthingsasshadows,clouds,gases,flames,rainbows,images,etc.Austinthinksthedichotomyisatypicalphilosophers’oversimplification.Itwouldbejustasconfusedtosaythatallweperceivearematerialobjectsasitistosaythatallweperceivearesense-data.2Inreallifethereareallsortsofdifferencesinthecharacterofourexperiences.Dreams,forexample,aredifferentfromwakingexperiencesinallsortsofways;andeventhestock-in-tradewaking-lifeexamplesoftheepistemologistaremisdescribed.Forexample,thestickinwaterwhich“looksbent”doesnotlooklikeabentstickoutofwater,andeveninwateritneednotlooktobebent.3Itdoesnotfollowfromtheargumentsaspresentedthatallweeverseearesense-data,andindeeditisquitearbitraryforthephilosopherstoselect“sense-data”or“materialobjects”astheobjectsofperception.Itisimportanttoemphasizethatincriticizingthesense-datatheory,Austinisnotdefendingtheideathatallweseearematerialobjects.Hethinksthatideaisjustascrudeasitsopposite.Inthecourseofhisdiscussionheintroducestheideaofwhathecalleda“trouserword,”andwhatsomephilosopherssubsequentlycametocall“excluders.”Somewordsgettheirmeaninginacontextfromthewordsthattheyareopposedtointhatcontext.Thusrealcreamisopposedtoartificialcream,butarealduckisopposedtoatoyduckoradecoyduck,andrealteethareopposedtofalseteeth.Theword“real”isanexcluderthatgetsitsmeaningincontextfromwhatitisopposedto.Thereisnocommonpropertyofrealitywhichtheword“real”invariablyandliterallyservestoascribe.Adecoyduckforexample,thoughnotarealduck,maynonethelessbearealdecoy,asopposedforexampletoapapermodelofadecoyduck.Whentheepiste-mologisttalksaboutrealityandperceptionsofreality,hefailstoappreciatethenatureoftheconcept.InareplytoAustin,Ayerclaimedthathismainpointscouldsurvive,evenifheacceptedallofAustin’sspecificobjections.Hismainpointisthatwecouldhavealloftheexperienceswedohaveandstillbemistakeninourclaimsaboutobjectsandstatesofaffairsintheworld.TothisIthinkAustinwouldhavereplied,firstthatthisdoesnot224\nJ.L.AUSTINshowthatallweeverperceivearetheexperiences,the“sense-data.”And,second,itdoesnotshowthattherelationbetweentheexperiencesandtheobjectstheyaretakentobeexperiencesof,isoneofevidence.Itdoesnotshowthattheexperiencesareevidenceforthepresenceoftheobjects.Whateverthemeritsofthedebate,IthinkAustin’scritiqueprovedimmenselyinflu-entialhistorically.Oneusedtohearalotaboutthesense-datatheoryofperception;onedoesnothearmuchaboutitanymore.OtherworksAustinwroteanumberotherimportantworkswhichlimitationsofspacepreventmefromexploringinanydetail,butImustmentiontheminpassing.“OtherMinds”pre-sentsacriticismoftraditionalepistemologywhichisverymuchinthespiritofSenseandSensibilia.“Truth”and“UnfairtoFacts”presentaversionofthecorrespondencetheoryoftruthandaresponsetoStrawson’scriticismsofit.IthinkAustinisrightthatthefundamentalnotionoftruthiscorrespondence,buttheparticularversionthathepresentsdoesnotsurviveStrawson’sobjections.“IfsandCans”isagemofphilosoph-icalanalysis.Asfarasthehistoryofphilosophyisconcerned,itsmainpointistorespondtothoseversionsofcompatibilismaboutthefreewillproblemwhichmaintainthattosaythatIcouldhavedoneotherwisejustmeansthatIwouldhavedoneother-wiseifIhadsochosen,andtosaythatIcandosomethingjustmeansthatIwilldoitifcertainotherconditionsaremet.ButAustinmakesalargenumberofotherpointsaboutrelatedconceptualissues.CharacterandintellectMyownimpressionsofAustinaresomewhatdifferentfromthoseofmanypeoplewhothoughtofthemselvesashiscloseassociatesandfollowers.Onetraitthatwewouldallagreeonwashisimmensecarefulnessandprecision.Notonlywhendoingphilosophy,buteveninthemostcasualconversation,Austinspokeandthoughtwithgreatpreci-sion,andhedidnottolerateloosenessinhisstudentsorcolleagues.Theworstcon-demnationthathecouldmakeofsomethinghewasreadingwouldbetoshakehisheadsadlyandsayinhisthin,preciseway,“It’sjustloose.”Indeed,onseveraloccasionshesaidtomeintonesmoreofsadnessthananger,“Thereisalotloosethinkinginthistown.”Forthemostpart,Austin’scolleaguesregardedhimwithakindofawe,anditseemedtomethattosomeextenttheywereeventerrifiedinhispresence.Certainly,hispresenceinseminarsandmeetingshadaprofoundeffectonthebehaviorofotherspar-ticipating.InoticedthiswhenIwentbacktoOxfordsomeyearsafterhisdeathandfoundthatmanyoftheprofessionalphilosopherswerebehavinglikeschoolboysduringrecess.TheyweremuchlesscautiousthantheywouldhavebeeninAustin’spresence.AtfirstIcouldnotunderstandthesourceofAustin’sinfluence,becauseitseemedtomethatIcouldIbeathiminargument;andlikealotofundergraduatesIthoughtthetestofaphilosopherwashowgoodhewasinthegiveandtakeofphilosophicalrepartee.Austin’stechniqueindiscussionwasalwaystotakeeverythingdeadliterally,andthentoinsistoncertainlinguisticdistinctionsthathethoughtwerebeingover-looked.So,forexample,whenWittgenstein’sPhilosophicalInvestigationswaspublished,225\nJOHNR.SEARLEIandsomeotherundergraduatephilosophersinsistedthatwediscussWittgenstein’sprivatelanguageargumentinAustin’sinformalinstructionforundergraduates.Austin’stechniquewastorefusetograntWittgensteinanyleewayatall.AtonepointwhenwewerediscussingWittgenstein’sfamousexampleofthebeetleinthebox,Austinsaidsarcastically,“Allright,forournextsessioneveryonebringaboxwithabeetleinit.”AtonepointinWittgenstein’sdiscussion,hesaystheremightbenothingatallinthebox.AustinthoughtWittgensteinwassimplycontradictinghimself.“Firsthesays,theredefinitelyisabeetleinthebox,andthenhesaystheremightbenothinginthebox,aplaincontradiction.”Austin’shabitofinsistingonthehighestlevelofprecision,bothinhisprofessionalactivitiesandeveninordinaryconversations,seemstomeoneofthemainreasonsfortheterrorthatheinspiredinhiscolleagues.Atthetime,muchofthesourceofAustin’sinfluencederivedfromhisschoolmasterlystyle.MostOxfordphilosophersofthetimehadbeenstudentsatBritishboardingschools,andAustinwas,sotospeak,theulti-mateschoolmaster.IfhewerereadingapaperthatIhadwritten,atypicalquestionheputtomewouldbe,“Whyexactlydidyouusethesubjunctive?”Or,onanotherocca-sion,“Intheverb‘suppose,’whatdoesthe‘sup’mean?”Austinwasfamousatthetimeforhisattentiontotheminutedetailsofordinarylanguage,butitseemedtomethen,anditseemstomenow,thathisrealcontributiontophilosophywasnotsomuchinthedetails.Austindidindeedhaveageniusforspottinglinguisticdifferencesanddis-tinctionswheremostpeoplewouldhavethoughttherewerenone,thoughinthedetailshisviewsweresometimesmistaken.Hismostimportantcontributiontophilosophy,Ibelieve,isinhisoverallvisionoflanguage.Thoughhewasregardedasterrifyingbymanyofhiscolleagues,Icansaythattoundergraduateshewasimmenselykind,patient,helpfuland,inhisreservedway,evenfriendly.Hiscontemptwasreservedforpeoplehethoughtofaspretentious,self-important,pompous,andaboveallobscurantist.WhenAustin,alongwithseveralotherOxfordphilosophersandstudents,wentofftoRoyaumontinFranceforanEnglish–FrenchphilosophycolloquiumheconsideredthepretentiousnessofMerleau-Ponty,thenthemostinfluentialFrenchphilosopher,quiteridiculous.“ThatMerleau-Ponty,heisjustalittletingod.Hewillnevergetanywhere.”Austinwouldhavehatedthe“deconstructionists”and“postmodernists”whocurrentlypretendtoadmirehiswork.Hedidnotthinkhisbrandofphilosophywastheonlycorrectwaytodothesubject,buthedidtrytoextenditsinfluencewithanalmostmissionaryzeal.Again,whenwewereinRoyaumont,andhesawmeindiscussionwithanelderlydistinguishedFrenchphilosopher,hetookmeasideandsaid,“Don’twasteyourtimeontheaged.Talktotheyoung!”ThisremarkannoyedmeatthetimebecauseIdidnotthinkofmyselfthereforanyotherreasonthantopracticephilosophy.Austin,Ibelieve,wouldhavethoughtitawasteoftimeforustogotoFranceifwedidnottrytospreadthetruth.Ineverheardanyoneotherthanhiswifeaddresshimbyhisfirstname,andwhenoneofhiscolleagueshadthetemeritytoaddresshimas“John,”Austinisreportedtohavesaidevenly,“‘Austin’isalsoaChristianname.”Hedidhaveahabitofholdingone’sattentionindiscussion.So,forexample,whenmakinganinvolvedpointhewouldtakehispipeinonehandandlightamatchwiththeother.Nevertakinghiseyesoffhisinterlocutor,hewouldallowthematchtoburneverclosertohisbarefingersuntil226\nJ.L.AUSTINatthelastmillisecondhewouldflickhiswristtoextinguishthematch,whereuponhewouldstartthewholeprocessoveragainwhilecontinuinghisrelentlessdiscourse,eyesalwaysonhislisteners.Austin’sreluctancetopublishwaspartofthecultureofOxfordatthetime,butalsoitwaspartlycharacteristicofhisownattitudes.Oxfordhadalongtraditionofnotpub-lishingduringone’slifetime,indeeditwasregardedasslightlyvulgartopublish.Peoplewhodidpublishalot,likeA.J.Ayer,wereregardedasremissforhavingpublishedtoomuchtoosoon.Asfarashavingacareerandmakingareputationwereconcerned,theattitudeinOxfordwasthattheonlyopinionsthatreallymatteraretheopinionsofpeopleinOxford,andperhapsafewinCambridgeandLondon,andtheywillknowaboutone’sworkanyway.Onedoesnotneedtopublish.Whatonedoesnotwantisalotofgraduatestudentssomewhere,pickingoverone’shalf-bakedpublishedtextsand–horrorofhorrors–findingmistakes.SoIthinkAustin’sreluctancetopublishwaspartlyduetohisextremecarefulness,butitwaspartlyduetohissheervanity;hedidnotwantanyintellectualinferiorpointingouterrors.Atatimewhenanti-AmericanismwasverycommoninBritain,especiallyamongtheintellectualclasses,AustinsimplyadoredtheUnitedStates,andespeciallyitsuniversitysystem.HewouldnottoleratecriticismsoftheUnitedStates,andtheonlysubjectonwhichIhaveeverheardhimshowuncriticalenthusiasmwasAmerica.Indeed,heoncesaidtome,“ThefuturelieswithAmerica,”andonanotheroccasion,“Thereareunplowedfieldsinthatcountry.”OnceHerbertHartwascriticizingAmericancooking,andAustinsaidevenly,inhisdiscussion-endingway,“Itisnotsobad.”IoftenreadhowmuchAustinwasinfluencedbyWittgenstein.Nothingcouldbefurtherfromthetruth.AustinhadnosympathywhateverforWittgenstein,andIthinkhewasincapableoflearningfromsomeonewhosestylewasso“loose.”HetypicallyreferredtoWittgensteininthestyleofEnglishschoolboyslangofthetimeas,“Witters,”pronounced“Vitters.”HethoughttherewerenooriginalideasinWittgenstein.IndeedheoncesaidtomeaboutWittgenstein’sphilosophy,“It’sallinMoore,”oneoftheleastaccuratethingsIhaveeverheardAustinsay.IfAustinhadaninspirationalmodel,itwasMoore(seeMOORE).ItwillseemaparadoxicalfeatureofAustin’scareerthathearousedsuchpassion-atecontroversybothproandcon.Onthesurface,atleast,hispresentationsareinvari-ablymodest,cautious,andself-effacing.ThoughIdidnotregardmyselfasoneofhisfollowers,Ifounditeasytoseewhytheyregardedhimwithsuchenthusiasm.Heofferedthemanewconceptionofphilosophy,andwithit,anewresearchprogram.Butitismorepuzzlingtotrytounderstandwhyhewashatedsomuch.Ithinkinordertounderstandthehostilitythathearoused,wehavetocomparehiscareerwiththatofSocrates.HewashatedformuchthesamereasonthatSocrateswashated:heseemedtodestroyeverythingwithoutleavinganythingsubstantiveinitsplace.LikeSocrateshechallengedorthodoxywithoutpresentinganalternative,andequallycomforting,orthodoxy.AllAustinoffered,againlikeSocrates,wasanewmethodfordoingphilosophy.Austin’ssubstantiveachievement,especiallyinthetheoryofspeechacts,hassur-vivedhisdeathnowfornearlyahalfcentury,andwill,Ibelieve,continuetobeafocusofresearch.Buthisofficialdoctrineastohowphilosophymightbepursuedhaswaned227\nJOHNR.SEARLEconsiderably.Ithasveryfewfollowersandpractitioners.Why?Well,partoftheansweristhatitisjusttoodifficult.ThesortofverycarefulanalysisofminutelinguisticdistinctionsthatAustinurgedustoundertakeissimplytoomuchworkformostphilosophers.Austinthoughtthatweoughttobemorepatientandhardworking.Ifentomologistscanclassifyamilliondifferentkindsofinsects,surelyphilosophersoughttohavethepatiencetoclassifythefewdozensorfewhundredorevenafewthousanddifferentsortsofusesofdifferentsortsofwords.Buttheproblemisthatthemotiva-tionthattendstomakeoneaphilosopherseemstobequitedifferentfromthemotiva-tionthatmakesoneanentomologist.Philosopherswantverygeneralanswerstoverylargequestions,whereasAustinthoughttheyhadfirstbettergetclearaboutthedistinctionsamonganumberofadverbs,workingthemselvesuptoundertakeananalysisofafewverbs.TherearecertainlimitationsonAustin’smethods,which,paradoxicallybuttohiscredit,wecanuseAustin’stheoryofspeechactstoexpose.(1)SometimesAustinconfusesthetruthconditionsofaterm,thatis,thecondi-tionsunderwhichitisafactthatsomeobjectsatisfiesthatterm,withtheconditionsforappropriatelyassertingthatthetermapplies.Thus,totakeanAustin-styleexample,Austinpointsoutthatwewouldn’tnormallyassertthatamanwalkedacrosstheroomintentionally,ifhejustwalkedacrosstheroominanordinary,unexceptionalway.“Nomodificationwithoutaberration,”Austintellsus.Nonethelessitmaybetruethatthemanwalkedacrosstheroomintentionally,itisjustnotappropriatetoassertitunlessthereissomethingunusual.Itmayjustbetooobviousthattheactwasdoneintentionally.(2)Relatedtothefirstmistakeisthemistakeofconfusingthemeaningofatermwiththeillocutionaryforcethatcharacteristicallyaccompaniestheassertionthatthetermappliestoanobject.Thus,totakeanotherexamplefromAustin,hepointsoutthatwhenwesaythatweknowsomething,weareoftengivingourguaranteeforwhatweclaimtoknow,thattheclaimtoknowhascertainfeaturesincommonwithaper-formative,suchas“Ipromise.”Austiniscarefulnottosaythat“know”isaperforma-tive,buthedoesthinkthattheassertionthatoneknowshasaperformative-likeguaranteeingforce.Butonceagainthisdoesnottellusthemeaningoftheword“know,”becauseitcannotaccountfortheoccurrenceofthiswordinothercasessuchasconditionalsornegations.Soevenifanutteranceoftheform“Iknowthatp”meanssomethinglike“Iguaranteethatp,”still,anutteranceofaconditionaloftheform“ifIknowthatpthenq,”doesnotmeananythingatalllike“ifIguaranteethatp,thenq.”(3)Evenafteryouhavedoneacarefullinguisticanalysisandshownthatthestandardphilosophicalpositionsrestonamisuseofwords,still,youcanoftenstatethepositionagainwithoutusingthosewords.Theproblemremainsevenafterthemisuseofwordshavebeencorrected.Thus,totaketheproblemoffreewill,Austinpointsoutthatwhenwesaythatanactwasdonefreely,“freely”functionsasanexcluder,exclud-ingallofthevariouswaysinwhichanactmaynothavebeendonefreely,suchas,forexample,doneunderduressorundercompulsion.ButevenifAustinisrightaboutthis,andheprobablyis,youstillhaveafreewillproblemleftover.Hereitis:Areallhumanactssuchthattheperformanceoftheacthasantecedentcausalconditionswhicharecausallysufficienttodeterminetheact?IfIwalkacrosstheroom,andIwalkacross228\nJ.L.AUSTINtheroominawaywhichisnotunderduressorcompulsion,allthesamethereisstillthequestionaboutfreewill.WeretheantecedentconditionspriortotheonsetofmyactionofwalkingacrosstheroomsufficienttodeterminethatIwasgoingtowalkacrosstheroom?Thatquestionremainsevenafterwehavebecomeclearaboutallofthevarioususesoffreely,voluntarily,etc.(4)Austinsaysthatordinarylanguageembodiesallofthedistinctionsthathumanshavechosentomakeovermillennia.Butthereisasenseinwhichthatisnotquiteright.Wecanindeedstateinordinarylanguageallofthedistinctionsthathumanshavechosentomake,andindeedwecanstatealotthattheyhavenotchosentomake.Butitisnotthecasethateveryrealdistinctionthathumanshavemadeismarkedbyalexicaldistinction,bytwodifferentwords,inordinaryusage.ThustotakeoneofAustin’sexamples,theword“pretend”doesnotmarkadistinctionbetweenthosecasesofpretensewhicharegenuinelyintendedtodeceive,andthosecasesofpretensewhichareputonoraremock-performances,butnotdesignedtodeceive.SoifIpretendtobethePresidentoftheUnitedStatesinordertobeadmittedtotheWhiteHouse,Ihavepretendedinthedeceptiveway.ButifIpretendtobethePresidentoftheUnitedStatesaspartofagameofcharades,thereisnointentiontodeceive.Thisisanobviousandimportantdistinction,asAnscombepointedoutinthesymposiumwithAustinonpretending,butwedonothavetwoverbswhosemeaningsare“pretenddeceptively”and“pretendnon-deceptively.”ConclusionJ.L.Austinwasoneofthemostimportantphilosophersofthetwentiethcentury.Inexamininghiscontributionweneedtodistinguishbetweenthephilosophyoflanguageandlinguisticphilosophy.Thephilosophyoflanguageistheattempttogiveanaccountofcertainverygeneralfeaturesofthestructure,use,andfunctioningoflanguage.Linguisticphilosophyistheattempttosolvephilosophicalproblemsbyusinglinguisticmethods.Austinmadeimportantcontributionstoboththephilosophyoflanguageandlinguisticphilosophy.Duringhislifetimehewasfamousasalinguisticphilosopher,butnotforhisphilosophyoflanguage.Sincehisdeathithasemergedthathismostimpor-tantcontributiontophilosophyhasbeenhisphilosophyoflanguageasexpoundedinhistheoryofspeechacts.Attheconclusionof“IfsandCans,”Austinexpressesthehopethatthenextcenturymayseethebirthofacomprehensivescienceoflanguage.Ibelievethathethoughthistheoryofspeechactswasacontributiontowardthatfuturescience.BibliographyWorksbyAustin1962:HowToDoThingsWithWords,Cambridge,MA:HarvardUniversityPress.1964:SenseandSensibilia,Oxford:OxfordUniversityPress.1979:PhilosophicalPapers,3rdedn.,Oxford:OxfordUniversityPress.(2ndedn.,1970.)Articlesinthiscollectionaregivenbelow,withfirstpublicationdetailswhereapplicable:“AgathosandEudaimoniaintheEthicsofAristotle,”fromAristotle:ACollectionofCriticalEssays,ed.J.M.E.Moravcsik,NewYork:Doubleday,1967,andLondon:Macmillan,1968.“APleaforExcuses,”ProceedingsoftheAristotelianSociety57(1956–7).229\nJOHNR.SEARLE“AreThereAPrioriConcepts?,”ProceedingsoftheAristotelianSociety,AnnualConference,1939.“HowtoTalk,”ProceedingsoftheAristotelianSociety53(1952–3).“IfsandCans,”ProceedingsoftheBritishAcademy42(1956).“OtherMinds,”ProceedingsoftheAristotelianSociety,suppl.vol.20(1946).“PerformativeUtterances.”“Pretending,”ProceedingsoftheAristotelianSociety,suppl.vol.32(1957–8).“TheLineandtheCaveinPlato’sRepublic.”(Notin1stand2ndedns.)“TheMeaningofaWord.”“ThreeWaysofSpillingInk,”PhilosophicalReview75/4(1966).“Truth,”ProceedingsoftheAristotelianSociety,suppl.vol.24(1950).“UnfairtoFacts.”WorksbyotherauthorsAnscombe,E.(1957–8)“Pretending,”ProceedingsoftheAristotelianSociety,suppl.vol.32.Ayer,A.J.(1940)FoundationsofEmpiricalKnowledge,London:Macmillan.——(1973)“HasAustinRefutedSense-data?,”inEssaysonJohnL.Austin,ed.I.Berlinetal.,Oxford:ClarendonPress.Fann,K.(ed.)(1969)SymposiumonJ.L.Austin,London:RoutledgeandKeganPaul.Holdcraft,D.(1978)WordsandDeeds,Oxford:OxfordUniversityPress.Searle,J.(1969)SpeechActs:AnEssayinthePhilosophyofLanguage,Cambridge:CambridgeUniversityPress.——(1979)ExpressionandMeaning:EssaysintheTheoryofSpeechActs,Cambridge:CambridgeUniversityPress.——(1989)“HowPerformativesWork,”LinguisticsandPhilosophy12,pp.535–58.Travis,C.(1975)SayingandUnderstanding,Oxford:BlackwellPublishers.Warnock,G.(1991)J.L.Austin,ArgumentsofthePhilosophersseries,NewYork:Routledge.230\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200118NormanMalcolm(1911–1990)CARLGINETIntroductionNormanMalcolmwasbornonJune11,1911,inSelden,Kansas,anddiedinLondononAugust4,1990.HisundergraduateyearswereattheUniversityofNebraska,whereO.K.Bouwsmawasoneofhisteachers.HisPh.D.,grantedin1940,wasfromHarvard,butthemostimportantphilosophicalinfluencesonhimduringhisgraduateyearswereG.E.MooreandLudwigWittgenstein,withwhomhestudiedduringafellowshipatCambridgeUniversityin1938–9.HewasaninstructoratPrincetonbeforejoiningtheUSNavyin1941.AfterthewarhespentanotheryearinCambridge,1946–7,study-ingwithMooreandWittgenstein.In1947hejoinedtheSageSchoolofPhilosophyatCornell,whereheremaineduntilhisretirementin1978.DuringthelasttwelveyearsofhislifehelivedinLondonandwasappointedaVisitingProfessorandFellowatKing’sCollegeLondon,wherehegaveaweeklyseminarmainlydevotedtothephilosophyofWittgenstein.MalcolmcreditedMoorewithbeingthefirsttoemploythetechniqueofrefutingparadoxicalphilosophicalstatementsbypointingoutthattheygoagainstordinarylanguage–thattheyimplythatordinaryusesoflanguageareincorrectuses–whichisMalcolm’sownfavoritetechnique.FromWittgensteinhetooktheideathataphilo-sophicalproblemisessentiallyaconfusioninourthinkingthatistoberemediedbyremindersoftheactualuseoflanguage,andbyreconstructingandcriticizingtheanalogiesandreasoningthatbewitchthevictimofthepuzzle.MalcolmwasamajorexpounderandendorserofWittgenstein’slaterphilosophy.HedevotedseveralarticlesexplicitlytoexplainingWittgenstein’sthought.TheearliestandprobablymostinfluentialofthesewashisdiscussionofWittgenstein’sPhilosophicalInvestigationspublishedinThePhilosophicalReviewin1954,whichpromptedagooddealofinterestinWittgenstein’sargumentagainstthepossibilityofaprivatelanguage.WittgensteinvisitedMalcolminIthacain1949andtheirdiscussionsthereofknow-ledgeandcertaintystimulatedthethinkingthatledWittgensteintohislastmajorwork,OnCertainty.Lateinhislife,in1986,Malcolmpublishedabook,NothingisHidden,theaimofwhichistoexpoundWittgenstein’slatercriticismofhisTractatusLogico-Philosophicus.Andathisdeathheleftamonograph,FromaReligiousPointofView?(publishedposthumously,editedandwitharesponsebyPeterWinch)inwhichhe231\nCARLGINETsummarizesmuchofWittgenstein’swork,earlyandlate,inanattempttoseeinwhatsenseWittgenstein’sremarkthatheapproachedeveryproblemfromareligiouspointofviewmightbetrue.ButnearlyallofMalcolm’swork,fromtheearly1950son,isshotthroughwithapprovingreferencetoremarksofWittgenstein’s.ItwouldnotbeanexaggerationtosaythatheaimednearlyallofhisworkatgettingacrossinsightsheowedtoWittgenstein(seeWITTGENSTEIN).Malcolm’swritingisremarkableforitsclarityandvigoranditsfreedomfromtech-nicaljargon.Itiscrammedwithdown-to-earthexamples.Thesesometimeshelptogiveaconcretegraspofanabstractidea,buttypicallytheyservetoremindhisreadersofhowordinarylanguageisactuallyused.Theyalwaysgivehiswritingconsiderablecharm.WhatfollowsarebriefexpositionsofsomeoftheviewsMalcolmarguedfor(Ifindnoevidenceinhiswritingsofanymajorchangeinhisviews),placedunderfourhead-ings:knowledge,mind,memory,andphilosophyofreligion.Thissampleisfarfromcomprehensive,butIhopeitgivesagoodideaofthebreadthandcharacterofMalcolm’swork.KnowledgeIntwoimportantearlypapers,“CertaintyandEmpiricalStatements”(1942)and“TheVerificationArgument”(1950),Malcolmrebuttedtheclaimmadebysomephiloso-phers(e.g.C.I.Lewis,Carnap,Russell,Ayer)thatitisimpossibleforanempiricalstate-ment(acontingentstatementaboutmaterialobjects)tobeknownwithcertainty.InLewisandCarnaphefindsanargumentforthisparadoxicalclaim,whichhecallstheVerificationArgument.Hearrivesatthefollowingformulationoftheargument(“TheVerificationArgument”,in1963a:26):I.AnyempiricalstatementShasconsequences(notinthesenseofentailmentbutinthesenseinwhichitisaconsequenceof“Yesterdaythephrase‘thestreamofthought’wasonpage224ofvol.IofmycopyofJames’sThePrinciplesofPsychology”thatifIweretolookonthatpagenowIwouldseethatphrase).II.TheconsequencesofSareinfiniteinnumber.IIIa.ItisnotcertainthattheconsequencesofSwilloccur.IVb.Ifanyempiricalstatementcanbeconclusivelyestablishedastrueorfalse,thenifasufficientnumberoftheconsequencesofSshouldfailtooccurthenitwouldbeabsolutelyconclusivethatSisfalse.Va.IfatanytimeitshouldbeabsolutelyconclusivethatSisfalsethenatnoprevi-oustimedidanyonemakeabsolutelycertainthatSistrue.Itdoesfollowfromthesepremisesthat,foranyempiricalstatementS,nooneevermadeabsolutelycertainthatSistrue.MalcolmdeniespremiseIIIa.HearguesthatitwasacceptedbyproponentsoftheVerificationArgumentonlybecausetheythoughtitfollowsfromIII:ItispossiblethattheconsequencesofSwillfailtooccur.ButtheythoughtthisonlybecausetheyfailedtodistinguishamongdifferentinterpretationsofIII.Forthosesensesof“possible”inwhichIIIistrue(“TheconsequencesofSwillfailtooccur”isnotself-contradictory;noconsequenceofSisentailedbythegroundsforholdingittrue),IIIdoesnotentailIIIa;andforthosesensesof“possible”inwhich232\nNORMANMALCOLMIIIentailsIIIa(thereissomereasontobelievethattheconsequencesofSwillnotoccur;thereisnoreasontothinkthattheconsequencesofSwilloccur;thegroundsforholdingthattheconsequencesofSwilloccurarenotabsolutelyconclusive),IIIisnottrue.Inhisearlierworkonknowledge,Malcolmseemedtosharethecommonassump-tionthatstatementsoftheform“Sknowsthatp”stateafactaboutS,afactaboutwhosenecessarycomponentsphilosophersmighthopetosaysomethinginformative,e.g.,thatSmustbelievep,thatpmustbetrue.Butinmanyofhislaterwritings,Malcolmseemstotreatsuchsentences,particularlyinthefirst-personpresent-tenseform,aslikeper-formativeswhoseuseis,notprimarilytoreportafactaboutthesubject,onethatisthereindependentlyofanyutterance,butrathertoachievesomeaimofthespeaker,sothatthecorrectnessandintelligibilityofitsusedependheavilyonthecontextoftheuse.(Therewere,however,alreadyintimationsofthisideainearlypapers.In“DefendingCommonSense”(1949)hesaidthatMooremisused“know”inmakingsuchassertionsas“IknowthatIamahumanbeing”or,whenholdinguphishand,“Iknowthatthisisahand,”inthecontextsinwhichhemadethembecausetherewasnotanydoubtordisagreementaboutthematterthatwouldgiveapointtosuchasser-tions.In“PhilosophyforPhilosophers”(1951),forsuchreasonsasthatasentencelike“IknowIfeelhot”isalmostneverseriouslyused,thatthenormalusageof“Iknow”isinformativeandconnectedupwithinvestigating,findingout,makingsure,producingevidence,withaskingandanswering“Howdoyouknow?”,hesaidthat“Inthesenseof‘knowledge’inwhichknowledgeiscontrastedwithbelief,wedonot(andcannot)haveknowledgeofourownsensations”(p.336).)His1976paper“MooreandWittgensteinontheSenseof‘IKnow’,”saysthat“Iknow”doesavarietyofjobsinordinarylanguageuse;forexample,“itisusedtoclaimthepossessionofevidence,orexpertise,orability;itisusedtocomfort,reassure,expressagreement;itisusedtosaythatonehasthoroughlycheckedsomething,orthatonecanbereliedon,orthatonedoesn’tneedtobereminded”(1977b:192).Forsuchreasonsheoftensaysthat,exceptinveryspecialcontexts,itmakesnosensetosay“Sknowsthatheisinpain”andwhatitmeanswhenitdoesmakesenseisverydifferentfromwhatismeantby“Sknowsthatthereisagashinhishand.”Herefusestoacceptthatthereisinthiscaseanydistinction(suchashasbeensuggestedbyGrice,Searle,andothers)betweentruth-conditionsandrequirementsfortheaptnessofasserting(seeGRICEandSEARLE).Itisunclearwhathisviewwouldbeaboutwhetherthereissuchadistinctionforsuchsentencesas“Mooreisahumanbeing”or“WhatMooreisholdingupisahand.”MindTherearetwoconnectedprinciplesaboutpsychologicalconceptsthatarefundamen-talforMalcolm.Oneis:Onlyofalivinghumanbeingandwhatresembles(behaveslike)alivinghumanbeingcanonesay:ithassensations;itsees;isblind;hears;isdeaf;isconsciousorunconscious...thinks(seeWittgenstein,PhilosophicalInvestigations§282,§360).Weascribementalpropertiestoothersonthebasisofobservablebehavioralcriteriathatarenon-contingentlyconnectedtotheconceptsofthoseproperties;itispartofhavingtheconceptsofthepropertiestoknowbehavioral233\nCARLGINETcriteriathatjustifyascribingthepropertiesintherightcircumstances.Thesecondis:Itisnotonthebasisofanycriteriathatweascribementalproperties(current,consciousones)toourselves;ourself-ascriptionsareanalogousto,andinsomecasessimplyreplace,naturalmanifestationsofmentalstates,suchastheexpressionofpainincryingormoaning;theyserveothersascriteriaforascribingthepropertiestous.“Firstpersonutterances,andtheirsecondandthirdpersoncounterparts,”hesays,“arelinkedinmeaningbyvirtueofbeingtied,indifferentways,tothesamebehavioralcriteria”(1971:91).Cartesianphilosophyofmind,whichhesometimescalls“introspectionism,”runsafoulofthefirstprinciple.Behaviorismasaphilosophyofmindrunsafoulofthesecond.AgainstthethesisthatmentalstatesorprocessesareidenticalwithbrainstatesorprocessesMalcolmmarshalsseveraldifferentarguments.In“ScientificMaterialismandtheIdentityTheory”(1964)hearguesagainstJ.J.C.Smart’sclaimthatasuddenthoughtiscontingentlyidenticalwithabrainprocessasfollows:weattachnomeaningtodeterminingthebodilylocationofathought;so,ifxisidenticalwithyonlyifxandyoccuratthesameplaceandtime,andtheidentityiscontingent,thentherecanbenowayofestablishingthatthissamelocationconditionissatisfied.(Inthesamepaperheremarksthatthesenselessnessofthesuppositionthataseparatedbraincouldhavethoughtsorsensations“seemssoobviousthatIfindithardtotakeitseriously”(p.124).)In“FunctionalisminPhilosophyofPsychology”(1980)heimaginesMr.Asayingtohiswife,“Areyoualwaysontime?”andarguesthatoneknowsthatMr.Ameanthisutterancesarcasticallyandnotadmiringlyonlybyknowingsomethingofthepreviouscourseoftheirlivestogether,sothatthereisnowayinwhichthepres-enceoftheoneintentionratherthantheothercanbeaccountedforbysomestoryaboutneuralfiringsorelectricpotentialswithinMr.Aatthetimeofhisutterance.InConsciousnessandCausality(1984)hearguesthatmentalstateswithoutgenuinedura-tion(abilities,dispositions,intentions,beliefs)cannotbeidenticalwithbrainstateswhichdohavegenuineduration,andhearguesthat,sincehavinganintentionwithacertaincontententailshavingtheconceptsrequiredtounderstandthatcontent,itisimpossibletoidentifytheintentionwithabrainstate,becausepossessionofthosecon-ceptswould,presumablybeidentifiedwithotherbrainstatesanditisonlycontingentthattheyoccurinthesamebrain.Inhis“TheConceivabilityofMechanism”(1968),whichhasbeenmuchcitedinsubsequentdiscussionsofmentalcausation,Malcolmarguesthatacompletelymecha-nisticexplanationofapieceofhumanbehavior–oneentirelyintermsofphysicalstatesandprocessesintheorganism–isincompatiblewithanyintentionalorpurpo-siveexplanationofit.Hefindsuntenablebothofthetwowaysheseesoftryingtomain-taintheircompatibility:maintainingthatintentionalconceptscanbedefinedintermsofnon-intentionallyspecifiedbehavioraldispositionsandmaintainingthatintentionalstatesoreventsarecontingentlyidenticalwithneurophysiologicalstatesorevents.Ifallhumanbehaviorhadsufficientmechanisticcausesthen,heargues,humanbeingswouldhavenointentionsordesires.And,heobserves,therewouldthereforebeapragmaticparadoxinanyone’sassertingthatallhumanbehaviorismechanisticallyexplicable:sincetheasserter’sutterancecouldcountasanassertiononlyifhehascertainintentionsaboutit,hisassertingthiswouldconstituteacounterexampletowhatheasserts.234\nNORMANMALCOLMInspiredbyaremarkofWittgenstein’saboutdreaming,Malcolmnotoriouslyargued,inapaper“DreamingandSkepticism”(1956)andabookDreaming(1959),thatdreamscannottakeplaceduringsoundsleep,inthesenseofoccurringatdefinitetimesandhavingdefinitedurations.Heinfersthis(andthestrongerconclusionthattherecanbenomentalactivityduringsoundsleep)fromthepremisethattheconceptofsoundsleepprecludesthesubject’smanifestinganymentalactivitywhilesoundasleep.Hesaysthat,ifwefoundacorrelationbetweensomephysiologicalprocessduringsleepandreportsonawakingofdreamsandusedthatasabasisforlocatingdreamsinobjectivetime,thatwouldbetoadoptadifferentuseof“dreaming”thanwenowhave,anewmeaningfortheterm.“Asthingsare,”hesays,“thenotionsofdura-tionandtimeofoccurrencehavenoapplicationinordinarydiscoursetodreams.Inthissense,adreamisnotan‘occurrence’andthereforenotanoccurrenceduringsleep”(1956:30).MalcolmmadeasignificantcontributiontothestudyofDescartes’sphilosophyofmindintwopapers,“Descartes’ProofthatHisEssenceisThinking”(1965)and“Descartes’ProofthatHeisEssentiallyaNon-materialThing”(1975).Thefirstcon-jecturesthatDescartesarguesasfollows:“xismyessenceifitisthecasethat(a)ifIamawareofxthen(necessarily)Iamawareofmyself,and(b)ifIamawareofmyselfthen(necessarily)Iamawareofx.Thinkingsatisfiestheseconditions.Ergo,thinkingismyessence”(1977b:32).Thisargument,Malcolmsuggests,couldbeDescartes’sreasonforthinkingthathehasaclearanddistinctideaofhimselfasathingwithnocorporealcharacteristics.Malcolm’scriticismoftheargumentisthat,although(a)and(b)aretruewhen“thinking”issubstitutedforx,thisisnotbecauseofanynecessaryconnectionbetweenmyselfandthinking.Hesays:“(a)istruesolelybecausethestate-ment‘Iamnotawareofmyself’isself-defeating...(b)istruebecausetheawarenessofanythingisthinking,andalsobecauseofDescartes’doctrinethatonecannotthinkwithoutbeingawareofthinking”(1977b:36).Inthesecondpaper,Malcolm,respondingtoasuggestionfromRobertJaeger,findstextualsupportinDescartesforthefollowingargument:“IthinkIambreathingentailsIexist.IthinkIambreathingdoesnotentailIhaveabody.Therefore,IexistdoesnotentailIhaveabody.”Malcolmrejectsthesecondpremise.Itis,hesays,conceptuallyimpossibleformetoexistwithouteverhavinghadabody,orformindstoexistwithoutthereeverhavingbeenbodies,becausetheprimaryuseof“Hethinksheisbreathing”presupposesbehavioralcriteriaofitstruth(andsecondaryusesinspeakingofghostsordisembodieddeitiescouldnotexistwithoutprimaryuses).HepointsoutthatDescartescouldbehoistwithhisownpetardhere,forIambreathingentailsIexistbutdoesnotentailIamthinking.In“ThoughtlessBrutes”(1972presidentialaddresstotheEasternDivisionoftheAmericanPhilosophicalAssociation)MalcolmarguesthatthereasonDescartesclaimsthatanimalsdonothave“real”sensationsisthatheinsiststhat“whenwemeanby‘sensation’somethingotherthanmerephysiologicalprocesses,thensensation[has]propositionalcontent”andhethinks,rightly,thatpropositionalrepresentationsdonotoccurinthe“lower”animals.Malcolmcomments,“Whenweseetheenormityof[Descartes’s]exaggerationofthepropositionalinhumanlife,ourunwillingnesstoascribepropositionalthinkingtoanimalsoughtnolongertomakeusrefusetoattributetothemapanoplyofformsoffeeling,ofperception,ofrealization,ofrecognition,that235\nCARLGINETare,moreoftenthannot,nonpropositionalinthehumancase”(1972:53).Headds,“Weneedtoavoididentifyingthoughtswiththeirlinguisticexpression.Atthesametimeweshouldrejectthesuggestionthatitispossiblethatlanguage-lesscreaturesshouldhavethoughts...[F]oritismeaningfultosupposethatapersonmighthavehadathoughttowhichhegavenoexpression,onlybecausethispersonspeaksorspokealanguageinwhichthereisaninstitutionoftestifyingtopreviouslyunexpressedthoughts”(p.55).MemoryMalcolm’sworkonmemoryisfoundinhis“ThreeLecturesonMemory”(1963b)andabook,MemoryandMind(1977a).Inthefirstlecture,“MemoryandthePast,”hearguesagainstRussellthatthehypothesisthattheworldbeganfiveminutesagocom-pletewithmisleadingrecords,delusorymemories,etc.,isnot“logicallytenable.”Hismainargumentisthatalinguisticcommunitycanbesaidtohavemasteredthepasttense,andthereforemakepasttensestatementsandhavepasttensebeliefs,onlyifnotalloftheirpasttensestatementsarefalse.Healsoassertsthat,ifourapparentmemo-rieslargelyagreewitheachotherandwiththerecordsthentheapparentmemorieswouldbeverifiedastrue,and“iftheapparentmemorieswereverifieditwouldnotbeintelligibletoholdthat,nevertheless,thepasttheydescribemaynothaveexisted”(1963a:199).Inthesecondlecture,“ThreeFormsofMemory,”hedistinguishesfactualmemory(rememberingthatp),personalmemory(rememberingsomethingonepreviouslyperceivedorexperienced),andperceptualmemory(personallyrememberingsome-thingbyformingamentalimageofit).Hesaysthatwhileapersonalorperceptualmemoryalwaysentailssomefactualmemory,therecanbeafactualmemorythatdoesnotentailanyperceptualorpersonalmemory(contrarytoRussellandothers).Therecould,hesays,beapersonwholackedperceptualmemoryaltogetherbuthadmoreorlessnormalfactualandpersonalmemories,buttherecouldnotbeacreaturewewouldrecognizeasahumanbeingwhoaltogetherlackedfactualorpersonalmemory.Inthethirdlecture,“ADefinitionofFactualMemory,”hesuggeststhefollowingdefinition:“Aperson,B,remembersthatpfromatime,t,ifandonlyifBknowsthatp,andBknewthatpatt,andifBhadnotknownattthatphewouldnotnowknowthatp”(1963a:236).Concerningthethird,counterfactualconjuncthere,hesays,“Whetherornotitmakessensetopostulateaspecificbrain-stateorneuralprocessper-sistingbetweenthepreviousandthepresentknowledgethatp,suchapostulationisobviouslynotrequiredbyananalysisoftheconceptofremembering,”andguesses“thatourstrongdesireforamechanismofmemoryarisesfromanabhorrenceofthenotionofactionatadistance-in-time”(1963a:237–8).Inthebookhemaintainsthatitisanerrortothinkthatthecausalingredientinmemoryrequirestheassumptioneitherofatemporallycontinuouschainofcausationorofcausallaws.Hearguesagainsttheideathattheremustbearepresentationinrememberingandtheideathattheremustbeastructuralisomorphismbetweenanoccurrentmemory,whatisremembered,andaninterveningbrainstateorprocess:whatoneremembersofarememberedexperiencecouldnotbeenumeratedinaclosed236\nNORMANMALCOLMsetofitemsofthesortneededtomakeoutanisomorphism,anditwouldbeimpos-sibletodeviseakeyofisomorphismthatcouldprovideanyreasonableprospectforthediscoveryofaone-to-onecorrelationofcomponentelementsbetweenanymentalstateandanyneuralstate.PhilosophyofreligionMalcolm’spaper“Anselm’sOntologicalArguments”(1960)provokedconsiderablediscussion.InithesaysthatAnselmputforwardtwodifferentontologicalproofsoftheexistenceofGod.Thefirst,inProslogion2,usestheprinciplethatathingisgreaterifitexiststhanifitdoesnotexist.Thesecond,inProslogion3,usesthediffer-entprinciplethatathingisgreaterifitnecessarilyexiststhanifitdoesnotnecessar-ilyexist.Thefirstisfallaciousbecauseitisanerrortoregardexistenceasapropertyofthingsthathavecontingentexistence,butitdoesnotfollowthatitisanerrortoregardnecessaryexistenceasapropertyofGodandasaperfection.Ashortsummaryofthesecondproof:IfGodexists,Hisexistenceisnecessary;thusGod’sexistenceiseithernecessaryorimpossible;assumingthattheconceptofGodisnotself-contradictoryorinsomewaylogicallyabsurd,itfollowsthatHenecessarilyexists.Malcolmremarks,“Ishouldthinkthereisnomoreapresumptionthat[theconceptofGod]isself-contradictorythanistheconceptofseeingamaterialthing.Bothconceptshaveaplaceinthethinkingandthelivesofhumanbeings”(1963a:160).BibliographyofMalcolm’sworkAcompletelistofMalcolm’sarticlespublishedthrough1981maybefoundinCarlGinetandSydneyShoemaker(eds.)KnowledgeandMind(Oxford:OxfordUniversityPress,1983).ThecollectionWittgensteinianThemes(1995)containsfourteenofhisessayswritteninthelasttwelveyearsofhislife.1942:“CertaintyandEmpiricalStatements,”Mind51,pp.18–46.1949:“DefendingCommonSense,”PhilosophicalReview58,pp.201–21.1950:“TheVerificationArgument,”inPhilosophicalAnalysis,ed.M.Black,Ithaca,NY:CornellUniversityPress.(ReprintedwithrevisionsandadditionalfootnotesinMalcolm1963a.)1951:“PhilosophyforPhilosophers”(intendedtitle:“PhilosophyandOrdinaryLanguage”),PhilosophicalReview60,pp.329–40.1956:“DreamingandSkepticism,”PhilosophicalReview65,pp.14–37.1959:Dreaming,London:RoutledgeandKeganPaul.1960:“Anselm’sOntologicalArguments,”PhilosophicalReview69,pp.41–60.(ReprintedwithnewfootnotesinMalcolm1963a.)1963a:KnowledgeandCertainty,EnglewoodCliffs,NJ:Prentice-Hall.1963b:“ThreeLecturesonMemory,”(“MemoryandthePast,”“ThreeFormsofMemory,”and“ADefinitionofFactualMemory”),inMalcolm1963a.(“MemoryandthePast”firstpublishedinTheMonist45(1962),pp.247–66.)1964:“ScientificMaterialismandtheIdentityTheory,”Dialogue3,pp.115–25.1965:“Descartes’ProofthatHisEssenceisThinking,”PhilosophicalReview74,pp.315–38.(ReprintedinMalcolm1977b.)1968:“TheConceivabilityofMechanism,”PhilosophicalReview77,pp.45–72.237\nCARLGINET1971:ProblemsofMind,NewYork:HarperandRow.1972:“ThoughtlessBrutes,”Presidentialaddress,ProceedingsoftheAmericanPhilosophicalAssociation46,pp.5–20.(ReprintedinMalcolm1997b.)1975:“Descartes’ProofthatHeisEssentiallyaNon-materialThing,”PhilosophyForum14(1975).(ReprintedinMalcolm1977b.)1976:“MooreandWittgensteinontheSenseof‘Iknow’,”inEssaysinHonourofG.H.vonWright,inActaPhilosophicaFennicaed.JaakkoHintikka,28,1–3,pp.216–40.1977a:MemoryandMind,Ithaca,NY:CornellUniversityPress.1977b:ThoughtandKnowledge,Ithaca,NY:CornellUniversityPress.1980:“FunctionalisminPhilosophyofPsychology,”ProceedingsoftheAristotelianSociety,newseries80,pp.211–29.1984:ConsciousnessandCausality:ADebateontheNatureofMindwithD.M.Armstrong,Oxford:BlackwellPublishers.1986:Wittgenstein:NothingisHidden,Oxford:BlackwellPublishers.1994:Wittgenstein:AReligiousPointofView?,ed.witharesponsebyP.Winch,Ithaca,NY:CornellUniversityPress.1995:WittgensteinianThemes:Essays1978–1989,ed.G.HenrikvonWright,Ithaca,NY:CornellUniversityPress.238\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200119WilfridSellars(1912–1989)JAYF.ROSENBERGLifeandworkHadWilfridStalkerSellarsneverwrittenanoriginalphilosophicalword,hisaccom-plishmentsasaneditorwouldlikelybesufficienttoearnhimaplaceofhonorinthehistoryofpostwaranalyticphilosophy.In1950,heandHerbertFeiglco-foundedPhilosophicalStudies,thefirstscholarlyjournalexplicitlydevotedtoanalyticphilosophy,whichtheyeditedjointlyuntil1971andSellarstheneditedaloneforthreemoreyears.Ayearearlier,FeiglandSellarshadalreadypublishedaseminalanthology,ReadingsinPhilosophicalAnalysis;ReadingsinEthicalTheory,co-editedbySellarsandJohnHospers,followedthreeyearslater.The“philosophicalanalysis”representedinthesevolumeshadbeentransplantedfromitsoriginsandearlydevelopmentatCambridgeandOxfordandenrichedbygenerouscross-fertilizationfromthelogicalempiricismofanexpatri-ateViennaCircle,mostnotablybytheworkofRudolfCarnap,andindigenousstrainsofpragmatism,criticalrealism,andevolutionarynaturalism.Fromsuchseeds,the“analytic”styleofphilosophizinganditsagendaofproblemsgrewtodominateAmericanacademicphilosophy,definitivelychangingitsintellectuallandscape.The“logico-linguisticturn”becamethenewmethodologicalcenterofphilosophicalinquiry,andregionalphilosophiesoflogic,language,mind,andsciencefirstjoinedandthengraduallybegantosupplantmorebroadly-conceivedtraditionalmetaphysicalandepistemologicalstudies,whilenormativeethicalinquiriesgavegroundtoissuesinmetaethicsandmoralpsychology.ButWilfridSellarsinfactwrotemanyanoriginalandimportantphilosophicalword,andsonotonlyhelpedtostimulatethegrowthofpostwaranalyticphilosophy,butalsobecameoneofitsmostdistinguishedandinfluentialpractitioners.Hisacademictra-jectorytookhimfromstudiesattheUniversityofMichiganand,asaRhodesScholar,atOxfordUniversity,tofacultypositionsattheUniversityofIowa,theUniversityofMinnesota,whereheservedasChairduringthemid-1950s,andYaleUniversity,beforehebecameUniversityProfessorofPhilosophyandResearchProfessorofthePhilosophyofScienceattheUniversityofPittsburgh,apositionwhichheheldfrom1963untilhisdeathin1989.Hisintellectualtrajectorymeanwhilecarriedhimfromanearlyperiod,duringwhichheworkedouthisfundamentalphilosophicalideasinovertwodozendialectically-challengingessays,throughamiddleperiodcharacterizedbythe239\nJAYF.ROSENBERGdevelopmentandexpositionofasystematicphilosophicalvisionofremarkablescopeanddepth,intoalateperiodofconsolidation,refinement,anddeepeningofmaturethesesandinsightsthatweresimultaneouslycomingtobemorefullyappreciatedandexplicitlyappropriatedbyanewphilosophicalgeneration.ThecritiqueofgivennessSellars’srevolutionary1956essay“EmpiricismandthePhilosophyofMind,”immedi-atelyacknowledgedasacontemporaryclassic,marksthebeginningofhisexception-allyproductiveandinfluentialmiddleperiod.(Thiscanbetreated,somewhatarbitrarily,asculminatingin1972withthepublicationofhis1970AmericanPhilosophicalAssociationEasternDivisionPresidentialAddressontheKantiantext,“thisIorheorit(thething)whichthinks.”)Thecentralthemeof“EmpiricismandthePhilosophyofMind”isathoroughgoingandgeneralcritiqueofwhatSellarsfamouslycalledthe“mythofthegiven,”aperennialandpolymorphicphilosophicalmotifmanifestedinteraliaintheidea,characteristicofclassicalsense-datumtheory,thatempiricalknowl-edgerestsonafoundationof“immediateawarenesses”andontheassumptionthatthe“privacy”ofthementalandone’s“privilegedaccess”toone’sownmentalstatesareprimitivefeaturesofexperience,logicallyandepistemologicallypriortoallintersub-jectiveconceptspertainingtoinnerepisodes.Sellarscriticizedsense-datistandothertraditionalepistemicfoundationalismsforfailingproperlytodistinguishnon-conceptualstatesofsensingfromconceptuallystructuredperceptualtakings.Perceivingalwaysinvolvestakingsomethingsensorilypresenttobethisorthat,andso,asKantrecognized,hasajudgmentalformwhichmobilizesandapplies,correctlyorincorrectly,descriptiveandclassificatoryconcepts.Perceptionproperlyso-calledisconsequentlyanormativebusiness,andtheabilitytoengageinitrequiresmorethanreliabledifferentialdispositionstorespondtosensorystimuli.“Theessentialpoint,”Sellarswrote,“isthatincharacterizinganepisodeorastateasthatofknowing,wearenotgivinganempiricaldescriptionofthatepisodeorstate;weareplacingitinthelogicalspaceofreasons,ofjustifyingandbeingabletojustifywhatonesays”(1963a:169).Aperceptualjudgmentmayconsequentlybedirectorimmediateinthesenseofbeingaunmediatedresponsetostimulation,i.e.notitselfinferredfromotherjudgments,buttheepistemicauthorityofsuchjudgmentsdependsupontheirbeingappropriatelycaughtupintheintersubjectivegameofhavingandgivingreasons,andsocannotbeindependentoftheirinferentialrelationshipstootherjudgments.Sellarsthusnotoriouslyadvocatedastrongepistemicinternalism,accordingtowhich“observationalknowledgeofanyparticularfact...presupposesthatoneknowsgeneralfactsoftheformXisareliablesymptomofY”(1963a:168).EpistemicauthorityConsistentwithhisstronginternalism,Sellarsinterpretedevenfirst-personepistemicauthoritywithrespecttothesensoryaspectsofone’sownexperienceasbuiltonandpresupposinganintersubjectivestatusforsensoryconceptsperse.Correlatively,hedeci-sivelyrejectedtheideathatsensoryconsciousnesssuppliesaformofempiricalknowl-edgethat(1)isimmediate(i.e.non-inferential);(2)presupposesnoknowledgeofother240\nWILFRIDSELLARSmattersoffact,particularorgeneral;and(3)constitutestheultimatecourtofappealsforallfactualclaims(1963a:164).Althoughapersoncanjustifiablybelieveanempiri-caltruthwithouthavinginferreditfromotherpropositions,noempiricalbeliefsareself-justifying,self-warranting,orself-authenticating.Instead,Sellarsargued,“tosaythatsomeonedirectlyknowsthat-pistosaythathisrighttotheconvictionthat-pessen-tiallyinvolvesthefactthatthe[thought]that-poccurredtotheknowerinaspecificway”(p.188).Theepistemicauthorityofanon-inferentialperceptualbeliefcanbetracedtothefactthat,inthecourseoflearningperceptuallanguage,thebelieverhasnotonlyacquiredpropensitiesforthereliableuseoftherelevantconceptsinperceptualsitua-tionsbutalsohascometoknowwhatisinvolvedinlearningtouseperceptualsen-tencesreliablyinperceptualcontexts.Thus,assumingthathehasmasteredtheuseoftherelevantwordsinsuitableperceptualsituations,apersonwhocandidlyandspon-taneouslythinks-out-loud“Lo!Hereisaredapple”–Sellars’scustomarymodelofaperceptualtaking–isjustifiedinreasoning:Ijustthought-out-loud“Lo!Hereisaredapple”(nocountervailingconditionsobtain);so,thereisgoodreasontobelievethatthereisaredappleinfrontofme.(1975d:341–2)This“trans-level”reasoning,asSellarscalledit,doesnothavetheoriginalperceptualjudgmentasitsconclusion,butisratheraninferencefromthecharacterandcontextoftheoriginalnon-inferentialexperiencetotheexistenceofagoodreasonforaccept-ingitasveridical.CentraltoSellars’sthoroughgoingepistemicinternalism,indeed,ishisconvictionthatthereasonablenessofacceptingevenfirstprinciplesisamatteroftheavailabilityofgoodargumentswarrantingtheiracceptance.Whatisdefinitiveoffirstprinciples(FP)istheunavailabilityofsoundargumentsinwhichtheyarederivedasconclusionsfromstill-more-basicpremises,thatis,argumentsoftheform:(A1)P1,P2,...,PnFP.Here,too,Sellarsappealstothenotionofa“trans-level”justificatoryinference,point-ingoutthattheabsenceofgoodargumentsoftheform(A1)isentirelycompatiblewiththeexistenceofsoundargumentsoftheform(A2)P1,P2,...,PmItisreasonabletoacceptFP.Sellarsinterpretedtheconclusionof(A2)asaclaimtotheeffectthataparticularcourseofepistemicconduct,acceptingtheprincipleFP,canbesupportedbyadequatereasons.Thissuggestedthein-principleavailabilityofyetanotherargument,specifi-callyapieceofsoundpracticalreasoning,whoseconclusionexpressesanintentiontoengageinsuchconduct:(A3)IshallachievedesirableepistemicendE.AchievingEimpliesacceptingprinciplesofkindK.TheprincipleFPisofkindK.Therefore,IshallacceptFP.OnSellars’sview,suchpatternsofpracticalreasoningalsogovernthewarrantedacceptanceoflawlikegeneralizationsandtheoreticalsystems.Adoptingasystematic241\nJAYF.ROSENBERGtheoreticalframeworkisultimatelyjustifiedbyappealtotheepistemicendof“beingabletogivenon-trivialexplanatoryaccountsofestablishedlaws”(1975c:384),andadoptingnomologicalclaimsthatprojecttheobservedstatisticalfrequency(includingafrequency=1)ofsomepropertyinanopenclasstofurtherunobservedfinitesamplesoftheclass,isultimatelyjustifiedbyappealtotheepistemicendof“beingabletodrawinferencesconcerningthecompositionwithrespecttoagivenpropertyYofunexam-inedfinitesamples...ofakindX,inawaywhichalsoprovidesanexplanatoryaccountofthecompositionwithrespecttoYofthetotalexaminedsample,K,ofX”(1975c:392).Whatiscrucialisthattheseendsconcerntherealizingofalogicallynecessaryconditionofbeingintheframeworkofexplanationandprediction,i.e.beingabletodrawinferencesconcerningtheunknownandgiveexplanatoryaccountsoftheknown.(1975c:397)Sellarsargued,inshort,thatinductivereasoningdoesnotneedtobevindicated,thatis,showntobetruth-preserving,butisratheritselffundamentallyaformofvindica-tivereasoning,justifyingourengagingindeterminateepistemicconducts.Itsends-in-viewmustconsequentlybeofasortthatcanbeknowntoberealizedorobtain.UnlikesuchReichenbachianendsasbeinginpossessionoflimit-frequencystatementswhicharewithinacertaindegreeofapproximationofthetruth(wheresuchlimitsexist),theendsofbeinginpossessionofexplanatorylawsandprinciplesenvisionedbySellarssatisfedthatconstraint.Self-knowledgeSellarsfamouslyengagedtheCartesianpictureofdirectandincorrigibleself-knowledgewithhis“mythofJones,”astorysetinafictionalcommunitywhosemembersspeakahypotheticalsophisticated“Ryleanlanguage.”Thefundamentaldescriptivevocabularyofthislanguagepertainstopublicspatiotemporalobjects,andwhileitincludeslogicaloperators,subjunctiveconditionals,andeventhefundamentalresourcesofsemanticdiscourse(enablingitsspeakerstosayoftheirpeers’utterancesthattheymeanthisorthat,standinvariouslogicalrelationstooneanother,andaretrueorfalse),itneverthelesslacksanyresourcesforspeakingofinnerepisodes,whetherthoughtsorexperiences.Inthiscommunity,then,agenius,Jones,developsatheoryaccordingtowhichovertutter-ancesarebuttheculminationofaprocesswhichbeginswithcertaininnerepisodes....[His]modelfortheseepisodeswhichinitiatetheeventswhichculminateinovertverbalbehavioristhatofovertverbalbehavioritself.(1963a:186)Jones’stheoryearnsitsepistemiccredentialsbyenablinghim,andhisfellowRyleanswhomasterit,successfullytoexplainandanticipateintelligentbehaviorconductedinsilence,thatis,unaccompaniedbyovertverbalepisodesofthesortthatwewouldrecognizeasexpressinganagent’sconduct-rationalizingbeliefsanddesires.ThenewidiomsofJones’stheory,forexample,“isthinking‘...’,”initiallyhaveapurelytheoreticaluse,beingascribedonthebasisofinferencesfromobservable242\nWILFRIDSELLARSbehaviorinobservablecircumstances.Butcrucially,Sellarsargued,whatbeginaspurelytheoreticaltermscanacquireafirst-personreportingrole.ForitturnsouttobepossibleforJonestotrainhiscompatriots,inessencebyaprocessofSkinnerianoperantconditioning,torespondreliablyanddirectly(i.e.non-inferentially)totheoccurrenceofsuchan“innerepisode”withajudgmenttotheeffectthatitisoccurring.Thatis,theycanrespondtoonethoughtwithasecond(meta-)thoughttotheeffectthattheyarethinkingit;thismatchesthedefactophenomenologyoffirst-person“privilegedaccess”sufficientlytoaccountfortheCartesianillusionofmentaltransparency.TheJoneseanstorythusshowshowtheessentialintersubjectivityoflanguagecanbereconciledwiththe“privacy”ofinnerepisodes.ScientificrealismSellars’snovelappealtoformsoftheoreticalreasoninginhismythofJonesreflectedhisbroaderphilosophicalconcernwiththenature,structure,androleoftheoriesinthenaturalsciences.Onthereceived,positivist,viewexplanationwasidentifiedwithderivation.Singularmattersofempiricalfactweretobeexplainedbyderivingdescrip-tionsofthemfrom(“inductive”)empiricalgeneralizations(alongwithappropriatestatementsofinitialconditions),andthese“empiricallaws”inturnweretobeexplainedbyderivingthemfromtheoreticalpostulatesandcorrespondencerules.Onthepositivistview,inconsequence,theories(e.g.microtheories)explainobservationalmattersoffactonlyindirectly,byimplyingthe(observation-language)generalizationsthatexplainthemdirectly.Sellars,incontrast,arguedthatthis“levelspicture”oftheorieswasfundamentallymisleading.Theoriesdonotexplainlawsbyentailingthem.Rather,“theoriesexplainlawsbyexplainingwhytheobjectsofthedomaininquestionobeythelawsthattheydototheextentthattheydo”(1963c:123).[Thatis,]theyexplainwhyindividualobjectsofvariouskindsandinvariouscircum-stancesintheobservationframeworkbehaveinthosewaysinwhichithasbeeninduc-tivelyestablishedthattheydobehave.Roughly,itisbecauseagasis...acloudofmoleculeswhicharebehavingincertaintheoreticallydefinedways,thatitobeystheempiricalBoyle–CharlesLaw.(1963c:121)OnSellars’sview,then,“theoreticalentities”arenotmerelyconvenientfictions,enablingustoabbreviatecomplicatedandunwieldystoriesaboutentitiesthatwehavegood,(observational)reasonstobelieveactuallyexist.Theoreticalentitiesareratherthoseentitieswejustifiedlybelievetoexistforgoodandsufficienttheoreticalreasons.Sellarsthusadvocatedarobustlyrealisticepistemologyofscientificinquiryand,cor-relatively,anunderstandingofitsultimateoutcomesasontologicallydefinitive:“Inthedimensionofdescribingandexplainingtheworld,scienceisthemeasureofallthings,ofwhatisthatitis,andofwhatisnotthatitisnot”(1963a:173).(Thisishis“scientiamensura.”)Scientifictheoriesindeedexplanatorily“savetheappearances,”buttheydosopreciselybydescribingtherealityofwhichtheappear-ancesareappearances.Thisrobustrealism,combinedwithathoroughgoing243\nJAYF.ROSENBERGnaturalism,infactsetSellars’smetaphilosophicalagendaforpostwaranalyticphilosophyperse.MetaphilosophicalviewsSellarssawcontemporaryphilosophyasconfrontedbytwo“images,”eachofwhichpurportedtobeacompletepictureofman-in-the-world,whichneedtobeunifiedintoasinglesynopticvision.The“manifestimage”istheconceptionoftheworldandtheplaceofpersonsinitthathasdescendedfromthegreatspeculativesystemsofancientGreece,throughthedialecticsofa“perennialphilosophy,”tothedimensionsofcon-temporaryAnglo-Americanthoughtthatemphasize“ordinaryusage”and“commonsense.”Itsprimaryobjectsarepersons,beingswho,interalia,reflectivelyconceiveofthemselvesasbeingintheworldbothassentientperceiversandcognitiveknowersofit,andasagentscapableofaffectingitthroughdeliberateandrationalelectiveconduct.Incontrast,the“scientificimage”isthecomplexnewunderstandingofman-in-the-worldthatisstillintheprocessofemergingfromthefruitsoftheoreticalreasoning,inparticular,fromtheprocessesofpostulationaltheoryconstruction.Althoughthisimageis“methodologicallydependentontheworldofsophisticatedcommonsense,”Sellarsarguesthatitpurportstobeacompleteimage,i.e.todefineaframeworkwhichcouldbethewholetruthaboutthatwhichbelongstotheimage.Thusalthoughmethodologicallyadevelop-mentwithinthemanifestimage,thescientificimagepresentsitselfasarivalimage.Fromitspointofviewthemanifestimageonwhichitrestsisan“inadequate”butpragmaticallyusefullikenessofarealitywhichfirstfindsitsadequate(inprinciple)likenessinthescientificimage.(1963c:57)Theleadingchallengeforcontemporaryphilosophy,heconcluded,istoshowhowtheinevitabletensionsbetweenthesetwoimagescanberesolvedbya“stereoscopicunder-standing”inwhichtheycometobe“fused”intoasinglesynopticvisionofman-in-the-world.Sellars’sphilosophyisusefullyviewedasafullerarticulationofthiscon-frontationofthetwoimagesandadetaileddialecticalengagementwiththephilo-sophicalagendatowhichitgivesrise:thatplacesbefoundwithinthecontextofathoroughgoingnaturalismfortheintentionalcontentsoflanguageandthought,forthenormativedimensionsofknowledgeandaction,andforthesensuouscontentsofperceptionandimagination.Consonantwithsuchanaturalism,thesoughtsynopticstorymustfindaplaceformindwithoutassigninganindependent,autonomous,andirreducible,ontologicalstatustointentionalstatesorentities,anditmusteschewanyontologicalviewofabstractaasrealthatfailstodeliveranadequateaccountoftheirrolewithinthecausalorder,broadlyconstrued.SemanticmeaningThecenterpieceofSellars’sresponsetobothofthesenaturalisticchallengeswasasophisticatedtheoryofconceptualroles,concretelyinstantiatedintheconductsoflin-guisticcommunitiesandsociallytransmittedbymodesofculturalinheritance.Atthe244\nWILFRIDSELLARSheartofthistheorywasanincreasinglyrefinedaccountofmeaningasfunctionalclassification,moreprecisely,ofthe“meaning”idiomas,inthefirstinstance,acontextoftranslationintermsofwhichstructurallydistinct“natural-linguisticobjects”(e.g.utteringsorinscribings)areclassifiedintermsoftheirrolesorfunctionsvis-à-vistheorganizedbehavioraleconomiesoffamiliesofspeakingorganisms.Inshort,Sellarsinterpreted“means”asaspecializedformofthecopula,tailoredtometalinguisticcon-texts,accordingtowhichtherightsideofthesuperficiallyrelationalform“–means...”isalsoproperlyunderstoodasmentioningorexhibitingalinguisticitem.Ordinaryquotation,arguedSellars,suffersfromasystematicambiguityregardingthecriteriaaccordingtowhichlinguistictokensarecorrectlyclassifiableasbelongingtothisorthatlinguistictype.Hethereforeintroducedthestraightforwarddeviceoftwoseparatestylesofquotationmarks–star-quotesanddot-quotes–todifferentiatebetweentwodifferentwaysofsortingandindividuatinglexicalitems.Star-quotesformcommonnounsthataretrueofitemsbelongingtothespatiotemporalcausalorder(“tokens”)whichareappropriatelystructurallyisomorphictothetokensexhibitedbetweenthem,whiledot-quotesformcommonnounstrueoftokensthat,insomespec-ifiedlanguage,areappropriatelyfunctionallyisomorphicto(i.e.playtheroleordothejobperformedby)thetokensexhibitedbetweentheminourlanguage.Sellarsthenproposedtotranscribesuchsemanticclaimsas(1s)(InFrench)“rouge”meansred,and(2s)(InGerman)“Esregnet”meansitisraining,bythemoreperspicuousformulations••(1*)(IntheFrenchlinguisticcommunity)*rouge*sareredsand••(2*)(IntheGermanlinguisticcommunity)*Esregnet*sareitisrainings.RolesandrulesTosortandclassifydescriptivelyindividuatedfamiliesofnaturalitemsintermsoftheirlinguisticjobs,roles,orfunctions,istosortandclassifythemnormatively.Sellars’soverallphilosophicalagendathusrequiredacomplementarythoroughlynaturalisticaccountofthenormativedimensionoflanguage,andheindeedofferedone,basingitonthenotionofwhathecalledpattern-governedbehavior.Thebasicconceptofpattern-governedbehavioristheconceptofbehaviorwhichexhibitsapattern,notbecauseitisbroughtaboutbytheintentionthatitexhibitthispattern,butbecausethepropensitytoemitbehaviorofthepatternhasbeenselectivelyreinforced,andthepropensitytoemitbehaviorwhichdoesnotconformtothispatternselectivelyextinguished.(1974:423)Pattern-governedbehaviorcanarisefromprocessesofnaturalselectiononanevolu-tionarytime-scaleasacharacteristicofaspecies,forexample,thedanceofthebees,245\nJAYF.ROSENBERGbutitcanalsobedevelopedinindividuals,“trainees,”bydeliberateandpurposiveselec-tiononthepartofotherindividuals,the“trainers.”Inthisconnection,Sellarsintro-ducedadistinctionbetween“rulesofaction”and“rulesofcriticism.”Rulesofactionspecifywhatsomeoneoughttodo,forexample,“Ceterisparibus,oneoughttosaysuchandsuchifincircumstancesC.”Theycanbeefficaciousinguidinglinguisticactivityonlytotheextentthattheirsubjectsalreadypossesstherelevantcon-cepts,suchasconceptsof“sayingsuch-and-such,”of“beingincircumstancesC,”and,indeed,ofobeyingarule(i.e.doingsomethingbecauseitisenjoinedbyarule).Rulesofcriticism,incontrast,specifywhatoughttobethecase,forexample,“Westminsterclockchimesoughttostrikeonthequarterhour”(1975d:95).Theitemswhoseperformancesmaylegitimatelybeappraisedaccordingtosuchrules,however,neednotthemselveshavetheconceptofarulenor,indeed,evenbecapableofhavinganyconceptsatall.Thustrainerscanbeunderstoodinthefirstinstanceasactinginaccordancewithrulesofconductwhoseauthorityderivesfromrulesofcriticism,thatis,asaimingatbringingaboutthepattern-governedbehaviorswhichtheirtrainees’conductoughttomanifest:“Patterned-behaviorofsuchandsuchakindoughttobeexhibitedbytrainees,hencewe,thetrainers,oughttodothisandthat,aslikelytobringitaboutthatitisexhibited”(1974:423).Iftrainingissuccessful,then,inconsequenceoftheconductsoftrainersundertheguidanceofsuchrulesofaction,thebehaviorofalanguage-learnercancometoconformtotherelevantrulesofcriticismwithouthis,inanyothersense,graspingthemhimself.[The]membersofalinguisticcommunityarefirstlanguagelearnersandonlypotentially“people,”butsubsequentlylanguageteacherspossessedoftherichconceptualframeworkthisimplies.Theystartoutbeingthesubjectmatteroftheought-to-be’sandgraduatetothestatusofagentsubjectsoftheought-to-do’s.(1975d:100)Themodesofpattern-governedbehaviorrelevanttosemanticmeaningand,corre-spondingly,therelevantfamiliesofrulesofcriticism,Sellarsproposed,arethreefold:1LanguageEntryTransitions:Itoughttobethecasethatspeakersrespondtoobjectsinperceptualsituationsandtocertainstatesofthemselveswithappropriatelinguistic-conceptualactivity.2Intra-linguisticMoves:Itoughttobethecasethatspeakers’linguistic-conceptualepisodestendtooccurinpatternsofvalidinference(theoreticalandpractical,formalandmaterial),andtendnottooccurinpatternswhichviolatelogicalprinciples.3LanguageDepartureTransitions:Itoughttobethecasethatspeakersrespondtosuchfirst-personlinguistic-conceptualepisodesas“Ishallnowraisemyhand”withanupwardmotionofthehand,etc.(Cf.1974:423–4)Thesetransitionsarerespectivelythecoreelementsofperceptualtakings,inferences,andvolitions,and,althoughtheyareacts(quabothactualitiesandactualizationsofbehavioraldispositions),Sellarsinsistedthattheyarenotthemselvesactions.Theyareacquiredas,andremain,pattern-governedactivities,butformthebasisoflinguistic246\nWILFRIDSELLARSactionsproperas“thetraineeacquiresnotonlytherepertoireofpattern-governedlin-guisticbehaviorwhichislanguageaboutnon-linguisticitems,butalsothatextendedrepertoirewhichislanguageaboutlinguisticaswellasnon-linguisticitems”andthusbecomes“abletoclassifyitemsinlinguistickinds,andtoengageintheoreticalandpracticalreasoningabouthislinguisticbehavior”(1974:425).Linguisticrolesorfunctionsareindividuatedintermsofthestructureofpositiveandnegativeuniformitiesgeneratedinthenaturalorderbythesepattern-governedactivitiesofperception,inference,andvolition.AnalyzingsamenessoflinguisticroleassamenessofplaceinthecomplexrelationalstructuregeneratedbyconductsthathavebeencausallyshapedinthesewaysbysystemsofespousedlinguisticnormsequippedSellarswithafunctionalconceptionofsemantics.Thisconceptionneitherpresupposednorunavoidablyledbackintothedomainsofeitherontologicalabstractaorirreduciblyintentionalmentalentities.TheintentionalityofthoughtInstead,Sellars’salternativeaccountofthedistinctiveintentionalityofthoughtwasitselfdrawnintermsoftheformsandfunctionsofnaturallinguisticitems,modeledbywhathecametocall“verbalbehaviorism”(VB):AccordingtoVB,thinking“that-p,”wherethismeans“havingthethoughtoccurtoonethat-p,”hasasitsprimarysense[aneventof]saying“p”;andasecondarysenseinwhichitstandsforashorttermproximatepropensity[disposition]tosay“p.”(1974:419)TherootsofSellars’smatureverbalbehaviorismreachbacktothemythofJonesin“EmpiricismandthePhilosophyofMind.”Thethought-episodespostulatedbyJonesonthemodelofovertverbalbehaviorareintroducedbyapurelyfunctionalanalogy.Theconceptofanoccurrentthoughtisnotthatofsomethingencounteredpropriapersonabutratherthatofacausally-mediatinglogico-semanticrole-player,whosedeterminateontologicalcharacterisinitiallyleftentirelyopen.SinceonSellars’saccounttheconceptofathoughtisfundamentallytheconceptofafunctionalkind,noontologicaltensionisgeneratedbytheidentificationofitemsbelongingtothatfunctionalkindwithstatesandepisodesofanorganism’scentralnervoussystem.Themanifestimage’sconceptionofpersonsasthinkerscanconse-quentlyfusesmoothlywiththescientificimage’sconceptionofpersonsascomplexmaterialorganismshavingadeterminatephysiologicalandneurologicalstructure.Sellars’sconvictionthatthefundamentalcharacteristicofsemanticdiscourseisitsine-liminableappealtofunctionalconsiderations,andhiscorrelativepioneeringanalysesoftheintentionalcategoriesofthementalintermsofepistemologicallytheoreticaltranspositionsofthesemanticcategoriesofpubliclanguageearnhimthetitleofthefirstfunctionalistincontemporaryanalyticphilosophyofmind.Thisfunctionalistviewisonewhoseimplicationsandinfluencehavenotyetbeguntobeexhausted.CategorialontologyTheparallelsbetweensemanticdiscourseandtheclassicalontologicalidiomsofPlatonisticdiscourse,ostensiblydesignatingabstractentities,havenotgoneunnoticed.247\nJAYF.ROSENBERGConsistentwithhisglobalcommitmenttonaturalism,Sellarsexploitedtheseparallelstoconstructhisownuniquevariantoflinguisticnominalism,aviewaccordingtowhichtheabstractentitieswhicharethesubjectmatterofthecontemporarydebatebetweenplatonisticandanti-platonisticphilosophers–qualities,relations,classes,propositions,andthelike–arelinguisticentities.(1967a:229)Infirstapproximation,Sellarsproposedtoanalyzetheostensiblyrelationalontologicalclaims(1o)(TheFrenchword)“rouge”standsforredness,and(2o)(TheGermansentence)“Esregnet”expressesthepropositionthatitisraining,preciselyashehadanalyzedthecorrespondingsemanticclaims(1s)and(2s).Thesetwowillbeanalyzedas(1*)and(2*),claimsthatspecifythefunctionalrolesofdeter-minatefamiliesofstructurally-individuatedtokens.Thisstrategyofunderstandingtraditionalontologicaldiscourseasclassificatorydiscoursewithinafunctionalmeta-languagetransposedintothe“materialmodeofspeech”hadbeenpioneeredbyCarnap(seeCARNAP).ButunlikeCarnap,Sellarsrefusedtoidentifytheformallydefinablecon-structsofa“pure”syntaxorsemanticswiththecorrespondingsyntacticalandseman-ticaltermsineveryday,pre-philosophicalusage.Suchafacileinterpretationoftherelationshipbetween“pure”and“descriptive”syntacticandsemanticdiscourses,heargued,failstodoproperjusticetotheessentialnormativedimensionofthelatter.Thus,whileSellarswaspreparedtointerpretsuchparadigmaticontologicaltermsas“universal,”“individual,”“kind,”“quality,”“proposition,”and“fact”byappealingtosyntacticandsemanticcounterparts(e.g.“predicate,”“singularterm,”“commonnoun,”“monadicpredicate,”“sentence,”and“truesentence”)heinsistedthatthesesyntacticandsemantictermshaveaconceptualrolewhichisnomorereducibleto[non-syntacticaland]non-semanticalrolesthantheroleofprescriptivetermsisreducibletonon-prescriptiveroles....[The]empirical(inthebroadsense)characterofstatementsindescriptive(historical)[syntaxand]semanticsdoesnotentailthat[syntacticaland]semanticalconcepts,properlysocalled,aredescriptive.(1975a:459)LikeSellars’saccountofthedistinctiveintentionalityofthought,then,hisaccountofdiscourseostensiblyabouttheentitiesandcategoriesofclassicalontologyisalsodrawnintermsoftheformsandfunctionsofnaturallinguisticitems.“Abstractentities,”too,consequentlyconstitutenoobstacletothesoughtfusionofthemanifestandscientificimages.SensationsSurprisinglyitiswhenSellarsturnsfromintentionalthoughtandontologicalabstractatosensationsthatsignificantcomplicationstohissynopticprojectfirstcomeintoview.248\nWILFRIDSELLARSLikeKant,SellarsrejectedtheCartesianpictureofasensory-cognitivecontinuum.The“of-ness”ofasensation–e.g.itsbeingofaredtriangleorofasharpshootingpain–heinsisted,isnottheintentional“of-ness”(“aboutness”)ofathought.“The‘rawness’of‘rawfeels’,”hewrote,“istheirnon-conceptualcharacter”(1967c:376).Consequently,althoughSellars’sepistemologicalstoryaboutsensationsalsobeginswithastrategicappealtotheuniqueepistemicstatusofpostulatedtheoreticalentities,hisaccountoftheontologyofsensationsdivergessignificantlyfromhissemanticandfunctionalistaccountofintentionalthoughts.InafinalepisodeoftheMythofJonesthehero...postulatesaclassofinner–theoreti-cal–episodeswhichhecalls,say,impressions,andwhicharetheendresultsoftheimpingementofphysicalobjectsandprocessesonvariouspartsofthebody.(1963a:191)Jonespostulatesimpressionsaselementsofanexplanatoryaccountoftheoccurrenceinvariouscircumstancesofperceptualcognitionswithdeterminatesemanticcontents.Inthiscase,however,themodelforJones’stheoryisnotfunctionally-individuatedfam-iliesofsentencetokens,butrather“adomainof‘innerreplicas’which,whenbroughtaboutinstandardconditionssharetheperceptiblecharacteristicsoftheirphysicalsources”(1963a:191).Althoughtheentitiesofthismodelareparticulars,theentitiesintroducedbyJones’stheoryarenotparticularsbutrathernon-conceptual(non-intentional)statesofaperceivingsubject.Thus,whiletalkofthe“of-ness”ofsensa-tions,likethatofthe“of-ness”ofthoughts,isfundamentallyclassificatory,theclassi-ficationofsensationsisnotfunctional,butratherbasedonanalogiesthatareinitiallyextrinsicandcausal,andultimatelyintrinsicandcontentive.Inthefirstinstance,then,theconceptofaperson’shavinganof-a-red-trianglesen-sation–anadjectivalidiomcontrivedtohighlighttheclassificatoryroleof“of-ness”–ortheconceptofaperson’ssensing[redtriangle]Sly–anadverbialidiomcontrivedtoreflectthefactthat“sensation”isa“verbalnoun”–istheconceptofherbeinginthesortofstatethatisbroughtaboutinnormalperceiversinstandardconditionsbytheactionofredtriangularobjectsontheeyes.Thepointofthemodelof“innerreplicas,”however,istoinsistthatsuchstatescandischargetheirexplanatoryjobsinrelationtocognitiveperceptualtakings(especiallynon-veridicalperceptualjudgments)onlyiftheyareconceivedashavingthemselvesdeterminateintrinsiccharactersand,inpar-ticular,asresemblinganddifferingfromothersensorystatesinamannerformallyanal-ogoustothewayinwhichobjectsofthe“replica”model(e.g.“wafers”ofvariouscolorsandshapes)areconceivedtoresembleanddifferfromoneanother.Sellarsproceededtodevelopthiscoreaccountofsensationsintwodifferentdirec-tions,inconsequenceofwhichithascometoberegardedasoneofthemostdifficultandcontroversialaspectsofhisphilosophy.Thefirstlineofdevelopmentturnedonhisconclusionthatthefundamentalconceptpertainingtocolorwithinthemanifestimageistheconceptofakindofstuff.Itistheconceptofaquantumofredinspace,anexpanseorvolumeconsistingofred,andisabasicconceptinthesensethatthereis“no...determinatecategorypriortotheconceptofredasaphysicalstuff,asamatterforindividuatedphysicalthings”(1981,I:84).Whendialecticalpressuresgenerateworriesabouttheontologicalstatusoftheredwhichoneostensiblyseeswhenitisnotaconstituentrednessofaphysicalobject,however,noalternativecategorycansimply249\nJAYF.ROSENBERGbe“readoff”fromanintrospectivescrutinyofcolorquanta.Theideathatapersonisalwaysalsoawareoftheactualcategorialstatusoftheitemsthatheencountersinper-ceptionorintrospection,Sellarssuggests,isonlyaparticularlyperniciousformofthemythofthegiven.Allthatisavailableissuchtranscendentalsas“actual,”“something”and“somehow.”Theredissomethingactualwhichissomehowaportionofredstuff,somehowthesortofitemwhichissuitedtobepartofthecontentofaphysicalobject,butwhich...isnot,inpointoffact,aportionofphysicalstuff.(1981,I:90)Itthenbecomesthejobofanalogicalthinkingtoconstructnewcategorialformsofconceptspertainingtocolor.ThefirstcomplicationofSellars’saccountofsensationresultedfromhisconvictionthat,inthecaseofsensations,Jones’stheorytakesthisinterpretiveform.Itdoesnotintroducenewdomainsofentities,butrathernewformsofconcepts.[Jones’s]theoryofsenseimpressions...reinterpretsthecategorialstatusofthecubicalvolumesofpinkofwhichweareperceptuallyaware.Conceivedinthemanifestimageas,instandardcases,constituentsofphysicalobjectsandinabnormalcases,assomehow“unreal”or“illusory,”theyarerecategorizedassensorystatesoftheperceiverandassignedvariousexplanatoryrolesinthetheoryofperception.(1981,III:44)ThecruxoftheJoneseantheory,inotherwords,isthethesisthattheverycolorquantaofwhichweareperceptuallyawareasbeinginspaceareinsteadactuallystatesofpersons-qua-perceivers.Itfollowsthat,alreadywithinthemanifestimage,theonto-logicalstatusultimatelyaccordedtosensory“contentqualia”isinfactincompatiblewiththeiractuallybeinginstantiatedinphysicalspace.“[The]esseofcubesofpinkispercipior,tousealessambiguousterm,sentiri.Ofcourse...wearenotperceptuallyawareof[them]asstatesofourselves,thoughthatisinpointoffactwhattheyare”(1981,III:66).AbsoluteprocessesThesecondcomplicationofSellars’saccountofsensationsthenarosefromhisfurtherconclusionthatthescientificimage’scommitmenttotheideathatperceiversarecomplexsystemsofmicro-physicalparticlesconstitutesabarriertoanystraightfor-wardsynopticassimilationofthismanifestimageconceptionofsensorycontentsasstatesofperceivingsubjects.Ontheonehand,SellarsobservedthatJones’sanalogicaltreatmentofsensorycontentsasstatesofperceiversformallypreservesthe“ultimatehomogeneity”ofthosecontentsasoriginallyconceivedasspace-fillingstuffs.Nodefinedstatesofasystemormultiplicityoflogicalsubjects,heargued,couldcontinuetodoso.Ontheotherhand,Sellarscontended,wecannotsimplyadopta“reductivematerialist”viewaccordingtowhich“whatreallygoesonwhenapersonsensesa-cube-of-pinklyconsistsin[acertain]systemofmicro-physicalparticlesbeinginacomplexphysical-2state”(1981,III:79),where“physical-2”statesarethosedefinable250\nWILFRIDSELLARSintermsoftheoreticalpredicatesnecessaryandsufficienttodescribenon-livingmatter.(Tobe“physical-1,”incontrast,issimplytobelonginthespace-timenetwork.)Forsuchreductivematerialismamountstotherejectionoftheideathata(Jonesean-theoretical)stateof,forexample,sensinga-cube-of-pinklyisitselfsomethingactualinanycategorialguise,andthisfailsproperlytorespectthephilosophicaldemandsofanadequatesensoryphenomenology.WhatSellarsnotoriouslyconcludedwasthatsensorycontentscouldbesynopticallyintegratedintothescientificimageonlyifboththeyandthemicro-physicalparticularsofthatimageaswellweresubjectedtoyetanotherontologicalreconceptionintermsofacategorially-monisticframeworkwhosebasicentitieswereall“absoluteprocesses.”Onlywhenperceiversthemselveshadbeenreconceivedassystemsor“harmonies”ofabsoluteprocesses,includingtheultimateconceptualdescendantsofsensorycontents,wouldthewaybeclearedforaunificationofthetwoimages.Thoughtofasabsoluteprocesses,sensingswouldbephysicalnotonlyintheweaksenseofnotbeingmental(i.e.conceptual),fortheylackintention-ality,butintherichersenseofplayingagenuinecausalroleinthebehaviorofsentientorganisms.Theywould...bephysical-lbutnotphysical-2.Notbeingepiphenomenal,theywouldconformtoabasicmetaphysicalintuition:tobeistomakeadifference.(1981,III:126)IntentionandactionIncontrasttotheintegrativechallengesposedbythoughtsandsensations,thechallengeofintegratingactionsproperlysocalled,thatis,conductsinformedbyinten-tionandvolition,intothescientificimageisnotfundamentallyontological.Althoughtheyexhibitquitespecialfeatureswhenconsideredfunctionally,regardedfromtheontologicalperspective,intentionsandvolitionsaresimplyspeciesofoccurrentthought-episodes.Whatmakessuchthoughtspracticalaretheirspecialrelationshipstoconductorbehavior,analogoustothewayinwhichtheirstatusasnon-inferentialresponsestosensationsconfersonparticularthoughtsthefunctionalroleofpercep-tualjudgments.Sellarscharacteristicallysignalsthespecialconduct-determiningroleofpracticalthoughtsbyacontriveduseoftheauxiliaryverb“shall”asanoperatoronthoughtcontentsexpressedassentences.Categoricalintentionsaretemporallydeterminatefirst-personfuture-tensedpracticalthoughts.Theyhavethecanonicalform(INT):Shall(IwilldoXatt).Volitions(“actsofwill”)arespecialcasesofsuchintentions,whosetimedeterminationistheindexicalpresent,thus,(VOL):Shall(IwillnowdoX).Suchpracticalthinkings,onSellars’sview,mediatebetweendeliberativereasoningandovertbehaviorbybeingappropriatelycaughtupinanetworkofacquiredcausalpropensitiesthatguarantee,roughly,thatintentionsoftheform(INT)regularlygiverise,attimet,tovolitionsoftheform(VOL),which,absentparalysisandthelike,inturnregularlygiverise,thenandthere,tobodilymovementsthatare(furthercircum-stancesbeingappropriate)theinitialstagesofadoingofX.SuchpracticalthinkingsaregovernedaccordingtoasingleprinciplewhichunitespracticalandtheoreticalÈ˘È˘È˘È˘reasoning:Ifpimpliesq,thenShall(p)impliesShall(q).251\nJAYF.ROSENBERGPersonsHereagain,then,Sellarsconcluded,whatthemanifestimagecontainsisnottheconceptofsomethingwithadeterminateintrinsiccharacterwithwhichweareacquaintedbutratherthefunctionalconceptionofacausally-mediatinglogico-semanticrole-player.Practicalthinkingscanthusbeontologicallyaccommodatedwithinthescientificimagealongthelinesalreadysketchedforcognitivethoughtsingeneral.Buthere,hecontinued,ontologicalaccommodationcannotbetheendofthesynopticstory.Ifwetakeseriouslytheideathatthescientificimagepurportstobeacompleteimageofman-in-the-worldandacandidateultimatelytoreplacethemanifestimage,thenthelatter’scategoriespertainingtopersonswillneedtoreappearwithinthesoughtsynopticfusionassuch.Weneedtoreconcile“theideathatmaniswhatsciencesaysheis”with“thecat-egoriespertainingtomanasapersonwhofindshimselfconfrontedbystandards(ethical,logical,etc.)whichoftenconflictwithhisdesiresandimpulses”(1963c:38).OnSellars’sview,thebasicconceptofapersonisirredeemablysocial.Tothinkofanentityasapersonisessentiallytothinkofitasactuallyorpotentiallyamemberofacommunity,“anembracinggroupeachmemberofwhichthinksofitselfasamemberofthegroup”(1963c:39),anditisthemostgeneralsharedintentionsofitsmembersthatfundamentallydefinethestructureofnormsandvaluesintermsofwhichtheirconductscometobeappraisedas“correct”or“incorrect”or“right”or“wrong.”Thustheconceptualframeworkofpersonsistheframeworkinwhichwethinkofoneanotherassharingthecommunityintentionswhichprovidetheambienceofprinciplesandstandards(aboveall,thosewhichmakemeaningfuldiscourseandrationalityitselfpossible)withinwhichweliveourownindividuallives.(1963c:39–40)Aswehaveseen,Sellarsinterpretedtheframeworkofthoughtsasfoundedwithinthemanifestimageonaseriesofontologicallynoncommittalfunctionalanalogiestowhichwecanreadilyimagineanemergingscientificunderstandingprogressivelysup-plyingstructural(e.g.neurophysiological)form.Incontrast,hearguedthataccom-modatingthemanifestimage’ssensorycontentswithinasynopticfusionwouldrequiretheconceptualtranspositionofsomeofitsontologicallybasicentitiesintonewcate-gorialforms.Unliketheframeworksofthoughtsandsensations,however,Sellarscon-tendedthattheconceptualframeworkofpersonsassuch“isnotsomethingthatneedstobereconciledwiththescientificimage,butrathersomethingtobejoinedtoit”(1963c:40).Toachieveagenuinelysynopticvisionofman-in-the-world,hecon-cluded,thescientificimageneedstobeenrichednotwithmore[ordifferent]waysofsayingwhatisthecase,butwiththelanguageofcom-munityandindividualintentions,sothatbyconstruingtheactionsweintendtodoandthecircumstancesinwhichweintendtodotheminscientificterms,wedirectlyrelatetheworldasconceivedbyscientifictheorytoourpurposes,andmakeitourworldandnolongeranalienappendagetotheworldinwhichwedoourliving.(1963c:40)Suchanultimateunificationofthemanifestandscientificimages,theworldofpersonswiththeworldofscience,wasthecontrollingvisionofSellars’sphilosophy.252\nWILFRIDSELLARSWhatmakeshimoneofthetoweringfiguresofpostwaranalyticphilosophy,however,isnotjustthegrandscopeofhisenterprise,buttheprofoundoriginalityofhisspecificconclusions,thesophisticateddialecticallyarguedandhistoricallyinformedreasoningwithwhichhesupportedthem,andtheexemplarythoroughnesswithwhichhepainstakinglydevelopedsomethingveryrareintheanalytictradition,aprincipledandconsistentsystematicphilosophicalview.BibliographyofworksbySellars1963a:“EmpiricismandthePhilosophyofMind,”in1963d,pp.127–96.(FormerlypublishedinTheFoundationsofScienceandtheConceptsofPsychoanalysis,MinnesotaStudiesinthePhilosophyofScience,vol.I,ed.H.FeiglandM.Scriven,Minneapolis:UniversityofMinnesotaPress,1956.)1963b:“Phenomenalism,”in1963d,pp.60–105.1963c:“PhilosophyandtheScientificImageofMan,”in1963d,pp.1–40.(FormerlypublishedinFrontiersofScienceandPhilosophy,ed.RobertColodny,Pittsburgh,PA:UniversityofPittsburghPress,1962.)1963d:Science,PerceptionandReality,London:RoutledgeandKeganPaul,andNewYork:HumanitiesPress.(ReissuedbyRidgeviewPublishing.)1963e:“TheLanguageofTheories,”in1963d,pp.106–26.(FormerlypublishedinCurrentIssuesinthePhilosophyofScience,ed.H.FeiglandG.Maxwell,NewYork:Holt,RhinehartandWinston,1961.)1967a:“AbstractEntities,”in1967b,pp.229–69.ReprintedinReviewofMetaphysics16(1983).1967b:PhilosophicalPerspectives,Springfield,IL:CharlesC.Thomas.(Reissuedin2vols,PhilosophicalPerspectives:HistoryofPhilosophyandPhilosophicalPerspectives:MetaphysicsandEpistemology,byRidgeviewPublishing.)1967c:“TheIdentityApproachtotheMind–BodyProblem,”in1967b,pp.370–88.(FormerlypublishedinReviewofMetaphysics18(1965).)1968:ScienceandMetaphysics:VariationsonKantianThemes,London:RoutledgeandKeganPaulandNewYork:HumanitiesPress.(ReissuedbyRidgeviewPublishing.)1974:“MeaningasFunctionalClassification,”Synthese27,pp.417–37.1975a:“EmpiricismandAbstractEntities”,in1975b,pp.245–86.(FormerlypublishedinThePhilosophyofRudolphCarnap,ed.P.A.Schilpp,LaSalle,IL:OpenCourt,1963.)1975b:EssaysinPhilosophyanditsHistory,Dordrecht:Reidel.1975c:“InductionasVindication,”in1975b,pp.367–416.(FormerlypublishedinPhilosophyofScience31(1964).)1975d:“TheStructureofKnowledge,”inAction,Knowledge,andReality:StudiesinHonorofWilfridSellars,ed.H.N.Castaneda,Indianapolis:Bobbs-Merrill,pp.295–347.1975e:“LanguageasThoughtandCommunication,”in1975b,pp.93–117.(Formerlypub-lishedinPhilosophyandPhenomenologicalResearch29(1969).)1979:NaturalismandOntology,Reseda,CA:RidgeviewPublishing.1980:PurePragmaticsandPossibleWorlds:TheEarlyEssaysofWilfridSellars,ed.J.F.Sicha,Reseda,CA:RidgeviewPublishing.(AlsocontainsalongintroductoryessaybySichaandanextensivebibliographyofSellars’sworkthrough1979.)1981:“TheCarusLecturesfor1977–78,”TheMonist64/1.(Citationsbylectureandnumberedparagraph.)1989:TheMetaphysicsofEpistemology,LecturesbyWilfridSellars,ed.P.Amaral,Reseda,CA:RidgeviewPublishing.(AlsocontainsacompletebibliographyofSellars’spublishedworkthrough1989.)253\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200120H.P.Grice(1913–1988)STEPHENNEALELifeHerbertPaulGricewasbornonMarch13,1913,inBirmingham,England.HeattendedCorpusChristiCollege,Oxford,graduatingin1936.From1938until1967heheldvariousfellowshipsandlectureshipsatStJohn’sCollege.HistimeatOxfordwasinter-ruptedbynearlyfiveyears’wartimeserviceintheRoyalNavy,firstintheNorthAtlanticandlaterinAdmiraltyintelligence.In1967,hemovedtotheUniversityofCalifornia,BerkeleyasProfessorofPhilosophy.HewaselectedtotheBritishAcademyin1966,andgavetheWilliamJamesLecturesatHarvardin1967,theJohnLockelecturesatOxfordin1978,andtheTannerLecturesatStanfordin1980.HediedinBerkeleyinAugust1988,shortlybeforethepublicationofhisfirstbook,StudiesintheWayofWords.Gricewasoneofthemostgiftedandrespectedphilosophersofthesecondhalfofthetwentiethcentury.Hesetimpossiblyhighstandardsandwasalwaysreluctanttogointoprint–heroiceffortswererequiredbyeditorsandfriendstoextractfromhimthehandfulofpapershedeemedworthyofpublication–yetheexertedconsiderableinflu-encethroughseminarsandinvitedlectures.HeworkedontopicsinAristotle,meta-physics,ethics,andphilosophicalpsychology;buthisstrongestinfluencewasinthephilosophyoflanguage,wherehisthoughtcontinuestoshapethewayphilosophers,linguists,andcognitivescientiststhinkaboutmeaning,communication,andtherela-tionbetweenlanguageandmind.Withrespecttoaparticularsentence,X,andan“utterer”U,Gricestressedtheimportanceofseparating(1)whatXmeans,(2)whatUsaidonagivenoccasionbyutteringX,and(3)whatUmeantbyutteringXonthatoccasion.Second,heattemptedtosaywhatmeaningisbyprovidinganalysesofutterer’smeaning,sentencemeaning,andwhatissaid.Third,hetriedtoexplainhowwhatUsaysandwhatUmeanscandiverge.Fourth,hedefendedconceptualanalysisandsomeformofanalytic/syntheticdistinction.Fifth,bycharacterizingthedistinctionbetweenthe“genuinelysemantic”and“merelypragmatic”implicationsofastatement,Griceclarifiedtherelationshipbetweenclassicallogicandthesemanticsofnaturallanguage.Sixth,hedeployedhisnotionof“implicature”todevastatingeffectagainstoverzealousstrainsof“ordinary-languagephilosophy,”withoutabandoningtheviewthatphilosophymustpayatten-254\nH.P.GRICEtiontothenuancesofordinarytalk(seeAUSTIN).Seventh,Griceundercutthemostinfluentialargumentsforaphilosophicallysignificantnotionof“presupposition.”Eighth,hemadesignificantcontributionstodebatesaboutthesemanticsofpropernames,definitedescriptions,andpronouns.Ninth,hesketchedaphilosophicalpsy-chologyandatheoryofvaluethatpromisetoprovidethebasisoffutureworkonactions,mentalstates,andmoralphilosophy,andtoexplaintherelationshipbetweenmindandlanguageinherentinhisphilosophyoflanguage.Meaning,use,andordinarylanguageTheviewthattheonlyusefulthingtosayaboutthemeaningofanexpressionisthatitisusableinsuch-and-suchcircumstances,exercisedapowerfulinfluenceonphiloso-phyinpostwarOxford.Austin,Ryle,andothersundercutphilosophicalpositionsordis-posedofphilosophicalproblemsbypointingtoamisuseofsomeexpressionplayinganessentialroleinthepresentationofthepositionorproblem.Considerattemptstoanalyzeknowledgeintermsofbeliefalongthefollowinglines:Aknowsthatpifandonlyif(1)Abelievesthatp,(2)p,and(3)Aisjustifiedinbelievingthatp.Itmightbechargedthatitisafeatureoftheuseof“believe”thatonedoesnotuseitifonecansincerelyuse“know”instead.Suchaclaimmightbesupportedbyobservingthatitwouldbeinappropriateforamantosay“IbelieveSmithisdead”whenheknowsSmithisdead.Andsoitmightbeconcludedthattheproposedanalysismustbediscardedbecauseclause(1)conflictswiththeordinaryuseoftheverb“believe.”Griceacceptedthatatheoryofmeaningmustbesensitivetouseandattemptedtoexplicatethemeaningofanexpression(oranyothersign)intermsofwhatitsusersdowithit,thatis,intermsofwhatitsusers(could/would/should)meanbyitonpar-ticularoccasionsofuse.Twoimportantideascameoutofthissensitivitytouse.Thefirstisthatthelocutionbyutteringx,Umeantthatpcanbeanalyzedintermsofcomplexaudience-directedintentionsonthepartofU.Thesecondisthatthemost“basic”notionofmeaningisthatofanuttererUmeaningsomethingbydoingsomethingonaparticularoccasion;allothernotionsofmeaningarederivative.WhatUmeansbyproducingxonagivenoccasionisafunctionofwhatUintends,inacomplexway,togetacrosstohisaudience.Thebasicideais,veryroughly,thatforan“indicative-type”utterance,thelocutionbyutteringx,UmeantthatpexpressesatruthiffUutteredxintendingtoproduceinsomeaudienceAthebeliefthatpbymeansofA’srecognitionofthisintention.Sentencemeaningistobeanalyzedintermsofregularitiesovertheintentionswithwhichutterersproducesentencesongivenoccasions.Byutteringasentenceoftheform“porq,”Umaywellimplythathehasnon-truth-functionalgroundsforhisassertion;butthisisnotpartofwhatthesentence(orthestatementmade)implies.Gricewantedanyadequateexplanationofthepossibilityofpragmaticimplicationstoflowfromacompletelygeneraltheory.Todemonstratethedef-initeexistenceofpragmaticimplicationsdistinctfromsemanticimplications,Gricecon-sideredanextremeexample.SupposeAasksUforanevaluationofhisstudentMr.X.AllUsaysis“Mr.Xhasexcellenthandwritingandisalwaysverypunctual.”IfUleavesitatthat,thosepresentarelikelytoconcludethatUthinksMr.Xisnotmuchgoodatphilosophy.ThereissurelynotemptationtosaythatthepropositionthatMr.Xisnotmuchgood(orthatUthinksMr.Xisnotmuchgood)atphilosophyis(orisaconse-255\nSTEPHENNEALEquenceof)thestatementUmade.ThesentenceUutteredhasaclearlinguisticmeaningbasedonthemeaningsofitspartsandtheirsyntacticalarrangement;anditseemsquitewrongtosaythat,whenheutteredthatsentence,UmadethestatementthatMr.Xisnotmuchgoodatphilosophy.Ontheotherhand,itseemsquitenaturaltosaythat,inthecircumstances,whatUmeant(orpartofwhatUmeant)bymakingthestate-mentheinfactmadewasthatMr.Xisnotmuchgood(orthatUthinksMr.Xisnotmuchgood)atphilosophy.Thisissomethingtheuttererimpliedbymakingthestate-menthedidinthiscontext,notsomethingimpliedbythesentenceutteredorbythestatementUmadebyutteringthesentence.ThetheoryofconversationWithrespecttowhatUmeansbyalinguisticutterance,GriceproposedtoseparatewhatUsaysandwhatUimplicates(e.g.implies,indicates,orsuggests).WhatUsaysistobecloselytiedtotheconventionalmeaningofthewordsuttered,whichbothfallsshortandgoesbeyondwhatissaid.ItfallsshortbecauseaspecificationofwhatUsaidonaparticularoccasionmusttakeintoaccountnotonlytheconventionalmeaningofthesentenceusedbutalso(e.g.)thereferencesofreferringexpressions(e.g.propernames,demonstratives,andindexi-cals)andthetimeandplaceofutterance.WhatUsaidistododutyforwhatUstatedorthepropositionexpressedbyU.Wherethesentenceutteredisofthetypeconventionallyassociatedwiththespeechactofasserting(i.e.whenitisinthe“indicativemood”)whatissaidwillbestraightforwardlytruth-conditional.Whenthesentenceutteredisintheimperativeorinterrogativemood,whatissaidwillnotbestraightforwardlytruth-conditional,butitwillbesystematicallyrelatedtothetruthconditionsofwhatUwouldhavesaid,inthesamecontext,byutteringtheindicativecounterpart(oroneoftheindicativecounterparts)oftheoriginalsentence.Theconventionalmeaningofasentencealsogoesbeyondwhatissaidbecauseofdevicesthatsignaltheperformanceof“noncentralspeechacts”parasiticupontheper-formanceofthe“centralspeechacts”ofasserting,questioning,andordering.Suchdevices,althoughtheyplayapartindeterminingwhatUmeant,playnopartindeter-miningwhatUsaid.IfUutters(1)ratherthan(2),1Sheispoorbutsheishonest2SheispoorandsheishonestverylikelyUwillbetakentobeimplyingthatthereis(orthatsomeonemightthinkthereis)somesortofcontrastbetweenpovertyandhonesty(orherhonestyandherpoverty).ThistypeofimplicationisnopartofwhatUsaysbecauseitdoesnotcon-tributeinanywaytothetruthconditionsoftheutterance.Byuttering(1),Uhassaidonlythatsheispoorandsheishonest;andthisdoesnotentailthatthereisany(e.g.)contrastbetweenpovertyandhonesty(orbetweenherpovertyandherhonesty).TheimplicationinquestionGricecallsaconventionalimplicature.AccordingtoGrice,byuttering(1)Uisperformingtwospeechacts,sayingthatsheispoorandsheishonestandindicating(orsuggesting)thatsomeone(perhapsU)hasacertainattitudetowardwhatissaid.Gricedidnotdevelopthisidea;hejustleftuswiththeclaimthataconventionalimplicatureisdetermined(atleastinpart)bythe(con-256\nH.P.GRICEwhatUmeantwhatUwhatUconventionallynon-conventionallymeantmeantwhatUwhatUwhatUwhatUsaidconventionallyconversationallynon-conversationallyimplicatedimplicatednon-conventionallyimplicatedFigure1ventionsgoverning)thewordsused.Hedoesstress,however,thatthesortofimplica-tionwehavejustbeenconsideringisnotapresupposition(asoriginallydefinedbyStrawsonandadoptedbyothers).BisapresuppositionofA,justincasethetruthorfalsityofArequiresthetruthofB.(IfthetruthofArequiresthetruthofB,butthefalsityofAdoesnot,BisanentailmentofA.)Moreprecisely,ifApresupposesB,AlacksatruthvalueifBisfalse.ButasGricepointsout,anutteranceof(1)canbefalseeveniftheimpliedpropositionisfalse,effectivelyscotchingtheideathattheimpli-cationispresupposition(atleastnotonthestandardsemanticconceptionofthatnotion).ItisGrice’sviewthatanyallegedpresuppositioniseitheranentailmentoranimplicature.Forsomethingtobe(partof)whatUsays,itmustalsobe(partof)whatUmeant,thatis,itmustbebackedbyacomplexintentionofthesortthatformsthebackboneofGrice’stheoryofmeaning(seefigure1).IfUuttersthesentence“Billishonest”iron-ically,onGrice’saccountUwillnothavesaidthatBillishonest:UwillhavemadeasiftosaythatBillishonest.ForitisGrice’sviewthatastatementoftheform“byutteringx,Usaidthatp”entailsthecorrespondingstatementoftheform“byutteringx,Umeantthatp.”SoonGrice’saccount,onecannotunintentionallysaysomething(afactthathasinterestingconsequencesfor,forexample,slipsofthetongueandmisusedexpressions).Grice’sworkprovidesabreakdownofwhatUmeantasshowninfigure1.WhatUconventionallyimplicatesandwhatUsaysarebothcloselytiedtothecon-ventionalmeaningofthesentenceuttered,andtheyaretakenbyGriceasexhaustingwhatUconventionallymeans(i.e.meansbyvirtueoflinguisticconvention).Letusnow257\nSTEPHENNEALEturntowhatUnon-conventionallymeans.Consideragain,theexampleconcerningProfessorU’sevaluationofMr.X.Byutteringthesentence“Mr.Xhasexcellenthand-writingandisalwaysverypunctual,”Usaid(ormadeasiftosay)thatMr.Xhasexcel-lenthandwritingandisalwaysverypunctual.Inaddition,onGrice’saccountUconversationallyimplicatedthatMr.Xisnotmuchgoodatphilosophy(thereisaconver-sationalimplicaturetotheeffectthatMr.Xisnotmuchgoodatphilosophy).Conversationalimplicatureisaspeciesofpragmatic(non-semantic,non-conventional)implicationandistobecontrastedwiththe(atleastpartlysemantic)implicationthatGricecallsconventionalimplicature.Theprincipaldifferencebetweenaconventionalandaconversationalimplicatureisthattheexistenceofaconventionalimplicaturedependsuponthepresenceofsomeparticularconventionaldevice(suchas“but,”“moreover,”“still,”“yet,”orheavystress)whereastheexistenceofaconversationalimplicaturedoesnot.GriceproposestoexplainthepossibilityofadivergencebetweenwhatUsaysandwhatUmeans(oratleastbetweenwhatUconventionallymeansandwhatUmeans)byappealtothenatureandpurposeofrationalinteraction.Conversationisviewedbyhimasacharacteristicallypurposefulandcooperativeenterprisegovernedbywhathecalls:TheCooperativePrinciple:Makeyourconversationalcontributionsuchasisrequired,atthestageatwhichitoccurs,bytheacceptedpurposeordirectionofthetalkexchangeinwhichyouareengaged.Subsumedunderthisprinciple,Gricedistinguishesfourcategoriesofmorespecificmaximsandsubmaximsenjoiningtruthfulness,informa-tiveness,relevance,andclarity:Quantity:Makeyourcontributionasinformativeasisrequired(forthecurrentpurposesoftheexchange);donotmakeyourcontributionmoreinformativethanisrequired.Quality:Trytomakeyourcontributiononethatistrue.Specifically:(1)Donotsaywhatyoubelievetobefalse;(2)Donotsaythatforwhichyoulackadequateevidence.Relation:Berelevant.Manner:Beperspicuous.Specifically:(1)Bebrief;(2)Beorderly;(3)Avoidambigu-ity;(4)Avoidobscurityofexpression.Grice’sbasicideaisthatthereisasystematiccorrespondencebetweenwhatUmeansandtheassumptionsrequiredinordertopreservethesuppositionthatUisobservingtheCooperativePrincipleandconversationalmaxims.InthecaseofProfessorU’sevaluationofMr.X,onthesurfacetheCooperativePrincipleoroneormoreofthemaximsisintentionallyandovertlynotfulfilled.Bysaying“Mr.Xhasexcellenthandwritingandisalwaysverypunctual”(inthisparticularcontext),UseemsnottohavefulfilledoneofthemaximsofQuantityorthemaximofRelation.(IfMr.XisoneofU’sstudents,UmustbeinapositiontovolunteermorerelevantinformationthanjudgmentsaboutMr.X’shandwritingandtimekeeping;furthermore,Uknowsthatmoreinformation,ormorerelevantinformation,isrequired.)ThehearerisnaturallyledtotheconclusionthatUistryingtoconveysomethingelse,somethingmorerelevanttothepurposesathand.Inthecircumstances,ifUthoughtMr.Xwasanygoodatphilosophyhewouldhavesaidso.SoUmustthinkMr.Xisnogoodatphilosophyandbeunwillingtosayso.AndsoUhasconver-sationallyimplicatedthatMr.Xisnogoodatphilosophy.258\nH.P.GRICEOneinterestingfeatureofthisexampleisthatitmightwellbethecasethatonlywhatisimplicatedismeant(i.e.backedbyU’scommunicativeintentions).UmayhavenoideawhatMr.X’shandwritingislikebecauseMr.XhasshownUonlytypedmanuscriptsofhiswork(orbecausehehasnevershownUanything),andUmayhavenoopinionastowhetherornotMr.Xispunctual.Insuchaversionoftheenvisionedscenario,UhasonlymadeasiftosaythatMr.XhasexcellenthandwritingandisalwaysverypunctualbecauseUhadnointentionofinducing(oractivating)inhisaudiencethebeliefthat(Uthinksthat)Mr.Xhasexcellenthandwritingandisalwaysverypunctual.Thetruth-valuesofwhatUsaid(ormadeasiftosay)andwhatUconversationallyimplicatedmayofcoursediffer.Mr.Xmayhavequiteatrocioushandwriting,andUmayknowthis;butgiventherelevanceofwhatisconversationallyimplicated,Umaycareverylittleaboutthetruth-valueofwhathehassaid(ormadeasiftosay).Theprimarymessageistobefoundatthelevelofwhatisconversationallyimplicated.Ingeneral,Griceclaims,aspeakerconversationallyimplicatesthatwhichhemustbeassumedtothinkinordertomaintaintheassumptionthatheisobservingtheCooperativePrinciple(andperhapssomeconversationalmaxims),ifnotatthelevelofwhatissaid,atleastatthelevelofwhatisimplicated.Atsomeoverarchinglevelofwhatismeant,UispresumedtobeobservingtheCooperativePrinciple.ThewordingofGrice’smaximssuggeststhatsomeconcernonlywhatissaid(e.g.“Donotsaywhatyoubelievetobefalse”)whileothersconcern,perhaps,whatismeant(e.g.“Berelevant”).Weshouldprobablytreatthisassomethingofanuncharacteristicloosenessofexpres-siononGrice’spart.ExceptforthemaximsunderManner(whichcanapplyonlytowhatissaid)itseemsreasonabletounderstandGriceasallowingamaximnottobefulfilledatthelevelofwhatissaidtobelicensedoroverriddenbyadherenceatthelevelofwhatisimplicated.Onsuchaview,blatantlyviolatingamaximatthelevelofwhatissaidbutadheringtoitatthelevelofwhatisimplicatedwouldnotnecessarilyinvolveaviolationoftheCooperativePrinciple.Someimportantquestionsarestillunanswered:Howaresayingandimplicatingtobedefined?Howareimplicaturescalculated?WhatisthestatusoftheCooperativePrincipleandmaxims?Whathappenswhenaspeakercannotsimultaneouslyobserveallofthemaxims?ItisimportanttoseehowGriceattemptstofacesuchquestions.NooneshoulddenythatintheexampleoftheevaluationofMr.XthereisanintuitiveandobviousdistinctiontobemadebetweenwhatUsaidandwhatUconver-sationallyimplicated.ButinviewofthesortsofexamplethatreallybotherGrice–“theFisG,”“porq,”“ifpthenq,”etc.–hecouldnotrestwithanintuitivedistinction.TheexampleconcerningtheevaluationofMr.Xisclear-cut,obvious,anduncon-tentious.Andhereinliestheproblem.TheexamplesofpurportedconversationalimplicaturethatmostinterestGricearephilosophicallyimportantoneswithrespecttowhichmanyphilosophershavenotfelttheneedtoinvokesuchadistinction.Thismightbebecauseitisnotatallobviousthatthereissuchadistinctiontobemadeinthecasesinquestion(orifthereis,howrelevantitis),orbecauseadherencetosomeformofthe“meaningisuse”dogmahasblindedcertainphilosopherstothepossibilityofsuchadistinction.SoGriceultimatelyneedsanalysesof“whatissaid”and“whatisconversationallyimplicated”inordertogetphilosophicalworkoutofthesenotions.259\nSTEPHENNEALEGricehopestoanalyzethenotionofsayingintermsofutterers’intentions.Thispro-posalwillbeexaminedafterdiscussionofGrice’stheoryofmeaning.Heattemptstodefineconversationalimplicatureintermsofan,asyet,undefinednotionofsaying.Thefollowingschemaissupposedtobeafirststep:Someonewho,by(in,when)saying(ormakingasiftosay)thatphasimplicatedthatq,hasconversationallyimplicatedthatq,providedthat(1)heistobepresumedtobeobservingtheconversationalmaxims,oratleasttheCooperativePrinciple;(2)thesuppositionthatheisawarethatqisrequiredinordertomakehissayingormakingasiftosaypconsistentwiththispresumption;and(3)thespeakerthinks(andwouldexpectthehearertothinkthatthespeakerthinks)thatitiswithinthecompetenceofthehearertoworkout,orgraspintuitively,thatthesuppositionmentionedin(2)isrequired.Weappeartohavehereasetofnecessaryconditions.Theconditionsarenotsufficientbecauseconventionalimplicaturesarenotexcluded.Wheneverthereisaconversationalimplicature,oneshouldbeabletoreasonsomewhatasfollows:(i)Uhassaidthatp;(ii)thereisnoreasontosupposethatUisnotobservingtheCooperativePrincipleandmaxims;(iii)Ucouldnotbedoingthisunlesshethoughtthatq;(iv)Uknows(andknowsthatIknowthatUknows)thatIcanseethatUthinksthesuppo-sitionthatUthinksthatqisrequired;(v)Uhasdonenothingtostopmethinkingthatq;(vi)Uintendsmetothink,orisatleastwillingtoallowmetothink,thatq;(vii)andso,Uhasimplicatedthatq.IneachofthecasesGriceconsiders,itdoesseemtobepossibletojustifytheexistenceoftheimplicatureinquestioninthissortofway.Butnoticethatqissimplyintroducedwithoutexplanationinstep(iii),soGricehascertainlynotstatedanysortofmethodorprocedureforcalculatingthecontentofconversa-tionalimplicatures.AgooddealofworkneedstobedoneonthecalculationofparticularimplicaturesifGrice’sevidentinsightsaretoformthebasisofafinallyacceptabletheory.Anecessaryconditiononconversationalimplicaturesthatisintimatelyconnectedtocondition(3)isthattheyareintended.Thisfollows,ifnotfromcondition(3),atleastfromthefactthat(a)whatUimplicatesispartofwhatUmeans,and(b)whatUmeansisdeterminedbyU’scommunicativeintentions.Ahearermaythinkthat,bysayingthatp,Uhasconversationallyimplicatedthatq(Amayevenhavereasonedexplicitlyinthemannerof(i)–(vii)above).ButifUdidnotintendtheimplicationinquestionitwillnotcountasaconversationalimplicature.Wehave,then,fourconditionsthatarenecessarybutnotsufficientforclassifyinganimplicationasaconversationalimplicature.Entailmentsdonotseemtohavebeenexcluded.Inorderthatwemaystayfocusedontherelationbetweenthespeakerandcertainpropositions,letusmakeaharmlessadditiontoGrice’sterminology.Ifthepropositionthatpentailsthepropositionthatq,thenifUisacompetentspeakerwhosaysthatp,Utherebysays*thatq.SoifUisaperfectlycompetentEnglishspeakerwhohassincerelyutteredthesentence“Johnisabachelor,”notonlyhasUsaid(andsaid*)thatJohnisabachelor,hehasalsosaid*thatJohnisunmarried.Itseemsdesirablethatnopropositionbebothanentailmentandaconversationalimplicatureofthesameutterance.Butitisnotobviousthattheconditionslaiddownthusfaronconversationalimplicatureactuallyruleoutentailments.Furthermore,Gricecannotjustimposeafurtherconditiontothedefinitiontotheeffectthatnoentailmentisaconversational260\nH.P.GRICEimplicature.OneofGrice’savowedaimsistowardoffcertainordinarylanguageargu-mentsbyinvokingasharpdistinctionbetweenwhatwearenowcallingconversationalimplicatureandentailment;soitisnotgoodenoughforhimtousethenotionofanentailmentinadefinitionofconversationalimplicature.AfifthconditionGriceimposesonconversationalimplicaturesseemstohelp.Unlikeanentailment,aconversationalimplicatureissupposedtobecancelableeitherexplic-itlyorcontextually,withoutcontradiction.IfUsays*thatp,andpentailsq,thenUcannotgoontosay*thatnot-qwithoutcontradiction.Forexample,Ucannotsay“JohnisabachelorandJohnismarried.”ButifUsays*thatp,andtherebyconversationallyimplicatesthatq,Ucangoontosay*thatnot-qwithoutcontradiction.ConsideragainthecaseofU’sevaluationofMr.X.Afteruttering“Mr.Xhasexcellenthandwritingandisalwaysverypunctual,”Umight(withoutirony)continue“Moreover,Mr.X’srecentmodalproofoftheimmortalityofthesoulisabrilliantandoriginalcontributiontophilosophy.”Inthelightofthefirstcomment,thisadditionmightberatherodd,butitwouldnotresultinUcontradictinghimself.(Inadditiontodistinguishingconversa-tionalimplicaturesfromentailments,thecancelabilitytestisalsosupposedtodistin-guishconversationalfromconventionalimplicatures.Althoughitwillnotleadtocontradiction,attemptingtocancelaconventionalimplicaturewillresultinagen-uinelylinguistictransgressionofsomesort.Thisispreciselybecausethereisadistinctsemanticcomponenttoconventionalimplicatures.)Puttingthesefiveconditionstogether,wecomeascloseaswecanwithGrice’smachinerytoasetofnecessaryandsufficientconditionsonconversationalimplicature.PhilosophicalpsychologyForGrice,theprinciplesinvolvedinanaccountofconversationalimplicaturearetobegroundedinaphilosophicalpsychologythatexplicatesthepurportedlyhierarchicalrelationshipsthatholdbetweenthevarioustypesofpsychologicalstatesweascribetocreaturesthatcanreasonandformcomplexintentions.Thebeginningsofthislineofthoughtcanbetracedtotheendofhis1957paper“Meaning.”Itcontainstheseedsof(1)theviewthattheCooperativePrincipleandconversationalmaxims(inparticu-larthemaximenjoiningrelevance)aretoplayacentralrolenotonlyinanaccountofpossibledivergencesbetweenwhatUsaidandwhatUmeantbutalsoinanaccountoftheresolutionofambiguities,and(2)theviewthattheuseoflanguageisoneformofrationalactivityandthattheprinciplesatworkintheinterpretationoflinguisticbehaviorare(orareintimatelyrelatedto)thoseatworkininterpretingintentionalnon-linguisticbehavior.Twoquestionsspringtomindimmediately:(1)Whataretherelativerankingsofthemaximsincaseswhereitishard(orimpossible)forUtoobserveallofthem(orallofthemtothesamedegree),andwhy?(2)Whatisthebasisfortheassumptionthatspeak-erswillingeneral(ceterisparibusandintheabsenceofindicationstothecontrary)proceedinthemannerprescribedbytheCooperativePrincipleandmaxims?GriceisexplicitaboutthepositionofatleastoneofthemaximsofQualityinanyhierarchy.SupposeAisplanninganitineraryforavacationtoFrance.AwantstoseehisfriendC,ifsodoingwouldnotrequiretoomuchadditionaltraveling.AasksB261\nSTEPHENNEALE“WheredoesClive?”Breplies“SomewhereinthesouthofFrance.”BknowsthatAwouldlikemorespecificinformationbutheisnotinapositiontobemorespecific.SoBisfacedwithnotfulfillingeitheramaximofQualityoramaximofQuantity.Qualitywinsout.ThemaximsofQualityhaveaveryspecialstatuswithinGrice’soveralltheoryandGriceentertainstheideathatthefirstmaximofQualityshouldbepartofsomebroaderbackground;theothermaximscomeintooperationonlyontheassumptionthatthemaximofQualityissatisfied.ThemaximsofQuality(oratleastthefirstmaximofQuality)shouldnotbethoughtofasadmittingofdegreeorvaryingacrosscultures.Insomesensethisisanempiricalmatter;butunlikethemaximsofQuantityandManner,itdoesnotseemveryplausibletosupposethattherearethrivingculturesinwhichstandardlypeopledonotbehave(forparticularreasonstobedeterminedbyanthropologists)asiftheyareobservingthemaximsofQuality.Gricewasnotsatisfiedwiththeideathatitisjustawell-recognizedempiricalfactthatpeopledobehaveinaccordancewiththemaximsandtheCooperativePrinciple,thatinchildhoodtheylearnedtodosoandhavenotlostthehabit.Hewantedtofindabasisthatunderliesourbehaviorandbelieveditwouldhaveamoraldimen-sion:notonlydoweinfactbehaveintherequiredway,butitisreasonableforustodoso,andthepracticeissomethingweshouldnotabandongivenourcommonpurposesorgoals.ConversationisoneamongarangeofformsofrationalactivityforGrice.ObservanceoftheCPandmaximsisreasonable(rational):anyoneconcernedaboutthegoalscentraltocommunicationmustbeexpectedtohaveaninterest,givensuitablecircumstances,inparticipationininformationalexchangesthatwillbeprofitableonlyontheassumptionthattheyareconductedingeneralaccordancewiththeCPandmaxims.OnGrice’sview,valuepredicatessuchas“proper,”“correct,”“optimal,”and“rele-vant”cannotbekeptoutofanaccountofrationalactivitybecausearationalcreatureisessentiallyacreaturethatevaluates.Whetheravalue-orientedapproachtotheinter-pretationofintentionalbehaviorcanbedevelopedinafruitfulwayremainstobeseen.ButasGrice’sunpublishedworkonethicsandphilosophicalpsychologybecomesmorewidelyavailable,therewilllikelybearesurgenceofinterestinthematterofthepreciselocationofthetheoryofconversationwithinalargerscheme.ThelogicofnaturallanguageOnetaskofsemanticsistoprovideasystematiccharacterizationofjudgmentscon-cerningtruth,falsity,entailment,contradiction,andsoon.Inthelightoftheoreticalconsiderations,aninitialjudgmentof,say,entailmentmightberejectedonthegroundsthattheperceivedimplicationisanimplicatureratherthananentailment.Sofar,wehaveconsideredonlyexamplesofwhatGricecalls“particularized”conversationalimplicature,examplesinwhichthereisnotemptationtosaythattherelevantimpli-cationisanentailment(ora“presupposition”).Ofmorephilosophicalinterestare“generalized”conversationalimplicatures,thepresenceandgeneralformofwhichdependlittleupontheparticularcontextualdetails.ExamplesdiscussedbyGriceincludethoseattachingtoutterancesofsentencescontainingintentionalexpressionslike“look,”“feel,”and“try,”and“logical”expressionssuchas“and”,“or,”“if,”“every,”“a,”and“the.”262\nH.P.GRICEAccordingtoGrice,philosopherswhoseedivergencesinmeaningbetween“formaldevices”suchas“&,”,”“…,”“("x),”“($x),”and“(ix)”andtheirnaturallanguagecounterpartstendtobelongtooneoftwocamps,whichhecalls“formalist”and“infor-malist.”TheinformalistpositionisessentiallytheonetakenbyStrawson(andothersofthe“ordinary-languagemovement”).Theformalistcampisdominatedbypositivistsandotherswhoviewnaturallanguageasinadequatetotheneedsofthescienceandphilosophyofanageofprecision.Atypicalformalistrecommendstheconstructionofan“ideal”or“logicallyperfect”languagesuchasthelanguageoffirst-orderquantifi-cationtheorywithidentity(orsomesuitableextensionthereof).Sincethemeaningsofthelogicalparticlesareperfectlyclear,usinganideallanguage,philosopherscanstatepropositionsclearly,clarifythecontentsofphilosophicalclaims,drawthelimitsofintelligiblephilosophicaldiscourse,drawthedeductiveconsequencesofsetsofstate-ments,andgenerallydeterminehowwellvariouspropositionssitwitheachother.Griceviewstheformalistsandinformalistsasmistakenintheassumptionofseman-ticdivergence.Bothsideshavetakenmerepragmaticimplicationstobepartsofthemeaningsofsentencesofnaturallanguagecontaining“logical”expressions.Thecaseof“and”highlightssomeimportantmethodologicalconsiderations.Althoughitisplausibletosupposethat“and”(whenitisusedtoconjoinsentences)functionsseman-ticallyjustlike“&,”therearecertainlysentencesinwhichitappearstofunctionratherdifferently:1JackandJillgotmarriedandJillgavebirthtotwins.2Neroyelledandtheprisonerbegantotremble.Someonewhouttered(1)wouldtypicallybetakentoimplythatJackandJillgotmarriedbeforeJillgavebirthtotwins.Andsomeonewhouttered(2)wouldtypicallybetakentoimplythatNero’syellingcontributedinsomewaytotheprisoner’strem-bling.Thusonemightbeledtotheviewthat“and”isnotalwaysunderstoodas“&,”thatitis(atleast)threewaysambiguousbetweentruth-functional,temporal,andcausalreadings.ThepostulationofsemanticallydistinctreadingslooksextravagantandGricesug-gestsitisgoodmethodologicalpracticetosubscribeto“modifiedOccam’srazor”:sensesarenottobemultipliedbeyondnecessity.Giventheviabilityofthedistinctionbetweenwhatissaidandwhatismeant,ifapragmaticexplanationisavailableofwhyapar-ticularexpressionappearstodivergeinmeaningindifferentlinguisticenvironments(orindifferentconversationalsettings)thenceterisparibusthepragmaticexplanationispreferabletothepostulationofasemanticambiguity.Grice’sideaisthattheimpli-cationoftemporalsequenceattachingtoanutteranceof(1)canbeexplainedintermsofthefactthateachoftheconjunctsdescribesanevent(ratherthanastate)andthepresumptionthatUisobservingthesubmaximofMannerenjoiningorderlydeliveries.ItseemstobeGrice’sview,then,thatbyuttering(1)Uwillconversationallyimplicate(ratherthansay)thatJackandJillgotmarriedbeforeJillgavebirthtotwins(ifthisiscorrectthenwhatisconversationallyimplicatedwouldappeartoentailwhatissaidinthiscase).Similarly,theimplicationofcausalconnectionattachingtoanutteranceof(2)isapparentlytobeexplainedintermsofthepresumptionthatthespeakerisbeingrelevant.Beforelookingatproblemsforthisproposal,Iwantfirsttogetclearaboutitsstrengths.263\nSTEPHENNEALEConversationalexplanationstrumpsemanticambiguitiesongroundsoftheoreticaleconomyandgenerality.Aconversationalexplanationisfree:themechanismsappealedtoarealreadyinplaceandindependentlymotivated.Thegeneralitylostbypositingseveralreadingsof“and”isquiteconsiderable.First,implicationsof(e.g.)temporalpriorityandcausalconnectionattachtousesofthecounterpartsof“and”acrossunrelatedlanguages.Second,implicationsofthesamesortswouldsurelyariseevenforspeakersofalanguagecontaininganexplicitlytruth-functionalconnective“&.”Third,thesameimplicationsthatattachtoutterancesof“pandq”wouldattachtoanutteranceofthetwosentencesequence“p.q”notcontaininganexplicitdeviceofconjunction.Onmethodologicalgrounds,then,pragmaticaccountsofthetem-poralandcausalimplicationsin(1)and(2)arepreferabletoaccountsthatappealtosemanticambiguity.Griceopposespostulatingidiosyncraticpragmaticruleswithwhichtoderivegener-alizedimplicatures.ConversationalimplicaturesmustbeexplicableintermsoftheCooperativePrincipleandmaxims,construedasgeneralantecedentassumptionsabouttherationalnatureofconversation.Tocallanimplicature“generalized”ratherthan“particularized”isonlytoacknowledgethefactthatthepresenceoftheimplicatureisrelativelyindependentofthedetailsoftheparticularconversationalcontext,afactthatistobeexplainedbythecooperativenatureofconversation.AsecondchallengetoclassicallogicsemanticscamefromStrawson,whochallengedRussellonthegroundsthatthetheorydoesnotdojusticetoordinaryusage:speakersusedescriptionstorefer,nottoquantify,andhenceRussell’stheoryisopentoanumberofobjections(seeSTRAWSON).ButaccordingtoGrice,anumberofStrawson’sobjec-tionscanbedefusedbydistinguishingsentencemeaning,whatissaid,andwhatismeant.InGrice’sterminology,oneofStrawson’smaincomplaintsagainstRussellisthathistheoryconflatesthemeaningofasentence“theFisG”andwhatUsaysbyutteringthissentence(andsimilarlythesubsententialcounterpartsofthesenotions)andsocannotexplainthefactthatUmaysaydifferentthingsondifferentoccasionsbyutter-ingthesamesentence.GriceisrightthatStrawsoncangetnomileageoutofRussell’sfailuretoseparatesentencemeaningandwhatissaidinhisdiscussions.Uponreflec-tionitisclearthatRussell’sconcerniswithwhatissaidratherthansentencemeaning.IfRussellwerebeingmoreprecise,hewouldnotsaythatthesentence“theFisG”isequivalenttothesentence“thereisexactlyoneFandeveryFisG”;rather,hewouldsaythatwhatUsaysbyuttering“theFisG”onaparticularoccasionisthatthereisexactlyoneFandeveryFisG(occurrencesof“F”intheforegoingmay,ofcourse,beellipti-cal).Thefactthatadescription(oranyotherquantifiednounphrase)maycontainanindexicalcomponent(“thepresentkingofFrance,”“everymanhere,”etc.)doesnotpresentaproblem:allthismeansisthattherearesomedescriptionsthataresubjecttotheTheoryofDescriptions(seeRUSSELL)andatheoryofindexicality.Griceissurelyright,then,thatalthoughweneedasharpdistinctionbetweensentencemeaningandwhatissaid(andtheirsubsententialcounterparts),Strawson’sappealtothisdistinctionwhenchallengingRussellisempty.GriceneatlydisposesoftheviewthatdescriptionsareambiguousbetweenRussellianandreferential(oridentificatory)readings.Whenadescriptionisusedtoidentifysome-thing,whatUmeansdivergesfromwhatUsays.WhatUsaysisgivenbytheRussellian264\nH.P.GRICEexpansionbutUalsointendstocommunicateinformationaboutsomeparticularindi-vidual,andalthoughthisispartofwhatUmeans,itisnotpartofwhatUsays.ThisprovidesaperfectlysatisfactoryaccountofwhatisgoingonwhenUusesadescriptionthatdoesnotfititstarget,butsuchcasesarenotneededtoseeGrice’sdistinctionsatwork.AccordingtoGrice,whenadescriptionisusedinanidentificatoryway,therewillalwaysbeamismatchbetweenwhatUsaysandwhatUmeans(evenwherethedescrip-tionuniquelyfitstheindividualthespeakerintendstocommunicateinformationabout)becausewhatissaidis,onRussell’saccount,analyzableasaquantificationalpro-position,whereaswhatismeantwillalwaysincludeasingularorobject-dependentproposition.Again,methodologicalconsiderationsstronglyfavortheGriceanaccountofrefer-entialusageoveranaccountthatpositsasemanticambiguity:(1)IfweweretaughtexplicitlyRusselliantruthconditions,referentialusagewouldstilloccur;(2)exactlyparallelphenomenaoccurwithindefinitedescriptionsandotherquantifiednounphrases;(3)modifiedOccam’srazorenjoinsustooptforthesimpleroftwotheories,otherthingsbeingequal.Subsequently,farmoredetaileddefensesofRussellalongGriceanlineshavebeenproposedbyotherphilosophers,butthedebtstheseworksowetoGriceareconsiderable.Moregenerally,adebtisowedtoGriceforrejuvenatingthepositionthatclassicallogicisaremarkablyusefultoolasfarasthesemanticsofnaturallanguageisconcerned.ThetheoryofmeaningGriceattemptedtoanalyseorexplicatewhatissaidandwhatisimplicatedintermsofintention,belief,desire,andrecognition.Analyzinglocutionsoftheforms“XdidYintentionally,”“XcausedY,”“Xistrue,”“XentailsY,”andsoon,hasbeenseenbymanyphilosophersasacentraltaskofphilosophy.Grice’sanalysesof“byutteringX,Umeantthatp,”“Xmeans‘p’,”and“byutteringX,Usaidthatp”seemtohaveareduc-tiveandexplicativeflavorinthatitappearstobehisviewthatlocutionsoftheformscanbewhollyexplicatedwithoutappealingtosemanticalconcepts.Hebeginswithwhatpeoplemeanratherthanwithwhatthisorthatexpression,sign,oractionmeans,seekingtoanalyzethisintermsofcomplexaudience-directedinten-tionsonthepartoftheutterer,andtoanalyzeutterance-typemeaning(e.g.sentencemeaningandwordmeaning)intermsofutterer’smeaning.AlthoughGriceaimstoneutralizemanyordinarylanguagemaneuverswithhissaying/implicatingdistinction,oneofthedrivingforcesbehindhisworkisstilltheideathatthemeaningofanexpressionisafunctionofwhatitsusersdowithit.AbstractingawayfromcertaindetailsthatIwillgettolater,thedirectionofanalysisforGriceisshowninfigure2.Theidea,then,istobeginbyprovidingananalysisof(1)utterer’smeaning,andthentousethisanalysisinananalysisof(2)utterance-typemeaning.(3)Whatissaidisthentobedefinedintermsofanearcoincidenceofutterer’smeaningandutterance-typemeaning(forcertainutterance-types);andfinally(4)conversationalimplicatureistobedefinedintermsofsayingandutterer’smeaning.AlthoughGricedoesnotaddressthispointdirectly,itisclearthatthetaskofexpli-catingthelocution“byutteringx,Usaidthatp”takesonsomeurgencyforhim,265\nSTEPHENNEALE(2)utterance-typemeaning>>(1)utterer’smeaning>(3)whatissaid>>(4)whatisconversationallyimplicatedThe“__Æ__”isunderstoodas“__(oritsanalysis)playsaroleintheanalysisof__(butnotviceversa).”Figure2becausethesaying/implicatingdistinctionissocentraltohisattemptstocounterordi-narylanguageargumentsofthesortexaminedearlier.AdirectanalysisofsayingappearsoutofthequestionbecauseGriceopenlydeclaresthatheisusing“say”inaspecialsense,andthisprecludessystematicappealtointuitionsaboutordinaryusage.Bycontrast,whenitcomestopronouncingonthetruthofinstancesof“byutteringx,Umeantthatp,”Gricebelieveshecanhelphimselftosuchintuitions,manyofthemquitesubtle.Strictlyspeaking,then,sayingistobedefinedratherthananalyzed.Tosomephilosophersandlinguists,Grice’sprogramseemstoconstitutesomethingofasnubtoseriouscompositionalsemantics.Theideathatsentencemeaningistobeanalyzedintermsofutterer’smeaninghasbeenfelttoconflictwith(1)thefactthatknowingthemeaningofasentenceistypicallyanecessarystepinworkingoutwhatUmeantbyutteringthatsentence,i.e.forrecoveringU’scommunicativeintentions,and(2)thefactthatthemeaningofasentenceisdetermined,atleastinpart,bythemeaningsofitsparts(i.e.wordsandphrases)andthewaythepartsareputtogether(syntax).BothofthesechargesarebasedonmisunderstandingsofGrice’sproject,aswillbecomeclear.Utterer’smeaningThebasicGriceananalysisofutterer’smeaningisthis:I.“Byutteringx,Umeantsomething”istrueiffforsomeaudienceA,Uutteredxintending:(1)Atoproducesomeparticularresponser,(2)AtorecognizethatUintends(1),and(3)A’srecognitionthatUintends(1)tofunction,inpart,asareasonfor(1).Toprovideaspecificationofr,saysGrice,istosaywhatUmeant.Wherexisan“indicative”utterance,risA’sbelievingsomething.II.“Byutteringx,Umeantthatp”istrueiffforsomeaudienceA,Uutteredxintending:(1)Atobelievethatp,(2)and(3)asabove.266\nH.P.GRICEThistypeofcomplexintentionGricecallsan“M-intention”:byutteringx,UmeantthatpiffforsomeaudienceA,UutteredxM-intendingAtobelievethatp.TwogeneralproblemsfaceII.ThefirstisthatGriceprovidesanumberofexamplesinwhichitwouldbecorrecttosaythatUmeansthatpbutincorrecttosaythatUintendsAtobelievethatp(1989:105–9).SupposeUisansweringanexaminationquestionandsays“TheBattleofWaterloowasfoughtin1815.”HereUmeantthattheBattleofWaterloowasfoughtin1815;butUdidnotM-intendtheexaminertothinkthattheBattleofWaterloowasfoughtin1815(typically,Uwillbeundertheimpres-sionthattheexamineralreadyknowstheanswer).Inresponsetothisandrelatedexamples,Gricesuggeststhatclause(1)ofII.bechangedto(11):(11)AtothinkthatUthinksthatp.Adistinctionisthenmadebetweenexhibitiveutterances(utterancesbywhichUM-intendstoimpartthebeliefthatUhasacertainpropositionalattitude)andprotrepticutterances(utterancesbywhichUM-intends,viaimpartingabeliefthat[U]hasacertainpropositionalattitude,toinduceacorrespondingattitudeinthehearer).Thesuggestedrevisionmaynotseemtocomportwiththecommonlyheldviewthattheprimarypurposeofcommunicationisthetransferofinformationabouttheworld;ontherevisedaccount,theprimarypurposeseemstobethetransferofinformationaboutone’smentalstates.Anotherworryisthateveniftheproposedrevisionisanimprovement,itdoesnotweakentheanalysisinsuchawayastoletincasesofremind-ing(somecasesofwhichbringupanotherproblem).SupposeUknowsthatAthinksthatpbutneedsreminding.SoUdoessomethingbywhichhemeansthatp.Notonlydoesitseemincorrecttosay(astheoriginalanalysiswouldrequire)thatUintendsAtothinkthatp–UknowsthatAalreadythinksthatp–italsoseemsincorrecttosay(asthemodifiedanalysisrequires)thatUintendsAtothinkthatUthinksthatp(UmayknowthatAalreadythinksthatUthinksthatp).Whatseemstobeneededhere,saysGrice,issomenotionofanactivatedbelief:(1)needstobechangednotto(11)buttosomethingmorelike(12):(12)AactivelytobelievethatUthinksthatp.Butthereseemsstilltobeaprobleminvolvingreminding.SupposeAhasinvitedBoverfordinnertonightatseven-thirty.BhasagreedtocomebutUdoubtsBwillshowupandsaysasmuchtoA.Atseveno’clock,UandAaredeepinphilosophicalconver-sationandU,realizingthatAhaslosttrackoftime,says“Bwillbehereinhalfanhour.”Thistypeofexamplesuggestswearebetteroffwithsomethinglike(13),atleastforsomecases:(13)Aactivelytobelievethatp.Soperhapsadisjunctiveclauseisgoingtoberequiredinanyfinallyacceptableanalysis.PerhapstheproblemwiththefirstclauseofII.isaninstanceofamoregeneraldif-ficultyconcerningthecontentoftheintention(orM-intention)characteristicofcom-municativebehavior.ThisseemstobetheviewofSearle(1969).OnewayofputtingSearle’sgeneralpointisasfollows:bypayingtoomuchattentiontoexamplesinwhichUintendstoinduceinAsomepropositionalattitudeorother,Gricehasmistakenlytakenaparticulartypeofintentionthatdoesinfactaccompanymanyutterances–the267\nSTEPHENNEALEsubintentionspecifiedinclause(1)–tobeanessentialingredientofcommunicativebehavior.Buttherearejusttoomanycasesofmeaninginvolvinglinguistic(orother-wiseconventional)utterancesinwhichUdoesnotseektoinduceinanaudienceanypropositional(oraffective)attitude.Searlebringsupthreeproblems:first,itisnotatallclearwhatattitudeIM-intendtoimpartwhenmakingapromisebyutteringasentenceoftheform“Ipromiseto__”;second,sometimesIdon’tcarewhetherIambelievedornot;Ijustfeelitismydutytospeakup;third,onlyanegocentricauthorintendsmetobelievethatpbecausehehassaidso.ThesearegenuinedifficultiesforGrice’sanalysisasitstands,buttheydonotseemtowarrantabandoningGrice’sproject;rathertheysuggestthatthespecificationofthetypeofresponsementionedinthefirstclauseneedstobeweakenedtosomethinglikethefollowing:(14)Aactivelytoentertainthebelief/thought/propositionthatp.Ofcourse,inmanycasesUalsointends(oratleastwouldlike)Atogoontobelievethatp,butthisfactwouldnotenterintotheanalysisofutterer’smeaning.ArevisionalongtheselinesmightprovidethebeginningofawayoutofSearle’sproblems.Thesecondproblemisthatclause(3)ofII.seemsproblematic.Theoriginalmotiva-tionforclause(2)isclear.Itisnotenough,Gricepointsout,forUtomeanthatp,thatUutterxintendingAtothinkthatp.UmightleaveB’shandkerchiefnearthesceneofthemurderwiththeintentionofgettingthedetective(actively)toentertainthethoughtthatBisthemurderer.Butthereisnotemptationtosaythatbyleavingthehandkerchief,UmeantthatBisthemurderer.Henceclause(2),whichrequiresUtointendAtorecognizetheintentionspecifiedinthefirstclause(howeverstated).Butwhatofclause(3)?Gricewantsthisinordertofilteroutcasesinwhichsomenaturalfeatureoftheutterancemakesitcompletelyobviousthatp.Heworriedaboutcaseslikethis:inresponsetoaninvitationtoplaysquash,Billdisplayshisbandagedleg.AccordingtoGrice,wedonotwanttosaythatBillmeantthathislegwasbandaged(thoughwemightwanttosaythathemeantthathecouldnotplaysquash,oreventhathehadabadleg).Manypeople’sintuitionsarelessrobustthanGrice’shere.Heseemstobeworriedthatincaseslikethesethereissomethingapproximatingnaturalmeaningthatinter-fereswiththeideaofBillnon-naturallymeaningthathehasabandagedleg.GiventhelinksGriceseekstoforgebetweennaturalandnon-naturalmeaning,itisnotclearwhytheputativepresenceofnaturalmeaningissupposedtobeproblematic,andsoitisnotclearwhythethirdclauseofII.isneeded.Gricehimselfbringsupcasesthatseemtocreateaproblemforthethirdclause.Supposetheanswertoacertainquestionis“onthetipofA’stongue.”Uknowsthis;thatis,UknowsthatAthinksthatpbutcan’tquiteremember.SoUremindsAthatpbydoingsomethingbywhichhe(U)meansthatp.Insuchascenario,evenifUhastheintentionspecifiedinthefirstclause(howeverstated),itdoesnotseemtobethecasethatUhastheintentionspecifiedinthethirdclause.ItisnoteworthythattheexamplesGriceusestojustifythethirdclauseinvolvenon-linguisticutterances(Grice’s“JohntheBaptist”and“bandagedleg”cases).However,itispossibletoconstructcasesinvolvingproperlylinguisticutterancesinwhichthefactthatpismadejustasobviousbytheutteranceasinGrice’snon-linguisticcases.Consideranutterancebymeof(e.g.)“I’mrighthere”yelledinthedirectionof268\nH.P.GRICEsomeoneknowntobelookingforme.HerethereisastronginclinationtosaythatIdidnotmeanwhatIsaid.ProblemsawaitGriceifhedoesnotconcedethethirdclauseisoverlyrestrictive.Ultimately,hewantstodefinelocutionsoftheform“byutteringx,Usaidthatp”;butoneoftheconjunctsinhisproposeddefiniensis“byutteringx,Umeantthatp.”Soifherefusestoallowthat(e.g.)IcanmeanthatIcanspeakinasqueakyvoicebyutter-ing,inasqueakyvoice,“Icanspeakinasqueakyvoice,”GricewillbeforcedeithertoconcludethatIhavenotsaidthatIcanspeakinasqueakyvoice,orelsetoabandontheideaofdefiningsayingintermsofutterer’smeaning(hecannot,ofcourse,saythatinsuchascenarioIhaveonly“madeasiftosay”thatIcanspeakinasqueakyvoice).Itwouldseem,then,thatthethirdclausewillhavetobediscarded(oratleastmodified)ifsayingrequiresmeaning.Onepositiveresultofdiscardingthethirdclausewouldbethedisappearanceofthe“tip-of-the-tongue”problem.AnotherwouldbethatBillcouldmeanthathehadabandagedleginthescenarioabove,whichisnotobviouslyincorrect.Whenitcomestolinguisticutterances,theremightwellbeanotherinterestingconsequence.Typically,linguisticutterancesdonotseemtobeunderwrittenbyintentionsascomplexasM-intentions.Weakeningtheanalysansbytheremovingclause(3)goesalongwaytowardquietingthisworry;however,therearegroundsforthinkingthattherelevantintentionwillhavetobemorecomplexthantheonespecifiedbyclauses(1)and(2).Thefollowingtypeofexampleshowsthatclauses(1),(2),and(3)donotspecifyarichenoughintention(orbatchofintentions).SupposeA,afriendofmine,isabouttobuyahouse.Ithinkthehouseisrat-infested,butIdon’twanttomentionthisoutrighttoAsoIletratslooseinthehouseknowingthatAiswatchingme.IknowthatAdoesnotknowthatIknowthatheiswatchingmedothis.IknowAwillnottakethepresenceofmyratstobenaturalevidencethatthehouseisrat-infested;butIdoknow,indeedIintend,thatAwilltakemylettingratslooseinthehouseasgroundsforthinkingthatIintendtoinduceinhimthebeliefthatthehouseisrat-infested.Conditions(1)–(3)ofII.abovearefulfilled.ButsurelyitisnotcorrecttosaythatbylettingratslooseinthehouseImeanthatthehouseisrat-infested.Theproblemisthatinthisexamplemyintentionsarenotwhollyovert.Onepossibleremedyinvolvesaddingafourthclause:(4)AtorecognizethatUintends(2).Butthesamesortofcounterexamplecanstillbegenerated,andthenweneedafifthclause,thenasixth,andsoon.Griceproposedtoblockaninfiniteregressbyaddingaconditionthatwouldprohibitany“sneaky”intention:insteadofaddingadditionalclauses,hisideawastoaddasecondparttotheanalysis,theroughimportofwhichisthatUdoesnotintendAtobedeceivedaboutU’sintentions(1)–(3).AslongasUdoesnothaveadeceptiveintentionofthissort,Uisdeemedtomeanthatp.Somethinglikethefollowingisbestseenasthecharacterizationofutterer’smeaningthatGriceleftustoexploreandrefine:III.Byutteringx,UmeantthatpiffforsomeaudienceA,269\nSTEPHENNEALE(1)UutteredxintendingAactivelytoentertainthethoughtthatp(orthethoughtthatUbelievesthatp)(2)UutteredxintendingAtorecognizethatUintendsAactivelytoentertainthethoughtthatp(3)UdoesnotintendAtobedeceivedaboutU’sintentions(1)and(2).SentencemeaningandsayingTheideaofusingutterer’smeaningtoexplicatesentencemeaningisthoughtbysomephilosopherstoconflictwiththeideathatthemeaningofasentenceisafunctionofthemeaningsofitsparts(i.e.wordsandphrases)andtheirsyntacticalorganization.Grice’sprojectgetssomething“backwards”itisclaimed:surelyanyattempttomodelhowweworkoutwhatsomeonemeansonagivenoccasionwillprogressfromwordmeaningplussyntaxtosentencemeaning,andfromsentencemeaningpluscontexttowhatissaid,andfromwhatissaidpluscontexttowhatismeant.AndthisclasheswithGrice’sviewthatsentencemeaningisanalyzableintermsofutterer’smeaning.Butthisisincorrect.SupposethereisasentenceYofalanguageLsuchthatYmeans(pre-theoreticallyspeaking)“NapoleonlovesJosephine”(e.g.ifLisEnglish,thenthesentence“NapoleonlovesJosephine”willdo).WhenL-speakerswishtomeanthatNapoleonlovesJosephinetheyaremorelikelytouseYthanasentenceZthatmeans(pre-theoreticallyspeaking)“Wisdomisavirtue.”TosaythisisnottosaythatitisimpossibleforUtomeanthatNapoleonlovesJosephinebyutteringZ,it’sjusttosaythatnormally(usually,typically,standardly)UhasamuchbetterchanceofgettingacrosstheintendedmessagebyutteringY.Thusitmightbesuggestedthatanarbitrarysen-tenceXmeans(inL)“NapoleonlovesJosephine”iff(roughly)byutteringX,optimally,L-speakersmean(would/shouldmean)thatNapoleonlovesJosephine.GriceisnotcommittedtotheabsurdpositionthatahearermustworkoutwhatUmeantbyutteringasentenceXinordertoworkoutthemeaningofX.ToseethisasaconsequenceofGrice’stheoryistoignoretheconnectionbetweenthetheoryofcon-versationandthetheoryofmeaning.ItisGrice’sviewthattypicallythehearermustestablishwhatUhassaid(ormadeasiftosay)inordertoestablishwhatUmeant;anditisbytakingintoaccountthenatureandpurposeofrationaldiscoursethatthehearerisabletoprogress(via,forexample,conversationalimplicature)fromwhatUhassaid(ormadeasiftosay)towhatUmeant.Ananalysisofsentencemeaningintermsofutterer’sintentionsdoesnotconflictwiththisidea.Wemustdistinguish(1)accountsofwhatUsaidandwhatUmeantbyutteringXand(2)accountsofhowhearersrecoverwhatUsaidandwhatUmeantbyutteringX.WhatUmeantbyutteringXisdeterminedsolelybyU’scommunicativeintentions;butofcoursetheformationofgenuinecommunicativeintentionsbyUisconstrainedbyU’sexpectations:UcannotbesaidtoutterXM-intendingAtoøifUthinksthatthereisverylittleornohopethatU’sproductionofXwillresultinAø-ing.IfUM-intendsAactivelytoentertainthebeliefthat(Uthinks)NapoleonlovesJosephine,andUandAarebothEnglishspeakers,UmaywelluttertheEnglishsentence“NapoleonlovesJosephine.”TosaythisisnottocommitGricetotheviewthatsentencesthatarenotdirectly(orsodirectly)connectedtothepropositionthatNapoleonlovesJosephinemaynotbeemployedtothesameeffect.270\nH.P.GRICEOnthecontrary,thetheoryofconversationissupposedtoprovideanexplanationofhowthisispossible(intherightcircumstances).OntheassumptionthatUandAarebothoperatinginaccordancewiththeCooperativePrincipleandmaxims,theremaywellbefactsaboutthecontextofutterance,thetopicofconversation,backgroundinformation,andsoonthatmakeitpossibleforUtomeanthatNapoleonlovesJosephinebyutteringaverydifferentsentence.U’sconceptionofsuchthingsasthecontextofutterance,thetopicofconversation,backgroundinformation,andA’sabilitytoworkoutwhatUisuptomayallplayrolesintheformationofU’sintentions;butthisdoesnotunderminetheviewthatwhatdetermineswhatUmeansareU’scom-municativeintentions.Wecanputaside,then,thequestionoftheconceptualcoherenceofGrice’sana-lyticalprogram;theinterestingquestionsconcerntheadequacyofhisconcreteproposalsforexplicatingsentencemeaningandsaying.Thebasicideaistoanalyzesen-tencemeaningintermsofutterer’smeaning,andthendefinesayingintermsofanearcoincidenceofutterer’smeaningandsentencemeaning.SentencemeaningforGriceisaspeciesofcompleteutterance-typemeaning,therelevantanalysandumforwhichis“Xmeans‘p’,”whereXisanutterancetypeandpisaspecificationofX’smeaning.Griceputsforwardthefollowingasindicativeofthegeneralapproachheisinclinedtoexplore:IV.ForpopulationgroupG,completeutterance-typeXmeans“p”iff(a)atleastsome(many)membersofGhaveintheirbehavioralrepertoirestheprocedureofutter-ingatokenofXiftheymeanthatp,and(b)theretentionofthisprocedureisforthemconditionalontheassumptionthatatleastsome(other)membersofGhave,orhavehad,thisprocedureintheirrepertoires.Foralanguagecontainingnocontext-sensitiveexpressions,thetechnicaldifficultiesinvolvedinGrice’suseofthevariable“p”bothinandoutofquotescanberemediedeasilyenough.Butonceweturn(aswemust)tocompleteutterance-typemeaningforalanguagethatcontainsindexicalssuchas“I”and“you,”demonstrativessuchas“this”and“that,”andanaphoricpronounssuchas“him”and“her,”itisclearthatsomeworkisneededtotransformIV.intosomethingacceptable.Thisis,Ithink,averyseriousmatter;forwithoutsuchatransformation,Gricesimplywillnotbeabletoprovideananalysisofutterance-typemeaningforalanguagelikeEnglish,andconse-quentlyhewillnotbeabletoprovidethesortofdefinitionofsayinghewants.What,then,isthepreciserelationbetweensentencemeaningandsayingforGrice?Itmightbethoughtthatifweabstractawayfromtheproblemsraisedbyindexicalsandotherexpressionsthathighlightthegapbetweensentencemeaningandwhatissaid,wewillbeabletomovedirectlyfromwhenutteredbyU,Xmeant“p”tobyutteringX,Usaidthatp.Buttherearetwoproblemshere.First,onlywhereanutterance-typehascertainfeaturesdowewanttosaythatatokenofthattypemaybeusedtosaysome-thing.Amotoristdoesnotsayanything,inGrice’ssense,whenheindicatesaninten-tiontomakealeftturnbysignalling.Second,certaincasesinvolving,forexample,ironyorconversationalimplicaturecanbeusedtoshowthatwecannotmaketherelevantmovedirectly.IfUuttersthesen-tence“Smithisanhonestman”ironically,althoughitwouldbetruetosaythatthesentenceinquestionmeans“Smithisanhonestman,”itwouldnotbetruetosaythat271\nSTEPHENNEALEUissayingthatSmithisanhonestman.OnGrice’saccount,sinceUdoesnotmeanthatSmithisanhonestman(UhasnointentionofgettingAtobelievethat(hebelievesthat)Smithisanhonestman)UisonlymakingasiftosaythatSmithisanhonestman.(ParallelremarkscouldapplyinthecaseofProfessorU’sutteranceofthesentence“MrXhaswonderfulhandwritingandisalwaysverypunctual.”)OnGrice’saccount,whatissaidistobefoundintheareawheresentencemeaningandutterer’smeaningoverlap.Abstractingawayfromcontext-sensitiveexpressionsonceagain,itlooksasthoughsomethinglikethefollowingpreliminarydefinitionisontherighttrack:V.ByutteringX,Usaidthatpiff(1)byutteringX,partofwhatUmeantwasthatp(2)Xconsistsofasequenceofelements(suchaswords)orderedinawaylicensedbyasystemofrules(syntacticalrules),and(3)Xmeans“p”invirtueoftheparticularmeaningsoftheelementsinX,theirorderandtheirsyntacticalstructure.Grice’sunhappinesswithV.derivesfromtheexistenceofconventionalimplicatures.RecallthatGricedoesnotwanttoallowthesortsofimplicationsthatresultfromtheuseofwordssuchas“but,”“yet,”“still,”“even,”and“moreover,”tocountaspartofwhatissaid.Forexample,ifU(sincerelyandnon-ironically)uttersthesentence“Sheispoorbutsheishonest,”Udoesnotsaythatthereissomesortofcontrastbetweenpovertyandhonesty(orbetweenherpovertyandherhonesty).Rather,Uperformsa“centralspeechact,”bywhichUsaysthatsheispoorandsheishonest,andperformsinadditiona“noncentralspeechact,”bywhichUconventionallyimplicatessomesortofattitudetowardwhatissaid.PuttingtogetherwhatUsaysandwhatUconvention-allyimplicateswegetwhatUconventionallymeans(seefigure1).SoforGrice,atbestthethreeconditionsinV.definebyutteringX,UconventionallymeantthatpratherthanbyutteringX,Usaidthatp.ThisisasfarasGricegoes.Heleavesuswiththenon-trivialtaskofseparatingwhatUsaysandwhatUconventionallyimplicates,aratherdisappointingterminus.ThenotionofwhatissaidisforGriceafundamentallyimportantnotioninphilosophy.Ifthisorthatphilosopherisunclearaboutwhatheissaying(asopposedtowhatheorsheisimplicating)thenthatphilosopherisliabletomakeallsortsofmistakes,asisborneout,Gricethinks,bythecrudewayinwhich,forinstancethecausaltheoryofperceptionandthetheoryofdescriptionshavebeenwrittenoffbyphilosopherscon-cernedwiththenuancesofordinarylanguage.Furthermore,notuntilwhatissaidandwhatisconventionallyimplicatedareseparatedcanwhatisconversationallyimplicatedbedefinedinthemannerexaminedearlier.SoforGrice,ananalysisofsayingtakesonsomeurgency,anditisunfortunatethathedoesnotgetanyclosertoonethanhedoesinproducingV.above.However,itmaywellbethatGricehasbroughtusasfaraswecangowithoutcrossingourownpaths.Recallthathewantswhatissaidtocomprisethetruth-conditionalcontentofwhatisconventionallymeantbysomeonemakingastatement;buthecannotappealdirectlytotruthconditionsforfearofunderminingonepartofhisproject.Theremaybenosimplewayoutofthis.Atthesametime,onlyonepartofGrice’sprojectisthreatened:thepossibilityofprovidingadefinitionofsayingintermsofutterance-typemeaning272\nH.P.GRICEandwhatismeant.Noappealtotruth-conditionalcontentisneededinanalysesofutterer’smeaningorutterance-typemeaning,andtothatextentGricehascertainlyilluminatedtheseimportantnotions.Insodoing,hehasalsoalertedustoahostofimportantdistinctionsthatphilosophers,linguists,cognitivescientists,andliterarytheoristsignoreattheirperil.BibliographyWorksbyGrice1989:StudiesintheWayofWords,Cambridge:CambridgeUniversityPress.1991:TheConceptionofValue,Oxford:ClarendonPress.WorksbyotherauthorsAvramides,A.(1989)MeaningandMind,Cambridge,MA:MITPress.Grandy,R.E.andWarner,R.(eds.)(1986)PhilosophicalGroundsofRationality:Intentions,Categories,Ends,Oxford:ClarendonPress.Martinich,A.P.(1984)CommunicationandReference,NewYork:WalterdeGruyter.Schiffer,S.(1972)Meaning,Oxford:ClarendonPress.Searle,J.R.(1969)SpeechActs,Cambridge:CambridgeUniversityPress.Sperber,D.andWilson,D.(1986)Relevance,Oxford:BlackwellPublishers.273\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200121G.H.vonWright(1916–)FREDERICKSTOUTLANDGeorgHenrikvonWrightwasbornandeducatedinHelsinki,Finland,wherehisgraduateworkwassupervisedbyEinoKaila,adistinguishedFinnishphilosopherwhowasconnectedwiththeViennaCircle,andwhoencouragedvonWrightinthatdirec-tion.Becauseofconditionsonthecontinent,however,vonWrightwenttoCambridgein1939tostudywithC.D.Broad,andtherehemetWittgensteinandMoore,who,togetherwithKaila,becamethemajorinfluencesonhisphilosophicaldevelopment.HereturnedtoHelsinkiafterayearandin1941receivedaPh.D.forhisthesis“TheLogicalProblemofInduction.”Afterthewar,hereturnedtoCambridgeasalecturerandin1948wasappointedtotheChairofPhilosophyinCambridge,whichWittgensteinand,beforehim,Moorehadheld.In1951heresignedtoreturntoHelsinkiandresumetheprofessorshiptowhichhehadbeenappointedin1946.In1961hewasappointedtotheAcademyofFinland,whichatthetimecomprisedtwelvepersons,whoselifetimemembershipintheAcademyfreedthemtodotheirownwork.Hehascontinuedtoteachandlecturearoundtheworld,notablyasProfessor-at-LargeatCornell,ingiving(amongothers)theGifford,Tarner,Woodbridge,Tanner,andLeibnizLectures,andinengagingphilosophersfrommanycountriesinphilosophicalconversationofthehighestquality.Heisthesubjectofavolumeinthe“LibraryofLivingPhilosophers”(1989),whichcontainsabibliographyofover400papersandbooks(severaldozenmorehaveappearedsince),essaysonhisworkbythirty-onephilosophers,andanotablyinstructiveandinterestingautobiography.VonWright’sphilosophicalcareerhasbeenmarkedbyhisworkingthroughonesetofissues,thenleavingittofocusonanotherset,andsoon,andIhaveorderedmydis-cussionofhiscontributionstoreflectthosestagesinhiscareer.Theorderisbynomeansexact,however,becausecertainthemeshavepersistedthroughouthisworkandbecauseheoftenreturnstoissueswhenhehasnewthingstosay.InductionandprobabilityIssuesrelatedtoinductionandprobabilitywerethefocusofhisworkfromhisstudiesforhisPh.D.untilthelate1940s.Thetopicofhisdoctoraldissertationwasthe“canonsofinduction,”whichBaconandMillusedtoascertaincausesandeffects,andwhichvonWrightreplacedwiththemoreprecisenotionsofnecessaryandsufficientcondi-274\nG.H.VONWRIGHTtionsinordertorestateandre-evaluatetheclassicalcanons.Thiswasextended,cor-rected,andrelatedtoprobabilitytheoryinanumberofpapersandtheninhisTreatiseonProbabilityandInduction,writtenin1948butnotpublisheduntil1951.Hesumsupthisworkasfollows:Ihavetriedtoshowhowtheprobabilifyingeffectofevidenceonagivenhypothesisisanisomorphicreflectioninnumericaltermsofaprocessofeliminatingmembersfromaclassofhypothesesinitiallycompetingwiththegivenone.ThiseliminativeprocedureisthelogicalcoreofMill’scanons.Myaimcanthusbesaidtohavebeenaunificationofthetwomainbranchesofinductivelogic:inductionbyeliminationinthetraditionofBaconandMill,andinductionbyconfirmationinthetraditionfoundedbytheCambridgelogiciansJ.M.Keynes,C.D.Broad,andW.E.JohnsonandlatercontinuedbyCarnapandothers.(1993:114)Hispublicationsfromthisperiodalsoincludestudiesoftheparadoxesofconfirmation,atopictowhichhehasreturned(see1983b),ashehastoinductionandprobabilityinwritingarticlesoneachforthe1959editionoftheEncyclopediaBritannica.PhilosophicallogicVonWright’sworkoninductionandprobabilitywasnotintendedasacontributiontomathematicaltheorieslikethoseofvonMisesorKolmogorov,butasaphilosophicalstudywhichusedformalmethodsasawayofunderstandingandimprovingthecon-ceptsweemployinevaluatingcertainkindsofempiricalhypotheses.Thesameistrueofhisworkinlogic:itaimednottobeacontributiontomathematicallogicinthestyleofGödel,Tarski,orChurchbuttobeaphilosophyoflogicandaphilosophicallogic(seeTARSKI,CHURCH,GÖDEL).Hetookthetaskoflogicgenerallytobe“todescribeandsys-tematizetheprinciplesusedinargumentation,inference,andproof,”theaimofhisownworkbeingphilosophicalreflectiononthoseprinciplesandtheconceptstheyinvolve.Inthecourseofthatwork,anewaimemerged:toextendtheapplicationoflogic,asithaddevelopedsinceFrege,tosubjectmattersthattraditionallymadenoexplicituseoflogicalsymbolsandmethods.Thefocusofthisworkwastheconceptoflogicaltruth,whichhebegantoinvesti-gatebyconsideringhowfartheTractariannotionoflogicaltruthastautologicalcouldbeextendedtoquantificationallogic.Heshowedthatthiscouldbedoneforsimplerquantificationalstructures,andinsodoinginvented“distributivenormalforms,”whichothersappropriatedfortechnicaluses(whichincludedshowingpreciselyhowfarthenotionofatautologycouldbeextended).Thisworkledtosuchrelatedtopicsasconditionals,entailment,negation,andthelogicalantinomies,onwhichvonWrightwroteclarifyingandstimulatingpapersofthehighestquality,fromwhichmuchmaybelearned.(TheearlierworkisinLogicalStudies(1957)andthemorerecentin“PhilosophicalLogic”(1983b).)Whileworkingonthequantifiers,henoticedaparallelismbetweenthestructureof“some,”“none,”and“all”andthatofthemodalterms“possible,”“impossible,”and“necessary.”Justasthenegationof“SomeSareP”isequivalentto“NoSareP,”sothenegationof“Pispossible”isequivalentto“Pisimpossible,”andjustasthenegationof275\nFREDERICKSTOUTLAND“SomeSarenotP”isequivalentto“AllSareP,”sothenegationof“Not-Pispossible”isequivalentto“Pisnecessary,”andsoon.Alittlelater,henoticedthesameparal-lelismforthedeonticmodalterms“permissible,”“forbidden,”and“obligatory”:“Aisnotpermissible”isequivalentto“Aisforbidden”;“Not-Aisnotpermissible”isequiva-lentto“Aisobligatory,”etc.Hisprojectwastoarticulatethesemodalitiesinlogicalsystems,analogoustopropositionalandpredicatelogic,whichwouldhavepreciserulesforwell-formedformulaeandvalidinferences,whichwouldenableexactdetermina-tionofwhataclaimentailedandwhatitcontradicted,andwhichwouldpermitinves-tigationofmetalogicalissueslikeconsistencyandcompleteness.Hisideasonhowtodothisforthestrictmodalitieswerewrittenin1950(published1951b),whilehisproposalsforthesecondappearedinhispaper,“DeonticLogic,”inthefirstissueofMindfor1951(reprinted1957).VonWrightshowedthatfurtherconceptsexhibitedasimilarstructure,forexample,theepistemicmodalities“undecided,”“falsified,”and“verified”(ifPisundecided,thenPisnotfalsified;ifPisfalsified,thennot-Pisverified,etc.),andeventimeandcausal-itycanbeseenashavingakindofmodalstructure.VonWrightworkedonalltheseconcepts,showinghowlogicalprinciples,concepts,andmethodscouldbeappliedtothem,thusextendinglogicalinvestigationsintonewareas.(Forhissummary,see1993(essaysVIandVII)and1989.)Thisinspiredmanyotherstoapplylogicaltechniquesandsymbolismtodiversephilosophicaltopics,andwhilethisworkistoooftentechnicallyorphilosophicallyuninteresting,someofithasbeenverysignificantforbothlogicandphilosophy.Thelatter,significant,workhaspromptedtheviewthatthisisthebest–eventheonly–waytodoanalyticphilosophy,butthatisaviewvonWrightemphati-callyrejects.Ethics,norms,andvalues“DeonticLogic”isvonWright’smostfamouspaper,foritcreatedanewsubject,whichwashisowninaspecialsense.Itshowedhowtosymbolizevariousnormativeclaimsandpreciselydeterminetheirlogicalinterrelations,anditopenedupnewclaimsandrelationswhicharedifficulttonoticeoutsidealogicalsystem.This“logicofnorms”remainsofgreatinteresttolegalphilosophers,forinstance,whowrestlewithsuchquestionsaswhethertwolawsaremutuallyconsistentorwhether,andinwhatsense,asystemoflawsmaybeconsistentorcomplete.VonWright’sinterestinthelogicofnormswassustainedbecauseofitsconnectionwithissuesabouttruth.Hehadsincehisyouthbeencommittedtowhathecallsa“deepconceptualgapseparatingtheworldoffactsfromthatofnormsandvalues,”andhecontinuestoholdthatbeliefinitsradicalformastheviewthatnormativejudgmentsareneithertruenorfalse.Thisraisestheproblemofhowalogicofnormsisevenpos-sible,forsuchalogicassumesthattherearelogicalrelationsbetweennormsandthattherearedisjunctiveandconditionalnormsentences,andthoseassumptionsappeartorequirethatnormsentenceshavetruth-value.Indealingwiththisproblem,vonWrightappealstothedistinctionbetweensen-tenceswhichgiveanorm,andthusareprescriptive,andthosewhichstatethatanormhasbeengiven,andthusaredescriptive,whichyieldstwopossibilitiesforconstructingalogicofnormswithoutassumingthatnormativejudgmentsaretrueorfalse.Thefirst276\nG.H.VONWRIGHTistohavealogic,notofnormsassuch,butofthesentenceswhichdescribegivennormsandthushavetruth-values.Thesecondistohavealogicofnormsproperbutdonotrequirethatlogicallycomplexsentenceswithlogicalinterrelationsmusthaveatruth-value,sothat“LogichasawiderreachthanTruth.”VonWrighthaschangedhismindmorethanonceonwhichofthesewaysisbetter(see1963a,1968,1984),andmorerecentlyhassuggestedathirdpossibility,whichisthatdeonticlogicis“neitheralogicofnormsnoralogicofnorm-propositionsbutastudyofconditionswhichmustbesatisfiedinrationalnorm-givingactivity”(1993:111;cf.1996).NormandAction(1963a)andAnEssayinLogicandGeneralTheoryofAction(1968)extendtheselogicalinvestigationstotheactionswhichnormsgovernandtothechangesinvolvedinactions(whichIdiscussbelow).Butthestudyofnormsisalsopartofmoralphilosophy,whichvonWrightwaspursuingatthesametime,largelybecauseofhisteachingduties;itresultedinTheVarietiesofGoodness,hismainworkinmoralphilosophy.Itscentralclaimisthatmoralsensesof“good,”andof“right”or“duty,”arederivativeonnon-moralusesoftheterms.Itisaconceptualinquiry,whichaimsatmaking“fixedandsharpthatwhichordinaryusageleaveslooseandundetermined,”butsinceitdeniesthatthereisanycleardistinctionbetweenethicsandmetaethics,itsinquiriesintohowtoimproveourmoralconceptsarealsomeanttobeinquiriesintowhatmoralpointofviewweoughttoadopt.Itassumesthattherecanbe“aphilo-sophicalpursuitdeservingthename‘ethics’,whichshareswithacommonconceptionof‘meta-ethics’thefeatureofbeingaconceptualinvestigationandwithacommonconceptof‘normativeethics’thefeatureofaimingatdirectingourlives”(1963b:6).ThebookisAristoteleaninitsinsightfulclassificationofthevarietiesofgoodness,initstakingseriouslythenotionofvirtue,initstakingthehumangoodtobemorebasicthanduty,andinitscommitmenttotherebeingsuchathingaspracticalrationality,whichisnotinferiortotherationalityoftheexactsciences.Thesenotionsarenotraretoday,buttheywereaudaciousin1963,anditisunderstandablethatvonWrightregardsTheVarietiesofGoodnessasthebestarguedofallhisworksandthemostfuntowrite.PhilosophyofactionVonWright’sworkinphilosophyofactionstemmedfromhisattempttogetasymbol-ismadequatetodistinguishthevariouskindsofactiongovernedbynorms:todistin-guish,forexample,whatagentsdo(eitherbybringingsomethingaboutorpreventingsomething)fromwhattheydonotdo(eitherintentionallyorbysimplydoingnothing).Itstruckhimthatsuchdistinctionscouldnotbeexpressedwithoutasymbolismfordif-ferentkindsofchange,andhenceheworkedatembeddingalogicofactioninalogicofchange,aprojecttowhichhehasreturnedanumberoftimes(1963a,1968,1973).BroaderissuesaboutactionwereimplicitinthisworkandbecameexplicitinExplanationandUnderstanding(1971).Itarguedthatexplanationsofintentionalactionarelogicallydistinctfromexplanationsinthenaturalsciencesbecausethelatterappealtocausallawswhiletheformerinvokeconceptualconnectionsbetweenanagent’sreasonsandhisactions.Explanationofactionis,therefore,necessarilyconnectedwithpracticalreason(see1983a),andtheintentionalattitudesthatfunctionasreasonsshouldbeseen,notasinternalstateswithcausalpowers,butaswaysofunder-277\nFREDERICKSTOUTLANDstandingandarticulatingwhatagentsmeanbytheirbehavior.HeconnectedthisviewwiththeAristoteleantradition,contrastingitwiththeGalileantraditionwhichassim-ilatedexplanationofaction(andofhistoricalevents)tothelaw-basedmodelofthenaturalsciences.ExplanationandUnderstandingreinforcedtherejectionofacausalmodelofactionbyarguingthattheconceptofcauseisinseparablefromtheconceptof(experimental)interventioninnature,andhenceinseparablefrom,indeedderivativeon,conceptsofactionandagency.Thisimpliedthatthekindofdeterminismwhichthreatenstounder-minethepossibilityofgenuinehumanagencyisself-defeating,apointvonWrighthasarticulatedinanumberofways(1973).Indeed,determinismhasbeenacoreinterestofhissinceExplanationandUnderstanding,whichhasbeenexpressedpartlyinlogicalinvestigationsintoissuessuchaskindsofnecessityandknowledgeandtruthaboutthefuture(1984),partlyindiscussionsofexplanationandfreewill(1973),partlyincriticismsofthewaydeterminismencouragesa“reifiedconception”ofpeopleandsocialinstitutionsasgovernedbyuniversallaws(1983a,1993).PhilosophyofmindVonWrightfirstencounteredphilosophyasayouththroughthemind–bodyproblem,buthewrotenothingaboutituntilhisTannerLectures,“OfHumanFreedom”in1984(reprintedinIntheShadowofDescartes,1998),whenreflectionontherelationofneuro-physiologicalexplanationandactionexplanationledtosomeremarksonpsycho-physicalparallelism.Hethenbegantowriteextensivelyonthephilosophyofmindbutpublishedlittleuntilhis1998book,whichhaspapersontherelationofphys-iologicalandactionexplanation,ontheconceptsofqualityandthing,andonpercep-tionandsensation(withspecialattentiontosounds).Thepapersdonotformaunityanddonotdiscusscurrentliteratureandcontroversies(ofwhichvonWrightisbynomeansignorant),buttheymanifestanintenseefforttogetclearonsomeofthemostdifficultandfundamentalissuesinphilosophyofmind,andtheycontainawealthofdistinctionsandobservationswhichmayproveproductive.WittgensteinVonWrightknewWittgensteinwell,wasnamedasoneofhisliteraryexecutors,andhasdevotedenormoustimeandefforttohislargeandextraordinarilycomplexNachlass.Hehassearchedforlostmaterial,interviewedpersonswhoknewWittgenstein,organizedandindexedthepapers(seeWittgenstein,vonWright1982),andeditedseveralvolumesofhiscorrespondenceandmanuscripts.WhilehehaswrittenrelativelyfewpapersonWittgenstein’sthought,thesevolumesincludesomesplendidstudies,whichhavebeenpublished(alongwithhismoving“BiographicalSketch”)inhis1982book.WhilevonWrightwasdeeplyimpressedbyhisencounterswithWittgenstein,whonodoubtinfluencedhisworkinnumerousways,hehaschosenhisownpath.HisworkonphilosophyofactionandofmindarethemostWittgensteinianincontent(hislogicalworkistheleast),butnoneofthiscanbetracedbackinanydirectwaytoWittgenstein’swriting.Allofhiswork,however,manifestsWittgenstein’sstressoncon-278\nG.H.VONWRIGHTceptualmultiplicityandhisdistrustofsimpleanswersandphilosophicaltheses,anditsharesWittgenstein’saversiontosuperficial,careless,orpretentiouswriting(seeWITTGENSTEIN).HumanismVonWrightalsoshareswithWittgensteinasensitivitytothelargercontextofphilos-ophy,whosepresentcharacterhealsotendstothinkofas“thedarknessofthistime.”Hehaswrittennumerousessays(mostlyinSwedishorFinnish)onwriterssuchasSpengler,Toynbee,Dostoevsky,andTolstoyandontopicsincludingeducation,thestateofthehumanities,theimageofscience,themythofprogress,andtheideaofrevolu-tion:essaysaimedatclarifyingthemeaningofhislifeasahumanbeingratherthanasaprofessionalphilosopher.(SeeTheTreeofKnowledge,1993.)Hebeganasabelieverinthe“spiritofscientificrationality,”whichwasinspiredaboveallbypost-Cartesianmathematicsandphysics,andwhichinturninspiredthephilosophyofKailaandthelogicalempiricists,thenewlogic,andanalyticphilosophy.Hewasneverabelieverintheideathatscientificrationalitywouldresultininevitableprogress–evenhisyouthful“aesthetichumanism”wasstronglyaffectedbySpengler–buthewasabelieverintheidealofphilosophyasoneoftheexactsciences.Thisattitudechangedwhenhebeganteachingmoralphilosophy,andundertheinfluenceofJaeger’sPaideaheabandoned“aesthetichumanism”foranindividualistic“ethicalhumanism,”whichrecognizedapracticalrationalityonaparwiththeratio-nalityoftheexactsciences.ThiswasshatteredbytheeffecttheVietnamwarhadonhim,awarheprotestedineloquentandeffectivewaysandwhichforcedhimtothinkaboutthehumanconditioninsocialandpoliticalterms.Theresultwasthe“socialhumanism”towhichheremainscommittedandwhichhehasexpressedinpowerfulessaysonthethreattohumanlifeoftheveryscientificrationalitywhichinspiredhisearlyeffortsinphilosophy,essayswhichhavemadehimScandinavia’smostprominent–ifcontroversial–publicintellectual.Hiscriticalattitudetowardtheidealofscientificrationality,aboveallforthemajorroleithasplayedincreatingsocialandtechnologicalconditionswhichboththreatenthenaturalenvironmentandinspireirrationalandunjustwaystotrytomeetthethreat,havealsochangedhisconceptionofphilosophy.Hethinksofit,notasanexactscience,butascriticalreflectionontheunderlyingassumptionsofourthinking,judging,andacting–reflectionwhich,sincetheseassumptionsareembeddedinsocialinstitutionsandtraditions,mustinevitablyscrutinizethefoundationsofsociety(see“IntellectualAutobiography,”1989).AlthoughvonWrightthinksofhishumanistessaysasparalleltoratherthanpartofhisphilosophicalwork,theyexhibitmanyofthesamevirtues.Theyarerichwithilluminatinginsights,unexpectedobservations,andstimulatingsuggestionsforfurtherthought,andtheyarewritteninastyleperfectlysuitedtotheirsubjectandtotheirauthor.Hisstrictlyphilosophicalworkaddssuchfurthervirtuesasexactandintricateargument,systematicdistinctions,andtechnicalsophistication.Thesearethevirtuesprizedbyanalyticphilosophy,whichvonWrighthasnotabandoned,inspiteofreject-ingscientificrationalityastheidealofphilosophyandbeingskepticalaboutitsroleasanidealforanyinquiry.Whatheregardsasthechiefvirtueofphilosophyhasbeen279\nFREDERICKSTOUTLANDbestexpressedintheepigramtohisautobiography,whichisfromMelville’sMobyDick:“Alldeep,earnestthinkingisbuttheintrepideffortofthesoultokeeptheopenindependenceofhersea.”BibliographyofvonWright’swork1941:TheLogicalProblemofInduction,Helsinki:ActaPhilosophicaFennica(2ndrevisededn.,Oxford:BlackwellPublishers,1957).1951a:ATreatiseonInductionandProbability,London:RoutledgeandKeganPaul.1951b:AnEssayinModalLogic,Amsterdam:NorthHolland.1957:LogicalStudies,London:RoutledgeandKeganPaul.1963a:NormandAction:ALogicalInquiry,London:RoutledgeandKeganPaul.1963b:TheVarietiesofGoodness,London:RoutledgeandKeganPaul.1968:AnEssayinDeonticLogicandtheGeneralTheoryofAction,Amsterdam:NorthHolland.1971:ExplanationandUnderstanding,London:RoutledgeandKeganPaul.1973:CausalityandDeterminism,NewYork:ColumbiaUniversityPress.1981:FreedomandDetermination,Helsinki:ActaPhilosophicaFennica.1982:Wittgenstein,Oxford:BlackwellPublishers.1983a:“PracticalReason,”PhilosophicalPapers,vol.1,Oxford:BlackwellPublishers.1983b:“PhilosophicalLogic,”PhilosophicalPapers,vol.2,Oxford:BlackwellPublishers.1984:“Truth,Knowledge,andModality,”PhilosophicalPapers,vol.3,Oxford:BlackwellPublishers.1989:“IntellectualAutobiography,”inThePhilosophyofGeorgHenrikvonWright,TheLibraryofLivingPhilosophers,LaSalle,IL:OpenCourt.1993:TheTreeofKnowledgeandOtherEssays,Leiden:Brill.1996:SixEssaysinPhilosophicalLogic,Helsinki:ActaPhilosophicaFennica.1998:IntheShadowofDescartes:EssaysinthePhilosophyofMind,Dordrecht:KluwerAcademicPublishers.280\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200122RoderickChisholm(1916–1999)PartI:EpistemologyRICHARDFOLEYPartII:MetaphysicsDEANZIMMERMANRoderickChisholm’sworkspanssixdecadesandanimpressiverangeofsubjects.HisbooksandarticlesonBrentanoandMeinong,togetherwithhisworkasatranslatorandeditor,andasdirectoroftheBrentanoFoundation,broughtAnglo-Americanana-lyticphilosophybackintocontactwiththerichesoftheAustrianphilosophicaltradi-tion.Hewroteseveralimportantpapersonthefoundationsofethicsandaxiology(e.g.Chisholm1963,1974).ButChisholmisbestknownforhismanycontributionstoepistemologyandmetaphysics.PartI:EpistemologyThemostimportantofRoderickChisholm’swritingsonepistemologyarePerceiving,TheFoundationsofKnowing,andthethreeeditionsofTheoryofKnowledge.Intheseandinhisotherworks,Chisholmaddressedvirtuallyeverymajorprobleminepistemology.Attheheartofhisepistemologicalsystemisasetofepistemicprinciplesthatareintendedtogenerateintuitivelyplausibleresultsaboutthedegreetowhichvariouspropositionsarejustifiedforanindividual.ThekeytoChisholm’sepistemologyisunder-standinghowtheseprinciplesfittogetherandalsounderstandingboththeirepistemol-ogicalstatus(howisitthatwecancometoknowthem?)andtheirmetaphysicalstatus(aretheynecessaryorcontingent,andwhatisitthatmakesthemtrue?).TermsofepistemicappraisalInformulatinghisepistemologicalprinciples,Chisholmpresentsasetoftermsofepis-temicappraisal,whichhedefinesusingabasic,prephilosophicalnotionofjustification.ThefollowingaresimplifiedversionsofthedefinitionsthatappearinthethirdeditionofTheoryofKnowledge(1989a).281\nRICHARDFOLEYCertain:ApropositionpiscertainforanindividualSifandonlyifnootherproposi-tionismorejustifiedforStobelieve.Evident:ApropositionpisevidentforSifandonlyifSisatleastasjustifiedinbelievingpaswithholdingjudgmentonthatwhichiscounterbalanced.Beyondreasonabledoubt:ApropositionpisbeyondreasonabledoubtforSifandonlyifSismorejustifiedinbelievingpthanwithholdingjudgmentonp.Epistemicallyintheclear:ApropositionpisepistemicallyintheclearforSifandonlyifSisatleastasjustifiedinbelievingpaswithholdingjudgmentonp.Probable:ApropositionpisprobableforSifandonlyifSismorejustifiedinbelievingpthandisbelievingp.Counterbalanced:ApropositionpiscounterbalancedforSifandonlyifSisasjustifiedinbelievingpasbelievingnot-p,andvice-versa.Chisholmintendsthefirstfiveofthesetermstobesuchthatthehigheronesimplythelowerones,andheintroducesaxiomstoensurethis(1989a:12,13,17).So,ifapropositioniscertainforsomeone,itisalsoevidentforthatperson,andifitisevidentfortheperson,itisalsobeyondreasonabledoubt,andsoondownthelist.EpistemicprinciplesMakinguseoftheabovetermsofepistemicappraisal,Chisholmproposesasetofepistemicprinciples.Theprinciplesareexpressedasconditionals,whoseantecedentsdescribesufficientlogicalconditionsfortheapplicationofthesetermsofepistemicappraisal.Inthemoststraightforwardcase,aprinciplewillassertthatifcertainnon-epistemicconditionsaresatisfied(e.g.conditionsaboutwhatsomeoneisexperiencing,believing,etc.),thenapropositionphasacertainepistemicstatusfortheperson(e.g.itisevidentorbeyondreasonabledoubt).Chisholm’sprojectinformulatingtheseprinciplescanbecomparedtoatraditionalprojectinethics.Acentralaimoftheoreticalethicsisthatofdescribingasetofnon-moralconditionsthatissufficienttomakeanactionmorallyright.Accordingtoutilitarians,thenon-moralconditionsareoneshavingtodowiththeproductionofpleasureandtheavoidanceofpain.Ifofallthealternativesavailabletome,alternativeXwillproducethegreatestbalanceofpleasuresoverpains,thenIamrequiredtodoX.Utilitariansclaimthatthisisthefundamentalprincipleofmoral-ity.Forthem,thereisbutonesourceofmoralobligation.Othersdisagree,insistingthatthereareothersourcesaswell,onesthatarenotdirectlyconcernedwiththemaximizationofhappiness.Equalityandfairnessareamongtheusualcandidates.IfdoingXwouldproduceafairresult,then,accordingtothisview,Ihaveaprima-facieobligationtodoXevenifdoingsowouldnotmaximizehappiness.Theremaybeothersourcesaswell,andcorrespondingtoeachofthesesourceswillbeanethicalprinciple,assertingthatthesourceinquestionproducesaprima-faciemoralobligation.ThislatterviewisthecounterpartofChisholm’sviewinepistemology.Hethinksthatthereismorethanonesourceofepistemicjustification,andcorrespondingtoeachofthesesourcesisanepistemicprincipledescribingtheconditionsunderwhichthesourceproducesjustification.However,Chisholmbelievesthatsomeofthesesourcesproducejustificationonlyinconjunctionwithothersources.Thus,theepis-282\nRODERICKCHISHOLMtemicprinciplescorrespondingtothesesourcesmustmakereferencetotheworkingsofotherprinciples.Theresultisacollectionofprinciplesthatareinterdependentincomplexways.BelowaresomeofthemostimportantoftheepistemicprinciplesthatChisholmdefends:1IfFisaself-presentingpropertyandifShasFandifSbelieveshimself(herself)tohaveF,thenitiscertainforSthathe(she)hasF.2IfitisevidenttoSthathe(she)isappearedtof-lyanditisepistemicallyintheclearforSthatsomethingisappearingtohim(her)inthisway,thenitisevidentforSthatsomethingisappearingftohim(her).3IfitisevidenttoSthatSisappearedtof-lyandifSbelievesthatitisaGthatisappearingtohim(her)inthiswayandifthispropositionisepistemicallyintheclearforS,thenitisbeyondreasonabledoubtforSthathe(she)perceivesaG.4IfSbelievesapropositionthatisnotdisconfirmedbythesetofpropositionsthatareevidentforS,thenthepropositionisprobableforS.(AccordingtoChisholm,pdisconfirmsqamountstoptendstomakenot-qprobable.)5IfSbelievesapropositionthatisnotdisconfirmedbythatwhichisprobableforS,thenthepropositionisepistemicallyintheclearforS.6Iftherearethreeormoreconcurrentpropositionsandifeachofthemisepistem-icallyclearforSandifinadditiononeofthemisbeyondreasonabledoubtforS,thentheyallarebeyondreasonabledoubtforS.7IftherearethreeormoreconcurrentpropositionsandifeachofthemisbeyondreasonablefordoubtforSandifinadditiononeisevidentforS,thentheyareallevidentforS.Principle(1)makesreferenceto“self-presentingproperties,”whichChisholmtakestobepurelypsychologicalproperties.Thesepropertiesarenon-relational,inthesensethatfromthefactthatIhaveapropertyofthissort,nothinglogicallyfollowsabouthowIamrelatedtothenon-psychologicalworld.Forexample,fromthefactthatIhavethepropertyofthinkingaboutsailingmyboat,itdoesnotfollowthatIaminfactsailingmyboat.IdoesnotevenfollowthatIhaveaboat.Nor,saysChisholm,doesanythingelsefollowaboutthenon-psychologicalworld.Onotherhand,fromthefactthatIhavethepropertyofbeingstuckinatrafficjamandthinkingaboutsailingmyboat,some-thingdoesfollowaboutthenon-psychologicalworld.Itfollowsthattherearetrafficjams,thatIaminone,andsoon.So,thispropertyisnotaself-presentingone.Chisholmdistinguishestwokindsofself-presentingproperties:intentionalproper-ties(waysofthinking,hoping,fearing,wondering,wishing,desiring,intending,etc.)andsensibleproperties(waysofbeingappearedtobythevarioussenses).Principle(1)saysthatifIhaveaself-presentingpropertyandifIbelievethatIhaveit,thenthepropositionthatIhavethepropertyismaximallyjustifiedforme.Nothingismorejus-tifiedformetobelieve.Theself-presentingprovidesafoundationonthebasisofwhichothercontingentpropositionscancometohavejustification.Principle(2)describesonewayofthishap-pening.IfitisevidenttomethatIamhavingavisualexperienceofthesortthatisinvolvedinseeingacatandifinadditionitisepistemicallyintheclearformethatsomethingisappearingtomeinthisway,thesetwothingscombinetomakeitevident283\nRICHARDFOLEYformethatsomethingisappearingtomeinthisway.Itmaynotbeevidenttomewhetheritisacatoradogorabushthatisappearingtome,butitisevidentformethatsomethingisdoingso.Itisevident,inotherwords,thatIamnothallucinating.Chisholmalsoproposesprinciplesof“perceptualtaking.”Forexample,theaboveprinciple(3)impliesthatifitisevidenttomethatIamappearedtoinacertainwayandifIbelievethatitisacatthatisappearingtomeinthiswayandifmoreoverthispropositionisepistemicallyintheclearforme,thenthesethreethingscombinetomakeitbeyondreasonabledoubtformethatIperceiveacat.Thereisalsoaprincipleanalo-gousto(3)formemory,expressedintermsofwhatIseemtoremember(1989a:68).Theantecedentsofprinciples(2)and(3)makereferencebothtopropositionsthatareevidentandpropositionsthatarebeingepistemicallyintheclear.Principle(1)describeshowpropositionscanbecomeevidentforme,butChisholmbelievesthatthesetofpropositionsthatareepistemicallyintheclearismuchlargerthanthesetofevidentpropositions.Thus,theremustbesomeothersourceofepistemicjustificationforthem.Whatisthisothersource?Chisholmsaysthatitisbeliefitself,thatonewayinwhichapropositioncanobtainadegreeofepistemicjustificationisbybeingbelieved.Principles(4)and(5)aremeanttodescribehow.Accordingto(4),ifIbelieveapropositionthatisnotdisconfirmedbythesetofpropositionsthatareevidentforme,thenthepropositionisprobableforme.Alargenumberofpropositionscanbecomeprobableformeinthisway.Theywillhavethisweaklyfavorableepistemicstatusevenifthereisnootherpositivesourceofjus-tificationforthem–fromself-presentation,perception,ormemory,forexample.Moreover,principle(5)allowsthesepropositionstorisetoanevenhigherepistemicstatus.Accordingto(5),ifIbelieveapropositionthatisnotdisconfirmedbythesetofotherpropositionsthatareprobableforme,thenthispropositionisepistemicallyintheclearforme.Whatarethesepropositionsthatareatleastprobableforme?Inlargepart,theyarepropositionsthatsatisfytheantecedentofprinciple(4),namely,believedpropositionsthatarenotdisconfirmedbythatwhichisevidentforme.So,(4),asitwere,createsmuchofthematerialfor(5)todoitswork.Principles(4)and(5)areprinciplesofnegativecoherence.Togethertheyimplythatifabelievedpropositionisnotincoherentwiththesetofotherpropositionsthatareprobableforme(manyofwhichgetthisstatusbythefactthatIbelievethemandtheyarenotdisconfirmedbythatwhichisevidentforme),thenitisacceptableformetobelievetheproposition.Withtheseprinciplesinhand,reconsiderthequestionofhowthepropositionsmen-tionedintheantecedentoftheseprinciples(2)and(3)getthestatusofbeingepistemi-callyintheclearforme.Principles(4)and(5)provideananswer.Theycangetthisstatusbybeingbelievedbyme.IfIbelieveapropositionofthesortmentionedintheantecedentof(2),saythepropositionthatsomethingisappearingtomeinacat-likeway,andifthepropositionsthatareprobableformedonotdisconfirmthisproposition,thenthepropositionisepistemicallyintheclearforme.Andthen,principle(2)saysthatthisinconjunctionwiththefactthatitisevidenttomethatIamappearedtoinacat-likewaymakesitevidentthatsomethingisappearinginacat-likewaytome.ItisevidentthatIamnothallucinating.Similarlyforpropositionsofthesortmentionedintheantecedentof(3):ifIbelievethatitisarealcatappearingtomeinacat-likewayandifthispropositionisnotdis-284\nRODERICKCHISHOLMconfirmedbythesetofotherpropositionsthatareprobableforme,thentheproposi-tionisepistemicallyintheclearforme.Andthen,principle(3)saysthatthisincon-junctionwiththefactthatitisevidentformethatIamappearedtoinacat-likewaymakesitbeyondreasonabledoubtthatitisacat–andnot,say,adogorabush–thatIamperceiving.Chisholmalsothinksthatrelationsofpositivecoherenceamongasetofproposi-tions,orwhathecalls“concurrencerelations,”areanimportantsourceofjustifica-tion.Asetofpropositionsisconcurrentjustifthepropositionsarelogicallyindependentandmutuallysupportive,inthesensethateachpropositioninthesetissuchthattheotherstendtomakeitprobable.Chisholmdefendstwoprinciplesofconcurrence.Principle(6)saysthatifthereisasetofconcurrentpropositionseachofwhichisepistemicallyintheclearformeandatleastoneofwhichisalsobeyondreasonablefordoubtforme,thentheyallbecomebeyondreasonabledoubtforme.Principle(7)sayssomethingsimilarforconcurrentpropositionsofthenexthighestepistemicstatus.Accordingto(7),ifthereisacon-currentsetofpropositionseachofwhichisbeyondreasonablefordoubtformeandatleastoneofwhichisevidentforme,thentheyallbecomeevidentforme.So,despitehisreputationastheleadingfoundationalist,Chisholmisalsoacoher-entist.Butunlikeapurecoherentist,hedoesnotthinkthatpositivecoherencerelationsaretheonlysourceofempiricaljustification.Together,theaboveprinciplesdescribewhatChisholmtakestobesomeoftheprin-cipalsourcesofempiricaljustification:namely,self-presentation,perception,memory,beliefcoupledwithalackofnegativecoherence,and,finally,positivecoherenceamongpropositionswithsomeantecedentpositiveepistemicstatus.TheepistemologicalandmetaphysicalstatusoftheprinciplesAccordingtoChisholm,wehaveatleastavague,prephilosophicalideaofwhatitisforabelieftobejustified(1989a:5),anideawhichguidesusinidentifyinginstancesofbeliefsthatarejustified.Inturn,theseintuitionsaboutjustifiedbeliefsallowtheepis-temologicalprojecttogetofftheground.Chisholmisaparticularistwhenitcomestomattersofepistemologicalmethod(1989a:7).Hebeginsbyexaminingparticularinstancesofbeliefsthathetakestobejustified,andhethentriestoabstractoutoftheseinstancesgeneralconditionsofjustification,whichheexpressesintheformofepis-temicprinciples.Chisholmalsopresupposesthatwecanimproveandcorrectourbeliefsbyreflection,eliminatingthosethatareunjustifiedandaddingothersthatarejustified(1989a:1,5).Thispresuppositionactsasaconstraintwhenhetriestouseparticularinstancesofjustifiedbelieftoformulategeneralconditionsofjustification.Itforceshimtolookforconditionstowhichwehavereflectiveaccess,sinceotherwisetherewouldbenoreasontothinkthatwecouldeliminateunjustifiedbeliefsandaddjustifiedonessimplybybeingreflective.ThisisoneofthesensesinwhichChisholmisaninternalistaboutjustifica-tion,inanepistemicsense.Theprephilosophicalnotionofjustificationthatallowsepistemologytogetoffthegroundisvague,likemostordinarynotions,butitneednotremainso.Oneoftheben-eficialby-productsofformulatingandrefiningepistemicprinciplesisthatthebasic285\nRICHARDFOLEYnotionbecomesincreasinglyprecise,sothateventuallyepistemolgistsareinapositiontogiveageneralcharacterizationofit.AccordingtoChisholm,thecharacterizationistobegiveninethicalterms.Epistemicjustificationisultimatelytobeunderstoodintermsofethicalrequirementsonourbelievingsandwithholdings.Morespecifically,tosaythatanindividualSismorejustifiedinbelievingpthanwithholdingonpistosaythatSisrequiredtoprefertheformeroverthelatter(1989a:59).Chisholmgoesontoclaimthatrequirementstopreferarebestexplicatedinanegativeway.Therequirementtopreferbelievingpoverwithholdingonpisarequirementnottochoosebetweenbelievingandwithholdingwithoutchoosingtheformer,andthis,hepointsout,isarequirementthatcanbesatisfiedevenifonedoesnothavedirectcontroloverone’sbelievingsandwithholdings.Inaddition,Chisholmsaysthatthisrequirementisonethatsupervenesonnon-normativestates,specifically,onconsciousstates(1989a:60).Assuch,aproposi-tioncouldnothaveanepistemicstatusdifferentfromtheoneitdoeshaveforanindi-vidualwithoutthatindividual’spsychologicalstatesbeingdifferent.ThusChisholmtakeshisepistemicprinciplestoexpressnecessarytruths,andthetruthsthattheyexpressareultimatelyonesabouttherelationshipbetweenanindividual’scon-sciouspsychologicalstatesatatimeandanethicalrequirementonbelievingsandwithholdings.ThisillustratesanothersenseinwhichChisholmisaninternalistaboutjustification,ametaphysicalsense.Theconditionsthatmakeapropositionevidentorbeyondrea-sonabledoubtorprobableareinternalconditions.Theyarecurrent,psychologicalstates,notnon-psychological“external”states,andnotpastpsychologicalstates.Chisholm’sepistemicinternalismrequiressomethinginadditiontothis.Itrequiresthatwealwayshavereflectiveaccesstotheseinternalconditions.ThedefinitionofknowledgeTheepistemicprinciplesandthetermsofepistemicappraisalusedtoformulatetheprinciplesconstitutetheheartofChisholm’sepistemologicalsystem.TheyarethetoolsChisholmusestoaddressthemajorquestionsofepistemology.Amongthesequestions,nonehaspreoccupiedChisholmmorethanthequestion,Whatisknowledge?Overhiscareer,heproposedvariousdefinitionsofknowledge,mostofthemvari-antsoftheideathatknowledgeisnon-defectivelyevidenttruebelief.Likemanypro-poseddefinitionsofknowledge,Chisholm’sdefinitionwasaimedatcomingtogripswithapairofexamplespresentedbyEdmundGettier,whichweredesignedtoillustratethatknowledgecannotbeadequatelydefinedasjustifiedtruebelief.ThebasicideabehindbothcounterexamplesisthatonecouldbejustifiedinbelievingafalsehoodP,fromwhichonededucesatruthQ.InthiscaseonehasajustifiedtruebeliefinQbutdoesnotknowQ.Gettier’sexamplesinspiredahostofsimilarcounterexamples,andthesearchwasonforafourthconditionofknowledge,onethatcouldbeaddedtobelief,truth,andjustificationtoproduceanadequateanalysisofknowledge.ThetwomostdistinctiveaspectsofChisholm’sattempttohandleGettierproblemsare,first,hisinsistencethatabeliefmustbeevidenttocountasknowledge,andsecond,hisinsistencethatwhatmakesthebeliefevidentmustbenon-defective(Chisholmaddssomefurtherqualification;see1989a:98).286\nRODERICKCHISHOLM“Evident”isamongthestrongestofChisholm’stermsofepistemicappraisal,rankingonlybelowthatwhichiscertain.ApropositioniscertainforanindividualSonlyifitismaximallyjustified;nootherpropositionismorejustifiedforStobelieve.Amongthepropositionsthatcanbecertainaresimplenecessarytruths,forexample,theelementarytruthsofarithmetic,aswellascontingentpropositionsaboutself-presentingstates.Indefiningknowledgeintermsoftheevident,Chisholmisrejectingtheviewthatknowledgerequirescertainty.Ontheotherhand,heisinsistingthatknowledgeinvolvesaveryhighdegreeofjustification.ForChisholm,aparadigmaticrequirementonbelievingisthatofwithholdingjudgmentonthatwhichiscounter-balanced,forexample,thepropositionthatthenexttossofafaircoinwillturnupheads.ApropositionisevidentforS,inturn,onlyifSisatleastasjustifiedinbelievingitaswithholdingjudgmentonthatwhichiscounterbalanced.ForChisholm,thisrep-resentsaveryhighdegreeofjustification.Moreover,ifSistohaveknowledgeofaproposition,notonlydoesthepropositionhavetobeevidentforS,inadditionthatwhichprovidesthisveryhighdegreeofjusti-ficationmustbenon-defective,inthesensethatitmustnotmakeanyfalsehoodevidentforS.Toillustratetheintuitiveforceofthisrequirement,consideroneofGettier’sexam-ples.SmithhasverystrongevidenceforthepropositionthatJonesownsaFord,sinceSmithisawarethatJoneshasalwaysownedacar,thatthecarhasalwaysbeenaFord,thatJoneshasjustofferedSmitharidewhiledrivingaFord,andsoon.Fromthisevi-denceandthepropositionthatJonesownsaFord,Smithdeducesthedisjunctivepropo-sition,eitherJonesownsaFordorBrownisinBarcelona.HeinfersthispropositiondespitehavingnoideaofwhereBrownis.However,itturnsoutthatJonesdoesnotinfactownaFord(hehasbeendrivingsomeoneelse’scar)while,bychance,BrownisinBarcelona.SoSmithhasverystrongjustificationforthepropositionthateitherJonesownsaFordorBrownisinBarcelona;hebelievestheproposition;andthepropositionistrue.ButitseemsasifSmithdoesnotknowtheproposition.Why?Chisholm’sansweristhatalthoughthepropositionmaybebothtrueandevidentforSmith,theconsider-ationswhichmakesthepropositionevidentforhimalsomakesevidentafalsehood,namely,thatJonesownsaFord.PartII:MetaphysicsWhereasChisholm’sepistemologicalviewsconstituteaunifiedwholethatmaybeuse-fullyandconciselysummarized,themanymetaphysicalproblemsheaddressedformaheterogeneouscollectionthatdoesnotsubmitreadilytoconciseoverview.Furthermore,hisroleintherejuvenationofmetaphysicsduringthesecondhalfofthetwentiethcenturywouldnotbeconveyedbyasummaryofhisapproachestoparticu-larproblems.Onemustbackupabittoseewhyandwherehisinfluencewasgreat.Chisholm’simpactoncontemporarymetaphysicswouldbehardtooverestimate.Bytheendof1950shiscontributionswerealreadynumerous.Hemountedaninfluentialdefenseofthemeaningfulnessoftraditionalmetaphysicalquestionsinthefaceofdefla-tionarycritiquesfromthe“ordinarylanguage”philosophers(1951,1952,1964).Hehelpedbringdownthecurtainonphenomenalism(1948,1957b:Appendix).HedrewattentiontoFranzBrentano’scharacterizationofthepsychologicalintermsof“inten-tionalinexistence,”andattemptedtorehabilitateitasalogico-linguisticcriterionof287\nDEANZIMMERMANsentencesreportingintentionalmentalstates(1955–6).Hedefendedthethesisthatlin-guisticintentionalityistobeexplicatedintermsoftheintentionalityofthought,andnotthereverse(1955a,1957a).Andhehelpedfocusdebatesaboutcounterfactuals,dispositions,andlawsofnature(1946,1955b).Thisworkwaswidelyanthologizedinsubsequentdecades.Buthismostimportantcontributionstometaphysicscamesomewhatlater,ashebegantoconstructcomplex,evolving,andinterconnectedtheoriesofaction,persis-tencethroughtime,eventsandcausation,referenceandintentionality,andontologi-calcategories.Asummaryofthepositionshedefendedonthesetopicswouldgosomewaytowardexplaininghisimportanceasametaphysician.Inmanyareas,onestillfindsChisholm’sworkcitedascontainingtheparadigmaticformulationofanimportantposition,ortheoriginalstatementofaparadoxthatstandsinneedofresolution.Inactiontheory,forinstance,hisdefenseoftheincompatibilityoffreedomanddeterminism,andofagentcausation,areasfrequentlydiscussedasever.ButasimplesurveyofChisholm’sviewsonparticularmetaphysicalissueswouldmisstheforestforthetrees.Abetterpictureofhisplaceintwentieth-centurymetaphysicscanbegainedbyconsideringthestatusofmetaphysicsatthetimehiscareerbegan,andbycomparingChisholm’smethodol-ogywiththatofanotherimposingfigurefromthesamegeneration:W.V.Quine.Metaphysicsatmid-centuryMostphilosophicalworkthatbearsthe(sometimespejorative)label“metaphysics”ischaracterizedbyitsattentiontomattersofontology.Acentralpartofthedisciplinehasalwaysbeentheconstructionofcomprehensiveontologicalschemes,theoriesaboutthenatureofandrelationsamongthemostabstractcategoriesunderwhichabsolutelyeverythingfalls,togetherwiththeexplicituseoftheseontologicaldistinctionsinthefor-mulationofsolutionstophilosophicalproblems.Indeed,onecouldarguethatdistinc-tivelymetaphysicalproblemsalwaysinvolvetheveryabstractcategoriesappropriatetoontology;andthatanyphilosophicalproblembecomes,atleastinpart,ametaphysicalproblemassoonasontologicaldistinctionsbecomecentraltoitsstatementandresolution.Bymid-century,metaphysicsofthissorthadfallenonhardtimes.Theairhadgoneoutofdebatesabouttheontologicalstatusofuniversalsandparticulars,thedistinc-tionbetweenessenceandaccident,andsoon.RussellandMooreand,perhaps,thetrac-tarianWittgensteinhadtakensuchquestionsprettyseriously(seeMOORE,RUSSELL,andWITTGENSTEIN);butlogicalempiricism,Wittgensteinian“therapy,”andAustinian“ordi-narylanguagephilosophy”hadeclipsedthemetaphysicalpreoccupationsoftheearli-est“analyticphilosophers”(seeAUSTINandWITTGENSTEIN).Theproponentsoftheseinfluentialdoctrinesallthought(albeitfordifferentreasons)thatthetraditionalques-tionsofmetaphysicsweremisguided,unanswerable,nonsensical.Further,thereputationofmetaphysicswaspoorlyservedbytheobscurityofmanyofitsmostwell-knownpractitioners.ClarityofexpositionwasnotamongthevirtuesexemplifiedbyRoyce,Bradley,Bosanquet,Bergson,Whitehead–namesthatmeantmetaphysicsatthetime.Totheskeptical,itcouldeasilyseemthattherecipeforsuccessinmetaphysicswasthis:(1)inventyourownbaroqueontologicalscheme,usinganew,288\nRODERICKCHISHOLMpeculiarjargon;(2)claimthatitisradicallyopposedtoallprecedingmetaphysicalsystems;and(3)explainitsintricaciesbytheintroductionoffurtherundefinedtech-nicaltermsinaseriesofeverlongerbooks.Chisholm’schiefcontributiontocontem-porarymetaphysicswastoshow,bypreceptand,moreimportantly,byexamplethatitispossibletoconstructmetaphysicalsystemsonagrandscalewithoutfallingintothesevices.Hechampionedachastenedapproachtometaphysics,onethatneithershiesawayfromthetraditionalproblemsofontology,norfallsbackintothearcane,unteth-eredsystem-buildingthathadgivenmetaphysicsabadname.ThecomparisonwithQuineW.V.QuinebeganteachingatHarvardwhileChisholmwasagraduatestudent.QuineprovidedsomethingthatwouldprovecrucialtoChisholm’smetaphysicalprogram:theapproachtoquestionsofontologicalcommitmentdefendedin“OnWhatThereIs”(1948),butalreadyinplaceby1939,whenChisholmwasastudent(Quine1939).ChisholmtookQuine’scriterionofontologicalcommitmenttoamounttothefollow-inginjunction:Ifoneaffirmsastatementusinganameorothersingularterm,oraninitialphraseof“existentialquantification,”like“Therearesomeso-and-so’s”(seeQUINE),thenonemusteither(1)admitthatoneiscommittedtotheexistenceofthingsansweringtothesingulartermorsatisfyingthedescription,or(2)providea“para-phrase”ofthestatementthateschewssingulartermsandquantificationoverso-and-so’s.BothQuineandChisholmagreethatMeinong,whoaffirmstruthsaboutallsortsofthingswhichhethenadmitsdonotexist,istryingtohavehiscakeandeatittoo;Meinongmustberesistedifmetaphysicsistobekepthonest.Chisholm’smetaphysicslooksnothinglikeQuine’s,however.ForQuine,itisthedeliverancesofsciencealonethatneedbetakenintoaccountwhenattemptingtoworkoutone’sontologicalcommitments;heidentifiestheprojectof“limningthetrueandultimatestructureofreality”withthatofworkingoutthemostontologicallyaustereregimentationofthelanguageofthehardersciences.ThisenablesQuinetokeephisontologylean,includingnothingbutthemostwell-understood,sharplydemarcatedthings:ultimately,nothingbutconcretespatiotemporalentitiesandtheabstractbutwell-definedworldofsettheory.Butthecostisgreat:therepudiationofquitealotofwhatwewouldordinarilyregardastruismsaboutbeliefs,desires,andotherintentionalattitudes;aboutwhatmustormightbethecase;aboutwhatwouldhavehappenedif...;andsoon(see,e.g.,Quine1960).Chisholm,however,asks:Whynotassume,intheseminarroom,thesamethingswetakeourselvestoknowinevery-daylife?Why,whenwedophilosophy,shouldweappealtonothingbutwhatwefindinourphysicsandchemistrytextbooks?ChisholmrejectsQuine’sskepticismtowardallbutscience;anontologicalschememustshowitsadequacyonamuchbroaderplayingfield.BothChisholmandQuineagreethatontologicalschemesaretobejudgedbythecompetingdesiderataofsimplicityandsufficiencyofscope.Oneschemeissimplerthananotherifitpositsfewer,andbetterunderstood,typesofentities.Oneschemeissupe-riortoanotherinscopeinsofarasitallowsforthestatementofsatisfactoryphilosophi-caltheoriesonmoresubjects,theoriesthatpreserve,sometimesinthefaceofapparentcontradictionorphilosophicalpuzzlement,mostofwhatwetakeourselvestoknow.289\nDEANZIMMERMANQuine’saustereontologicalnaturalismispurchasedatthecostofsevererestrictionsonthescopeofwhatwemayreasonablytakeourselvestoknow.Althoughonecannotacceptthemathematicsneededforsciencewithoutsettheory,nofurther“queerenti-ties”needberecognizedbyonewhoaffirmsnothingbutthedeliverancesofthe(sufficientlyhard)sciences.Chisholm,however,hasmanymoretruthstoconsider;forhim,balancingthecompetingdesiderataofsimplicityofschemeandsufficiencyofscopeismuchtrickier.Theadequacyofanontologicalschemecomestoturnuponitsroleintheresolutionofthetraditionalproblemsofphilosophy,mostofwhichQuinewasabletosidestepbyrejectingthecommonsensicalconvictionsfromwhichtheproblemsarise.Itisnosurprise,then,tofindthetwophilosophersdifferingdrasticallydespitetheirinitialpointofagreement.Chisholmfindsthatonecannotarriveatmetaphysicalthe-oriessatisfyingbothdesiderataofsimplicityandscopewithoutmakingreferencetothingsnotfoundinQuine’sontology,suchas“intensionalobjects.”Hecanfindnoonto-logicallyperspicuoustheorythatdoesjusticetowhatweknowaboutpersonswhileeschewingirreduciblyintentional(psychological)notions(e.g.“conceiving,”“attribut-ing”).Ultimately,heconcludesthatpersonsmustbeveryspecialindeed:theyhavecausalpowersunlikethosefoundelsewhereinnature,theycan“grasp”orconceiveofabstractobjects,andtheirpersistenceconditionsaremysteriouslydifferentfromthoseofordinaryphysicalobjects.Quine,andmanyothernaturalisticallyinclinedphiloso-phers,willfindsuchconclusionsfantastical.Bethatasitmay,thetheoriesChisholmconstructsoffersolutionstoahostofphilosophicalproblems;andhismetaphysicalprogramstandsasachallengetobemetbythosewhowouldbemorenaturalisticornominalisticthanChisholm,butwhoarenotpreparedtoretreatintoaskepticismasradicalasQuine’s.Chisholmianmethodologyillustrated:statesofaffairsasnecessarythingsChisholm’sontologicalviewsunderwentfrequentrevision,asoneoranotherschemeprovedinadequateinscope,unabletomakeroomforenoughofwhatwetakeourselvestoknow;orashethoughtofsomewaytokeepaplausiblephilosophicaltheoryinplacewhilesimplifyingitsontologicalcommitments.Oneofthemoreradicalchangeswastherejectionof“statesofaffairs”intheearly1980s.ItprovidesagoodexampleofChisholm’sefforttomakesystematicmetaphysicsresponsiblebytyingontologytotheresolutionofawidespectrumofphilosophicalproblems.Inthiscase,thechangewasbroughtaboutbyproblemsofself-reference.ThegreatestontologicaldivideinChisholm’stheoryofcategoriesisbetweennecessarythingsandcontingentthings.ThestatesofaffairssocentraltoChisholm’sontologythroughoutthe1960sand1970sweretakentobenecessarythings.Chisholmadvancedseveralwaysofmarkingthedistinctionbetweennecessaryandcontingentthings.Hehopedtorestricthismodalprimitivestothoseexpressiblebymeansofonelocution:“xisnecessarilysuchthatitisF,”where“F”canbereplacedbyanypredicate,andthephraseisequivalentto“xisnecessarilysuchthatitexistsifandonlyifitisF.”(SomethingispossiblyF,ofcourse,ifandonlyifitisnotneces-sarilynotF.)Butthen,evenif“exists”wereallowedasapredicatesubstitutableforF,replacingFwith“exists”yieldsonlyasenseof“necessarilyexisting”accordingto290\nRODERICKCHISHOLMwhicheverythingexistsnecessarily.Oneproposalformakingthedistinctionwithinhisrestrictedvocabularyisthis:contingentthingsarepossiblysuchastobecomingintoexistenceorpassingaway(i.e.possiblysuchastohavehadnopropertiesandpossiblysuchastobegoingtohavenoproperties)andnecessarythingsareneither(cf.1989b:164,and1996:127).Thispresupposesthattherearenothingsthatcouldhavefailedtoexistbutthat,giventhattheydoexist,cannotpossiblybecreatedordestroyed.Somemighthavedoubtsaboutthisassumption.Chisholmmayhavedoubtedithimself,sincehetriedotherwaysofmakingthedistinction.Anotherproposedmarkofthenecessary/contingentdivideisthis:xisnec-essaryifandonlyifxhasapropertythatisessentialtoitandthatnothingelsecouldpossiblyhave(i.e.apropertythatisanindividualessenceofx);andevery-thingissuchthatsomethinghasthatproperty(1986:26).Thispresupposesthateverynecessarythinghasanindividualessence.PerhapsChisholmhaddoubtsaboutthisassumption,too;forhislastattempttoformulateacriterionforthenecessaryexistenceofathing,x,was:“Thereisanattributethatissuchthat(1)everythingisnecessarilysuchthatthereissomethinghavingthatattribute,(2)xisnecessarilysuchthatithasthatattribute,and(3)thatattributeisnotnecessarilyhadbyeverything”(1996:17).Counterexamplesaregeneratedifoneallowsfordisjunctivepropertiessuchasbeingeitherananimalorprime.ButChisholmalsodevelopedtheoriesofthestructureofproperties,includingaccountsofwhatitisforapropertytobeadisjunctionoftwoothers.Perhapshewouldhavefoundtheresourcestheretorefinehislastdefinitionsoastoruleoutsuchcounterexamples.Chisholmlongheldthattherewereatleasttwosortsofnecessarything:statesofaffairsandpropertiesorattributes.(Healwaysatleastleftanopeninginhistableofthecategoriesforathird,aswell:God,anecessarysubstanceuponwhichallelsedepends.)Headvocatesan“intentionalapproach”tobothstatesofaffairsandproperties;thatis,heclaimsthattheircriteriaofidentityandstructuralfeaturescanonlybeadequatelydescribedusingintentionalterms,suchas“believing”and“con-ceiving.”Statesofaffairsaredefinedasthosethingswhichonemaybelieve(1976:117),propertiesasthosethingswhichonemaybelievetobeexemplifiedbyotherthings(1996:29).Botharegivenintentionalcriteriaofidentity.Astateofaffairspisidenti-calwithastateofaffairsqifandonlyif,necessarily,(1)p“obtains”or“occurs”ifandonlyifqdoes(1976:118);and(2)whoeverbelievespbelievesq,andviceversa.Aprop-ertyFisidenticalwithapropertyGifandonlyif,necessarily,(1)somethingexempli-fiesFifandonlyifitexemplifiesG,and(2)whoeverconceivesFconceivesG,andviceversa(1989b:145).Propositionsareidentifiedwithstatesofaffairsthateitheralwaysobtainorneverobtain,eventswithobtainingstatesofaffairsthatarenotpropositionsandthatentailtheexemplificationofacertainsortofproperty–aproperty“rootedin”thetimeatwhichitisexemplified.Chisholm’sontologyis“realist”inseveralsensesoftheterm.Itincludespropertiesthat,likePlato’suniversals,existwhetherornottheyareexemplified.Itisopposedtopsychologismandlinguisticismaboutthesubjectmatteroflogic.Logicdiscoversnec-essaryrelationsamongpropositions:necessarilyexistingstatesofaffairs,innosensemind-dependentorlanguage-dependent.Indeed,thetruepropositionsarenottobedis-tinguishedfromfacts.AndsoChisholmadvocateswhatissometimescalledan“iden-291\nDEANZIMMERMANtitytheoryoftruth”:truepropositions“correspondwithfactsinthefullestsensethatispossible,fortheyarefacts”(1977:88).ThefirstpersonandtherejectionofhaecceitiesThroughoutthe1960sand1970sstatesofaffairsfigureprominentlyinChisholm’smetaphysics,epistemology,andmetaethics.Herearesomeexamplesofthemanydutiestheyperform.Statementsaboutparticularoccurrencesofagiventypeofeventaretobeparaphrasedintermsofthe“obtaining”ofabstract,eternalstatesofaffairs.Andacausalrelationbetweenapairofeventsisreallyamatteroftwostatesofaffairsbeingcausallyrelatedrelativetoacertaintime.Onthisapproach,thereisnoneedtorecognizeanontologicalcategoryof“tropes”or“particularizedproperties”inadditiontostatesofaffairsandpropertiesconceivedasPlatonicuniversals(1976:ch.IV).“Times”aregivenaglossmuchlikeA.N.Prior’s:theyaremaximal,consistentstatesofaffairs,com-pletewaystheworldcouldbe“allatonce”(1979a:357).Beliefandotherpropositionalattitudesaresaidtoberelationsbetweenthinkersandstatesofaffairs(1976:ch.IV).Moregenerally,arelativelysimpleontologyofproperties,statesofaffairs,andcontin-gent,persistingindividualthingsappearstobeadequatetotheformulationofphilo-sophicaltheoriesacrossthewholerangeofsubjectsChisholmaddressedinthisperiod.Thephenomenonoffirstpersonreferencesubjectedtheontologyofstatesofaffairstoconsiderablestrain.Ifpropositionalattitudesarerelationsbetweenthinkersandstatesofaffairs,whatstatesofaffairsareimplicatedinthoseattitudesexpressedusingthefirstpersonpronoun?ErnstMachcatchessightofhimselfinamirrorwithoutreal-izingwhoitis,andthinks:“Thatisashabbypedagogue,”withoutthinking:“Iamashabbypedagogue.”Thecontentsofthetwoattitudesdiffer;buthowisthisdifferencetobereflectedasadifferenceinthestructuresofstatesofaffairs,asitmustbeonChisholm’stheory?Sincestatesofaffairsarenecessarythings,theirconstituents,too,mustbenecessaryexistents.Theonlyway,then,forastateofaffairstobeaboutsomecontingentthingisforittocontainan“individualconcept”ofthatthing:apropertyonlyonethingcouldhave,andonethatishadbythatthing.Butwhatindividualconceptisinvolvedinmyfirstpersonthoughts?WhatextrapropertyisthereinthestateofaffairsIamashabbypedagoguethatisnotpresentinSomeoneisashabbypedagogue?SurelyIneednotknowanythingaboutmyrelationstootherthingsinordertothinkafirstpersonthought;soitmustbesomeintrinsicproperty,peculiartome,thatenablesmetothinkofmyselfinthisway.AndsoChisholmisledtoacceptthenotionthateachpersonhasan“haecceity,”anindividualessencepeculiartohimorher,and“repug-nantto”everythingelse(1976:ch.I).Chisholmgraduallycametofeelthatintroducinghaecceitiesforthispurposewasacheat.Althoughextraordinarilyuseful,haecceitiesremain,atbottom,utterlymysteri-ous.Wecannotrestcontentwithsimplypositingtheirexistenceonthebasisoftheirusefulness,sinceapartoftheiruseissupposedtobetheiraccessibilitytointellectualgraspbythethinkingthingsthatexemplifythem:Ifthisessentialistictheoryweretrue,theneverytimeapersonexpresseshimselfbymeansofanI-sentencehegraspshisownessenceorhaecceity.But,onewonders,doIeverthusgraspmyownindividualessenceorhaecceity?IfIdoevergraspit,shouldn’tIbeableto292\nRODERICKCHISHOLMsingleoutitsvariousmarks?...[I]fIcangraspmyindividualessence,thenIoughtalsotobeabletosingleoutinitthosefeaturesthatareuniquetoit.Ifbeingmeismyindi-vidualessenceandbeingyouisyours,then,presumably,eachanalysesintopersonhoodandsomethingelseaswell–onesomethingelseinmycaseandanotherinyours.ButIhaven’tthefaintestideawhatthissomethingelsemightbe....IthinkthatBrentanowasrightaboutthispoint.Hesaidthat,whenweconsiderthenatureofourselves,wenevergraspanypropertiesthatareindividuating.AnypropertyIknowmyselftohaveisonewhichissuchthatsomeentityotherthanIcouldalsohavethatproperty.(1979a:322)ThephenomenologicalinadequacyofthehaecceitytheoryledChisholmtorethinkproblemsofself-reference,lookingforanhaecceity-freetheorythatwouldallowforthedistinctionsweactuallymakeamongself-directedbeliefs.Whatresultedwasthe“directattribution”theoryofbelief:theobjectsoftheso-calledpropositionalattitudesarereallyproperties,andthethingsthataretrueandfalse(inatleastoneprimarysense)aredirectattributionsofpropertiestooneself(Chisholm1979b,1981).(DavidLewisreachedthesameconclusionindependently:Lewis1979.)Forcedtoregardtheobjectsofbelief,hope,wonderment,etc.aspropertiesinatleastthosecasesascribablebymeansofanindirectreflexive(“she,herself,”“he,himself”),Chisholm(andLewis)advocatetreatingallbelieving,etc.asamatteroftheself-ascriptionofproperties.Whenapersonbelievesthatshe,herself,ismortal,sheself-ascribesthepropertyofmortality.Whenshebelieves,withrespecttoherfather,thatheismortal,whatishappeningisthis:sheself-ascribesapropertythatimpliesthatthereissomerelationholdingbetweenherandonlyoneotherperson,andthatpersonismortal;andherfatherinfactstandsinthatrelationtoher.Whenshebelievesthatsomeone(orother)ismortal,sheself-ascribesasimplerproperty:beingsuchthatsomeoneismortal.InTheFirstPerson,Chisholmworksoutinterpretationsofdemonstratives,ofpropernames,andofsenseandreference,intermsoftheself-ascriptionofproperties.AsinhiscorrespondencewithSellars(Chisholm1957a),hedefendstheprimacyofpsycho-logicalintentionalityoverlinguisticintentionality:weconceiveofandself-ascribeprop-ertiesthatallowustosingleoutotherthings(cf.SELLARS),andweusewordstocauseotherstoconceiveofandself-ascribepropertiesthatsingleoutthesamethings.InChisholm’sview,thereisnowaytoavoidpositinganirreducibleintentionalrelation,suchas“conceiving,”thatrelatesthinkerstoextramentalthings(propertiesand,indirectly,otherindividuals)–arelationthatcannotbeidentifiedwithanabilitytomanipulatewordsineitherapubliclanguageoraninner“languageofthought.”TheunravelingoftheontologyofstatesofaffairsAtfirst,thenewaccountofthepropositionalattitudessentrelativelyminorripplesthroughChisholm’ssystem,asheexaminedtheextenttowhichthechangecalledformodificationsofhisviewsinepistemology(1982:ch.1),actiontheory,axiology,andethics,andofhisresolutionoftheparadoxofanalysis(1986).Intheseareas,therewaslittlechangeinfundamentaldoctrine.Butafairlyradicalrethinkingofhistheoryofeventsandcausationwascalledfor.Theself-ascriptionaccountofthinkingsolvesproblemswiththeolder,propositionalaccountbyrejectingthereceivedopinionthattruthandfalsityare,atbottom,propertiesofpropositions.Inordertogiveaunified293\nDEANZIMMERMANtheoryoftruthandfalsehood,Chisholmadoptswhathecallsa“doxastictheoryoftruth,”notunlikeRussell’s“multiplerelationtheoryofjudgment”(Russell1910):itisbeliefsorjudgmentsthataretrueandfalseinthe“primarysense”;thetruthandfalsityofotherthingsistobeexplicatedintermsofthesenseinwhichbeliefsaretrueandfalse(1993,1986:23).Thisstripstheoldstatesofaffairsoftwooftheirmostimpor-tantfunctions:asthethingsthatare,atbottom,trueandfalse;andastheobjectsofpropositionalattitudes.Furthermore,nowthathaecceitieshavebeenrejected,nonec-essarilyexistentstateofaffairscan,inanyobviousway,implytheexistenceofcontin-gentparticulars;sostatesofaffairsareinadequatevehiclesoftruthandfalsehoodinallbutthemostabstractorgeneralcases.Statesofaffairsbecomeathirdwheelwithinthetheoryofthetrueandfalse,andareeventuallyjettisoned.Butstatesofaffairshadplayedadualrole,asbothobjectsofpropositionalattitudesand,whentrue,worldlyfactsandevents.Formerly,thebearersoftruthandfalsehoodwerepropositions,whichwereacategoryofstatesofaffairs;andstatesofaffairs(whentheyobtained)werenottobedistinguishedfromfacts.Whenthebearersoftruthandfalsehoodaredoxastic–actsofjudgment–nosimpleidentificationofthebearersoftruthwithfactsoreventsispossible.Manyfactsandeventshavenothingtodowithjudgmentsorthinkers.Somethingmustbeintroducedtoplaytherolesoffactandeventinthenewontology:thosethingsinvirtueofwhichactsofjudgmentaretrueorfalse,andthesortsofthingsthatarecausesandeffects.AndsoChisholmintroducesanewcategory,thatofstates:contingentlyexistingstructuresthataremadeoutofthingsandproperties,andthatexistonlyifthethingshavetherelevantproperties(1990,1996).BibliographyWorksbyChisholm1946:“TheContrary-to-FactConditional,”Mind55,pp.289–307.1948:“TheProblemofEmpiricism,”JournalofPhilosophy45,512–17.1951:“PhilosophersandOrdinaryLanguage,”PhilosophicalReview60,pp.317–28.1952:“Commentsonthe‘ProposalTheory’ofPhilosophy,”JournalofPhilosophy49,pp.301–6.1955a:“ANoteonCarnap’sMeaningAnalysis,”PhilosophicalStudies4,pp.87–9.1955b:“LawStatementsandCounterfactualInference,”Analysis15,pp.97–105.1955–6:“SentencesaboutBelieving,”ProceedingsoftheAristotelianSociety56,pp.125–48.1957a:“Chisholm–SellarsCorrespondenceonIntentionality,”MinnesotaStudiesinPhilosophyofScience,vol.II,pp.511–20.1957b:Perceiving:APhilosophicalStudy,Ithaca,NY:CornellUniversityPress.1963:“SupererogationandOffence,”Ratio5,pp.1–14.1964:“J.L.Austin’sPhilosophicalPapers,”Mind73,pp.1–26.1966:TheoryofKnowledge,EnglewoodCliffs,NJ:Prentice-Hall.1974:“PracticalReasonandtheLogicofRequirement,”inPracticalReason,ed.S.Körner,Oxford:BlackwellPublishers.1976:PersonandObject,LaSalle,IL:OpenCourt.1977:TheoryofKnowledge,2ndedn.,EnglewoodCliffs,NJ:Prentice-Hall.1979a:“ObjectsandPersons:RevisionandReplies,”inEssaysonthePhilosophyofRoderickM.Chisholm,ed.E.Sosa,Amsterdam:Rodopi.294\nRODERICKCHISHOLM1979b:“TheIndirectReflexive,”inIntentionandIntentionality,ed.C.DiamondandJ.Teichman,Ithaca,NY:CornellUniversityPress.1981:TheFirstPerson:AnEssayonReferenceandIntentionality,Brighton:Harvester,andMinneapolis:UniversityofMinnesotaPress.1982:TheFoundationsofKnowing,Minneapolis:UniversityofMinnesotaPress.1986:“Self-Profile,”inRoderickM.Chisholm,ed.R.J.Bogdan,Dordrecht:Reidel.1989a:TheoryofKnowledge,3rdedn.,EnglewoodCliffs,NJ:Prentice-Hall.1989b:OnMetaphysics,Minneapolis:UniversityofMinnesotaPress.1990:“EventsWithoutTimes:AnEssayonOntology,”Noûs24,pp.413–28.1993:“WilliamJames’sTheoryofTruth,”TheMonist75,pp.569–79.1996:ARealisticTheoryofCategories,Cambridge:CambridgeUniversityPress.WorksbyotherauthorsGettier,E.(1963)“IsJustifiedTrueBeliefKnowledge?,”Analysis23,pp.121–3.Lewis,D.(1979)“AttitudesDeDictoandDeSe,”PhilosophicalReview88,pp.513–43.Quine,W.V.(1939)“ALogisticalApproachtotheOntologicalProblem,”JournalofUnifiedScience9,pp.84–9(preprintsonly).(ReprintedinQuine1976,pp.196–202.)——(1948)“OnWhatThereIs,”ReviewofMetaphysics2,pp.21–38.——(1960)WordandObject,Cambridge,MA:MITPress.——(1976)TheWaysofParadox,revisededn.,Cambridge,MA:HarvardUniversityPress.Russell,B.(1910)“OntheNatureofTruthandFalsehood,”inPhilosophicalEssays,London:Longmans,Green&Co.295\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200123DonaldDavidson(1917–)ERNESTLEPOREDonaldDavidsonisoneofthemostimportantandinfluentialphilosophersofthesecondhalfofthetwentiethcentury.Hehasneverattemptedasystematicexpositionofhisphilosophicalprogram,sothereisnosingleplaceastudent,interpreter,orcriticcanseekitsofficialformulation.Hispublishedessays,takentogether,formamosaicthatmustbeviewedallatonceinordertodiscernanoverallpattern.Inaddition,theysometimesexhibitanenigmaticquality,withsubtleties,complexities,andcross-referencesthatoftencannotbeentirelyappreciatedexceptinconjunctionwitheachother.Allofthiscanrenderaccesstohisthoughtnotjustdifficultbutattimesfrustrating,despiteitsobviousimportance.Inanefforttohelpanoviceorevensomeonealreadystumped,thefollowingmajorthemesinDavidson’sphilosophywillbesummarized:(1)ReasonsandCauses,(2)EventsandCausation,(3)AnomalousMonism,(4)TheoryofMeaningandCompositionality,(5)RadicalInterpretation,(6)AdverbialModification,(7)TheMethodofTruthinMetaphysics,(8)AgainstFacts,(9)TruthandCorrespondence,(10)AnimalThought,(11)AlternativeConceptualSchemes,(12)Anti-skepticism,(13)Anti-CartesianismandFirstPersonAuthority,(14)TheRejectionofEmpiricism.Thoughthislistisnotexhaustive,itidentifiesbroadthemesthatstructureDavidson’sproject.Ofcourse,nothingshortofreadingDavidson’sownworkcansupplantadetailedcharacterizationofhistakeoneachindividualpoint,butIhopetotemptyoutojourneyintoDavidson’sphilosophicalworld.ReasonsandcausesWeoftentrytoexplainanother’sactionsbycitingherreasonsforsoacting.MaryoffersBilltheseatnexttoheronatrain.ShedoessobecauseshewantsBilltositdownandbelievesthatbyidentifyingthisopenseathewill.Shethereuponaskshimtositdown.WhatrelationmustobtainbetweenMary’sbehaviorandherreasonforthisbehaviorinordertocorrectlyconcludethatsheactedasshedidbecauseofherreason?UntilDavidson’s“Actions,Reasons,andCauses”(see1980),asdifficultasitnowseemstocomprehend,somethingclosetoaconsensushadformedinphilosophythatwhatevertherelationshipmightbeitcouldnotbecausal.Itwasbelievedthatanalleged“logicalconnection”betweenreasonsandactionsexcludedanycausalrelation296\nDONALDDAVIDSONbetweenthem.ThecentralpurposeofDavidson’sessaywas“todefendtheancient–andcommonsense–positionthatrationalizationisaspeciesofcausalexplanation”(1980:3).Muchofhisessayisdevotedtorefutingvariousarguments,thenpopular,thatpurportedtoshowthatreasonscouldnotcausetheactionstheyrationalize.Accordingtoeach,anecessaryconditionforcausalinteractioncannotbesatisfiedbyreasonsandactions.Theseargumentsaretoomanytobeproperlytreatedhere,butinpassingitshouldbenotedthattheywereinspiredbyremarksofthetwentieth-centuryphilosopherLudwigWittgensteinorbycertaininterpretationsoftheeighteenth-centuryBritishphilosopherDavidHume’sstricturesoncausation.Bywayofasingleexample,manyauthorsbelievethatHumehadmoreorlessestab-lishedthatinorderforonethingAtobecausallyrelatedtoanotherthingB,AandBmustnotbelogicallyconnected.So,forexample,whenonebilliardballmovingwithacertainmomentumhitsanother,then,unlesscircumstancesareunusual,thesecondwillbecausedtomove.Thesecondball’smovementoccursnotasamatteroflogic,butratherasamatterofthephysicalnatureofouruniverse.Itislogicallypossiblethatcausalinteractioninthisuniversehasbeengovernedbyaphysicalnaturedistinctfromwhatactuallygovernsit.Oneintuitionbehinddenyingthatreasonscanbecausesofactionsisthattheyarelogicallyrelated;so,ifIbelievethatsmokingisharmful,andIdesirenoharm,mustn’tI,asamatteroflogicorofmeaningalone,intendnottosmoke?Ifso,thenhowcouldmyreasonsforintendingnottosmokehavecausedmetohavethatintention?Forourpurposes,itsufficestosaythatDavidsonrepliedtothisargumentandothersbyshowingeitherthatreasonsandactionsindeedsatisfythenecessaryconditioninquestion,orthatthewould-benecessaryconditionforcausalinteractionisinfactnotoneatall.So,withrespecttoHume’sobservation,Davidsonarguedthatalogicalcon-nectionbetweendescriptionsofacauseandaneffectdoesnotbyitselfpreemptcau-sation,asisevidentin“Thecauseofeventecausedevente.”Noonewouldinferfromthefactthataneventcouldnothaveexistedunderthedescription“thecauseofe”withoutbeingthecauseoftheeventethatthefirstdidn’tcausethesecond.Similarly,evenifnoonecouldbedescribedashavingabeliefunderthedescriptionof“smokingisharmful”andhavingadesireunderthedescription“don’tdoanythingharmful”withouthavinganintentionunderthedescription“don’tsmoke,”itdoesn’tfollowthatthereisnocausalrelationbetweenthesereasonsandtheintentionalactionthatensues.EventsandcausationIn“TheIndividuationofEvents”(in1980),Davidsonarguesthatwemustrecognizethatwecandescribethesameactionindifferentways.Indeed,ifwedidnotdothis,wecouldnotmakesenseofperfectlynaturalclaimslike“Jonesmanagedtoapologizebysaying‘Iapologize’,”whereasingleactionisdescribedbothasamanagingtoapolo-gizeandasasaying“Iapologize.”Butwhatsortofobjectaretheseactionsthatadmitofre-description?Tables,chairs,andpeopleareconcrete,datedparticulars,thatis,unrepeatableentitieswithlocationinspaceandtime.Thesefeaturesalonedistinguishthemfrom,say,eithernumbersorGod.Butwhataboutactionsandevents,forexample,theaction297\nERNESTLEPOREofBobshootingBill,ortheeventofHurricaneFloyd,ortheeventofthestockmarketcrashofOctober1929?Anyonewhodoubtsthesethreeeventsexistneedonlyconferwithoneoftheirvictims.Butwhatsortofentitycananactionoreventbe?Wespeak,forexample,oftheeventofabaseballgamebetweentheYankeesandtheRangersaslastingthreehours,andastakingplaceinNewYork.Wemightconcludethatthiseventwasexcitingortoolongorevencrucialindeterminingachampion.Onthebasisofonsuchassessmentswhatcanweconcludeaboutwhatsortofthinganeventis?Davidson’schiefclaimaboutactionsandeventsisthatliketablesandchairstheyareconcrete,datedparticularsthatcanbedescribedinvariousnonlogicallyequivalentandnon-synonymousways.Whatdistinguishesthemfromothersortsofconcrete,datedparticularsistheirpotentialforcausalinteraction,andso,it’spartofthenatureofbeinganeventthatitcanstandinacausalrelationship.Davidsongoesontoindi-viduatethem,sothattwoeventsareidenticaljustincasetheyhavethesamecausesandeffects.SinceDavidsontreatscausationasarelationbetweenevents,andtakesactiontobebutaspeciesofevent,eventscomprisetheverysubjectmatterofactiontheory,aswellasscienceandethics.(Wewilltakeupbelowhisargumentfortheirexis-tenceandforspecificclaimsastotheirnatureinthesectiononthemethodoftruthinmetaphysics.)Davidsonholdsthateventsrelatedbycausationmustbesubsumableundersomelaworother,wherealawisageneralizationconfirmablebyitspositiveinstancesand,iftrue,supportscounterfactualstatements,andwhereaneventissubsumedbyalawjustincaseitinstantiatesthatlaw(1980:217).So,forexample,accordingtoBoyle’sLaw,thepressureofafixedmassofgasataconstanttemperatureisinverselypropor-tionaltothevolumeofthatgas.Aninstanceofthislaw–ifaisthepressureofafixedmassofgasataconstanttemperature,thenaisinverselyproportionaltothevolumeofthatgas–isconfirmedjustincasewhenitsantecedentistrue,soisitsconsequent.Tosayitsupportsitscounterfactualinstancesmeansthatevenifaisnotthepressureofafixedmassofgasataconstanttemperature,ifitwere,thenitwouldbeinverselyproportionaltothevolumeofthatgas(1980:215).Davidson’sviewsaboutthenatureofeventsandtheirrelationtolawsbroughthimtoastunningconclusionabouttherelationshipbetweenmindsandbodies,namely,histhesisofanomalousmonism,towhichweturnimmediately.AnomalousmonismMuchcanandhasbeensaidinfavorofeachofthefollowingthreeclaims:1Thementalandthephysicalaredistinct.2Thementalandthephysicalcausallyinteract.3Thephysicaliscausallyclosed.Theproblem,though,isthattheyseeminconsistent.Considertheirapplicationtoevents.(1)saysthatnomentaleventisaphysicalevent;(2)saysthatsomementaleventscausephysicalevents,andviceversa;aloudnoisereachingTom’searmaycausehimadesiretoturndownhisradio;andhisdesiretoturndownhisradiomaycausehisarmtomoveinsuchawaytoresultinthevolumeofhisradiobeinglowered.(3)saysthatallthecausesofphysicaleventsarethemselvesphysicalevents.Thedilemma298\nDONALDDAVIDSONposedbytheplausibilityofeachoftheseclaimsandbytheirapparentincompatibilityisthetraditionalmind–bodyproblem.Davidson’sresolution,asarticulatedin“MentalEvents”(in1980),“TheMaterialMind”(in1980),and“PhilosophyasPsychology”(in1980)consistsoftheses(4)–(6),whichtakentogethercomprisehisthesisofanomalousmonism:4Therearenoexceptionlesspsychologicalorpsychophysicallaws,andinfactallexceptionlesslawscanbeexpressedinapurelyphysicalvocabulary(1980:214–15,231).5Mentaleventscausallyinteractwithphysicalevents(1980:208).6Eventccauseseventeonlyifanexceptionlesscausallawsubsumescande(1980:208).Thethesisismonistic,sinceitassumesthereisbutonekindofstuffintheworld,physicalstuff,butitisanomalous,sincealthoughitsmonismcommitsittophysicalandmentalstuffbeingthesamestuff,itdeniesthatthatthereisastrictreductionoftheonetoother.Afull-blownexegesisanddefenseofthisviewandtheclaimswhichcompriseitisbeyondtheintendedscopeofthisessay,butafewwordsabouttheextra-ordinarythesis(4)areinorder.Thesis(4)isaversionof(1).Itiscommonlyheldthatwhateverpropertyamental-isticpredicateMexpressesisreducibletooneexpressedbyaphysicalpredicateP(where“M”and“P”arenotlogicallyconnected)onlyifanexceptionlesslawlinksthem(whereanexceptionlesslawisexactlywhatitsays,namely,onethatundernoconditionsadmitsofexceptions.Boyle’sLaw,statedabove,isobviouslynotanexceptionlesslaw).Accordingto(4),mentalandphysicalpropertiesaredistinct.Insketch,Davidsonbaseshisargumentagainstthepossibilityofexceptionlesslawslinkingmentalandphysicalpredicatesontheirsufficientlydistinctconstitutiveprinciples(1980:222,238).Measurabilityoflength,mass,temperature,andtimeareconstitutiveofthephysi-cal,inasmuchasthesefeaturesgoverntheapplicabilityofphysicalpredicates(1980:221).So,forexample,anythingphysicalmusthavelength.Suppose,though,uponinvestigation,wediscoverthatamongthreephysicalitems,thoughthefirstislongerthanthesecondandthesecondlongerthanthethird,thefirstisnotlongerthanthethird.Whatwouldweconclude?Wewouldassumethateitherwearemistaken,orthattheirlengthschangedduringthecourseofmeasurement.Whatwewouldnotconcludeisthatthetransitivityoflengthisfalse.Why?Because,nothreethingscouldhaveaphysicalpredicatetrueofthemunlesstheirlengthsrespecttransitivity;thatis,respect-ingthisconstraintisconstitutiveofbeingphysical.Withmentalitems,theirconstitutiveprinciplesincludeprinciplesofrationality,forexample,constraintsaboutconsistencyandrationalcoherence(1980:236–7).Forexample,thetransitivityofdesire–ifapersondesiresaoverbandboverc,heoughttodesireaoverc;ortheconsistencyofbelief–weassumethatifanagentbelievesthatp,heoughtnotalsotobelievethatnotp.Thistiesinwithearlierdiscussionofreasonsandactions;interpretingthebehaviorofanotherrequiresattributingbeliefsanddesires.Theseattributionsareintendedtoprovideanagent’srationaleforacting,buttheyfailthistaskunlessadegreeofrationalchoiceispresumed.Byvirtueofthispre-sumption,theconstitutiveprinciplesofthementalincludenormsofrationality.Davidsonclaimssuchconstitutiveprincipleshave“noechoinphysicaltheory”(1980:299\nERNESTLEPORE231).So,heconcludes,inanimportantsensepsychologycannot“bereducedtothephysicalsciences”(1980:259),namely,exceptionlesslawscannotlinkthetwosortsofsciences,becausethenormativerelationshipsamongmentalstatescannotbeexpressedinaphysicallanguage.Thiscompactdiscussionofanomalousnessleavesahostofquestionsunanswered,butitservestoidentifypointsofcontroversysurroundingDavidson’sthesis(4).Whatabouthisthesis(6)?In“CausalRelations”(in1980),Davidsonarguesthatthemostplausibleinterpretationofsingularcausalstatementslike“Theshortcircuitcausedthefire”treatsthemastwo-placepredicatestatementswiththeirsingularterms,inthiscase,“theshortcircuit”and“thefire”designatingevents.Thesis(6)saysthataneventccausesaneventeonlyiftherearesingulardescriptionsDofcandD¢ofe,andanexceptionlesscausallawLsuchthatLand“Doccurred”entail“DcausedD¢”(1980:158).But(6)andthesecondpartof(4)entailthatphysicaleventshaveonlyphysicalcauses,andthatalleventcausationisphysicallygrounded.Giventheparallelsbetween(1)–(3)and(4)–(6),itmayseemthatthelatter,too,areincompatible.Davidson,however,arguesthattheycanallbetrueif(andonlyif)indi-vidual(token)mentaleventsareidenticaltoindividual(token)physicalones(1980:215,223–4).Supposeaneventeisphysicaljustincaseesatisfiesapredicateof(basic)physicalscience,wheresuchpredicatesarethosethatoccurinexceptionlesslaws.SinceDavidsonassumesthatonlyphysicalpredicates(orpredicatesexpressingpropertiesreducibletophysicalproperties)occurinexceptionlesslaws(1984b:240),itfollowsthateveryeventthatentersintocausalrelationssatisfiesaphysicalpredicate.But,then,itfollowsthatamentaleventthatentersintoacausalrelationmustsatisfysomephysicalpredicate,andso,isitselfaphysicalevent.Hisargument,ifsound,establishesnomorethanthateveryconcretementaleventisidenticaltosomeconcretephysicalone.Sincethisidentityofmentalandphysicaleventtokensiscompatiblewithreject-ingsystematiclawsbridgingmentalandphysicalevent-types,thatis,withanomalousmonism,thelatterthesisonlypartiallyendorses(1).Thementalandphysicalremaintype-distinctinsofarasmentalandphysicaleventsarenotlinkedbyexceptionlesslawsundermentalandphysicaldescriptions.Davidson’saccountofreasonsandeventsimpactson,andisinturn,affectedby,hisradicalapproachtolanguage;toseethis,weturnnowtoexaminehistheoryofmeaning.TheoryofmeaningandcompositionalityOurdiscussionwillbelimitedprimarilytoideaspresentin“TheoriesofMeaningandLearnableLanguages”and“TruthandMeaning”(bothin1984b),whereDavidsonidentifiesanadequacycriterionfortheoriesofmeaningfornaturallanguages,andthenappliesitcriticallytoanumberofthenprominentanalysesofaspectsofnaturallan-guage(seeTARSKI,CHURCH,GÖDEL).Healsosketchesaprograminwhich,surprisingly,acertainausterestyleofmeaningtheorymeetsthisadequacycriterion.Whataim(s)shouldatheoryofmeaningseektoaccomplish?Thatwilldependonwhichlinguisticaspectsatheoristwantstoexplain.So,forexample,thoughnaturallanguagesarespokenbyfinitespeakerswithoutmagicalabilities,theystillhaveaninfinityofmeaningful(non-synonymous)sentences,eachofwhich,atleastpotentially,300\nDONALDDAVIDSONaspeakercouldunderstand(atagiventime).Forany(indicative)sentenceSofEnglish,anewonecanbeformedbyprefacingitwith“Itisbelievedthat.”Foranytwo(indica-tive)sentences,SandS¢,anewonecanbeformedbydisjoiningthemwiththeword“or”;andsoonforotherproductivemechanismsofourlanguage.Thenovelsen-tenceswhichtheseproductivemechanismsgiverisetoareintelligibletonormalspeak-ersiftheircomponentsare.Thiscapacityseemstorequirethatspeakershavelearned(afinitenumberof)rulesthatdeterminefromafinitesetofsemanticprimitiveswhatcountsasmeaningfulcompositions(whereanexpressionissemanticallyprimi-tiveifthe“ruleswhichgivethemeaningforthesentencesinwhichitdoesnotappeardonotsufficetodeterminethemeaningofthesentencesinwhichitdoesappear”(1984b:9)).Onthebasisofsuchconsiderations,Davidsonrequiresofatheoryofmeaningthatitspecifywhateverysentencemeansbyexhibitingitsmeaningasafunctionofthemeaningofitssignificantparts(based,presumably,ontheirarrangementinthesen-tence).Let’scallanysuchtheoryforalanguage“acompositionalmeaningtheory”forthatlanguage.Davidsonwasthefirstphilosophertobringtoprominencetheimpor-tanceoftherequirementthatatheoryofmeaningofourlanguageexhibititascom-positional(1984b:23).Therequirementfocusesattentionontheneedtouncoverstructureinnaturallanguages.Whilethisisclearlysomethingthatphilosophersfromtimeimmemorialhaveengagedin,thisprojecthadnotbeen,untilDavidson,clearlyseparatedfromtheancientbutnowdiscreditedprojectofconceptualanalysis.Davidson’spositivesuggestionforacompositionalmeaningtheoryforalanguageL,surprisingly,utilizesnoconceptofmeaningthatgoesbeyondtruth.Towit,histheoryofmeaningtakestheformofa(finite)theoryoftruththat,foreachsentenceSofL,entailswhatweshallcallaT-sentenceofform(T)SistrueinLif,andonlyif,p,where“p”specifies(inametalanguage)conditionsunderwhichSistrueinL.So,forexample,anadequatecompositionalmeaningtheoryforGermanshouldissueinatheoremlike(S):(S)“Schneeistweiss”istrueinGermanif,andonlyif,snowiswhite.WhydoesDavidsonchoosethisratheraustereformoftheoryover,say,onethatexpli-citlyinvokesmeaningbyissuingintheoremsoftheform(M)SinLmeansthatp,where“p”specifies(inametalanguage)whatSmeansinL?Hisreasonshavenothing,assometimessuggested,todowithreplacingthecomplexnotionofmeaningwithonemoretractableoreasilyunderstood.Davidson’sinquiryisnotguidedbyconceptualormetaphysicalqualmsaboutthenotionofmeaning,butsolelybythegoalofdevisingacompositionalmeaningtheory.Hearguesthatthisaimcanbeachievedwithacom-positionalmeaningtheorythatissuesintheoremsthattaketheformof(T),butnotbyonethatissuesintheoremsofform(M).(Hisreasonsarebasedonthefactthatunlikethelocution“istrueifandonlyif,”thelocution“meansthat”issemanticallyopaque,thushinderingthedevelopmentofacompositionalmeaningtheory.Whatsemanticopacityiswillbediscussedbelowinthesectiononanimalthought.)301\nERNESTLEPOREAcompositionaltheoryofmeaningforalanguageLthatissuesininterpretiveT-sentenceslike(S)issuchthatanyonewhoknowsitispositionedtounderstandeverysentenceofL.ByspecifyingthemeaningofasentenceSinLviasentencesofform(T),Davidsonisrequiringaspecificationthatenablesanyonewhounderstandsthelanguageinwhichthespecificationisgiven(themetalanguage)tounderstand(theobjectlanguagesentence)S.Theobservationthatnaturallanguagesarecom-positionalisthefoundationuponwhichDavidsonbuildshisprograminthetheoryofmeaning.Sincewedonotandcouldnotknowapriorihowtointerprettheexpressionsofanaturallanguageorhowtoassigntheminterpretivetruthconditions,anadequatecompositionalmeaningtheorymustbeempirical.Anysuchtheorymustfunctionasatheoryforanaturallanguage.Inthecaseofone’sownlanguage,ofcourse,thoughwhatoneknowsaboutitisnotapriori,nodifficultyarisesinidentifyingwhichsen-tencesofform(T)areinterpretive.Theproblemforatheoristconcerningalanguagehealreadyunderstandsissimplytofigureoutwhattheaxiomsofthetheoryoughttobeinordertoconstructproofsofthem.Aquitedifferentproblemconfrontsatheoristforforeignlanguages(eventosomedegreeforotherspeakersofone’sownlanguage,athemeemphasizedmoreinDavidson’slaterwork).Thisiswheretheimportantnotionofradicalinterpretationentersintohisdiscussion.RadicalinterpretationWhatjustifiesachoiceofonecompositionalmeaningtheoryforalanguageoveranother?Davidsonarguesthatanadequatecompositionalmeaningtheorymustbeempiricallywarrantedunderthepracticeofradicalinterpretation.Whatthismeansisthatcertainspecificempiricalconsiderationsmustberespectedinchoosingbetweendistinctbuttruecompositionalmeaningtheories,namely,inoptingforacompositionalmeaningtheoryforGermanthatissuesin(S)overonethatissuesin(W).(W)“Schneeistweiss”istrueinGermanif,andonlyif,grassisgreen.(W)is,asamatteroffact,true,but,unlike(S),itfailstointerpret“Schneeistweiss,”andsonocompositionalmeaningtheoryforGermanthatissuesin(W)canbeade-quate.Butforlanguageswedonotalreadyunderstand,acompositionalmeaningtheorymustbeselectedonthebasisof“evidenceplausiblyavailabletoaninterpreter,”thatis,“someonewhodoesnotalreadyknowhowtointerpretutterancesthetheoryisdesignedtocover”(1984b:128).Davidsonboldlyclaimsthatnothingcanbealanguageunlessacorrectcomposi-tionalmeaningtheorythatissuesinatrueandinterpretivesentencelike(S)foreachsentenceofthatlanguagecanbeselectedonthebasisofthesortsofobservationsplau-siblyavailabletoaradicalinterpreter.Inabitmoredetail,aradicalinterpreter,bydef-inition,isignorantofthelanguagesheistryingtointerpret,andalsolacksaccesstobilingualinformants,priordictionaries,andthelike.Aradicalinterpreterisgenerallyallowedtobeabletodeterminewhenaninformantholdsasentencetrue,eventhoughshefailstounderstandwhateversentenceisbeingheldtrue(1984b:135).So,ineffect,theprimarydataforradicalinterpretationareformulablebywhatmightbecalledsin-gularheldtruesentences,forexample,(E):302\nDONALDDAVIDSON(E)KurtbelongstotheGermanspeechcommunityandKurtholdstrue“Esregnet”onSaturdayatnoonanditisrainingnearKurtonSaturdayatnoon.Datalike(E)canbecollectedfromavarietyofspeakersacrossavarietyoftimesbysomeonewhodoesnotalreadyunderstandGerman,andwilleventuallyconfirmagen-erallyheldtruesentencelike(GE):(GE)ForanyspeakeroftheGermanspeechcommunityandforanytime,thatspeakerholdstrue“Esregnet”atthattimeifandonlyifit’srainingnearhimorheratthattime.Claimslike(GE)provideevidenceforaradicalinterpreterthatspeakersoftheGermanspeechcommunitytakesomeformofwordstoexpressaspecifictruth(1984b:135).Butwhatlicensesinferringfromdatalike(GE)atargettheoremlike(R)?(R)“Esregnet”istrueinGermanofaspeakersatatimetjustincaseit’srainingnearthatspeakeratthattime.CouldallGermanspeakersgetitwrong?Davidsondeniesanysuchpossibility,answer-ingthattheinferencefrom(GE)to(R)islegitimate,becauseacertainprincipleofcharityislegitimatelypresupposed(1984b:137).Accordingtothisprinciple,thefavoredcom-positionalmeaningtheoryforalanguageLmustentailsentencesofform(T)suchthatmostsentencesspeakersofLholdtrueareinfacttrue.Underradicalinterpretation,sentencesspeakersholdtruekeepturningouttobetrue.Thisisnoaccident;rather,it’sbecauseradicalinterpretationisaspecialsortofproject,namely,onethatisconstitutedbythisprincipleofcharity.So,oncedatalike(GE)arecollected,wecaninferitscorrespondingT-sentences(R)viathisprincipleofcharity.Toboot,byvirtueofsecuringasetofinterpretiveT-sentencesforcontextsensitivesentenceslike“Esregnet,”atruththeoryautomaticallyassignsinterpretiveT-sentencestocontextinsensitivesentenceslike“Schneeistweiss.”So,toaveryroughapproximation,supposethat,onthebasisofwhataradicalinterpretercollects,shedevisesforcontextsensitivesentences“Esschneit”and“Esistweiss”interpretiveT-sentenceslike(a)and(b):(a)Foranyspeakerandtime,ifthespeakerutters“Esschneit”atthattime,thenthespeaker’sutteranceof“Esschneit”istruejustincaseit’ssnowingnearhimorheratthattime.(b)Foranyspeaker,timeandobject,ifthespeakerutters“Esistweiss”atthattime,thenthespeaker’sutteranceof“Esistweiss”istrueatthattimeofthatobjectif,andonlyifthatobjectiswhiteatthattime.Butonthebasisofthesegeneralizations,supposethatsomehoworotheraradicalinter-preterconjecturesapplicationconditionsforsub-sententialexpressionslike“schnee”and“weiss,”asinboth(c)and(d):(c)Foranyobjectandtime,“schnee”istrueofthatobjectatthattimeif,andonlyif,thatobjectissnowatthattime.(d)Foranyobjectandtime,“weiss”istrueofthatobjectatthattimeif,andonlyif,thatobjectiswhiteatthattime.303\nERNESTLEPORESuchinformationwecanimaginebeingexploitedinattributingtruthconditions(e)toacontextinsensitivesentencelike“Schneeestweiss.”(e)“Schneeistweiss”istrueif,andonlyif,snowiswhite.TheupshotisthatonlythosecompositionalmeaningtheoriesthatentailT-sentenceslicensedbyradicalinterpretationareadequate.Inthenextfewsections,wewillexploresomephilosophicalramificationsofembrac-ingtruththeoriesasthecorrectformforcompositionalmeaningtheories,andofembracingradicalinterpretationastheirpresumedmannerofconfirmation.AdverbialmodificationWediscussedDavidson’sviewsaboutthenatureandroleofactionsandeventsabove.Yetthestrongestargumentheadvancesfortheexistenceofeventsandcertainviewsabouttheirnaturederives,remarkably,notfromanypieceofpuremetaphysicalrea-soning,butratherfromconstraintsoncompositionalmeaningtheories.Anytheoryofmeaningforalanguagemustembodyaviewoftherelationshipbetweenlanguageandreality.Davidson’sconvictionisthatacompositionalmeaningtheory,byprovidingaviewaboutthisrelationship,offerssubstantiveanswerstothemetaphysicalquestionsaboutrealityanditsnature.Inparticular,thebestcomposi-tionalmeaningtheoryforEnglish,forexample,willrequirepositingeventsinordertoexplainwhatwesayforsentencesaboutactions,events,andsingularcausalrelation-ships(like“Frank’spushingBillcausedhimtofall”).So,considertheaction/eventsentence(1)andanobviouscandidateforitsinter-pretivetruthcondition(2):1JohnhitBill.2“JohnhitBill”istrueif,andonlyif,JohnhitBill.In(2),linguisticexpressionsarebothusedandmentioned.Boththewords“hit,”“John,”and“Bill”andthecorrespondingaspectsoftheworld,thepeopleandtheactionwhichrelatesthem,arediscussed.Inthislimitedsense,(2)couldbesaid“hookup”lan-guageandreality.Thishook-upremainssilent,though,onthenatureofreality,sinceittellsusnomorethanwhattheEnglishsentence(1)requiresforitstruth,namely,thatJohnhitBill.However,sinceanadequatecompositionalmeaningtheorymustbefinite(1984b:4–15)inconstructingacompositionalmeaningtheoryfor,say,English(anunboundedlanguage),structuremustbereadintoitssentences.Consideractionsen-tences(3)–(5):3JohnhitBillatsix.4JohnhitBillatsixinthebedroom.5JohnhitBillatsixinthebedroomwiththestick.Thesearebutthreeexamplesofhowadverbialmodifierscanbeadded,intheformofprepositionalphrases,toanEnglishsentencewithoutcompromisingitsgrammatical-ityorintelligibility.ThereisnoobviousspecifiableupperlimituponthenumberofmodifiersEnglishallowsustosensiblyattachtothesesortsofsentences–“afterdark,”“onhisear,”“onaTuesday,”andsoon.Therefore,anycompositionalmeaningthat304\nDONALDDAVIDSONtreatseachofsuchsentencesasinvolvingadistinctprimitiverelationthreatenstooffendagainstthefinitudeconditiononacompositionalmeaningtheory.WerewetotrytodeviseacompositionalmeaningtheoryforEnglishaccordingtowhich(3)istruejustincasethethree-placerelationofhittingobtainsbetweenJohnandBillandsixo’clock,whatwouldwesayabout(4)?Isittruejustincasethedistinctfour-placerela-tionofhittingobtainsamongtwopeople,atimeandaplace?Andonitgoes.Thisstrat-egyfordevisinginterpretiveT-sentences,ineffect,windsuptreatingeachadverbialmodifierasintroducingadistinctandnovelrelationwithadistinctnumberofrelata.Acompositionalmeaningtheoryprohibitspositingindefinitelymanydistinctprimitivepredicatesinalanguage.Recall,theaimofacompositionalmeaningtheoryistoexplainhowtherecanbeindefinitelymanynon-synonymousmeaningsentencesgivenafinitebasisofmeaningfulcomponents.ThemethodoftruthinmetaphysicsOnthebasisofsuchconsiderations,Davidsonadvancesaproposalwhichsimultane-ouslyrevealscommonelementsinallthesemanydistinctsentences,issuesintheircorrectinterpretiveT-sentences,andvalidateslogicalimplicationsamongthem,forexample,that(4)logicallyimplies(3),andthatboth(4)and(3)imply(1).Hisidea,roughly,istointerpretsentenceslike(1)and(3)–(5)insuchawaythattheyare“revealed”toharboranexistentialquantifierthatrangesoverevents.Themetaphysicalpunchlineisthateventsmustexist,since,otherwise,afinitecom-positionalmeaningtheoryforEnglishwouldbeunattainable.(ThisishowDavidsonjustifiesthesortofontologicalcommitmenttoeventswesawatworkinthesection“EventsandCausation.”)Morespecifically,DavidsonarguesthatthebestcompositionalmeaningtheoryforEnglishwillinterpretactionsentenceslike(1)and(3)–(4)roughlyalongthefollowinglinesof(1¢)and(3¢)–(4¢):1¢ThereisaneventthatisahittingofBillbyJohn.3¢ThereisaneventthatisahittingofBillbyJohnanditoccursatsix.4¢ThereisaneventthatisahittingofBillbyJohnanditoccursatsixanditoccursinthebedroom.Ineachinterpretation,thereisexistentialquantificationoveranevent.Eachinvokesexactlythesamethree-placerelationofhittingandtherebyshowswhatwealreadyintuitivelyrecognizetobecommongroundamongthem.Andin(3¢)and(4¢),adver-bialmodificationistransformedintoapredicationofthispositedevent.Thisstrategyfordiscerningontologicalcommitmentsextendstoanylocutionwherequantificationandpredicationarerequiredinordertoconstructasatisfactorycompo-sitionalmeaningtheoryfornaturallanguage,andthusprovidesuswithageneralmethodforisolatingontologicalcommitmentsforspeakers.Notehowfarremovedthisstyleofargumentisfromalong-standingtraditionoftreatingmetaphysicsasaninde-pendentdiscipline,whichsomehowabstractsfromquestionsaboutmeaningoraboutscience,insteadtakingamiddlegroundbetweenscienceandanalysisfordiscerningthenatureofreality.Davidson,perhaps,couldbeinterpretedaschallengingtheneedforanysuchmiddleground.305\nERNESTLEPOREAgainstfactsAdoptingthemethodoftruthinmetaphysicsasastrategyfordiscerningontologicalcommitmentalsohasnegativeramificationsfortraditionalontologicalcommitmenttofacts.AviewgoingbackatleasttoearlyRussellandWittgensteinoftheTractatusisthattheworldispopulatedwithfactsandthattruesentencesaretruebecausetheycorrespondtothesefacts.Inmorerecentwritings,forexample,“TheMythoftheSubjective”(1989b)and“TheStructureandContentofTruth”(1990),Davidsonattemptstorefutetheclaimthatsentencesarerepresentationsofreality.Inthelatteressay,hearguesagainst“thepopularassumptionthatsentences,ortheirspokentokens,orsentence-likeentitiesorconfigurationsinourbrainscanproperlybecalled‘representations’,sincethereisnothingforthemtorepresent”(Davidson1990).Inpar-ticular,therearenofacts.Andiftherearenofacts,andsofactsfailtomakesentencestrue,thenwecanask,Inwhatsensearesentencesrepresentational?OnDavidson’sapproachtomeaning,thebestcompositionalmeaningtheorydoesnotrequiretreatingsentencesascorrespondingtofactsorpropositions.(Lookin(T)above,thecandidateinterpretationof“Schneeistweiss.”Onitsrighthandside,thereisanEnglishsentence,whichprovidesaninterpretationoftheGermansentencewithoutmakinganyreferencetoafactorproposition.)ThoughnotallofwhatDavidsonsaysinhisattackonrepresentationscanbeaddressedhere,enoughcanbesaidabouthisattackagainstfactsandcorrespondencetheoriesoftruthtoilluminatehismethodologicalapproach.Davidson’smainargumentagainstfactsandcorrespondencetheoriesoftruthappearsinhis“TruetotheFacts”(in1984b).Therewefindhisso-calledGreatFactargumentfortheconclusionthat,givencertainplausibleassumptions,therecanbeatmostonefact.TheassumptionsDavidsonpresumesare“thatatruesentencecannotbemadetocorrespondtosomethingquitedifferentbythesubstitutionofco-referringsingularterms,orbythesubstitutionoflogicallyequivalentsentences.”Fromtheseassumptionshearguesthatonecanprovethat“iftruesentencescorrespondtoany-thing,theyallcorrespondtothesamething”(1990:303).Rightlyorwrongly,Davidsontakestheseassumptionstoembodytraditionalwisdomaboutfacts.ThemainpointoftheGreatFactargumentisthatifacontextsatisfiesthesetwoassumptions,thenthatcontextistruthfunctional(thatis,thecontextlooksjusttothetruth-valueoftheelementsitrelates,ratherthantoricherfeatures,suchastheirmeaning).So,ifonesentenceismadetruebysomefactf,theneverysentencethatagreesintruth-valuewithitisalsomadetruebyf,andthus,theGreatFact.SinceDavidsonplacestheconceptoftruthtotheforefrontoftheprojectofgivingatheoryofmeaningforanaturallanguage,andsinceherejectsonetraditionalcor-respondencetheoryoftruth,it’sappropriatetoaskwhathisviewisaboutthiscentralconcept.Inthenextsection,wewillconsiderbrieflyDavidson’sposition.TruthandcorrespondenceAsnotedinthelastsection,Davidsonrejectsthetraditionalcorrespondencetheoryoftruth,correspondencewithfacts.Still,inaway,Davidsonhimselfisacorrespondencetheorist,sinceheexplainsthetruthofsentencesintermsofarelationbetweenlan-306\nDONALDDAVIDSONguageandsomethingelse.Recastingacorrespondencetheoryoftruththisbroadly,Davidson’sclaimisthatthesortofcompositionalmeaningtheoryforanaturallanguageheendorsesisacorrespondencetheoryoftruth(1984b:70),becauseitexplainswhatitisforsentencestobetrue,notbyrelatingsentencestoobjects,butbyrelatingpredicatesandreferringtermstoobjectsviatherelationsofsatisfactionandreference,andexhibitingtheconditionsunderwhichsentencesaretrueintermsofthoserelations.Recallfromourearlierdiscussionofadverbialmodification,thataccordingtoDavidson,thebestcompositionalaccountforexplainingtheunboundednessofmodi-ficationasin“JohnkissedMaryintheparkafterdark”willtreatthissentenceasquan-tifyingoveraneventandpredicatingofthiseventthatitisakissingbyJohnofMary,andthatittookplaceintheparkandthatitoccurredafterdark.IfDavidsonisright,theninorderforthissentencetobetrue,theremustbeanentityintheworld,anevent,that,sotospeak,makesthissentencetrue.Thiswayofthinkingaboutcorrespondencepromisestobemoreilluminatingthanthetraditionalcorrespondencetheory,becausewecanexhibitbywayoftheproofofaT-sentencefromtheaxiomsofatruththeoryhowthetruthconditionsofthesen-tencearearrivedatonthebasisofthesatisfactionconditionsforitssignificantparts;foreachnon-synonymoussentence,therewillbeadifferentroutetoitstruthcondi-tions.Moreover,itisclearthatthisapproachaffordsnowayofeliminatingtheseman-ticconcept(s)onwhichitisbased,thatis,satisfactionandreference,sothisapproachisnotaredundancytheory.(Theothersideofthiscoinisthattheapproachshedslittlelightonwhatitisforapredicatetobesatisfiedbyanobjectorforasingulartermtomakereferencetoanobject.)Itisinthissense,andonlyinthissense,thatDavidsonisacorrespondencetheorist.Davidson,inmorerecentwork,rejectsthislabelasmis-leading(1990),thoughnottheviewitself.Itisworthnotingthatbeingacorrespon-dencetheoristinthissenseisneutralwithrespecttotraditionaldisputesbetweenvariousbrandsofrealismandanti-realism.ThusfarwehavebeenexploringramificationsofDavidson’sviewsaboutmeaningandinterpretationforthenatureofanexternalreality;inthenextfewsectionsthedis-cussionturnsinwardstoexploreconclusionsDavidsondrawsaboutaninnerreality,aboutthenatureofmind.AnimalthoughtWhatcomesfirst,languageorthought?Intuitioncutsbothways:it’shardtofindapetownerwhobelievesthatherpetisathoughtlessbrute,regardlessofwhetherthatpethasanyfacility,tospeakof,withlanguage.Ifintuitionhasgotthisright,thenlanguageandthoughtareindependent.Yetitseemsalmostequallyintuitive,thatatleastwhenwethink,wethinkinournativetongue.Soatleastforwewhohavelanguageitseemsasifthetwoaremutuallydependent.Davidson,however,defendstheunrestrictedthesisthatanycapacitytothinkrequiresfacilitywithlanguage,andsoonlycreatureswithalanguagecanthink.Somuchforpetlovers!In“ThoughtandTalk”(1984b)and“RationalAnimals”(1986c),hebeginshisargumentforhiscontroversialthesisbynotingthatascriptionofpsy-chologicalstatestoothers,e.g.,belief,desire,intention,andthelike,exhibitsemantic307\nERNESTLEPOREopacity.So,inattributingtoadogthebeliefthatthecatwentupthetree,wouldtheproprietyofthisascriptionbeaffectedwerewetosubstitutefor“thetree”anotherexpressionthatreferstothesametree,say,“thechestnutinyourbackyard”?Giventhatthetreeisthechestnutinyourbackyard,wouldyoubesoinclinedtoascribetothedogthathebelievesthatthecatwentupthechestnutinyourbackyard?Ifnot,thiswoulddisclosethatyourattributionofbelieftothedogfallsshortofliteralness;thatis,itissemanticallyopaque(1986c:475).Davidson’smaincontentionisthatsemanticopacity,afailuretopreservetruthunderco-referentialsubstitution,existsonlywhenlanguageistiedtothought.Headvancestwodifferentlinesofargumentforhisconclusion.Thefirstappealstoaholis-ticthesisaboutbeliefascription,bywhichismeantthatwecouldneverhavegroundsforascribingasinglebelieftoanorganismexceptagainstthebackgroundofawidearrayofotherbeliefs.Since,asDavidsonargues,wecouldneverhavegroundsforascribingtherequiredarrayofbackgroundbeliefstocreaturesthatdidnothavealan-guage,wecouldneverbewarrantedinascribingtosuchcreaturesanythoughtatall.Theargumentrunsasfollows:1Beliefascriptionexhibitssemanticopacity,and2semanticopacityrequiresthatweregardbeliefsaspossessingsomedefiniteinten-tionalcontent,and3thepossessionofabeliefwithadefiniteintentionalcontentpresupposes“endless”furtherbeliefs(holism).4Therefore,acreaturetowhomwearewarrantedinascribingabeliefisonethatmustpossessasophisticatedbehavioralrepertoire;5butonlylinguisticbehaviorexhibitsthesortofcomplexpatternthatmightwarrantsuchascription.Thisargument,evenifsound,atmostestablishesthatweareunlikelyevertohavedeci-siveevidencethataspeechlesscreaturehasbeliefs.Buttheviewsforwhichwecancollectdecisiveevidencesandwhatwecanestablishastruthcancomeapart.Davidsonwantstodrawastrongerconclusion,that“unlessthereisactuallysuchacomplexpatternofbehavior,thereisnothought”(1986c:476).Hedoessobyarguingthat:6Propositionalattitudesrequireadensenetworkofbeliefs(holism),and7“inordertohaveabelief,itisnecessarytohavetheconceptofbelief”and8“inordertohavetheconceptofbeliefonemusthavelanguage,”thatis,onemustbeamemberofa“speechcommunity.”(1986c:478)ThebigquestionforDavidsonishowtogetfromtheubiquityofbelief(6)andtheviewthatbeliefsrequiresecond-orderbeliefs(7)tohisconclusionthat“acreaturemustbeamemberofaspeechcommunityifitistohavetheconceptofbelief”(8)(1984b:170).Hearguesasfollows:9Thepossibilityofbelieforthoughtgenerallyistakentodependontheconceptofarepresentationthatmightbetrueorfalse,and10aconceptoftruthandfalsityincludessomenotionofanobjective,publicdomain,and11this,inturn,ispossibleonlyforaninterpreter.(in1984b:170,1986c:480)308\nDONALDDAVIDSONDavidsonholdsthatonlyutterancescanaffordthefine-grainedstructurerequiredforattributingthought;foronlyacreaturewhosebehaviorexhibitsthekindofstructureimpliedbyacompositionalmeaningtheoryisacreatureinwhichsemanticallyopaquerepresentationscanmakeanappearance.AlternativeconceptualschemesEvenifthoughtrequireslanguage,isn’titpossiblethatdifferentpeople,communities,cultures,orperiodsview,conceptualize,ormaketheworld(ortheirworlds)indiffer-entways?Couldn’tanotherthinkerhaveconceptsorbeliefsradicallydifferentfromourown?Davidson,unsurprisingly,identifiesconceptualschemeswithsetsofintertrans-latablelanguages(1984b:185).Bysodoing,hetransformsthequestionaboutalter-nativeconceptualschemesintooneaboutwhethertherecouldbenon-intertranslatablelanguages.Butwhyshouldanyonebelievethatquestionsaboutconceptualrelativityhaveanythingtodowithtranslation?Davidson’sidentificationrequirestwoassumptions:first,thatspeakersalonehavethoughtsandsecond,thatanyconceptaspeakerpossessesandanythoughthecanentertainisexpressibleinhislanguage.(Bothassumptionswereevidentinourdiscus-sionaboveof“ThoughtandTalk,”and“RationalAnimals.”)Togethertheseentailthatadifferenceintheconceptualschemesoftwopeoplerequiresthataportionofthelan-guagethatonespeaksisnottranslatableintoanyportionoftheother’s.Onceconceptualschemesareidentifiedwithsetsofintertranslatablelanguages,thequestionofwhethersensecanbemadeofradicallydifferentconceptualschemesreducestowhethersensecanbemadeoftwonon-intertranslatablelanguages(or,muchthesame,towhetherornota“significantrangeofsentencesinonelanguagecouldbetranslatedintotheother”(1984b:185)).Davidsonarguesthatmakingsenseofsuchtalkrequiresacriterionforwhenaformofbehaviorcanbecountedbothasspeechbehaviorandasspeechthatisuntranslatableintoourown.Hethenarguesthatnosensecanbemadeofatotalfailureoftranslatabilitybetweenlanguages(1984b:185),andsonoonecouldbeinapositiontojudgethatothershadconceptsorbeliefsradicallydifferentfromhisown(p.197).Inshort,pressuresfromthenatureofradicalinterpretationtogetherwiththefactthat“allunderstandingofthespeechofanotherinvolvesradicalinterpretation”(1984b:125)forcehimtodrawhiscriticalconclusions.Anti-skepticismAnotherconsequenceofradicalinterpretationisanti-skepticism,thatis,theimpossi-bilityofmassiveerror.Inanumberofarticles,beginningwith“TheMethodofTruthinMetaphysics,”andincluding“EmpiricalContent,”and“ACoherenceTheoryofTruthandKnowledge,”continuingmorerecentlythrough“WhatisPresenttotheMind,”“TheConditionsofThought,”“ThreeVarietiesofKnowledge,”and“EpistemologyExternalized,”Davidsonargues,onthebasisofhisprincipleofcharity,thataninter-pretercannotfindspeakerstohavelargelyfalsebeliefs,evenifsheherselfhasnoopinionastothegeneraltruthandfalsityofthesebeliefs.Givenwhatbeliefsare,andhowtheircontentsaredeterminedonthisstory,Davidsoniscommittedtothe309\nERNESTLEPOREimpossibilitythat“allourbeliefsabouttheworldmightbefalse”(1991b:193).Aradicalinterpretermusthavebeliefsabouttheworldinordertosucceedinascribingtoothersbeliefsabouttheworld.But,asradicalinterpretationisconceived,shealsomustfindotherslargelyinagreementwithherinthosebeliefs.Davidson’santi-skepticalargumentfromradicalinterpretationrestsontwoassump-tions,namely,thattobeaspeakeristobeinterpretablebyothers,andthattobeinter-pretablebyothersrequiresbeinglargelyright,notonlyinone’sgeneralbeliefs,butinbeliefsaboutthelocalenvironment.Ontheassumptionthatradicalinterpretationispossible,theproperwaytostatetherequirementonaspeakeristhatherbeliefsaboutherenvironmentbemostlytrue.Thecrucialaspectofradicalinterpretationistheimportanceofcausalityindeterminingwhatsomeonemeansorbelieves.Wecannot“ingeneralfixwhatsomeonemeansindependentlyofwhathebelievesandindepen-dentlyofwhatcausedthebelief...Thecausalityplaysanindispensableroleindeter-miningthecontentofwhatwesayandbelieve”(1986:435).So,itisthecentralroleofcausationinfixingthecontentsofbeliefsthatensuresthatthetruthofeverythingwebelieveisnotingeneral“logicallyindependent”ofhavingthosebeliefs;andthatotherscannotdiffertoomuchfromusinwhattheybelieve.Thecentralclaimthroughoutisthathowthecontentsofbeliefsaredeterminedlimitstheextentoffalsityanddiversitydiscriminableinacoherentsetofbeliefs.Ininterpretinganother,aradicalinterpreterventureshypothesesastowhatinthecir-cumstancesinquestioncausesaspeakertoholdtruethesentenceinquestion.Thisissupposedtoprovidehim(ceterisparibus)withthemeaningofthatsentence.Ineverycasetherewillbedifferentcausalchainsleadingtothesameutterance.Aradicalinter-pretermustchooseone.Hedoessobyrespondingtosomethingintheenvironment,andsoconvergesonsomethingthatisacommoncausebothofhisownresponseandoftheutteranceofthespeaker,therebycorrelatingthetwoandthusgivingthecontentofthespeaker’sutterances.(ThisiswhatDavidsoncalls“triangulation”(1986c:480,1991b:159–60).Thecentralthoughtthatemergesisthatwithoutconstraintsonwhatacreatureisthinkingaboutinadditiontothoseprovidedsimplybytreatingitasaratio-nalcreaturecapableofthoughtandspeech,anyanswertothequestionwhatitisthink-ingaboutwillbesowildlyunderdeterminedthatwecangivenoclearcontenttotheideathatitisthinkinganythingatall.IsthedogthinkingthatthesquirrelranupthetreeorinsteadthatlargebrownthinginfrontofitorthetreethatUncleBillplantedtenyearsago,andsoon.Withoutanyclearwaytoruleoutoneoftheseattributionsovertheothers,whyassumethatthedogbelievesanyoneofthem?)Themethodofradicalinterpretation“enforces”onanysuccessfulinterpretertheconclusionthataspeaker’sbeliefsarelargelytrueandlargelylikeherown.Thus,globalskepticismisruledout.Anti-CartesianismandfirstpersonauthorityAcentralfeatureoftheCartesiantraditioninmodernphilosophyisthatatthefoun-dationofthestructureofourjustifiedbeliefsabouttheworldareourbeliefsaboutourownmentalstates,ourattitudes,experiences,andsensations.Aswehaveseen,Davidson’sapproachbothtomeaningandinterpretation,andtocentralissuesinepis-temology,isanti-Cartesianinasmuchasherejectsthisassumption.Aradicalinterpreter310\nDONALDDAVIDSONisrestrictedtobehavioralevidenceininterpretinganother.Fromthisstandpoint,Davidsontreatsthecentralconceptsemployedininterpretinganotherastheoreticalconceptsintroducedtokeeptrackofbehavior.Viewedfromhisperspective,theroleofatheoryofinterpretationistoidentifyandsystematizepatternsinthebehaviorofspeakersinrelationtotheirenvironment.Ifthisisright,wedonotfirsthaveaccesstofactsaboutspeakers’meaningsandattitudes,includingourown.Howheaimstoreconcilehistreatmentofthecentralconceptsofinterpretationastheoreticalwiththepresumptionoffirstpersonauthority,thatis,withthefactthatspeakersarenecessarilymoreauthoritativeingeneralabouttheirownattitudesandsensationsthanothersare,isacentraltopicinDavidson’slaterwritings.Itseemsscarcelyintelligiblethatanothercouldbeaswellplacedasyouarewithrespecttowhetheryoubelievethatyouarehungry,orinpain.Thisasymmetryinepis-temicpositionisconnectedwithadifferenceinthewayweknowourownmentalstatesandthewayothersknowthem.Inascribingmentalstatestoothers,werelyontheirbehavior(orrecordsoftheirbehavior);butinthecaseoffirstpersonascription,wedonot.Indeed,inourowncasewedonotrelyonevidenceatall,andsodonotconsultourbehavior.Althoughknowingsomethinginadifferentwayornotonthebasisofevidencedoesnotinitselfguaranteethatwhatisknownisknownbetter,wemayexpectthatthisdifferenceinhowoneknowsone’sownmentalstates(firstpersonknowledge)andhowothersdounderliesfirstpersonauthority.ThechallengethatfirstpersonauthoritypresentstoDavidson’sassumptionofthetheoreticalcharacteroftheconceptsofinterpretationisfirsttakenupinhis“FirstPersonAuthority,”whereheoffersanexplanationforthepresumptionthat“aspeakerisrightwhenhesincerelyattributesabelief,desire,orintentiontohispresentself”(1984a:101)bygroundingitintheassumptionsthataninterpretermustmakeinordertosucceedatinterpretation.Davidsonaimstoexplaintheasymmetrybetweenourknowledgeofourmentalstatesandourknowledgeofthementalstatesofothers(or,alternatively,theasym-metrybetweenourownknowledgeofourmentalstatesandtheknowledgeothershaveofthem)byexplainingacloselyrelatedasymmetry:whythereisa“presumptionthataspeakerisrightwhenhesincerelyattributesabelief,desire,orintentiontohispresentself,whilethereisnosuchpresumptionwhenothersmakesimilarattributionstohim”(1984a:101).Hisexplanationoffirstpersonauthorityrestsonanexplana-tionofanasymmetrybetweentheknowledgeaspeakerandinterpreterhaveofmeaningsofthespeaker’swords.Thisasymmetrybetweentheknowledgeonehasofone’sownwordsandaninterpreter’sknowledgeofthemeaningsofone’swordsismoststrikinginthecaseofaninterpreterwhoisnotamemberofone’sspeechcommunity.TheformofDavidson’sargumentisthefollowing:onespeaksalanguageonlyifoneisinterpretable;oneisinterpretableonlyifoneismostlyrightaboutthemeaningsofone’swords;therefore,onespeaksalanguageonlyifoneismostlyrightaboutthemeaningsofone’swords.Davidson’scentralmethodologicalassumptionisthatathirdpersonpointofviewonothers’utterancesandpsychologicalstatesisprimaryinthesensethatbehavioralevidenceformsouronlyevidencefortheapplicationoflinguisticandpsychologicalconceptsandtermstoothers,andthattheircontentistobeunderstoodwhollyinterms311\nERNESTLEPOREoftheirroleinaccountingforthebehavioralevidenceavailabletousfromthisstand-point.Hisshiftofviewpointissofundamentalthat,onceadopted,thewholelandscapeinthephilosophyoflanguageandmindlooksdifferent.IfDavidsonisright,thecentralmistakeofourphilosophicaltraditionistheassumptionoftheCartesianstandpoint,and,inparticular,thecentralplaceourtraditionaccordstotheepistemicpriorityofknowledgeofourmentalstatestoknowledgeoftheworldandotherminds.Oncethisassumptionisrelinquished,eachdomaininwhichwehaveknowledgewillbeseenasnecessaryfortheothers,butknowledgeoftheworldandbyextensionofothermindswillturnouttobeautonomousfromknowledgeofourownminds,inthesensethatitisnotexplicablebyappealtoinferencesfromabasisinknowledgeofourownminds.Inlightofthealternative,Davidson’spictureisattractive.Partofitsinterestandpowerliesinitspromisetolaytorestwhathavebeenperhapsthecentralproblemsofthetra-ditionfromthebeginningofthemodernperiod.Despitethedifficultiesitfaces,itisworthpursuing.TherejectionofempiricismAnotherconsequenceofhistakingwhatwemightcallathirdpersonperspectiveoftheradicalinterpreterasmethodologicallyfundamentalistherejectionofallformsoftraditionalempiricism.Essentialtotraditionalempiricismisitsattempttoaccountforourknowledgeoftheworldexclusivelybyappealtosensoryexperience.Whatisdis-tinctiveaboutempiricismisnotthethoughtthatsensoryexperiencecanplayaroleinjustifyingourbeliefsabouttheworldaroundus,butthatitplaystheroleofafounda-tionforourempiricalknowledge.Thisinturnentailsthatthefirstpersonpointofviewisfundamental,sinceeachperson’sexperienceistreatedasbeinghisownfoundationforhisempiricalknowledge.Inadoptingthethirdpersonpointofviewasfundamen-tal,then,Davidsonrejectsacentraltenetofallformsofempiricism,andthetraditionalprojectassociatedwithitofexplainingourempiricalknowledgebyappealtoexperi-ence.Rather,inDavidson’sview,ourknowledgeoftheworldaroundus,ofotherminds,andofourownminds,hasaunifiedsourceinournatureasrationalbeingscapableofcommunicatingwithoneanother.Inconclusion,Davidsonarguesthatlanguage,mind,andactionareinseparable.Toaccountforlanguage,thatis,toanswerthequestion,Whatismeaning?,headvancestheradicalideathatatheoryofmeaningcanbesatisfactoryonlyifitdiscoversafinitebasicvocabularyandrulesofcompositioninthelanguagetobeinterpreted.Theaimtoprovideacomprehensiveunderstandingofnaturallanguagesledhimtoatreatmentofthetheoryoftruthforalanguageasanempiricaltheory,andtotheadoptionofthestanceoftheradicalinterpreterasthestandpointforconfirmation,linkingthestruc-tureofarichtheorywithitsbasicevidence,andplacingthetheoryofmeaninginthecontextofatheoryofrationalagency.Adoptingthisstanceasfundamentalistanta-mounttotherejectionofCartesianismandempiricism,andsotheabandonment,amongotherphilosophicalmainstays,ofconceptualrelativism,globalskepticism,andrepresentationalism.Theoriesfrequentlyyieldinsightintoproblemsthattheywerenotspecificallydesignedtosolve.Aswithothersignificantphilosophers,acarefulreadingofDavidson’swritingsbearsoutbothhowbroadinscopehisphilosophicalaccom-plishmentsareand,moreimportantly,howwelltheycohere.312\nDONALDDAVIDSONNoteSomeofthisentry,inparticular,fromthesection“AgainstFacts”totheend,isadaptedfromKirkLudwig’sandmyforthcomingDonaldDavidson:Truth,Meaning,andReality,OxfordUniversityPress.BibliographyWorksbyDavidson1980:EssaysonActionsandEvents,Oxford:OxfordUniversityPress.Thecollectioncontainsthefollowingessays,listedherewiththeiroriginalpublicationdates,andpageextentsinthecollection.“Actions,Reasons,andCauses”[1963],pp.3–19.“CausalRelations”[1967],pp.149–62.“MentalEvents”[1970],pp.207–27.“PsychologyasPhilosophy”[1973],pp.229–44.“TheIndividuationofEvents”[1969],pp.163–80.“TheLogicalFormofActionSentences”[1967],pp.105–48.“TheMaterialMind”[1973],pp.245–59.1984a:“FirstPersonAuthority,”Dialectica38,pp.101–11.1984b:InquiriesintoTruthandInterpretation,Oxford:OxfordUniversityPress.Thecollectioncontainsthefollowingessays,listedherewiththeiroriginalpublicationdatesandpageextentsinthecollection.“InDefenceofConventionT”[1970],pp.65–76.“OntheVeryIdeaofaConceptualScheme”[1974],pp.183–98.“RadicalInterpretation”[1973],pp.125–39.“TheMethodofTruthinMetaphysics”[1977],pp.199–214.“TheoriesofMeaningandLearnableLanguages”[1965],pp.3–15.“ThoughtandTalk”[1975],pp.155–70.“TruetotheFacts”[1969],pp.37–54.“TruthandMeaning”[1967],pp.17–36.1986a:“ACoherenceTheoryofTruthandKnowledge”[1983],inLePore1986,pp.307–19.1986b:“EmpiricalContent”[1982],inLePoreandMcLaughlin1986,pp.320–32.1986c:“RationalAnimals”[1982],inLePoreandMcLaughlin1986,pp.473–80.1989a:“ConditionsonThought,”inTheMindofDonaldDavidson,issue36ofGrazerPhilosophischeStudien(ed.BrandlandGombocz),Amsterdam:Rodopi,pp.193–200.1989b:“TheMythoftheSubjective,”inRelativism:InterpretationandConfrontation,ed.M.Karusz,SouthBend,IN:UniversityofNotreDamePress.1989c:“WhatisPresenttotheMind,”inTheMindofDonaldDavidson,issue36ofGrazerPhilosophischeStudien(ed.BrandlandGombocz),Amsterdam:Rodopi,pp.3–18.1990:“TheStructureandContentofTruth,”JournalofPhilosophy87,pp.279–328.1991a:“EpistemologyExternalized,”Dialectica45,pp.191–202.1991b:“ThreeVarietiesofKnowledge,”inA.J.Ayer:MemorialEssays,ed.A.Phillips,Cambridge:CambridgeUniversityPress,pp.153–66.WorksbyotherauthorsFodor,J.andLePore,E.(1992)Holism:AShopper’sGuide,Oxford:BlackwellPublishers.Hahn,L.(ed.)(2000)ThePhilosophyofDonaldDavidson,LaSalle,IL:OpenCourt.313\nERNESTLEPORELePore,E.(ed.)(1986)TruthandInterpretation:PerspectivesonthePhilosophyofDonaldDavidson,Oxford:BlackwellPublishers.LePore,E.andMcLaughlin,B.(eds.)(1986)ActionsandEvents:PerspectivesonthePhilosophyofDonaldDavidson,Oxford:BlackwellPublishers.314\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200124G.E.M.Anscombe(1919–2001)ANSELMMÜLLERGertrudeElizabethMargaretAnscombe,Britishphilosopher,studiedGreatsatOxford(1937–41),andwentasaresearchstudenttoCambridge,whereshebecameapupilandclosefriendofLudwigWittgenstein.ShewasappointedResearchFellow(1946),Lecturer(1951),andTutorialFellow(1964)ofSomervilleCollege,Oxford.In1967shewaselectedFellowoftheBritishAcademy.SheheldtheChairofPhilosophyintheUniversityofCambridgefrom1970to1986.HerphilosophicaloutlookhasbeeninfluencedmostofallbyAristotleandbyWittgenstein.SheisoneofWittgenstein’sliteraryexecutors,andhastranslatedandeditedlargepartsofhiswork.Atthesametimesheshowsgreatoriginality,notleastinthewayinwhichshebringsWittgenstein’sideastobearontopicsthathedidnothimselfexplore.Manyofherpapersareremarkablealsofortheuniquelyappreciative,unsparing,andcreativemannerinwhichsheengageswithgreatmindsofthepast,suchasHume.Anscombehasagiftforspottingwhatismostbasicintraditionalprob-lems,andoftenhersolutionsseemtoopenone’seyestowhatlayunderone’snose.Herlanguageisforcefulandaustere,herthinkingunrestrictedbyconventionorfashion.Anearlyexampleofherindependenceofmindcanbeseenin“TheJusticeofthePresentWarExamined,”apamphletwrittenwithNormanDanielintheautumnof1939.HereshequeriedboththeaimsoftheBritishGovernment,andthemeanslikelytobedeployed,infightingthewaragainstGermany;shealreadyforesaw“areabombing,”foresightthatlaybehindheropposition,in1956,totheconfermentonPresidentTrumanofanhonorarydegreebytheUniversityofOxford(1981a,III:72–81;cf.viiand62–71).Anscombehascontributedtoallprincipalareasofphilosophy.Thefollowingsummaryaccountofherworkconsidersitundersevenmainheadings,coveringmostofherpublishedwritings.Language,thought,andrealityApartfromtwoarticlesentitled“SubjunctiveConditionals,”whicharguethat“if–then”is,roughlyspeaking,truth-functional(in1981a,II:196–207),and“OnPrivateOstensiveDefinition,”whichexpoundsanddefendstheviewthattherecanbenoprivateconfermentofmeaning,nosinglebookoressayofAnscombe’sisasystematic315\nANSELMMÜLLERtreatmentofquestionsinthephilosophyoflanguageandlogic.Shecontributestoitratherbyshowingupweakspotsinreceivedanswers,takingasherclue,forthemostpart,passagesfromancientGreekphilosophersorWittgenstein.Typicalisherremark:“WeareaccustomedtothinkthatPlatointheCratyluswasextraordinarilyblindinassumingthatphonemeshavemeaning-roles.Butthis,asoften,maybeafailureonourparttoseeaproblem”(1981b:150).Thus,“UnderstandingProofs,”animaginarycontinuationofMeno85d9–86c2(1981a,I:34–43)canbereadasachallengetogiveabetteraccountthanthePlatonictheoryofreminiscence,ofthefactthatsimpleconceptualtruthscannotbeunderstoodwithoutbeingbelieved.Chapter1ofAnIntroductiontoWittgenstein’sTractatus(1959)containsanelaborateargument,nottobefoundintheTractatusitself,insupportoftheviewthat“a(verylarge)classofmutuallyindependentpropositions”ispresupposedinthecommonexplanationoftruth-functionaltautologies.Inotherpartsofthisbookandinpaperssuchas“Parmenides,MysteryandContradiction”(1981a,I:3–8)and“TheEarlyTheoryofForms”(pp.9–20)wefindkeenobservationsonissuessuchasnegationandtheinternalstructureofsentences,truthandfalsehood,propositionandassertion,meaningandreference,useandmention,tenseandmodality,universals,classes,andpredication.Againandagain,Anscombereturnstoclassicalproblemsofhowwordsandpropo-sitions,conceptsandthoughtsarerelatedtotherealitiestheysignify.Inanearlymas-terpiece,“TheRealityofthePast”(1981a,II:103–19),sheshowsthatthequestion“Howisitthatstatementsaboutthepasthavemeaning?”mustbeansweredbydescrib-ingtheuseofthepasttenseratherthanappealingto“theexperienceofremembering”orgesturingatthepastthing“reachedbythought”or“seenthrough”thepresentevi-dence.SheisintriguedbythemoregeneralParmenideo–Platonicproblem:“Howcanwethinkwhatisnot?”,whichshecontrastswiththemodernquestion:“Howcouldtherebewhatwecannotthink?”TheParmenidespaperpointsoutthat,whileofcoursewecannotwithoutabsurditysayofanythingthatitisbutcannotbethought,wemayreasonablysupposethatsomethingisbutcannotbethought.Concerningmodernattempts“todeducewhatcouldbefromwhatcouldholdofthought,”Anscombebelievesthat“theancientshadthebetterapproach,arguingonlythatathoughtwasimpossiblebecausethethingwasimpossible”(1981a,I:xi).IsWittgenstein’slaterphilosophyaversionofthemodernapproach(“essenceisexpressedbygrammar,”thatis,bytheruleswhichgovernourapplicationofwordstothatwhoseessenceisinquestion)?“No,”wearetoldin“TheQuestionofLinguisticIdealism”(1981a,I:112–33),anexpositionanddefenseofWittgenstein’sviewsontherelationshipbetweenlanguageandreality(seeWITTGENSTEIN).“Itlooksasifeitherthegrammarcorrespondedtosomethingoftheobject,itsrealessence,whichithaswhetherthereislanguageaboutitornot,orthe‘object’wereitselfdependentonlan-guage”(p.113);bothseemunacceptable.Anscombe’ssolutionisthis:Ontheonehand,realitydoesnotforceonustheconceptsinwhichwerelatetoit(“Howcouldanexpe-riencedictatethegrammarofaword?”:p.114).Samenessofexperience,orofkindofobject,cannotdeterminetheshapeofaconcept,sinceitisthecorrect(re-)applicationofthecorrespondingexpressionwhich,inthefirstinstance,settleswhichsameness–andhence,whichexperienceorkindofobject–wehaveinmind.(TheremaybeatensionbetweenthisclaimandtherhetoricalquestionshequotesfromWittgenstein:316\nG.E.M.ANSCOMBE“Dowemakeaconceptwhereverweseeasimilarity?”)Thusalternativestooursetofconceptsareindeedpossible.Ontheotherhand,theexistenceofwhateverourcon-ceptsapplytodoesnotthereforedependonourbeingtheretoconceiveofit.“Theseessences,then,whichareexpressedbygrammar,arenotcreatedbygrammar”(p.114).Butthereisroomfor“apartialidealism”(p.118),ofwhichmoreinthesection“ExistencebyConventionandIntention,”below.Thesecondpartofthepaperadressesafurtherproblem.Thegrammarofourlan-guagegovernsthepermissibilityofthejudgmentsexpressedinit,bylayingdown,inparticular,whatcountsasdecisiveevidenceforthem.However,itlaysdownalsowhatistakenforgrantednotonthebasisofevidencebut,forexample,asimpliedinthewayswejudgeandargueandact,orasaresultofteaching.Thecorresponding“hinge”propositionsvaryvastlyinsubjectmatterandrole.Compare“MynameisL.W.”;“TheearthhasexistedforalongtimebeforeIwasborn”;“ThereisnoGod”;and“Caesarisahistoricalfigure”(“HumeandJuliusCaesar”:1981a,I:86–92).Can“assumptions”thatareinthiswayatthebottomofalinguisticpracticethemselvesberightorwrong?“Findinggrounds,testing,proving,reasoning,confirming,verifyingareallprocessesthatgoonwithin,say,oneoranotherlivinglinguisticpracticewhichwehave”(p.130).Anddivergenceinjudgmentonaccountofdivergencein“world-picture”or“knowl-edgesystem”isnotamatterofmistakebutrather“disagreementinthelanguage”used(p.131).However,AnscombealsotakesWittgensteintoholdthatsomeonewhocomestojettisonacertainkindofgroundlessassumptionoritscertaintymayberightorwronginbelievingherealizesthatformerlyhewasnotcompetenttojudge.Andfromthissheconcludes:“Thatoneknowssomethingisnotguaranteedbythelanguage-game”;thatis,evenwheretherulesofourlinguisticpracticeleavenoroomfordoubt,falsehoodisnotexcluded(p.132f.).SomeofAnscombe’sessaysconcernthenotionofamaterialsubstance,defendingitscoherence,andcriticizing“bareparticular”conceptions,comparablemisunder-standingsofthenotionofmatter,andempiricistobjectionsandalternativestoabasi-callyAristotelianview.(See“ThePrincipleofIndividuation,”1981a,I:57–65;“Substance,”II:37–43;and“Aristotle:TheSearchforSubstance,”inThreePhilosophers,1961(togetherwithP.T.Geach),pp.3–63.Time,necessity,andcausationAnscombe’s“firststrenuousinterestinphilosophywasinthetopicofcausality,”andmuchofherlaterworkinthisareaisanelaborationoftheideathatthefutureisunde-terminedinthesensethat,forexample,thereis“nosuchthingashowsomeonewouldhavespenthislifeifhehadnotdiedachild”(1981a,II:vii).“AristotleandtheSeaBattle”(I:44–56)alreadyshowsheranincompatibilist:“Ifwhatthetypewriterisgoingtodoisnecessary,Icannotdoanythingelsewiththetypewriter”(p.48).AnscombemaintainsAristotle’sview,canvassedearlierin“TheRealityofthePast”(1981a,II:112–16),“thatnothingwhatevercouldmakewhatiscertainuntrue”(I:52);and,forthisreason,that“whenpdescribesapresentorpastsituation,theneitherpisneces-sarilytrue,or-pisnecessarilytrue”(p.53).Thiskindofnecessityisfurtherexploredinthecelebrated1971inaugurallectureatCambridgeUniversity,“CausalityandDetermination”(1981a,II:133–47).From317\nANSELMMÜLLERAristotleonwards,almostallphilosophers(includingHume)haveseentheessenceofcausalityinnecessitation.OnAnscombe’salternativeaccount,thenotionofcauseisembodied,inthefirstinstance,intheuseofsuchverbsas“scrape,push,wet,carry,...”(p.137).ThisnotionisoneofAderiving,orcoming,fromB:ofsomethingthat(paceHume)isoftenobservable.And“ifAcomesfromBthisdoesnotimplythateveryA-likethingcomesfromsomeB-likethingorset-uporthateveryB-likethingorset-uphasanA-likethingcomingfromit;orthatgivenB,Ahadtocomefromit,orthatgivenA,therehadtobeBforittocomefrom.Anyofthesemaybetrue,butifanyis,thatwillbeanadditionalfact,notcomprisedinA’scomingfromB”(p.136).Thesecondpartofthelectureexaminesthenotionofdeterminationanditsapplic-abilitytophysicalevents.“Whenwecallaresultdeterminedweareimplicitlyrelatingittoanantecedentrangeofpossibilitiesandsayingthatallbutoneoftheseisdisal-lowed...[by]...somethingantecedenttotheresult”(1981a,II:141).WemayknowthatAhasbeencausedbyBwithouthavinganyreasontothinkitwas,inthatsense,determinedbyBoranythingelse.AsystemlikeNewtonianmechanicswould,itistrue,providesuchareason;andthesolarsystemoffersamisleadinglyundisturbedinstan-tiationofitslaws,whichcanmakeitlookasifallcausalityhadtomatchthismodel.Butthisappearanceisillusoryontwocounts.(1)Onlyifthissystem(oracomparableone)appliedtoarbitrarilysmallquantities(soastoexcludeevenminutecausalindeterminaciesandtheirmultiplicationovertime)–onlythenwouldtheresultof,say,manyballsinteractingwitheachotherforsometimeinstablesurroundings,be(notonlycausedbut)determinedbysomeinitialsituation.Andwherewecannotassumesuchasystem,weshallhavetoadmitnon-necessitatingcauses,liketheradioactivematerialwhich(inFeynman’sthoughtexperiment)mayormaynot,viasomeGeigercounter,causeabombtogooff(pp.144f.).Anscombehopesthatthisparticularkindofexamplemaypreventusfromgoingon“asifundeterminednesswerealwaysencapsulatedinsystemswhoseinternalworkingscouldbedescribedonlybystatisticallaws,butwherethetotalupshot,andinparticulartheoutwardeffect,wasasnearasmakesnodifferencealwaysthesame”(pp.146f.).(2)Evenaphysicistwhobelievesthat“theresultthathappensoughttobeunder-stoodastheonlyonethat[inthecircumstances]waspossiblebeforeithappened”(1981a,II:142)neednotbeadeterminist.SupposeNewton’slawswerevalidforarbitraryquantitativedimensionsandthusprovidedfornecessitatingcauses.Thissuppositiondoesnotyetruleoutpreventionandinterferencefromotherforces,and,inthissense,thepossibilityofalternativeresults.Determinisminvolvesmore,namelythebeliefthat“thewholeuniverseisasystemsuchthat,ifitstotalstatesattandt¢arethusandso,thelawsofnaturearesuchasthentoallowonlyonepossibilityforitstotalstateatanyothertime.”Anscombeseesnoreasonforbelievingthisand,moreover,thinksitincompatiblewiththefreedomofaction,whichafterallforthemostpartconcernsphysicalmovements:“ifthese...arephysicallypredeter-minedbyprocesseswhichIdonotcontrol,thenmyfreedomisperfectlyillusory”1(p.146).Inotherworks,Hume’saccountofcausalityisfoundwantingonthefollowingthreecounts.(1)Hearguesthatitisimaginableandthereforepossibleforabeginningofexistencenottohaveacause.But,Anscombeasks,canwedetermine,without318\nG.E.M.ANSCOMBEidentifyingacause,thatthereandthenanobjectstartedtoexist,ratherthanarrived(havingtravelled“sayasagas”:1981a,II:161)?(2)Humefailstotelluswhatkindofspecificationofacauseanditseffectistocountifwewanttoconvinceourselvesofhisclaimthattheideaoftheoneis“distinct”fromthatoftheother(p.150).(Suchdis-tinctnessseemstobeabsent,forexample,whenX’smotherissaidtobethecauseofX.)(3)Imayknowwithoutobservationofsomethingthatithascausedmetodosome-thing,aswhen“IthoughtIsawafaceatthewindowanditmademejump”(p.75).Thistypeofcause(whichAnscombecallsa“mentalcause”)doesnotlenditselftoHume’sexplanationintermsofregularsuccessionatall.Fromexperiencetoself-consciousnessAnscombeisbestknownforherinfluentialworkinthephilosophyofmind.ApartfromherclassicIntention,shehasproduced“casestudies”suchas“TheSubjectivityofSensation”(1981a,II:44–56),“CommentsonProfessorR.L.Gregory’sPaperonPerception”(pp.64–70),“OnSensationsofPosition”(pp.71–4),“Pretending”(pp.83–93,includinganaccountofanger),“OntheGrammarof‘Enjoy’”(pp.94–100),and“WillandEmotion”(1981a,I:100–7).Oneofhermaintargetsisatemptationtobringeverythingmentalundertheheadingof“experience,”thusassimilatingittosen-sationsandimages.(See“EventsintheMind”:1981a,II:57–63.)Infact,apsycho-logicalconceptmaymakeessentialreferencetoawidevarietyofthingssuchasantecedentandsurroundingconditions,behavioralandverbalexpression,actionsandaims,assumptionsandthoughts,capacitiesandtendencies.In“Memory,‘Experience’andCausation”(1981a,II:120–30),afterremindingusthatamentalimagecannotbetheessenceofmemorybecause(memory)beliefs“areinvolvedinreferringanimagetothepast”(p.126),Anscombearguesthatthereisnocoreexperiencetomemoryatall(asthereistoseeing,orhearing).Genuineremember-ingisgenerallyassumedtobecomposedofsomesuchexperience,M,plusan“appro-priate”causallinkbetweenwhatisrememberedandM.Consider,however,thecaseofX,whoknowsaboutapastevent,E,inhislifewithoutknowingwhetherhedoessobecauseheremembersEorbecausehehasbeentoldofit.Interestingly,thislackofknowledgedoesnotprecludeXfromknowingthatEhappened.Nor,Anscombeargues,doesitconsistinX’sfailingtoknowwhetherhispresentcertaintyofEwascausedbyE.SupposenowthatX’sknowledgeisnotinfactmemoryofE;then,ofcourse,Mdoesnotcomeintothepictureatall.If,ontheotherhand,itisacaseofmemory(“hemustberemembering”),againitcannotinvolveM;forMwassupposedtobea“memoryexperience,”leavingnoroomforquestionslike“Ismybelief,iftrue,acaseofremem-beringE?”HenceMisaphilosophers’fiction.AsAnscombepointsout,thisisaproblemnotonlyforCartesiansandEmpiricists,butalsoforanymaterialistidentitytheoryofthemind(p.128).AccordingtoAnscombe,thehumanmindischaracterizednotsomuchbyexperi-enceasbyself-consciousness,understoodnotassomekindofself-perceptionbutasthebasisofself-ascriptionofallsortsofthingsincludingexperiences.“Self-”herehasthesamefunctionasintheclaimthateveryoneuses“I”tospeakofhim-,orherselfand,indeed,astheword“I.”Doesthisstandforthehumanbeingwhousesit,orratherforaCartesianego?Neither,wearetoldin“TheFirstPerson”(1981a,II:21–36).“I”is319\nANSELMMÜLLERnot,sinceithasnotthesenseof,aname,demonstrative,orother“referringexpres-sion.”Suchasensewouldrequire“a‘conception’throughwhichitattachestoitsobject”(pp.28f;cf.n.2).Forthesamereason,NN,insaying“IamNN,”mayexpressknowledge(oramistake,orlie)butdoesnotmakeastatementofidentity.Supposethateverybodycouldavailthemselvesofeverypossibleconceptionofeveryhumanbeing.Theresultinginformationwouldleaveunansweredformebutnotforotherstheques-tionwhichoneofthesepeopleIam.HumanactionandpracticalthoughtIntentionfirstappearedin1957andhasbeeninfluentialeversince,perhapsevenmorethanisgenerallyrecognized.Wespeakofintentiontof,intentionalf-ing,and(further)intentioninf-ing.ThesecondoftheseformsisthecentraltopicofAnscombe’sinquiry.Intentionalactionscanbemarkedofffromnon-intentionalactionsasthosewhichtheagent,withoutrelyingonobservation,knowsheisperformingandtowhichheallowsthequestion“Why?”toapplyinaspecialsense.Thissenseismarkedofffromthesenseassumedinananswerthatwouldstateacauseand,inparticular,amentalcause.Arelevantanswerwillgiveareasonforf-ingwhicheitherlooksbackward(asinrevenge:“Ifbecausehehitme”)oramountstoafurtherintention(aninterpretativereason:“Myf-ingisasignalforNN”;oranend:“Iamf-inginordertofindX”);oritwillindi-catethatthereisnoreason(“IjustthoughtIwouldf”).Thelattercasemustbedis-tinguishedfromarejectionofthe“Why?”question(“IwasnotawareIwasf-ing”or“IobservedthatIwasf-ing”or“Idon’tknowthecause”).Knowledgeofwhatyouaredoing,notbyobservationbutinintention,maybecalled“practicalknowledge.”Discrepancybetweenthoughtandrealityishereblamedonthelatter.Ifyoudonotbuywhatisonyourshoppinglist,themistakeisinyourperformancenotinthelist,asitwouldbeinthelistofadetectivewhomadeamistakeintrackingyourproceedings.WhatyoupracticallyknowyouaredoingwilloftencoincidewiththeconclusionofanAristotelian“practicalsyllogism.”Ifthiscon-clusionistof(orwordssuchas“SoI’llf”),yourpremisesmention(1)somethingwanted(undersome“desirabilitycharacterization”orother)and(2)fasawayofachievingit.Suchpracticalreasoningneednot“necessitate”:f-ingmaynotbetheonlywayofrealizingthethingyouwant.Also,theconnectionsexhibitedinapracticalsyl-logismmaytruthfullybestatedinyouranswertothe“Why?”questioneventhoughyoudidnotgothroughthembeforeyouf-ed.Theseideasarefurtherdevelopedin“ThoughtandActioninAristotle:Whatis‘PracticalTruth’?”(1981a,I:66–77),“PracticalInference”(1995:1–34),and“VonWrightonPracticalInference”(1989:377–404).UnderadescriptionAsAnscombepointsout(1963:37–49),oneandthesameactionmaybeintentionalundersomedescriptionsandnotunderothers.Inasingleactionyoumaybeinten-tionallymovingyourarmandintentionallycuttingbreadbutunintentionally(thoughperhapsknowinglyandthereforevoluntarily)contractingtheseparticularmusclesandpointingtowardsXwiththebread-knife.In“TheTwoKindsofErrorinAction”(1981a,320\nG.E.M.ANSCOMBEIII:3–9),thisisbroughttobearonthelegalandmoralassessmentofactions.Since,forexample,“youmayconsenttosomethingunderonedescriptionandnotunderanother,thefactoffraudmaybeaproofthatacertainconsenthasnottakenplaceatall”(p.3).And“ifamangenuinelyandreasonably,butwrongly,thoughtthatthiswaspropertyhehadarighttotakeaway,thenwesay‘Thatwasnotstealingatall’”(p.5).Moregenerally,culpableandnon-culpableignorance,knowledgeandintentioncanberelevantinvariouswaysto(1)whatdescriptionsaretrueofanactionofyours,and(2)whetheryouareresponsiblefortheactionunderagivendescriptionwhichappliestoit.Herealsobelongsthedistinction,essentialtothedoctrineof“doubleeffect”(seebelow),betweendescriptionsunderwhichyouintendanactionanddescriptionsunderwhichyoumerelyknowittobeinvolvedinwhatyouaredoingintentionally.In“UnderaDescription”(1981a,II:208–19),Anscombedefendsthisexpressionagainstattacksandmisunderstandings.Sheextendsitsapplicationandremindsusthat“thedescriptionunderwhich[something]isaimedatisthatunderwhichitiscalledtheobject”(1963:66).Thedescriptionunderwhichanactionisintended/somethingisan2intentionalobject(e.g.ofsight),canbe(1)non-interchangeable,(2)indeterminable,and(3)existentiallynon-committal,inwaysthatmaybeelucidatedbythefollowingexamples.(1)“ImeanttocutbreadbutnottopointtheknifetowardsX.”/“Didn’tyouseetheblood?”“Well,Isawredpatchesonthefloor.”(2)Cuttingasliceroughly1cmthickwillbecuttingit0.90orcuttingit0.91or...cmthick.Butintendingtocutitroughly1cmthickisnotintendingtocutit0.90orintendingtocutit0.91or...cmthick./Thepeopleyousawwerethirteeninnumber;butasaneye-witnessyoumayhavetostickto“Isawquitealotofpeople”or“perhapsadozen.”(3)“Cuttingbread”describesmyintentionbutnotwhatIamactuallydoingiftheknifeishopelesslyblunt./“WastherearealflashoflightwhenIsawone,orwastheresomethingwrongwithmyeyes?”(cf.p.4).In“TheIntentionalityofSensation”(1981a,II:3–20),Anscombeshowshowdif-ferentaccountsofperceptionsufferfromacommonneglectofthistopic:phenome-nalism“misconstruesintentionalobjectsasmaterialobjectsofsensation,”while“‘ordinarylanguage’philosophy...doesnotallowforadescriptionofwhatisseenwhichise.g.neutralasbetweenitsbeingarealspot(astain)oranafter-image”(pp.11f.).In“CausalityandExtensionality”(pp.173–9)shecommentsontheinadequacyofacausalstatementlike“ThechilddiedbecausethetallestgirlintownisRhesus-negative”ascomparedwith“...becausehismotherisRhesus-negative.”OnAnscombe’sview,suchstatementsshouldbeunderstoodasnon-extensionalstate-mentsconnecting,non-truth-functionally,genuinepropositionalcomponents.Hereagain,wemaysaythatitisthedescriptionunderwhichtheelementsofcauseandeffectareidentifiedthatmatterstotheintelligibilityifnottruthofthecausalclaim.ExistencebyconventionandintentionAlternativedescriptionsareinplayalsowhere,invirtueofgivenconventions,acertaindistributionofinkonapieceofpaperisanEnglishsentence,oraparticularkillingcapitalpunishment.“Convention”herepointstoculturalconstitutionratherthanagreement,letalonearbitrariness.ItisapervasivethemeinAnscombe,itsvariationssurfacinginthephilosophyof(1)language,(2)knowledge,(3)action,(4)morality,and321\nANSELMMÜLLER(5)socialinstitutions.Theextensiverelevanceofthetopicismadeexplicitin“Rules,RightsandPromises”(1981a,III:97–103),whereAnscombementionsthe“naturalunintelligibility”(Hume’sphrase)ofpromises,contracts,rights,legalobligation,etiquette,rulesofgames,rulesofgrammarandlogic,infringement,andsacrilege.(1)In“ATheoryofLanguage”thequestion“whatabouttheoccurrenceofasoundconstitutesitasign”(1981b:150)isleftunanswered.Butlanguage-gamedescriptionsaresaidtogiveus,bywayofcomparisonandwithoutrecoursetothenotionofmeaning,anideaofthepossiblefunctioningofawordinuse:anideaofhowthegrammaticalconventionsofouractuallanguagework.Thereisnorightorwrongabouttheseconventions.Theycreateourconcepts,butnotwhattheseareconceptsof(cf.thesection“Language,Thought,andReality,”above),withthenotableexcep-tionof“promises...rulesandrights,[which]areessencescreatedandnotmerelycapturedorexpressedbythegrammarofourlanguages”(1981a,III:100;seealso(3)below).(2)Grammaris,however,supposedtodeterminenotonlycriteria(thetypeofevi-denceforthepresenceofXwhoseprima-facievalidityispartoftheconceptofX),butalsostandardsofcomparativecertaintyforpotentiallyincompatiblejudgments.Iscon-ventionatthebottomofthesestandards,too?Andisthere,incaseofconflictingstand-ards,anycourtofappeal?Thesequestions(alsodiscussedinthefirstsection,above)donotseemtoreceiveadefinitiveanswerinAnscombe’swork.(3)Inashortpaper“OnBruteFacts”(1981a,III:22–5)weareintroducedtotheideaoffacts,describableasA,which(inasocietywithcertaininstitutions,givenavaguelyspecifiablenormalcontextandtheabsenceofanindefiniterangeofdefeatingconditions)“amountto”factsdescribableasB.Forinstance,makingtheaboveassump-tions,thefact(A)thatXhasdeliveredaquarterofpotatoestoYamountsto(andis“brute”relativeto)thefact(B)thatYowesmoneytoX.Buthowcan,inthiscase,theeventdescribedasAconstitutetheobligationclaimedinB?Howcan,inawholeareaofcomparablecases,an“ought”derivefroman“is”?Thesequestionsaretakenupinanessay“OnPromising”(1981a,III:10–21).Undersuitableconditions,myuttering“Ipromiseyoutof”amountstoapromisetof.Oneoftheseconditionsismyintentiontopromise.But(a)howcanweinvokethisinten-tioninexplainingpromisingtof,ifanaccountoftheintentionhastomentionitscontent,i.e.thatverypromise?And(b)howcanI,merelybyutteringcertainwordswiththatintention,bringaboutrestrictions,whichdidnotexistbefore,onmypos-sibilitiesofacting?Anscombe’sanswertobothquestionsconsistsinahighlyoriginalapplicationofWittgenstein’sideaofalanguage-game.Sheimaginesthefollowingprac-tice(pp.15–17):Thereisaformofwords“Bump!I’llf.”AparticipantNNwhohasuseditisliabletobemadebyotherstof.ThepressuretheyputonNNmaybephysi-calor,atalessprimitivestageofthelanguage-game,conventional.Inthelattercase,theyaddress“stoppingmodals”toNN,like“youcan’t”and“youhaveto.”These“areatfirstwordsusedbyonewhoismakingyoudosomething(orpreventingyou),andtheyquicklybecomethemselvesinstrumentsofgettingandpreventingaction”(p.101).Finally,theyarecombinedwith“logoi”like“(but)youbumpedtof!”(pp.101f.,142f.):“reasons,”whoseconnectionwithf-ing“isitselfnothing,exceptthatitislin-guisticallyMADE”(p.140).Ifrecalcitrant,NNisreproachedforhavingusedthosewords322\nG.E.M.ANSCOMBEandnotf-ed.Thispracticeofbumpinghasthesignificanceofpromising,ifweassumethataparticipantwilltrytoextracta“Bump!I’llf”fromotherswhenhewantsthemtof,andusetheirhavingbumpedtofasaweaponinmakingthemf,etc.Howisthisananswertoourquestions?(a)Mypromisecannowbeunderstood,withoutcircularity,asinvolvingmyintentiontopromise.“Foritisclearthatwhatyoudoisnotamoveinagameunlessthegameisbeingplayedandyouareoneoftheplayers...Thatinvolves...appropriateexpectationsandcalculations”inconnectionwithyourproceedings(1981a,III:17).Thatis,inordertohavetherequisiteintentionofpromisingtof,Ineednotadministeranaccountofpromisingtomyself;ratherwhatIdoandthinkhastobeinlinewithacertainpractice.(b)Toseehowmerewordscancreaterealrestrictions,weneednomorethanlookattheimpossibility,whichissuesfromone’sbumpingtof,ofavoidingthedangerofunwelcomeconsequencesunlessonef’s.(4)Thisrestriction,however,isnotyet(a)moralrequirement,or(b)anecessitytorespectit.(Hencethatpaper’sfulltitle,“OnPromising,”isfollowedby“andItsJustice,andWhetherItNeedBeRespectedinForoInterno.”)Istherea(prima-facie)needtokeeppromisesandrespectrights–someofthemofcoursecreatedbypromises–whichgoesbeyondthenecessityinternaltothelinguisticpractice?(a)Thispracticeprovidesus,interalia,withawayof“gettingoneanothertodothingswithouttheapplicationofphysicalforce,”andthis“isanecessityforhumanlife”(1981a,III:18)inAristotle’ssenseof“thatwithoutwhichgoodcannotbeorcometobe”(p.15).Henceobligation:“arestriction”on“one’spossibilityofactingwell”(p.15).(b)Thepracticalnecessityarisingfromthecommongoodisnoteoipsoonefromthepointofviewofmyowngood.Rather,Anscombeholds,“ifsomeonedoesgenuinelytakeaproofthatwithoutdoingXhecannotactwellasaproofthathemustdoX,thenthisshows...thathehasapurposethatcanbeservedonlybyactingwell,assuch”(p.19).Notethatneither(a)nor(b)typenecessitiesarecreatedbyconvention,thoughthesecondis,inasense,broughtaboutbymyintention–anoverallpurposeinlife–unless“manhasalastendwhichgovernsall”(1995:34).(5)Anscombe’schiefcontributiontopoliticaltheoryandthephilosophyoflaw,“OntheSourceoftheAuthorityoftheState”(1981a,III:130–55),ispartlybasedonheraccountofrights.Politicalgovernment,tobedistinguishedboth“fromauthorityinvoluntaryco-operativeenterprises”andfrom“controlofaplacebyagangofbandits,”mustbecharacterized“byitsauthorityinthecommandofviolence”(p.132).Sinceauthorityisarighttogiveordersandmakedecisions,wemayhopetoexplainitbydescribingalanguage-gameinwhich“ItisN’srighttof”getsitsmeaning,origi-nallyasapreludeonlyto“Sohe/shecanf,”“Soyoucan’tf,”etc.,fromapracticeofpreventinganyonebutNfromf-ing,ofreproachingthemforinterferencewithN’sf-ing,andsoon(cf.(3)above).Giventhisexplanation,N’srighttofmayyetbemerelycustomary(“conventional”),andperhapsevenaninjustice.Awayofprovingthatitisnot(ifitisnot),istoshowthatf-ingisneededfortheperformanceofataskwhichitispracticallynecessarythatNperform.Forherewehaveanon-conventional“Nmust,”anditentailstheconventional“Ncan”whichascribesarighttoN.Anexistinggov-ernmentGmightthenbeshowntohavepoliticalauthority–arighttoenforceobedi-encebythethreatanduseofviolence–byshowingthat(a)suchenforcementisneeded323\nANSELMMÜLLERforgovernment,(b)(inviewofhowmentendtotreateachother)governmentisatasknecessaryforhumangood,and(c)itisGthatcustomaryrightorsomepracticalneces-sityrequiretogovern.Challengestocontemporarymoralphilosophy“Ingeneral,myinterestinmoralphilosophyhasbeenmoreinparticularmoralquestionsthaninwhatisnowcalled‘meta-ethics’”(1981a,III:viii).Someofthesequestionsrelatetotopicsinsocialethics:parentalauthority(pp.43–8,135);state,law,andpunishment(pp.51–60;123–55);war(pp.51–81).Othersconcerncontra-ception(pp.82–96),murder(pp.51–61),andtopicsinmedicalethics.SomeofAnscombe’sthemesare“topic-neutral.”Theyincludetheonesdiscussedhereonpp.320–4,theproblemof“AuthorityinMorals”(1981a,III:43–50),absoluteprohibi-tions,andtheanti-consequentialistprinciplethatyouarenotresponsibleforforesee-ablebutunintendedconsequencesofyouractionsinthewayyouareforchosenmeansandends(seepp.54–5,58–60,and78–9).MorewidelynoticedthanhertreatmentofparticularmoralquestionshasbeenAnscombe’s1958article“ModernMoralPhilosophy,”aspiriteddefenseofthreetheses:(1)Weshouldstopdoingmoralphilosophy“untilwehaveanadequatephilosophyofpsychology”(1981a,III:26).Foranadequateaccountofactingwellmustbebasedonaphilosophicalunderstandingofhumannatureandsuchconceptsasaction,pleasure,needandwant,intention,motive,andvirtue.(2)“Themoralsenseof‘ought’”isanillusionduetoreminiscencesofa“lawconception”ofethicswhichhaslongsincebeengivenup(pp.26,29–33).IfyoudonotbelieveinGodasalaw-giver(asStoics,Jews,andChristiansdo),whatremainsofanobligationtoactwellistheword,spokenwithaspecialemphasisandfeeling(plusvainattemptstogroundthe“morallaw”inindividualautonomyorsocialcontract).Hence“‘morallywrong’bothgoesbeyondthemerefactualdescription‘unjust’andseemstohavenodiscerniblecontentexceptacertaincompellingforce,whichIshouldcallpurelypsychological”(p.41).Withouttheassumptionofdivinelegislation,thereisindeedno“ought”from“is”thatisnotofthekinddiscussedintheprevioussection:necessitybyconventionandbythepracticalrequirementsofcommonorindividualhumangood,or“flourishing”(pp.38–42).(3)Thedifferencesbetweenthewell-knownEnglishmoralphilosopherssinceSidgwickare“oflittleimportance,”comparedwiththeircommon“consequentialism,”thatis,theirrejectionoftheHebrew–Christianconvictionthat“certainthings[are]forbiddenwhat-everconsequencesthreaten,suchas:choosingtokilltheinnocentforanypurpose,howevergood;vicariouspunishment;treachery,”etc.(pp.26,34–6).Anscombe’schallengestomoralphilosophyhavebeentakenupoverthelastdecadesbysomewhoconceiveofethicsonbroadlyAristotelianlines.Inotherquarters,however,hersubstantialandcriticalcontributionstothisastootherareasofphiloso-phyhavenotyetreceivedtheattentiontheydeserve.Notes1Cf.themoreelaborateargumentin“SoftDeterminism”(1981a,II:163–72).Cf.also“ChisholmonAction,”GrazerPhilosophischeStudien7/8(1979),205–13;and“The324\nG.E.M.ANSCOMBECausationofAction,”inKnowledgeandtheMind,ed.C.GinetandS.Shoemaker(NewYorkandOxford:OxfordUniversityPress,1983),pp.174–90.HereAnscombeexaminestherela-tionsbetweenaphysiologicalinvestigationintothecausesofhumanactions,anaccountofagencyintermsofintentions,andhistoricalexplanations.Shearguesthatthesecondandthirdareinsomesensesupervenientonly,ifdeterminismistrue.2Becauseofthecommonstructureofthetwocontexts,Anscombeherekeepstothisspelling,ratherthan“intensional,”remindingusoftheetymologicalbackground:“intenderearcumin”=“toshootat.”Inthephilosophyoflogic,thetopicofintensionalityistypicallytreatedindiscussionsofidentity,ormodalorbeliefcontexts,withinwhichanintensionalobjectinAnscombe’ssensecorrespondstothesenseof,e.g.,aname.BibliographyWorksbyAnscombe1959:AnIntroductiontoWittgenstein’sTractatus,London:Hutchinson.(Reprinted1971.)1961(withGeach,P.T.):ThreePhilosophers,Oxford:BlackwellPublishers.1963:Intention,Oxford:BlackwellPublishers.(Firstpublished1957.)1979:“ProlegomenontoaPursuitoftheDefinitionofMurder:TheIllegalandtheUnlawful,”DialecticsandHumanism6,pp.73–7.1981a:CollectedPhilosophicalPapers,vol.I,FromParmenidestoWittgenstein;vol.II,MetaphysicsandthePhilosophyofMind;vol.III,Ethics,ReligionandPolitics,Minneapolis:UniversityofMinnesotaPress.1981b:“ATheoryofLanguage,”inPerspectivesonthePhilosophyofWittgenstein,ed.I.Block,Oxford:BlackwellPublishers.1982a:“Medalist’sAddress:Action,Intentionand‘DoubleEffect’,”ProceedingsoftheCatholicPhilosophicalAssociation56,pp.12–25.1982b:“OnPrivateOstensiveDefinition,”inLanguageandOntology:Proceedingsofthe6thInternationalWittgensteinSymposium,ed.W.Leinfellner,Vienna:Hölder.1983:“SinsofOmission:TheNon-treatmentofControlsinClinicalTrials,”ProceedingsoftheAristotelianSociety,suppl.vol.57,pp.223–7.1987:Absicht(Germanedn.ofIntention),ed.J.M.ConnollyandT.Keutner,FreiburgandMunich:Alber.(ContainsalistofworksonAnscombe.)1989:“VonWrightonPracticalInference,”inThePhilosophyofGeorgHenrikvonWright,ed.P.A.Schilpp,LaSalle,IL:OpenCourt.1995:“PracticalInference,”inVirtuesandReasons:PhilippaFootandMoralTheory,ed.R.Hursthouse,G.Lawrence,andW.Quinn,Oxford:ClarendonPress.WorksbyotherauthorsDiamond,C.andTeichman,J.(eds.)(1979)IntentionandIntentionality.EssaysinHonourofG.E.M.Anscombe,Brighton:Harvester.(Containsa“Bibliographicnote”onAnscombeandalistofworksonher.)Gormally,L.(ed.)(1994)MoralTruthandMoralTradition:EssaysinHonourofPeterGeachandElizabethAnscombe,Blackrock:FourCourts.(ContainsalistofworksonAnscombe.)Thompson,M.(1998)“Anscombe,G.E.M.,”inRoutledgeEncyclopediaofPhilosophy,vol.I,ed.E.Craig,London:Routledge,pp.280–3.325\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200125R.M.Hare(1919–)WALTERSINNOTT-ARMSTRONGRichardMervynHarehaswrittenonawidevarietyoftopics,fromPlatotothephilosophyoflanguage,religion,andeducation,aswellasonappliedethics,butheisbestknownforhisgeneralmoraltheory.Hare’sviewsonethicshavedevelopedsincehisgroundbreakingbook,TheLanguageofMorals(1952),butthemainthrustofhispositionhasremainedfairlyconstant.DefinitionofmoraljudgmentsHaredefinestheclassofmoraljudgmentstoincludeanyjudgmentthatisprescriptive,universalizable,andoverriding(1981:53–7).Prescriptivitydistinguishesmoraljudg-mentsfromjudgmentsofnaturalscienceandhistory.Universalizabilityseparatesmoraljudgmentsfromparticularcommands,suchasbyarmysergeants,aswellasfromlegaljudgments(1963:36).Overridingnessdividesmoraljudgmentsfromaestheticvaluejudgments(1963:139).Hare’sdefinitionofmoraljudgmentsisformalinthatitdoesnotrequireanypar-ticularcontent.Itispossible,onHare’sdefinition,tomakeamoraljudgmentthatoneshouldneversteponcracksinthesidewalk,evenifsuchstepsharmnobody,breaknopromiseorlaw,andsoon.Somecriticsfindthisimplicationunpalatable,butHarerespondsthathisdefinitionstillcapturesonepossibleandusefulspecificationoftheterm“moral.”ManypeopleseektoformulateandjustifyasystemofjudgmentsthataremoralinHare’ssense.PrescriptivismHare’smoraltheory,then,startswithprescriptivism,whichistheviewthatmoralandothervaluejudgmentsaretypicallyprescriptive.Tocallajudgmentprescriptiveistosaythatitisusedtoprescribeaction,thatis,toperformsomespeechactinalargegroupthatincludescommanding,advising,encouraging,discouraging,andsoon,withrespecttosomeparticularactionorkindofaction.Theparadigmformofprescriptionistheimperative,soprescriptivismclaimsthatavaluejudgmentsuchas“HondasarebetterthanToyotas”areusedtoperformaspeechactsimilarto“ChooseaHondaoveraToyota.”Despitetheanalogiesbetweenvaluejudgmentsandimperatives,Harehas326\nR.M.HAREinsistedfromthestartthat“itisnopartof[his]purposeto‘reduce’morallanguagetoimperatives”(1952:2).Prescriptivismisbestunderstoodincontrastwithitspredecessorsinmetaethics.G.E.Moorecriticizednaturalism,whichistheviewthatmoralandothervaluejudgmentsascribeordenyevaluativeproperties,suchasgoodnessorrightness,thataresupposedtobereducibleinsomewaytopropertiesthatarenaturalapparentlyinthesensethattheycanbestudiedbythemethodsofnaturalscience.Inplaceofnaturalism,Mooreproposednon-naturalism,whichistheviewthatmoralandothervaluejudgmentsareaboutawhollydifferentkindofpropertythatcanbeknownonlybyintuition.HareacceptsMoore’sargumentsagainstnaturalismandaddsnewargumentsofhisown(1952:79–93).However,HarealsorejectsMoore’snon-naturalismasunneces-sarilymysterious.Moore’smistake,accordingtoHare,wastoretainabasicassump-tionofnaturalism,namely,descriptivism,whichistheviewthatvaluejudgmentsareusedtodescribevaluesorevaluativeproperties.Harearguesthatbothnaturalismandnon-naturalismshouldberejected,becausevaluejudgmentsarenotusedtodescribeineitherway.DescriptivismwasrejectedbeforeHarebyemotivists,includingA.J.AyerandCharlesStevenson(seeAYERandSTEVENSON).Emotivistsclaimedthatvaluejudgmentsareusedtoexpressandarouseemotions.Againstthisview,Harepointsoutthatonecanmakevaluejudgmentsdispassionately,withouthavinganyfeelingoremotiontoexpress.Moreover,justasadvicecanbegivenbutnotfollowed,soavaluejudgmentcansucceedinitsspeechactevenifitdoesnotarouseanyemotionorchangeanybehaviorintheaudience.ThesepointsenableHaretodistinguishvaluejudgmentsonhisviewfrompropagandaandtosubjectthemtorationalscrutinyofakindthatisoftenassumedtobeinappropriateforemotions(1952:9–16).Hare’sview,then,isthatvaluejudgmentsaretypicallyusedforadifferentspeechact,namely,prescribing.Hedoesadmitthatsomevaluejudgmentsarenotusedtopre-scribe.WhensomeonesaysabuildingisgoodGothicrevivaljusttoindicatethatitisthekindofbuildingthatwouldbejudgedgoodbypeoplewholikeGothicrevivalbuild-ings,Harecallsthisan“invertedcommasuse”(1952:124).Valuejudgmentscanalsobeusedironicallyandmerelytopaylipservicetoconventions.Past-tensevaluejudg-mentsareexplainedinsimilarways.WhenIsaythatitwasmorallywrongforJeffersontoholdslaves,IdonotprescribetoJeffersonthathedoanything,sinceitistoolateforthat;but,ifIhadbeenabletodoso,thenIwouldhaveprescribedthathenotholdslaves.Insuchcases,valuejudgmentsarenotusedtoprescribe,butsuchusesarestillparasiticon,becauseexplainedintermsof,prescriptiveusesbyotherpeopleorinothercircumstances.Hare’sclaimthatvaluejudgmentsare“typically”prescriptiveseemstomeanthatallvaluejudgmentseitherareusedtoprescribeorcanbeexplainedintermsofotherusesthatareprescriptive(1963:22n).Suchprescriptivismprovidesanaturalexplanationofmanyfeaturesofevaluativelanguage.Onecommondictumclaimsthat“ought”implies“can.”Ifajudgmentthatanagentoughttodoanactisusedtoprescribethattheagentdotheact,andthereissomethingwrongwithprescribingthatanagentdoanactwhentheagentcannotdoit,thenthisexplainswhythereissomethingwrongwithsayingthatanagentoughttodoanactthatshecannotdo(1963:51–66).327\nWALTERSINNOTT-ARMSTRONGAmoreoftenquestionedimplicationofHare’sprescriptivismisthatonecannotfullythinkthatoneoughttodoanactandyetnotdoit,ifonecandoitandnowisthetimetodoit.Thisseemstoruleoutweaknessofwill,butHareexplainsapparentcasesofweaknessofwillascaseswhereagentsdonotreallythinkatthetimethattheyoughttodorelevantparticularacts,useparasiticsensesof“ought,”psychologicallycannotbringthemselvestodowhattheythinktheyoughttodo,andsoon(1963:67–85).Ifsuchreinterpretationscanadequatelyexplainawayallapparentcasesofweaknessofwill,thenHare’sprescriptivismishardertorefutethanhiscriticsassume.UniversalizabilityDespitetheirprescriptivity,valuejudgmentsdifferfromsingularimperativesinacrucialrespect,accordingtoHare.Whenadrillsergeantcommands,“Totheleft,march!,”heneedsnoreasonforchoosingleftoverright,andnothinggoeswrongif,whenthesamecircumstancesariseagainlater,hethencommands,“Totheright,march!”Incontrast,Harearguesthatthereisalogicalinconsistencybetweensayingthatoneagentoughttodoanactanddenyingthatanotheragentoughttodoanactwithrelevantlysimilarpropertiesinrelevantlysimilarcircumstances(1952:81).Hare’sexplanationofthislinguisticruleisthat“allvaluejudgmentsarecovertlyuniversalincharacter,whichisthesameastosaythattheyreferto,andexpressacceptanceof,astandardwhichhasapplicationtoothersimilarinstances”(1952:129),thatis,theypresupposeageneralprinciple.Thisimplicituniversalitymakesitlegitimatetoaskforareasonwhytheagentoughttodotheact,orapropertythatmakesthethinggood.Italsoenablesvaluejudgmentstobeusefulforpublicteachingofstandards(1952:134).Inthisrespect,valuejudgmentsarelikedescriptivejudgments,accordingtoHare(1963:10–14).Thisdoctrineofuniversalizabilitymightseememptywithoutlimitsonwhichprop-ertiesarerelevant.Haredoesclaimthatreferencestoindividuals,asinpropernamesorindexicals,mustbeirrelevantinorderforthepresupposedstandardorprincipletobeuniversal.However,heinsiststhatnoadditionallimitsarepartofthesharedmeaningofevaluativeterms.Inhisview,wecannotdeterminewhichpropertiesaremorallyrelevantinadvanceofdeterminingwhichmoralprinciplesaredefensible(1981:62–4).ThismakesHare’sdoctrineofuniversalizabilitymuchweakerthanthatofhispredecessorKantonmostinterpretations.Nonetheless,Hare’sthesisofuniversalizabilityiscriticizedbyparticularists.Somearguethatevenanagent’sindividualidentityorparticularspatiotemporallocationmightbemorallyrelevant;oratleastthatthisisnotexcludedbylanguagealone,sotoassumeotherwiseistoadoptasubstantiveposition.Otherparticularistsarguethatthereasonswhyoneactismorallywrongmightnotmakeanotheractmorallywrong,becausetheforceofeachmorallyrelevantfeaturevarieswiththecircumstancesinwaysthatcannotbespecifiedinanygeneralprinciples.Harewouldrespondbyarguingthatsomefeatureofthepersonortimeorcircumstancesmustbeavailabletoexplainwhyoneactiswrongwhenanotherisnot.Wemightandneednotbeableinpracticetospecifyfullytherelevantpropertiesorunderlyingprinciples,butsomespecificationmustbepossibleintheoryforourvaluejudgmentstobelogicallyconsistent,inHare’sview(1963:18–20).328\nR.M.HARERationalityThenextprongofHare’stheoryisaparticularviewofrationality.Thelinguisticthesesofprescriptivismanduniversalizabilitylaydownsomelimitsonconsistencyand,hence,rationality.Harealsoassumesthat“anyrationalthinkingabout[moralques-tions]hastobedoneinthelightoffacts”(1981:87).Attimes,Harealsoseemstoassumethatlogicalconsistencyandknowledgeoffactsisallthereistorationality.Thisaccountofrationalitymightnotseemverycontroversial,butHareaddsthatonefactthatneedstobeconsideredformoraljudgmentstoberationalis“whatitislike”forpeoplewhoareaffectedbyouractions(1981:92).Inparticular,iftheysuffer,itisnotenoughformetoknowthattheysuffer.Ineedtoknowwhattheirsufferingisliketothem.Thiskindofknowledge,Hareargues,requiresmetohaveacertainmotivationorpreference.Thefollowingtwoclaimsaredistinctforanysituation:1InowpreferwithstrengthSthatifIwereinthatsituationxshouldhappenratherthannot.2IfIwereinthatsituation,IwouldpreferwithstrengthSthatxshouldhappenratherthannot.(1981:95)(1)isaboutmycurrentpreferencesregardingacounterfactualsituation,whereas(2)isaboutacounterfactualsituationinwhichIwouldhavepreferencesthatIdonotcur-rentlyhave.AlthoughHareadmitsthattheseclaimsaredistinct,hearguesthat“Icannotknowthat(2),andwhatthatwouldbelike,without(1)beingtrue,and...thisisaconceptualtruth,inthesenseof‘know’thatmoralthinkingdemands”(1981:96).Thistransferprinciple,asIwillcallit,impliesthatonemusthavecertainpreferencesinordertoberational.ThemasterargumentBycombiningprescriptivism,universalizability,andrationality,Hareclaimstobeabletoderiveakindofutilitarianism(1981:109–11).Theargumentstartswithanexample,whosefactsIhavemodifiedsomewhattoclarifythelogicoftheargument:1BcannotparkB’scarinthisparkingplaceunlessImovemybicycle.2BhasastrongpreferencetoparkB’scarinthisparkingplace.3Ihaveonlyaweakerpreferencenottomovemybicycle(ortoleaveitthere).UtilitarianismsuggeststhatImorallyoughttomovemybike,soHare’sgoalistoderiveanabsurdityfromthesuppositionthatanoppositebeliefisrational:4IrationallybelievethatImorallyoughttoleavemybicyclethere.ThissuppositiononHare’sviewofrationalityimplies5Iknowalloftherelevantfactsaboutleavingmybicyclethere,includingwhatitwouldbelikeforBifIleftmybicyclethere.Assumingthatrolesincludepreferences,sothat,whenweswitchroles,wealsoswitchpreferences,(2)and(5)imply329\nWALTERSINNOTT-ARMSTRONG6Iknowthat,ifBandIswitchedroles,IwouldstronglypreferthatBnotleavethebicyclethere.Hare’stransferprincipleplus(6)yields7Inowstronglypreferthat,ifBandIswitchedroles,Bwouldnotleavethebicyclethere.Assumingthatthisisthestrongestpreferencebetweenthesealternativesinthissituation,then,sincepreferencesareprescriptive,8Iaccepttheprescription:“ifBandIswitchedroles,letBnotleavethebicyclethere.”Buttheuniversalizabilityofmoraljudgmentsplus(4)imply9Ibelievethat,ifBandIswitchedroles,Bmorallyoughttoleavethebicyclethere.Thisplustheprescriptivityofmoraljudgmentsimplies10Iaccepttheprescription:“ifBandIswitchedroles,letBleavethebicyclethere.”Theprescriptionsin(8)and(10)areinconsistent,andinconsistencyexcludesrationality,so(4)isrefuted.Thismeansthat11IfIamrational,IcannotbelievethatImorallyoughttoleavethebicyclethere.Oncethisbeliefisruledoutasirrational,theoptionsarelimited:12IfIbelievethatImorallyeitheroughtoroughtnottoleavethebicyclethere,andifIamrational,thenImustbelievethatImorallyoughtnottoleavethebicyclethere.Assumingthatthisexampleisnotdifferentinanyrelevantrespectfromanyotherconflictinvolvingonlytwopeople,wecangeneralize:13Ineverytwo-personconflict,ifIbelievethatImorallyeitheroughtoroughtnottoanact,andifIamrational,thenImustbelievethatImorallyoughtnottodowhateverfrustratesstrongerpreferencesthanitsatisfies.Whatisbelievedintheconsequentof(13)isjustwhatpreferenceutilitarianismsaysaboutsuchtwo-personconflicts.Harenextextendshisargumenttochoicesthataffectanynumberofpeople.Then,tomakearationalchoice,anagentneedstoknowwhatitislikeforeachpersonwhoisaffected.Thatknowledge,accordingtoHare’stransferprinciple,requirestheagenttohavecurrentpreferencesfororagainsttheactbeingdoneifheweretoswitchroleswitheachofthemandtakeontheirpositiveornegativepreferences.Indeed,theagentmusthavepreferenceswiththesamestrengthsasthoseofthepeopleaffected.Thus,anagentmorallyoughttodowhathewouldprefertobedoneifhehadtooccupysuc-cessivelyeachoftherolesofeachpersonaffected.Thisconclusionisageneralversionofpreferenceutilitarianism.Hare’sargumentcouldbecriticizedfrommanyangles.First,amoralistsmightrefusetomakeanypositivemoraljudgmentsatall.Thentheydonotdeny(12)or(13),but330\nR.M.HAREtheydoblockanyderivationoftheirconsequents.Hareadmitsthatthereisnothing“logicallyinconsistentinthisposition”(1981:186),buthegives“reasonsofanon-moralsortwhyitshouldnotbechosen”(p.190).Adifferentproblemisraisedbyfanatics,whoclingtoidealsevenwhentheyknowthattheirmoraljudgmentsruncontrarytoutilitarianism.Harerespondsthatthefanatic’smoraljudgmentsareineffectanewmoralistickindofpreference,sofanaticsfaceadilemma(1981:181).Ononehornofthisdilemma,thefanaticclaimsthathismoralisticpreferencesarestrongenoughtooutweighallofthepreferencesofevery-oneelseaffectedbytheaction.Thisclaimwouldbeimplausible,and,evenifitweretrue,apreferenceutilitariancouldjustgrantthatthismoralisticpreferenceshouldbefulfilled,sinceitisstrongerthantherestputtogether.Thus,ifthefanatic’spositionistoremaincontrarytoutilitarianism,thefanaticmustgrabanotherhornofHare’sdilemma.Theonlyalternativeforthefanaticistoadmitthathismoralisticpreferenceisnotstrongenoughtooutweighallofthepreferencesofeveryoneelseaffectedbytheaction.Butthen,ifthefanaticisrational,hemustknowwhatitislikeforthosepeople’spreferencestobefrustrated.Bythetransferprinciple,hemustthencometohavecorrespondingpreferenceswiththesamestrengths.Thesepreferenceswillthenmakehimabandonhisfanaticism.Whathecannotdoisremainafanaticandalsoremainrational,accordingtoHare.Thetransferprincipleitselffacesproblems,however.Considersomeonewhoistryingtokillmesothattheycantakemyplaceonthebasketballteam.ShouldIstopthem?Formyanswertoberational,Ineedtoknowthat,ifweswitchedplaces(soIwantedtokilltheminordertogetontheteam),thenIwouldstronglypreferthattheynotstopmefromkillingthem.Accordingtothetransferprinciple,IcannotknowwhatthatwouldbelikeunlessInowstronglypreferthatifItriedtokillthemtogetontheteam,thentheywouldnotstopme.However,thiscurrentpreferencedoesnotseemneces-saryformetoknowallthatisgoingon.Icanknowwhytheywanttokillme,andwhatitislikeforthem,evenifmycurrentmoralbeliefskeepmefromhavinganycurrentpreferencethattheynotstopmeifItrytokillthem.Moregenerally,preferencesthatweseeasimmoralseemtobeonekindofpreferencethatrationalitydoesnotrequireustotakeoverfromothersinordertoknowwhatitislikeforthemtohavethatpreferencefrustrated.Thesameproblemariseswhenthetransferprincipleisusedagainstnon-utilitarianmoraltheories.Supposearetributivistbelievesthatthegovernmentmorallyoughttoexecuteaconvictedmurderer,evenifthisexecutionwouldnotmaximizepreferencesatisfaction.Thispositioniscontrarytoutilitarianism,soHarewouldlabelthisret-ributivistafanatic.Whateverheiscalled,sucharetributivistcanknowthat,ifhewereintheconvictedmurderer’ssituation,thenhewouldpreferstronglythattheexecutionnothappen.Hecanevenknowwhatitwouldbeliketobeexecuted,asmuchasanybodyelsecan.Buttheretributiviststillneednotcurrentlyhaveanypreferencethat,ifhewereinthesamesituationastheconvictedmurderer,heshouldnotbeexecuted.Inthatcounterfactualsituation,hehimselfwoulddeservetobeexecuted,accordingtohiscurrentretributivistviews;so,ifhestickswithhisviews,hewouldpreferthat,ifhehimselfhadcommittedsuchaheinousmurder,thenhehimselfshouldbepunishedlikeanyothermurderer.Suchretributivismseemsconceptuallycoherent,consistentwithbothprescriptivismanduniversalizability,andalsocommon.Thus,Hare’s331\nWALTERSINNOTT-ARMSTRONGtransferprincipleseemstofailintheverykindofcasewhereheneedsit.Ifso,hisargu-mentcannotprovehisversionofpreferenceutilitarianism.UtilitarianismEvenifHare’sargumentdoesnotprovehisversionofutilitarianism,thatconclusionstillmightbecorrectanddefensible.Hareisbasicallyanactutilitarianwithaprefer-ence-basedtheoryofvalue.Suchviewshavebeensubjectedtoagreatdealofcriticismandevenscorn,butHare’sversionhasdistinctivefeaturesthatmakeithardertorefutethanmanyothers.ThemostimportantinnovationisHare’sdistinctionbetweentwolevelsofmoralthinking.Thisdistinctionismotivatedbyadiscussionofapparentmoralconflicts.Inhismainexample(1981:27),alifelongfriendvisitsunexpectedlyfromAustraliaandaskstobeshownaroundOxfordontheverydaywhenHarepromisedtotakehischildrenforapicnicontheriver.ItseemsinitiallythatHaremorallyoughttokeephispromise,andhealsomorallyoughttoshowhisfriendaroundOxford.TheproblemisthatHarecannotdoboth,sothefactthathemorallyoughttokeephispromiseimpliesthathemorallyoughtnottoshowhisfriendaroundOxford.Sincemoraljudgmentsareprescriptiveandoverriding,so“ought”implies“can,”HaredeniesthepossibilitythathemorallybothoughtandoughtnottoshowhisfriendaroundOxford.Sohealsodeniesthatsuchmoralconflictsareeverpossibleintheend.Nonetheless,Harewantstoexplainwhymoralconflictsseemtobenotonlypossi-blebutcommon.Toexplainthisappearance,Haredistinguishestwousesof“ought”thatoccurattwolevelsofmoralthinking.Ontheintuitivelevel,Hare’sthinkingthathemorallyoughttoshowhisfriendaroundOxfordis“inseparablefrom”hisfeelingcompunctionbeforeandguiltafterhisfailuretoshowhisfriendaroundOxford(1981:30–1).Sinceagoodpersonwouldalsofeelcompunctionbeforeandguiltafterfailingtokeephispromisetohischildren,Hareadmitsthatconflictingmoralbeliefsattheintuitivelevelcanbothbejustifiedbytheirpracticalusefulness.Intuitivemoralbeliefsfacilitatemoralteachingandreduceerrorsinmoraljudgmentwheninformationisscarceortimeisshort.Thatexplainswhywedoandshouldsometimesthinkattheintuitivelevel,wheremoralconflictsseemtooccur.Butthisintuitivelevelofmoralthinkingisnotenoughbyitself.Weneedtodeter-minewhichintuitivemoralprinciplesshouldbetaughtandusedinmakingeverydaydecisions.Wealsoneedsomewaytodeterminethemorallyrightcourseofactionwhenintuitivemoraljudgmentsconflict.Bothoftheseproblems,accordingtoHare,canbesolvedonlyatadifferentlevelofmoralthinking,whichhecallsthecriticallevel.Atthatcriticallevel,tosaythatHaremorallyoughttoshowhisfriendaroundOxfordistomakeaprescriptiveandoverridingjudgment,sosuchcriticalmoraljudgmentscannotconflict.Moreover,criticalmoralthinkingmustconformtopreferenceutilitari-anism,accordingtoHare’smasterargument.Sinceincompatiblealternativescannoteachfrustratestrongerpreferencesthanitsatisfies,Hare’sutilitarianismimpliesthatanagentnevermorallyoughttoadopteachoftwoconflictingoptions.Suchcriticalthinkinginconformitywithutilitarianismiswhatultimatelydetermineswhichactismorallyright,accordingtoHare.ThatiswhyHarecanremainautilitarianandyetstillexplainwhymoralconflictsdoandshouldseemtooccur.332\nR.M.HAREThisdichotomybetweenlevelsofmoralthinkingalsoenablesHaretorespondtostandardcounterexamplesthataresupposedtorefuteutilitarianismbyshowingthatitconflictswithcommonmoralintuition:“Perhapsthe[proverbial]sheriffshouldhangtheinnocentmaninordertopreventtheriotinwhichtherewillbemanydeaths,ifheknowsthattheman’sinnocencewillneverbediscoveredandthatthebadindirecteffectswillnotoutweighthegooddirecteffects;butinpracticehewillneverknowthis”(1981:164).Hare’spointisthat,inordertocreateanytroubleforutilitarianism,thedetailsofsuchexamplesmustbesounrealisticthatwehavelittleornoreasontotrustourmoralintuitionsaboutsuchcircumstances.Ourmoralintuitionsarejustifiedbecausewewillactbetterinmostsituationsifweinculcatesuchmoralintuitionsdeeplyintoourcharacters,butthesamemoralintuitionsceasetobereliableguidesinunrealisticcircumstancesforwhichtheywereneverintended.Consequently,utilitarianismcannotberefutedbyanyappealtomoralintuitioninthisoranyotherunrealisticexample,accordingtoHare(1981:134).Thisresponsewillnotconvinceeveryopponent,andmanyotherobjectionscouldberaisedtoHare’sutilitarianismandtootheraspectsofhismoraltheory.Nonetheless,therangeofHare’sviews,hisattentiontodetail,theclarityofhiswriting,andhisabilitytobringabstractmoraltheorytobearonconcreteissuesofgreatimportanceprovideamodelforallmoralphilosopherstofollow.BibliographyWorksbyHare1952:TheLanguageofMorals,Oxford:ClarendonPress.1963:FreedomandReason,NewYork:OxfordUniversityPress.1971a:EssaysonPhilosophicalMethod,London:Macmillan.1971b:PracticalInferences,London:Macmillan.1972a:ApplicationsofMoralPhilosophy,London:Macmillan.1972b:EssaysontheMoralConcepts,London:Macmillan.1981:MoralThinking:ItsLevels,Method,andPoint,NewYorkandOxford:OxfordUniversityPress.1982:Plato,NewYorkandOxford:OxfordUniversityPress.1989a:EssaysinEthicalTheory,NewYorkandOxford:OxfordUniversityPress.1989b:EssaysonPoliticalMorality,NewYorkandOxford:OxfordUniversityPress.1992:EssaysonReligionandEducation,NewYorkandOxford:OxfordUniversityPress.1993:EssaysonBioethics,NewYorkandOxford:OxfordUniversityPress.1998:SortingOutEthics,NewYorkandOxford:OxfordUniversityPress.1999:ObjectivePrescriptionsandOtherEssays,NewYorkandOxford:OxfordUniversityPress.WorkbyotherauthorsSeanor,D.andFotion,N.(eds.)(1990)HareandCritics:EssaysonMoralThinking,Oxford:ClarendonPress.333\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200126P.F.Strawson(1919–)P.F.SNOWDONLifePeterFrederickStrawsonwaseducatedatSt.John’sCollege,Oxford,wherehereadPhilosophy,Politics,andEconomics,graduatingin1940.HethenservedforsixyearsintheBritishArmy,becomingacaptain.AfterashortperiodasalecturerattheUniversityofNorthWales,Bangor,hereturnedtoOxford,becomingaFellowatUniversityCollegein1948.In1968StrawsonwasappointedGilbertRyle’ssuccessorasWaynfleteProfessorofMetaphysicalPhilosophy.HewasmadeaFellowoftheBritishAcademyin1960,wasknightedin1977,andretiredin1987,thoughsincethenhehascontinuedhisphilosophicalactivities.Strawson’smajorpublicationsincludeIndividuals(1959),anexplorationinwhathecalled“descriptivemetaphysics,”TheBoundsofSense(1966),aconstructiveandcriticalstudyofKant,ScepticismandNaturalism(1985),astudyofbothgeneralskepticismandsomemorespecificvariants,andLogico-LinguisticPapers(1971),acollectionofhishighlyinfluen-tialpapersaboutlanguage,including“OnReferring,”thearticlethatfirstmadeStrawsonfamous.ThemesStrawsonestablishedhispre-eminenceinpostwarOxfordphilosophybytheextraordi-naryrangeanddepthofhiswork.Hehaswrittenaboutthephilosophyoflanguage,oflogic,metaphysics,epistemology,thehistoryofphilosophy,butalsoaboutthenatureofphilosophyitself.Andwithineachofthesebroadareashehasinvestigatedmanytopics.Thus,withinthephilosophyoflanguagehehaswrittenaboutreference,meaning,truth,thesubject/predicatedistinction,speechacts,themeaningofconnec-tives,andthenatureofgrammar.OneaspectofthedepthofStrawson’sworkhasbeenhisattempttoestablishexplanatorylinksbetweenthedifferentbranchesofphilosophy.Forexample,heilluminatinglylinksthemetaphysicaldistinctionbetweenparticularanduniversaltothelogicalsubject/predicatedistinction.Thespecialqualityofhisworkresidesinhisabilitytodeveloporiginalideasacrosssuchawiderange,thecareandingenuitywithwhichhedevelopstheseideas,andapersistenttendencytopursueissuestoadeeplevel.Strawsonalsowritesinastylish,distinctive,anduntechni-334\nP.F.STRAWSONcalmanner,conferringonhisworksaneleganceasliteratureunusualforrecentphilosophy.Strawson’sviewshavedevelopedandbeenmodified,andhehasneithertendedtorepeathimselfnorengagedoverlymuchwiththeextensivecriticaldiscussionsofhiswork.Inconsequence,andincontrastwithsomeotherleadingphilosophers,thereisnocoresetofrepeatedlydefendeddoctrineswhichmightbecalledStrawson’sphi-losophy.Thereare,however,certainabidingandrecurringthemesinhiswritingwhichdeservetobeformulated.Strawson’spictureofhumanthought,whichhaslinks,indifferentways,tothoseofHumeandKant,isthat,despiteitsimpressivedevelopmentovertime,with,forexample,theemergenceofscience,andtheimprovementofitsunderstandinginalldomains,thereisanabiding,fundamental,unrevisableframework.AshesaysintheIntroductiontoIndividuals,thereisamassivecentralcoreofhumanthinkingwhichhasnohistory–ornonerecordedinhistoriesofthought;therearecategoriesandconceptswhich,intheirmostfundamen-talcharacter,changenotatall.Obviouslythesearenotthespecialitiesofthemostrefinedthinking.Theyarethecommonplacesoftheleastrefinedthinking;andareyettheindis-pensablecoreoftheconceptualequipmentofthemostsophisticatedhumanbeings.(1959:10)Chartingthesecentralconceptsisonetaskforphilosophy,thetaskwhichStrawsoncallsdescriptivemetaphysics.Thistaskistheoreticalandconstructive,andrepresentsavisionofphilosophywhichcontrastssignificantlywithwhatwasamorenegative,critical,andpiecemealapproachassociatedwithAustin,anapproachwhichdominatedpartoftheOxfordlandscapeintheearlystageofStrawson’spresencethere.Strawson’sdevelopmentofhisownprogramsignificantlychangedthatlandscape.AccordingtoStrawson,onepartofthisframeworkisthephysicalworldofpercep-tiblebodies,constitutinganabidingframeworkinspaceandtime,withtheirmanifest,andnotsomanifest,causalproperties;anotherpartispersons,entitieswithbothbodilyattributes,suchasweightandheight,andalsopsychologicalattributes,suchasconsciousness,perception,thought,andaction,and,ofcourse,anunderstandingoftheveryconceptsbeingdescribed.Personsunderstandanduselanguagewhichisshapedtoexpressthebasicconcepts.Ouremploymentofthesecategoriescannotbeeliminatedinfavorofthoseofscience,sincethecategoriesofscienceitselfareacces-sibleonlyviathebasicframework.Further,thesecategoriesdonotearntheirrighttoemploymentbybeingjustifiedinthelightofargumentsbasedonexperienceastradi-tionallyconceivedbyempiricists,forthereisnosuchneutralexperiencedescribableinmorebasicterms.Therejectionoftheempiricistconceptionofexperience,advancedinarelativelysophisticatedformbyA.J.Ayer,representsonethemesharedbybothAustinandStrawson(seeAUSTIN;cf.AYER).Thesecategoriesdonotearntheirrighttoemployment,either,bybeingshowntobereducibletomorebasiccategories,fornosuchreductionispossibleorneeded.Asanontologist,therefore,wemightcallStrawson’sattitude“relaxedrealism.”Heendorsestherealityofsuchentitiesandtheirproperties–hence,therealism–withoutsupposingthatthisrequiressomestronguni-ficationbetweenthedifferentlevelsofthingorproperty–hencetherelaxedness.335\nP.F.SNOWDONFurther,althoughthegroundspresentedforsayingthishavechanged,Strawsonregardsskepticalcriticismsoftheframeworkasessentiallybasedonmisunderstand-ingsofoneformoranother,andStrawson’smain(althoughcertainlynotsole)episte-mologicalinteresthasbeentodisplaytheerrorsofskepticism.Inhisearlydiscussionofinduction,andalsoinIndividuals,heappearstosuggestthattheskepticalthoughtthatincircumstanceC(thebestcircumstancesthatcanobtain)wedonotknowthatP(ordonotreasonablybelievethatP)canbeshowntobeinconsistentwiththemeaningofP.InTheBoundsofSenseStrawsonsympatheticallyexploresastrategyoftranscendentalarguments,accordingtowhichthereisaninconsistencyintheskeptic’sattitude,inthattheconceptstheskepticispreparedtoapplypresupposetheapplica-tionoftheconceptsaboutwhichheisskeptical.MorerecentlyStrawsonhasattemptedtodiscreditskepticismonthebasisofitsinabilitytogenuinelypersuadeanyone.Aswellassuchextremetraditionalskepticism,Strawsonisalsoopposedtothemorelimitedskepticismofsomecurrentphilosophers,suchasQuine,whorejectpartsofourcon-ceptualscheme,inhiscasethosetodowithmeaningorpsychologicalstates.Strawsonarguesthattherearenolegitimatemetaphysicalrequirementswhichsuchnotionsfailtosatisfy,andtheyare,moreover,indispensabletoourthoughtandtotheveryinquirieswhicharesupposedtosupersedethem.Thereis,inStrawson’sapproach,aconceptualconservatismsimilartoWittgenstein’s,butStrawsonrevealsnosympathywithWittgenstein’soppositiontotheoreticalphilosophy.Onefundamentaltaskofphilosophyistodescribethisframework,todisplaycon-nectionsbetweenthebasiccategories,(saybetweenthecategoriesofperceptionandcausation),todisarmphilosophicalskepticismsorreductions,andtodescribe,inareal-isticway,thelanguagewehaveforexpressingtheseconcepts,withoutrestrictingthecategoriesweemployinthedescriptionofourlanguagetothosefavoredbyformallogi-cians,whosepurposesareratherdifferent.DefinitedescriptionsandreferenceStrawson’sfirstmajorpublicationwas“OnReferring,”which,in1950,establishedforhimmoreorlessimmediately,aninternationalreputation,andalsoinitiatedastillcontinuingdebateaboutthenatureofreference.Inthatarticle,Strawson’sprincipalaimwastocriticizeandreplaceRussell’sfamoustheoryofdefinitedescriptions.Noun-phrasesbeginningwiththedefinitearticle,forexample,“ThePrimeMinisterofEngland,”“Themanoverthere,”“Thecleverestmanintheworld,”arecalleddefinitedescriptions.Russell’stheoryproposesthatasentenceoftheform“TheFisG”(callthissentenceS),isequivalentto,oristobeanalyzedas,“ThereisoneandonlyoneFanditisG”(callthissentenceSR).Accordingtothisanalysistheoccurrenceofadefinitedescriptionsignalstheassertionofanexistentialclaim,“thereisanF,”andaunique-nessclaim,“andonlyoneF.”TheanalysisalsoimpliesthattheutteranceofsentenceSisfalseifthereisnosuchthingas“theF,”sinceSisanalyzedassayingthatthereisanF(seeRUSSELL).Strawsonmakesthreemaincriticisms.(1)Hefirstarguesthatthetheoryisunsup-ported.AccordingtoStrawson,RussellregardedSasequivalenttoSRbecausehethoughtthatsentencesoftheSformaremeaningfulevenifthereisnoF,whichtheycouldnotbeiftheyareofsubject-predicateform.Strawsoncommentsthatitisneces-336\nP.F.STRAWSONsaryinthinkingaboutlanguagetodistinguishbetweenroughlyasentenceandtheuseofasentence.Sentenceshavemeaning,butthatdoesnotrequirethateachuseofameaningfulsentenceexpressesatrue-or-falseassertion.Strawson’sdistinctionisimportantandhashadaprominentplaceinrecenttheoriesofindexicalsanddemon-stratives.However,theclaimthatRussell’soverlookingofthedistinctionis“thesourceofRussell’smistake”hasnotbeenwidelyaccepted.ThereasonisthatRussell’sownargumentsandintuitionsprimarilyrelatetowhatStrawsoncalls“uses.”ItisthecontentconveyedbyusestowhichRussellisattending.(2)StrawsonfurtherarguedthatSwouldnotberegardedasfalseincaseswherethereisnoF,ratherthequestionofitstruthorfalsitydoesnotarise.Thisbecameknownasthe“truth-valuegap”thesis.(3)Strawson’smainargument,however,isthatitisobviousthataspeakerwhouses“theF”issimplynotsayingthatthereexistssuchathingastheF;ratherthespeakerimpliesthatthereisanFbyemploying“theF,”inspeakingtotheaudience,torefertotheobject.Inthisrespecthecomparedtheuseofdefinitedescriptionstothatofdemon-stratives(suchas“that”and“this”).Arguments(2)and(3)werebothcontestedanddebated.Theexistenceoftruth-valuegapswasdenied,anditwasalsodeniedthatwecanjusttellwhatwearesayinginsuchcases.Further,evidencewasproducedthatsup-portedtheconclusionthatinmanycases“the”isnotadeviceofreference,forexample,asinthesentence“Thepersonthateachmanmostadmiresishismother.”InsubsequentworkStrawsonrefinedandlimitedhisaccount.InIntroductiontoLogicalTheoryheintroducedtheterm“presupposition”fortherelationthathethoughtexistedbetweensaying“TheFisG”andtheclaim“ThereisanF.”Roughly,presuppo-sitionholdsbetweenPandQifthetruthofPrequiresthetruthofQ,butthefalsityofQdoesnotrequirethefalsityofP.IfQisfalsethequestionofP’struthdoesnotarise.Thisterminologyandtheinvestigationofsucharelation(orrelatedrelations)hasbeenmoreprominentinlinguisticsthanphilosophy.Strawsonaddedtwootherimportantideas.Firstheinvestigatedthenatureofrefer-ence(bothinIndividualsand“IdentifyingReferenceandTruthValues”inStrawson1971)andprovidedanaccountormodelofwhathecalls“identifyingreference.”Theverycommonsensicalideaisthatbothspeakerandaudiencehavetheirrespectiveanddifferentknowledgeofobjectsintheworld,andinthecentralcasethespeakerchoosesareferringexpressionthathejudgesappropriatetoenabletheaudiencetoidentifyamongsttheobjectstheyknow,theonebeingspokenof.Thespeakerinvokesorratherreliesontheaudiences’sknowledgeoftheobjectbutdoesnotneedtoinformortelltheaudienceoftheobject’sexistence.ThisisanamplificationofStrawson’scentralin-tuitionabouttheuseofthedefinitearticle.Second,Strawsonpersuasivelyseparatedtheclaimthatatleastoneroleofdefinitedescriptionsistomakeidentifyingreferencefromtheclaimthatsentencescontainingemptydefinitedescriptionsareneithertruenorfalse(thetruth-valuegapthesis).Heclaimsthatidentifyingreferencecanbechar-acterizedwithoutimplyingthatsuchgapsexist.Heproposedinsteadthattheconse-quencefortruth-valueofreference-failureispartlydeterminedbytherelationbetweenthedefinitedescriptionandthedifferenttopicsofthediscourse.Wherethedescriptionaimsatfixingthetopictheresultofreferencefailureisatruth-valuegap;wherethedescriptionfiguresinasupplementingclaimaboutanothertopic,theresultisfalse-hood.Forexample,if,inatalkontheconstitutionofEngland,IstartbytellingyouaboutthepresidentofEngland,youwoulddismissmyremarksasconfused,butif,in337\nP.F.SNOWDONdescribingthevisitorstotheTowerofLondonIlistedthepresidentofEnglandyouwoulddismissthatasfalse.Thisproposalhasnotpersuadedeveryoneandthedebatecontinues,fueledbylaterimportantcontributionsbyKeithDonnellanandKripke.OnefundamentalquestionaboutStrawson’smorerecentaccountiswhetheritaccordsanover-centralroletothenotionofidentificationinunderstandingreference.Itisamark,though,oftheimpor-tanceofStrawson’scontributionthathisoriginalarticleisstillinfluential.TruthInanearlypaper,StrawsonendorsedaRamseystyleredundancytheoryoftruth,accordingtowhichthefundamentalcharacterizationoftruthisthattosayitistruethatpissimplytosaythatp(seeRAMSEY).Tothisheaddedsomeobservationsaboutthespeechactsstandardlyperformedbyuseof“true,”stressingthesetosuchanextentthathewasinterpretedasendorsingananalysisof“true”solelyintermsofthespeechactsitisusedtoperform,aso-called“performative”theoryoftruth.Inthe1950sand1960stheideaofaperformativeanalysis,bythenrenouncedbyStrawson,wasignoredandreplacedbyadebate,involvingStrawson,Austin,G.J.Warnock,andothers,intotherespectivemeritsoftheredundancytheorycomparedtoaversionofthecor-respondencetheorysuggestedbyAustin,andrefinedbyWarnock.Strawson’sapproachwastopropoundanddefendagainstcriticismtheredundancytheory,andtocriticizethecorrespondencetheory.ThiswasthemaindebateabouttruthuntilDummett,withhisanti-realistapproach,andDavidson,moveditinnewdirections(seeDUMMETT).Austinofferedthefollowinganalysisoftruth:astatementissaidtobetruewhenthehistoricstateofaffairstowhichitiscorrelatedbythedemonstrativeconventions...isofatypewithwhichthesentenceusedinmakingitiscorrelatedbythedescrip-tiveconventions.Hisaimwastoanalyzetruthasacorrespondencerelationbetweenstatementandworldwithoutexplainingthecorrespondencerelation,ascorrespon-dencetheoristshaveoftendone,intermsofastructuralisomorphismbetweenworldandrepresentation.Austin’sreferencetoconventionsismeanttoavoidsuchanotion.Strawson’scriticalresponsetoAustinisveryrich,butwecandistinguishthreemainlinesofcriticisms.(1)Manyattributionsoftruthcannotberegardedassayingany-thingaboutactualstatements,asAustin’saccountseemstoimplytheyhavetobe.Someonemight,forexample,beginatalkbysaying,“Althoughitistruethatp,q,”withoutitbeingnecessarythatthereisastatementthatpbysomeoneelsetobetalkedabout.(2)Strawsontreatstalkofstatesofaffairsasequivalenttotalkoffacts,forwhichheproposes,ineffect,itsownredundancytheory.Talkoffactscannotfigureinaseriousanalysisoftruth,sincetosaythatitisafactthatpisequivalenttosayingthatitistruethatp,bothsimplysayingp.AsStrawsonputsit;“Thereisnonuance,exceptofstyle,between‘That’strue’and‘That’safact’”(1971:196).(3)Strawson’smajorcriticismisthat“althoughweusetheword‘true’whenthesemanticconditionsdescribedbyAustinarefulfilled,”theword“true”patentlydoesnotstatethatthoseconditionsarefulfilled.Inusingtheword“true,”itis,accordingtoStrawson,obviousthatnothingisbeingsaidabouttheconventionsoflanguage.“Itistruethatp”isnomoreaboutlanguagethanis“p.”Thisargument,theinitialconcessioninwhichseemsnottobeentirelyconsistentwithobjection(2),resemblesStrawson’smainargumentagainst338\nP.F.STRAWSONRussell’stheoryofdefinitedescriptions.Inbothcases,Strawsonisrelyingonhissenseofwhatisbeingsaidorspokenaboutinparticularpartsofnaturallanguage.Strawson’scriticismsweregenerallytakenaspersuasive,butthedebatecontinuedinatleasttwoveryinterestingdirections.OnearoseoutofanobservationbyWarnockthat,evenifStrawson’scriticism(1)hadrevealedproblemsforAustin’saccount,itremainsplausibletoclaimthatinascribingtruthveryoftensomethingissaidaboutastatement.Indeed,Strawsonhimselfhadendorsedthecommonclaimthatstatementsratherthansentencesare,asitissaid,“thebearersoftruth.”Whatisit,then,thattheybear?Agreeingtothisis,though,difficultfortheredundancytheorysinceitdoesnottreat“...istrue”asexpressingapropertyofanything.Strawsonclarifiedthisissueandingeniouslyshowedthateveninaredundancytheoryanalysisitispossibletoincludereferencetostatements.Thus,“S’sclaimthatpistrue”canbetreatedas“AsSclaimed,p.”ItisdifficultnottofeelthatinthisdebatetherealproblemwhichWarnockwasgesturingatwaslost;theintuition(whetherrightorwrong)isnotsimplythatastatementisreferredto,butthatsomethingisascribedtoit,thatitisthebearerofsomething.Second,StrawsonrevisedhisearlierviewthatAustin’saccountofthetwotypesofconventionsisatleastanaccuratespecificationofwhenweuse“true.”Strawson,ineffect,arguesthatoncethereferentialconventionsattachingtocertainwordsinasen-tenceandthedescriptiveconventionsattachingtoothersareworkedthroughtodeter-minewhatissaid,thereisnodiscerningseparabledemonstrativeconventionattachingtothesentenceasawholetocontrastwiththedescriptiveconventionsalsogoverningthesentenceasawhole.ItisameasureofStrawson’ssuccessasacriticthatAustin’sversionofthecorrespondencetheorylackscurrentsupporters.Moreover,inthecourseofhisarticlesStrawsoncontributedmuchtotheamplificationofaredundancyview.(Searle’scontributiontoHahn1998,plusStrawson’sreply,illuminatethedebate.)LogicaltheorySomeaccountmustbegivenofStrawson’sfirstbook,IntroductiontoLogicalTheory(1952),butofallofhisbooksitistheonethathasdatedmostandsoIshallbebrief.Thebookhasthreemainaims:first,tobeanintroductorydescriptionofformallogic;secondtoprovideaphilosophicallyadequateanalysisoftheconceptscentraltothink-ingaboutlogic,inparticulartheconceptofentailment;and,third,todeterminehowfarthedevicesofartificialformallogicprovideanaccurateaccountofthesignificanceoftheexpressionsinnaturallanguage.ThislasttaskissimplyageneralizedformofwhatisatstakeinStrawson’sresponsetoRussell.Thefirsttaskiselegantlydoneinmanyrespects,forexampleinhisdiscussion,inchapter2,oflogicalform,butthereisnoseriousattentiontotheroleinformallogicofproofsystems,norisarigoroussemanticsdeveloped;andbecauseofthisthefun-damentalcontrastbetweensyntacticandsemanticnotionsisnotexplained,norarethenotionsofconsistencyandcompleteness.Strawson(inchapter1,part3)analyzesthepropositionthatAentailsBassayingthattheproposition(AandnotB)isself-contradictory,andaddsthatthedefectofself-contradictionisthatonedoesnotsayanythingbyutteringacontradiction.However,noclearaccountisgivenofself-con-tradiction,norisanyproperdefencegivenoftheclaimthatself-contradictorysentences339\nP.F.SNOWDONsaynothing(ratherthanthattheysaysomethingcontradictory).Itisinrelationtothethirdtaskthatthebookisstillrelevant.Inchapter3Strawsonundertakesacarefulcomparisonbetweenthesignificanceoftheformallogicalconstantsandtheirnaturallanguageanalogues.Hearguesthattherearesignificantdifferencesineachcase.Forexample,heclaimsthat“&”ispurelyconjunctive,whereas“and”cansometimesconveyinformationabouttemporalorder.Healsoarguesthat“PÆQ”istrueifPisfalse,butthat“IfPthenQ”isnotautomaticallytrueinthosecircumstances.Strawson’sargumentsareingeniousbuttheystimulatedGricetodevisehisowntheoryabouthowtodistinguishwhatisliterallymeantandwhatisotherwiseconveyedorimplied.InthelightofGrice’stheorysomeofStrawson’spointslookdisputable.However,StrawsonandothershavethemselvesdisputedelementsinGrice’stheory,andthedebate,inpar-ticularaboutconditionals,remainsopen.ThegeneralsloganthatStrawsonendorsesisthat“ordinaryexpressionshavenoexactandsystematiclogic”(1952:57).Strawsonmeansbythisthatitisnotpossibletogivetonaturallanguageexpressionsanabstractmeaningassignmentwhichexhaustswhattheycountasconveyingacrossallcontexts.Strawson’ssloganantici-patesrecentandveryfruitfulideasaboutlanguage.However,Strawsonhimselfdidnotembedthisintuitioninafulltheory.ThefinalchapterofIntroductiontoLogicalTheoryisafamousdiscussionaboutinduction,whichIshallconsiderwhendiscussingStrawson’sepistemology.MeaningandrelatednotionsOnlyabriefaccountofStrawson’sdiscussionofmeaningandrelatednotionsispos-sible.Hehasbeencriticalofatleastthreeapproachestomeaninginrecentphilosophy.OftheapproachassociatedwithQuine,whichisbroadlyskepticalaboutarangeofintuitivenotionsofmeaning,Strawsonhaswritteninnumerousplaces.Againstithemakesthefollowingpoints.First,theskepticismisgroundedonargumentswhichclaimthatthemeaning-notioncannotbeadequatelyexplainedincertainpreferredterms,saybehavioralones,butthereisnoreasontogroundthesemanticnotionsthatway.Second,thenotionscanbevalidatedbytheplainagreementbetweenpeopleinthejudg-mentstheymake.Third,thenotionsareindispensabletousaslanguageusers;thus,wesimplycannotspeakandthinkinthewaywedowithouttalkingofmeaningandsamenessofmeaning.Andfourth,thenotionsareindispensabletothetheoreticalstudyoflanguageandlogicaswell;thuslogiccannotdowithoutpropositions.In“MeaningandTruth”(inStrawson1971)Strawsoncriticizedanotherapproach,thatofDavidson,accordingtowhichatheoryofmeaningshouldbeatruthdefinition(seeDAVIDSON).Strawson’sargument,whichisrathercomplex,is,ineffect,thatthenotionoftruthissecondarytothenotionofsayingandthinking,andthattherefore,meaning,togetherwithtruth,hastobegroundedinarelationbetweensentencesandthecognitiverolesandcommunicativepurposesofspeakersandhearers.ThisidealedhimtoendorseamodifiedversionofGrice’sapproachtomeaning.InotherplaceshehascriticizedspecificproposalsbyDavidsonabouttheanalysisoflanguage.Finally,Strawsonhasrejectedtheanti-realistideasofDummettandothers,withtheirapproachestomeaning(seeDUMMETT).Strawsonseesanti-realismasaversionofrevi-sionarymetaphysics,whichisunbelievableandunsupported,anditcannotformthe340\nP.F.STRAWSONcoreofasatisfactoryaccountofthemeaningofordinaryjudgments.Whetherfinallysuccessfulornot,Strawson’sclearandelegantdiscussionsoftheseviewshavebeeninfluential.IndividualsIndividualswaspublishedin1959.Itprovedimmediately,andhasremained,bothcon-troversialandextremelyinfluential.Thebookisdividedintotwoparts,anditisthefirstpart,called“Particulars,”whichincludesthethreeveryfamouschapters,“Bodies,”“Sounds,”and“Persons,”thathasattractedspecialattention,andthatIshalldescribe.IntheIntroduction,Strawsonsetsouttheimportantdistinctionbetweendescriptiveandrevisionarymetaphysics.“Descriptivemetaphysicsiscontenttodescribetheactualstructureofourthoughtabouttheworld,revisionarymetaphysicsisconcernedtoproduceabetterstructure”(p.9).Strawsonmakesanumberofclaimsemployingthisdistinction;(1)Individualsisanexampleofdescriptivemetaphysics;(2)revisionarymetaphysicsis“attheserviceof”descriptivemetaphysics;(3)AristotleandKantaredescriptivemetaphysicians,Descartes,Leibniz,andBerkeleyarerevisionary;and(4)descriptivemetaphysicsbeinggeneralcannotavailitselfsolelyoftheresourcesgener-atedbyordinaryconceptualanalysis.Thisvividlyexpresseddistinction,whichiscer-tainlyvaluableforphilosophicaltaxonomy,andtheseclaims,haveoftenbeenaccepted,butqueriescanberaised,ofwhichIshallmentiontwo.First,Strawson’sownpracticeisnotpurelydescriptive.Heoffersexplanations,propoundsnecessities,andrejectscriticisms,aswellassimplydescribing.Second,andrelatedly,thedivisiondoesnotexhaustthetypesofmetaphysics.Thereisalsowhatwemightcall“anti-revisionarymetaphysics”whichamountstoadefenceoftheextantconceptualscheme,oracriticismofasuggestedrevision.Aswellasbeingdescriptive,Strawson’spracticeisalsoanti-revisionary.Inthefirstchapter,“Bodies,”Strawsonintroducessomeimportantconceptsandarguesthatmaterialbodiesarethe“basicparticularsfromthepointofviewofidenti-fication”(p.5).Thefirstconceptisthatofaspeakeridentifyingaparticularobjectforanaudience.Thisoccurswhenaspeakerreferstoanobjectandtheaudienceisabletoidentifytheobjectbeingreferredto.Thesecondconceptisthatofidentificationdepen-dence.Thus,itmightbethatourabilitytoidentify,inthefirstsense,onesortofpar-ticulardependsonourabilitytoidentifyanothersortofparticular,butnotviceversa.Ifso,thereisidentificationdependenceoftheformersortonthelattersort.Thethirdconceptisthatofreidentification;thisisStrawson’stermforanidentityjudgmentinwhichanitemencounteredononeoccasionisidentifiedwithanitemencounteredonanother.Armedwiththeseconcepts,Strawsonadvancesthreemainclaims.Thefirstcon-cernsthewayweareable,inthecourseofunderstandingreference,toidentifywhatitemsarereferredto.Strawson’spictureisthatwecandosoeitherbylocatingthemamongstthosesensiblypresenttous,roughly,perceivablebyusatthattime,orbypos-sessingidentifyingdescriptionswhichitemssatisfy.Seeingsuchatwo-foldstructuretoreferenceiswellknown.Russellspokelikewiseofknowledgebyacquaintanceandknowledgebydescription.Strawson’sviewofperceptionis,though,quitedifferentfromRussell’s,andsotheitemsdiscerniblethroughexperiencecanreallybeinspace.There341\nP.F.SNOWDONis,though,aworryaboutdescriptions:howdoweknowthatonlyoneitemfallsunderthem?Strawson’sansweristhattheparticularsweidentifyarelocatableuniquelyinthespatiotemporalframework,sayattheuniqueintersectionofvariousspatialcoor-dinates,andwecanthereforeguaranteeuniquenesswithinthatframework.Wecanrelatethegeneralframeworktothesegmentoftheworldwecurrentlyperceive.Second,Strawsonsuggeststhattheintelligibilityoflocatingitemsinthespatio-temporalframework(e.g.asthefountaininTrafalgarSquare)requiresrelativelyabidingstructuresofreidentifiableitems(e.g.theNationalGallery).Hence,totalkofobjectswithinthisframeworkisinconsistentwithskepticismaboutreidentification.Third,Strawsonarguesthatourabilitytoidentifybodiesdoesnotdependonanabilitytoidentifyparticularsofanyotherkind,butallotherkindofparticulars,forexample,privateparticulars,suchasthepaininmyleftfoot,unobservableparticulars,andparticularevents,dodependontheidentificationofbodies.“Materialbodies,there-fore,arebasictoparticularidentification”(1959:55).ManyfeaturesofStrawson’sargumenthavebeenchallenged,butIshallnoteonlytwothings.First,Strawsonarguesinthecaseofsomeothercandidates,notablyevents,thattheyarenotbasicbecausetheydonotpresentaregularenoughframeworktobethebasisfordefiningacoordinatesystem.Clearly,thisfailure(ifitisafailure)iscontingent.So,theconclusionabouttheuniquestatusofbodiesisitselfcontingent.Second,Strawson’snotionofidentificationisaninterpersonalone;itconcernswhatahearerisabletoidentifyasbeingreferredtobyanother.Itmust,therefore,berecog-nizedthatthedependencythesisneednotbetrueoftheindividualthoughtcapacitiesofasingleperson.Strawsoncontinueshisinvestigationoftheroleofspaceinourconceptualscheme,arguedtobefundamentalinchapter1,byseeingtowhatextentwecanimagineasubjectofnon-spatialexperiencewhoiscapableofapplyingconceptsofobjectiveandreidentifiableparticulars.Thecontrastthatmustbepresentinthisconceptualschemeisthatbetweentypeidentity–beingthesamesortagain–andnumericalidentity–beingthesameindividualagain.Strawsonchoosesacreaturewithpuresoundexperi-ence,whichheclaimswouldbenon-spatial.Heaskswhether(1)asubjectofsuchex-periencescouldmakesenseofnumericalidentityand(2)whetheritcouldmakesenseoftheself/non-selfdistinction,whichStrawsontakestobecentraltothinkingofobjects.Inconsidering(1)Strawson’sideaisthatananalogueofspaceisnecessary.Heimaginativelyproposestogeneratethatbyputtingintotheexperienceamastersoundofconstanttimbre,butvaryingpitchandloudness,changesinwhicharemeanttorep-resentmovement,alongwitharelativelyconstantcorrelationbetweenpointsonthemastersoundandcollectionsofothersoundtogeneratetheideaofreidentifiablepar-ticularobjectivesounds.Strawsondoesnotclaimthatthisissufficient,butonlythatitisnotobviouslyinsufficient.Inrelationto(2)Strawsonsuggeststhattherearenoveryhopefulgroundsforintroducingthedistinctioninsuchanimpoverishedexperienceworld.Strawson’simaginativeexerciseisbrilliantlydiscussedbyG.Evans(inVanStraaten1980),whoproposesthatStrawson’semploymentofthemastersoundunderestimatesthesignificanceofspaceandspace-occupationinourthoughtabouttheworld.Itcanbesaid,though,thatEvans’sfurtherinsightsrestonStrawson’spioneeringexplorations.342\nP.F.STRAWSONPersonsandstatesofmindChapter3ofIndividualsis,perhaps,themostdiscussedchapterofthebook,anditdeservesaseparatesection,bringingin,aswell,Strawson’slaterconsiderationofrelatedissuesinScepticismandNaturalism,chapter3.Strawson’sargumentbeginsbypickinguponathemethathademergedinrelationtothesoundworld.Howistheself/non-selfdistinctiondrawn?Toconsiderthis,thenatureofourbasicconceptofour-selvesneedstobedescribed.Strawsonclaimsthatthefundamentalaspectofthisconceptisthatthereisasinglethingtowhichweattributebothphysicalfeaturesandstatesofconsciousness(orpsychologicalpropertiesmoregenerally).ThusI,asinglething,amsixfeettallandinpain.Thequestionthenistransformedintotwo;whydoweascribestatesofconsciousnesstoanything,andwhytothesamethingasphysicalstates?Strawsonthenarguesthatwedonotanswereitherquestionbynotingthecausalimportanceofourbodiestothecharacterofourpsychologicalstates.Thisissimplynottherightkindoffacttoanswerthequestion.Twootheraccountsofourthoughtaboutourselvesareintroduced,andarguedtogiveincoherentaccounts.Thefirstclaimsthatcontrarytoappearanceswedonotascribementalstatestoanything–theso-called“no-ownershiptheory.”Butsuchanaccountmustexplainwhatisgoingonwhenweseemtoself-ascribestatesofconsciousness.Itproposesthatwearenotingthefactsofcausaldependencecitedearlier.But,asStrawsonpointsout,thecausaldependenceisofmyownexperiencesonthisbody,notofallexperiencesonthisbody,andsothedisavowedascriptionofexperiencestomyselfreappears.Theotheraccount,Cartesiandualism,deniesthatweascribethetwosortsofprop-ertiestothesamething;ratherweascribethephysicalsorttoourbodyandthementalsorttoourselves,anon-physicalego.Theproblem,accordingtoStrawson,withthisisthatself-ascriptionofmentalstatespresupposestheabilitytoascribesuchstatestoothers,sinceitisinthenatureofpredicatestohaveageneralapplication,andwithintheCartesianframeworkthereisnowaytopickoutothersubjectstomakesuchascrip-tions.TheproblemisthattopickoutanothersubjectImustdosoviatheideathatthesubjectrelatestoacertainbodyasIdotomine.ButthispresupposesIcanalreadythinkofmyself.Strawsonfurtherarguesthattheverynotionofanon-spatialparticular,suchasanegoissupposedtobe,lacksintelligibility,forhowcanweunderstandhowitispossiblethattherebetwosuchwhichareotherwisethesamewhentheycannotbedistinguishedspatially?Strawsonproposesthatwehavetotaketheconceptofapersonasprimitive,notillu-soryordecomposableintoelements.Hethendividesthepredicatesthatweself-andother-ascribeintoP-predicates,thosewhichareuniquetopersons,andM-predicates,thosewhichwesharewithmaterialbodies.Itfollows,hethinks,thatthecriteriaonthebasisofwhichweascribeP-predicatestoothersmustbelogicallyadequate,thatis,besuchthatnoskepticalproblemscanariseintheoptimalcase,onpainofnothavinganintelligiblestructureofconceptsatall.Sothephilosophicalproblemofothermindscannotarise.Strawsonaddstwothings:thattheexistenceofourpredicativepracticehereispartlyexplainedbythespecialnatureofaction,thefactthatitmixesthebodilyandthemental,andthattheincoherenceoftheCartesianmodeldoesnotimplythatwecannotimaginebecomingdisembodied.343\nP.F.SNOWDONStrawson’sdescriptionofthewaywethinkofourselves,as,thatis,double-sidedsinglethings,seemscompletelycorrectandveryimportant.Twolinesofcriticismordebate(amongstmany),however,deservementioning.Itisunclearwhatepistemologi-calimplicationscanbedrawnfromtheplausibleideathatpredicatesmusthaveanintelligiblepotentialapplicationtoarangeofthings.ItisalsounclearthatStrawsonshouldhaveallowedthatpersonscanbecomedisembodied.(FordeepcriticismsalongthesecondlineseeC.B.Martin1969.)Strawson’sdiscussionleavesquiteopenwhatshouldbesaidaboutaquestionwhichbecamecentralinthephilosophyofmindshortlyafterthepublicationofIndividuals,whichis:whatistherelationbetweenaperson’sphysicalstatesandhisorhermentalstates?Strawsonatmostinsiststhattheyarestatesofasinglething.InScepticismandNaturalism(ch.3)heconsidersthisquestionandarguesthatthereisacausalrelation,ratherthanoneofidentity.Hethereforerejectsmaterialismand,ineffect,espousesatypeoftheorythatusedtobecalled“doubleaspect.”Hisinterestingargumentisthattherewillbenowaytounifythementalandthephysicalstories,andthepointofiden-tityjudgmentsisunification.Thesecondpremisemight,ofcourse,bequestioned.ThechapteronPersons,togetherwiththatcitedabove,andhisdiscussionoftheParalogismsinTheBoundsofSense(forwhichseelater),constituteaprofoundandunifiedtreatmentofselves.SubjectsandpredicatesInPartTwoofIndividualsStrawsonprovidesatheoryofthesubject/predicatedis-tinction,ataskwhichheregardsasfundamentalandtowhichhehas,repeatedly,returned.HisfulltheoryisgivenmorerecentlyinSubjectandPredicateinLogicandGrammar(1974b),andIshallbrieflydescribeit.Ithasastrongresemblancetohisearlieraccount.Strawsonstartswithaseriesofwhatmightbecalledmarksofthesubject/predicatedistinction.Thus,predicateshaveanumberofplaces,whereassubjectexpressionsdonot.Predicatescanbenegatedandgenuinelycompoundedwhereassubjectexpressionscannot.Subjectexpressionsareopentoquantification,whereaspredicatesarenot,asQuinesuggests.Strawson’sattitudeisthatthesemarks(whichmayneedsomemodificationtoo)needtobeexplainedanddonotgivethebasicdistinction.Heproposestoexplainthembylinking,initiallyinacentralcase,thelogico-grammaticaldistinctiontoanontologicalone,namely,thedistinctionbetweenpar-ticularsanduniversals.Roughly,universalsrepresentwaysofclassifyingorcollectingparticulars.Withthisgoestheideathatuniversalsformstructures;thus,ifanobjectfallsunderoneclassificationitfollowsthereareothersunderwhichitdoesnotfall;or,ifitfallsunderoneclassificationitfollowsthereareothersunderwhichitdoesfall.Universalscomeinincompatibilityrangesorrequirementranges.Nothinganalagousappliestoparticulars.ThesuggestionwhichStrawsondevelopsisthatinalanguagesuchasEnglish,thelogicalfeaturesofpredicatesflowfromthefactthattheroleorfunctionofpredicatesistointroduceauniversal,togetherwiththesecondroleofindicatingthatthereferredtoitemexemplifiestheuniversal.Thustheideathatsubjectexpressionscannotbenegatedbutpredicatescanistobeexplainedbythefactthatuniversalsformlogicalstructures,whereasparticularsdonot.Theinaccessibilityofpredicateexpressionsto344\nP.F.STRAWSONquantificationistobeexplainedbythefactthattheyhaveadualrole:ofintroducingauniversalandofindicatingexemplification,forthelatterisnot,asonemightsay,asomething.Strawsonthenextendshisaccountbeyondthebasiccase.NoreceivedassessmentofStrawson’shighlyingeniousproposalhasemerged.TheboundsofsenseSevenyearsafterIndividuals,StrawsonproducedTheBoundsofSense.Init,heanalyzes,criticizes,anddevelopsthecentralideasofKant’sCritiqueofPureReason.Thetreat-mentofKantisunlikethatofmostcommentatorsinthatitisnotmarkedbyahagio-graphicalreverencetowardsKant,nordoesitsimplyrepeatKant’slanguagebywayofexplainingit,nordoesitaspiretothelengthoftheCritique.Strawson’saimispri-marilytoseparate,insofarasitispossible,Kant’sconstructiveandcriticalthesesfromthetranscendentalidealistframeworkinwhichKantplacesthem,andalsotoseparatethemfromtheoutdatedscienceandlogicofKant’stime.Strawson’smainclaimisthattranscendentalidealismisincoherent,butthattherearevariousthesesthataredefen-sibleandimportant,defensibleeitherinthelightofwhatKanthimselfoffersoronthebasisofotherargumentswhichStrawsonconstructs.What,finally,ismostdistinctiveofStrawson’streatmentisthebrilliantwayinwhichheattemptstoextractanddefendthesemetaphysicalandepistemologicalclaims.TheargumentagainsttranscendentalidealismoccursinPartOneandPartFour,reachingtheconclusionthatitisincoherentintwostages.Inthefirst,itisarguedthatnointerpretationofKant’sidealistclaims(accordingtowhich,ofcourse,theworldofspaceandtimeismerelyaformofappearance,contrastingwiththerealmofunknow-ablethingsinthemselves)issatisfactorywhichtreatsitassayingsomethingweakerthan:realobjectsaresupersensibleandwecanhavenoknowledgeofthem.Thesecondstagereasonsthatanymodelsustainingsuchaclaimmustbeincoherent.Strawsonarguesthisinvariousways,butoneistosuggestthatweourselvescannotcoherentlyfitintosuchapicture.If,asKantsays,wemerelyknowhowweappear,isthisnotagenuinetruthaboutourselves,henceitselfnotmerelyamatterofappearance?Typically,StrawsonisconcernedtodigdeeperandtoexplainKant’sadoptionofthemodel,andhearguesthatitssourceisadistortedresponsetothenot-at-allincoher-ent,indeed,accordingtoStrawson,centraltoourownthinking,ideathatexperiancesandrealobjectsarecausallyrelated.AllthreeelementsinStrawson’sresponsetotran-scendentalidealism–itsinterpretation,evaluation,andexplanation–havebeendis-puted.(See,forexample,Allison1983andWalker1978.)ButStrawson’sindependentassessmenthasbeenthestimulusofthisrenewedinterest.Ofmuchgreaterphilosophicalimportance,though,isStrawson’sattempttodetachthecentralconstructiveandcriticalthesesoftheCritiquefromtranscendentalidealismandtoassessthem.KantrepresentsthepurposeoftheCritiqueasexplaininghowthesyntheticaprioriispossible.Inhisconstructivereinterpretation,Strawsonreplacesthisbythequestion,Whatfeaturesareessentialtoanyconceptionofexperiencethatwecanmakeintelligibletoourselves?Hecallsthetaskofansweringthis“themetaphysicsofexperience.”TheKantianideaisthatfromthenotionofaself-conscioussubject,whocanself-ascribeitsexperiences,wecanderivesubstantialconditionsthatmustbemetbythecontentofexperiencesthusenjoyed.Thefirstconditionisthatit345\nP.F.SNOWDONmustincludeawarenessofwhatarerecognizablyindependentobjects(theobjectivitythesis).Further,theseobjectsmustberecognizablyspatial(thespatialitythesis),andtheymustsatisfyvariousprinciplesofpermanenceandcausation(thethesisoftheAnalogies).StrawsonarguesthatKant’sownthesisaboutpermanenceandcausationaretoostrong,butthatratherweakerclaimsaboutsemi-permanenceandthenecessaryapplicabilityofcausalnotionscanbedefended.However,themostcrucialandbrilliant,buttosomeextentobscure,partofStrawson’sreconstructionishisdefenseoftheobjectivitythesis.Whymustexperiencesometimesbeof,orasof,objects?Theargu-mentstarts,ofcourse,fromtheKantianassumptionthatwearedealingwiththeex-periencesofaself-conscioussubject,thatis,onewhocanascribetoitselftheexperiences.Fortheretobeanycontenttosuchascriptions,thatis,fortheclassifica-tionofthestatusofsuchoccurrencesasexperiencestohaveapoint,theremustbesomeunderstandingofthecontrastingstatusofnotbeinganexperience.Thiscontrastordistinctioncanbepresentonlyifthecreature’sexperiences,which,afterall,arewhatsustainitsconceptapplications,sustaintheapplicationofnon-experientialcategories.Butthatistorequirethatsomeoftheexperiencesmustpresent,orbeof,itemsofanon-experientialkind.Suchitemsmust,thatis,berecognizedasobjects.This,inaverycompressedstatement,seemstobeStrawson’sargument.Heconcludesthattherecannotbeagenuineproblemofjustifyingourbeliefinob-jects,forsuchaproblemrequiresavantagepointwherethereisself-ascriptionofexperienceswithoutanyobjectivejudgments,whichcollectivelyawaitjustification.Nosuchvantagepointisavailable.Strawson’sargumentclearlyhasaffinitieswithWittgenstein’sprivatelanguageargument.Ofcourse,ithasreceivedmuchcriticalexamination,aparticularlysubtleexampleofwhichisCassam’s(1995).Onepointisthattheargumentseemstorelyontheassumptionthatcategoriesonlyhavecontent-fulapplicationinvirtueoftherebeingcaseswhichdonotfallunderthem,butthisassumptionmayberejected.Anotherissueiswhetherathesisliketheobjectivitythesisrelatestohowexperiencemustseem,ortohowitsactualobjectsmustbe.InsubsequentchaptersStrawsonconsiderswhatcanbedefendedfromKant’sargu-mentsintheAnalogies.HerethestrongKantianthesiscannotbesustained,butweakerversionsare,accordingtoStrawson,defensible.Thus,forexperiencetoberecognizablyofobjects,theexperiencermustbeabletodistinguishthetemporalhistoryoftheobjectsfromthoseoftheexperiences,whichrequiresthattheobjectsasexperiencedyieldaframeworkintermsbywhichtokeeptrackofthem,bothspatiallyandtem-porally;andthisrequiresthattheexperiencesgroundtheapplicationofconceptsofenduringobjects(inonesense,thatis,ofsubstances).Further,asStrawsonputsit,objectshavetobeunderstoodasthegroundof“compendiaofcausallaws.”Sotheapplicationofcausalnotionsisalsorequired.ThelatterstagesofargumentherebothresembleanddeveloptheargumentsinIndividuals.TheBoundsofSensealsoanalyzesthecriticalprogramoftheDialectic.Itdoessoinadeepandveryilluminatingway,especiallyinconnectionwiththeillusionsofratio-nalpsychologyasexposedintheParalogisms.StrawsonseesKant’sachievementinthatsectionasrefutingattemptstoinferthatselvesarespecialnon-physicalthingsfromwhatareundoubtedlyspecialepistemologicalfeaturesofself-knowledge.ThemajorincompletenessinKant’saccountishisreluctancetosettleforanembodiedself,limited346\nP.F.STRAWSONasheisbyhistranscendentalidealism.TheresultisthatStrawson’sbookhashadanenormousimpactonbothKantianscholarshipandrecentmetaphysics.ResponsestoskepticismStrawsonhashadanumberofinterestsasanepistemologist.Hehasnotattemptedtoprovideananalysisofknowledge,buthehastriedtodescribeitsstructure,andespeciallytogiveanaccurateaccountoftheroleofperceptionwithinthatstructure.However,hisinfluencehasbeengreatestinhisroleasopponentofskepticism.Buthisattitudetoskepticismhasevolved,andIwanttodescribebrieflyfourmainstages.InIntroductiontoLogicalTheory,hearguedinrelationtotheparticularcaseofinduc-tion,thatwhentheskepticclaimsthatnojustificationhasbeenprovided,thereisnointelligibleandpossiblethingthatcanbeunderstoodby“justification.”Thecorrectresponsetoskepticismis,therefore,nottotrytoprovideajustificationbuttoseethattherecannotbesuchathing.Whycannottherebesuchathing?Theanswertothisemergeswhenoneofarangeoffullerspecificationsofthetaskisgiven.Thus,onewayofunderstandingthenotionofjustificationheretakesittorequireshowingthatinduc-tivesupportisreallydeductive.Clearlythisisabsurd,inthatitrequiresanobliterationoftheverydistinctivemethodofsupportthatraisestheprobleminthefirstplace.Anotherunderstandingisthatajustificationwouldamounttoshowinginductionisareasonableprocedure;butthisisnotsomethingthatneedstobeshown,since“beingreasonable”preciselymeansfollowinginducation.Anotherunderstandingisthatajus-tificationwouldbeaproofthatinductionisboundtowork;butsuchaproofisimpos-sible.Thereis,accordingtoStrawson,nocoherentdemandhere.ThestructureofStrawson’sargumenthereleavesroomforsomeonetofindacoherentinterpretationoftheskeptic’sclaim,andsubsequentdiscussionhaseitheraimedatdoingthat,oratdisputingtheanalyticclaimsStrawsonhimselfmakes.InIndividuals,Strawsonseemstohavearguedthatskepticismisincoherentinthatpossessionoftheconceptsthattheskepticneedstoidentifythetopicofhisownskep-ticismitselfrequiresanon-skepticalattitudetowardsthe(best)basesofapplicationforthoseconcepts.Forexample,havingmentalconceptsinvolvesunderstandingtheirapplicationacrossarangeincludingothersandthisrequiresthatthecriteriaofappli-cationtoothersarelogicallyadequate.Strawsonarguesinasimilarwayinrespectofacross-timeidentityjudgments.Themuch-debatedquestionaboutsuchclaimsiswhetheritisshownthattheadoptionofaskepticalattitudeisreallyinconsistentwithunderstanding.ThethirdsortofresponsethatStrawsonhasexplored,inTheBoundsofSense,aretran-scendentalarguments.Theideaisthattheskeptichimselfsupposesthatsomeconcep-tualapplicationsarepossible,forexample,anon-committaldescriptionofexperience.Butitisarguedthattheveryascriptionsthattheskepticispreparedtomakepresupposetheapplicationoftheconceptsthatheisskepticalabout.Thus,Strawsonargues,aswehaveseen,thattheself-ascriptionofexperiencesrequiresjudgmentsaboutobjects.Now,thereisnoapriorireasontoholdsuchanargumentcouldnotbecorrect,butanytran-scendentalargumentofmorethanminimallengthrunstheriskofoverlookingawayinwhichtheskeptic’sconceptscanhaveapplicationwithoutneedingtheonesindispute.347\nP.F.SNOWDONFinally,inScepticismandNaturalism,Strawsonsuggestedanotherresponse.ArguingashethinksinthespiritofHumebutespeciallyofWittgenstein,hedrawsadistinc-tionbetweenrealdoubtswhichareworthengagingwithandunrealdoubtswhichcannotassailanyoneandwhicharenotworthrespondingto.Thetraditionalskepticaldoubtsfallintothesecondcategory.Nooneeverseriouslywonderedwhetherthereisanexternalworld.Strawsonsuggeststhatwedonotneedthereforetoargueagainstskepticism.Thisveryboldandhistoricallyinformedresponsehasnotbeenpopular,thegroundbeingthatitisnotobviousthatthenon-persuasivenessofanargumentmeansitneednotbeengagedwith.ItmightalsobewonderedwhetherthehistoricalrootsforsucharesponsearenotmoreLockeanthanHumean(seeMOORE).Still,inpostwaranalyticalphilosophynoagreedresponsetoskepticismhasemerged,andStrawsoncanbecreditedwiththedevelopmentofanumberofthecandidatesstillbeinginvestigated.FreedomandresentmentStrawsonhaswrittenlittleaboutmoralphilosophy,hisattitudetoitbeing,perhaps,somewhatsimilartothatexpressedbyC.D.Broad,whenhereputedlysaidthatthewholeofmoralphilosophycouldbewrittenonthebackofapostagestamp.However,oneofStrawson’sessaysonmoralphilosophy,namely,“FreedomandResentment”(in1974a,butfirstpublishedin1962)hasbeenextremelyinfluential.Interestingly,Strawson’sargumentinthisarticlebearsacloserelationtothecaseagainstskepticisminhis1985book.Strawson’saimistofindapositioninthedebateaboutdeterminismandresponsibil-itythatavoidsincompatibilism(whichStrawsonrelabels“pessimism”)butwithoutdistortingthenatureofourmoralviewofourselvesandothersinthewayinwhich,accordingtoStrawson,standardcompatibilistsdo.Standardcompatibilistsobservethatourpracticesofpunishingandpraisingwouldhaveautilityevenifdeterminismwasaccepted.Strawsonclaimsthatviewingtheissuethiswayover-intellectualizesthebasisforsuchpractices,whichisnotreflectivelyshapedbyconsiderationsofutility,butratheristheupshotofcertaincentralreactiveattitudesengenderedinthecourseofordinaryhumanlife.Examplesareresentmentandgratitude,whicharedirectedatothers,butalsoguiltandremorse,whichareself-directed.Thereconcilingprojectisbasedonthefollowingclaims:suchattitudesandfeelingsareproducedinusinthecourseofournormalparticipationinhumanlife,withitsnecessaryengagementwithothers;theatti-tudescanbesuspendedinexceptionalcircumstances,forexample,whendealingwithpeoplewhoarepalpablymentallyabnormal.Theycannot,however,beuniversallysus-pended,becausetheyareintegraltohumanrelationshipsthatwecannotabandon.Andwhentheyaresuspendedinthelimitedcaseswherethisispossibleitisnotbecauseweseethecasesinthelightofageneralconvictionindeterminism,butratherbecauseofmorespecificreasons,whichvaryfromcasetocase.Strawsondrawsfromtheallegedimpossibilityofabandoningsuchreactions,andtheabsenceofadependenceofthemonarejectionofdeterminism,hiscentralcompatibilistconclusionthatdeterminismisnothreattotheirlegitimacy.Hedrawsfromtheclaimthattheyarenaturalandnotgov-ernedbyconsequentialistconsiderationstheconclusionthatstandardcompatibilistshavedistortedthecharacterofourmoralresponses.348\nP.F.STRAWSONStrawson’spaperwasimportantbecauseitrepresentedanovelcompatibilistapproach.Thecentralissueitraises,though,resemblesthatraisedbyhisobservationthatpeoplearenotpersuadedbyskepticalarguments.Doesthefactthatpeoplewillcarryonbelievingordoingsomethingshowthatitisbeyondcriticism,orlegitimateforthemtodoso?ConclusionThemoststrikingaspectofStrawson’sphilosophicalcareerhasbeenhisextraordinaryfertility,combinedwiththeconsistentdepthandclarityofwhathehasproduced.Inthepresentessay,therehasnotbeenspacetosurveymanyaspectsofhiswork.Afinalindicationofhisimportance,though,ishisinfluenceontheverybestphilosophersofthegenerationsafterhis,ofwhomIwishtomentiononlytwo.ThefirstisGarethEvans,whosebookTheVarietiesofReferenceisclearlycolossallyinfluencedbyStrawson.ThesecondisJohnMcDowell,whointheprefacetoMindandWorldpayseloquenttributetoStrawson.WhatmightbecalledaStrawsoniantraditionhasemerged.BibliographyWorksbyStrawson1952:IntroductiontoLogicalTheory,London:Methuen.1959:Individuals,London:Methuen.1966:TheBoundsofSense,London:Methuen.1971:Logico-LinguisticPapers,London:Methuen.1974a:FreedomandResentmentandOtherEssays,London:Methuen.1974b:SubjectandPredicateinLogicandGrammar,London:Methuen.1985:ScepticismandNaturalism:SomeVarieties,London:Methuen.1992:AnalysisandMetaphysics,Oxford:OxfordUniversityPress.1997:EntityandIdentity,Oxford:OxfordUniversityPress.WorksbyotherauthorsAllison,H.E.(1983)Kant’sTranscendentalIdealism,NewHaven,CT:YaleUniversityPress.Cassam,Q.(1995)“TranscendentalSelf-Consciousness,”inThePhilosophyofP.F.Strawson,ed.P.K.SenandR.R.Verma,NewDelhi:IndianCouncilofPhilosophicalResearch.Hahn,L.E.(ed.)(1998)ThePhilosophyofP.F.Strawson,LaSalle,IL:OpenCourt.(IncludesanintellectualautobiographybyStrawson,acollectionofessaysdiscussinghiswork,andhisreplies.)Martin,C.B.(1969)“People,”inContemporaryPhilosophyinAustralia,ed.R.BrownandC.D.Rollins,London:Allen&Unwin.VanStraaten,Z.(ed.)(1980)PhilosophicalSubjects,Oxford:ClarendonPress.(AdistinguishedcollectionofpapersonStrawsonwithhisreplies,thepapersincludingG.Evans’s“ThingsWithouttheMind,”hisprofoundresponsetoch.2ofStrawson’sIndividuals,andJ.McDowell’s“Meaning,CommunicationandKnowledge,”aresponsetoStrawson’slectureonmeaningandtruth.)Walker,R.(1978)Kant,London:RoutledgeandKeganPaul.349\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200127PhilippaFoot(1920–)GAVINLAWRENCEPhilippaFootisamongthehandfulofthetwentiethcentury’sverybestmoralphiloso-phers.Herachievementconsistsnotsomuchoftruthspresentedasofherdistinctivevoiceinphilosophy.Inthisway,sheislikeMooreorRawls,ormostpertinentlyWittgenstein.Toreadherisimmediatelytostrugglewiththerealstuffofthesubject,tothehigheststandards;thesubjectisnotthesameforoneagain.Herworkdividesintoseveral,diverselyoverlapping,strands:themajorthemesofethics,suchasitsobjectivityanditsrationality;middlerangeissues,suchasfreedomofthewill,virtuesandvices,thecritiqueofutilitarianism,andmoraldilemmas;morespecificethicaldistinctionsandproblems,suchasthedoctrineofdoubleeffect,abor-tion,euthanasia,andcapitalpunishment.Iwillfocusonthemajorthemes.Hertreatmentoftheissuesofmorality’sobjectivityandofrationalityfallsintothreephases.ThesephasesrelatetothreeHumean(orneo-Humean)orthodoxies:(1)thefact/valuedistinction,(2)thepracticalityofmorality,and(3)theend-relativeconcep-tionofpracticalreason.Roughly,Footstartsbyrejecting(1)whileaccepting(2)and(3).Shethenrejects(2)aswell.Finallysherejects(3)infavorofamoreAristotelianconceptionofpracticalreasonandcomestoreassert(2).Thefirstphase(1950stomid-1960s):theWittgensteiniandefenceofthepossibilityofnaturalismFromthefirst,FoothastakenmainstreamcontemporarymoralphilosophytobedominatedbytwoofthethreeHumeanpropositions:(1)Thefact/valuedistinction(anti-naturalism)assumes,againstthenaturalist,thatthereissomelogicalgapbetweenfactandevaluation–between“is”and“ought.”Evaluationsgobeyondthenaturalfacts.And,inthesubjectivistversion,theyrequireacontributionfromthesubject.Ifso,evaluativejudgments,unlikefactualones,arenotwhollyresponsibletotheworld,andevaluativeargumentmaybreakdowninawaythatfactualargumentcannot:twoopponentsmayagreeaboutallthefacts,andyetcommittheirwilldifferentlyandsobeleftinabareoppositionofwillorattitude,withoutanyrationalerror(seeANSCOMBE).Threeversionsof(1),“thebreakdowntheory,”particularlyconcernus.(1a)Radicalsubjectivismclaimsthatnoparticular,conceptuallyrestricted,rangeoffactsiseither350\nPHILIPPAFOOTnecessaryorsufficientevidenceforacertainevaluativepredicate(i.e.nocontentrestriction).(1b)Restrictedsubjectivismclaimsthat,whilenecessary,sucharangeoffactsisneversufficient.(1c)Partialsubjectivismclaimsthat,whilenecessaryandsome-timessufficient,sucharangeoffactsisnotalwayssufficient.(2)Thepracticalityofmoralityistheorthodoxythatmoralityissomehow“practical”oraction-guiding.Although(1)securesadivisionbetweenfactandvalue,itallowsonetoexplainthesecondorthodoxy.Itisbecausemoralityisamatterofvalue,notoffact,thatitcanbeaction-guiding.Itmaybethatthepracticalityofmoralitymaybeamattereitherofitsmotivationalefficacy,orofitsrationalityandthustheclaimthatmoralconsiderationsuniversallymotivate,orthattheyuniversallyconstitutereasons.(Thelatterraisesthefurtherissueofthenatureofpracticalreason,andthustherelevanceof(3).)Furtherdifferencesexistoverthemodalityofthethesis:doesmoralityjusthappentobeuniversallymotivating/reason-giving,orisitnecessarily,oressentially,so?andoverwhetherthemotivation/reasonyieldedbymoralityissupposedoverriding(orauthoritative).Inherfirstphase–mostnotablyin“MoralArguments,”“MoralBeliefs,”and“HumeonMoralJudgement”(allin1978)–Footarguesboththat(1)hasnotbeenmadeoutbyitsanti-naturalistproponents,andthat(2)doesn’tinfactrequireit.Herinitialtargetis(1a)radicalsubjectivism.Accordingtothisview,therearesup-posedlynocontentrestrictionsonwhatcanbeheldtobemorallygood,oramoralprin-ciple,oramoralcode.Consistencyapart,apersonisfreetodiscountthefactsorgroundsanyoneelsetakesasevidenceforsomething’sbeinggood,andisfreetocountasevidencefactsthatnooneelseacknowledgesasevidence.Thispersonalfreedomtodeciderelevantgrounds–whatFootterms“theprivateenterprisetheory”ofmorality–seeminglyrisksmakingevaluativepredicatesmeaningless.Isn’tapredicatethatcanbefreelypastedanywherenecessarilyuninformative?Whatinitiallysavesthesubjec-tivististhe“linguisticturn,”soinfluentialinthe1940sand1950s.Thisistherecog-nitionthatlanguageismulti-functional,andthat,inparticular,ithasotherpurposesbesidesthedescriptiveorinformational.Thus,theanti-naturalistneednotfollowMooreinholdingthat“good”isdescriptiveofanon-naturalproperty,ratherthanofanaturalone.Rathertheanti-naturalistholdsthattheprimaryuseof“good”isnottodescribetheworld,buttoexpressanattitudeortorecommend.Footarguesthatradicalsubjectivistsdonotprovetheircase.Shebeginsbyconsid-eringamiddlelevel,orthick,predicate,suchas“rude,”thatheropponentswouldlikelyconcedeisevaluative(i.e.tohaveanexpressiveoraction-guidingfunction).But,asshepointsout,ifoneusestheconceptofrudeness,oneisn’tfreetotakejustanythingonelikesasevidenceforrudeness(e.g.walkingslowlyuptoadoor),ortorejectjustany-thingeither(e.g.beingspaton),anymorethanoneisfreetodecidewhatisandisn’tevidenceforabraintumor.(Ofcourse,givenasuitablyspecialbackgroundstory,behav-iorthatnormallyisnotrudemayberude,forexamplepointedlywalkingslowlywhenaskedpolitelytohurry,andviceversa.)Footthenclaimsthat,foralltheoppositionhasargued,whatholdsfor“rude”mayholdforallevaluativeconcepts,includingthemoreabstract“thin”ones.Foralltheyargue,theremayyetbethe“verytightestofrelations”betweenfactandvalue.Evaluativeconceptsare,likeanyothers,criteria-governedcon-cepts:theyhave“definitionalcriteria”whichlaydownwhatisandisn’trelevantevi-denceforthem.Leavethesecriteriabehindandyouleavebehindtheconcept.(In351\nGAVINLAWRENCE“MoralBeliefs”sheputsthepointintermsofanobjectbeinginternallyrelatedtotheattitudeitisanobjectof.Onecannotfeelproudofjustanything:onehas,say,tobethinkingofitasanachievementofone’sown.)AtthispointsubjectivistscangoinoneoftwodirectionstoavoidFoot’sposition.First,theycanacceptFoot’spointfor“thick”evaluativeterms,butrejectitfor“thin”ones.Theymayclaimthatwearefreetodecidewhatcountsasbenefitorharm;ortheymayadmitthattherearecriteria,orrulesofevidence,forourexistingmoralcodeandourexistingmoralterms,butclaimwearealwaysfreetoinventnewmoralterms,newvirtues,ormoralcodes.Tothis,FootrepliesinamannerreminiscentofWittgenstein’sPhilosophicalInvestigations§261,thatfactualconstraintsapplyheretoo.Notjustanythingcancountasabenefitoraharm,norjustanythingcountasavirtueormoralcode.Theclaimthatnooneshouldlookathedgehogsinthelightofthemooncouldnotcountasamoralprinciplewithoutaspecialbackground.Moregenerally,eventheword“moral”iscontent-restricted.Notjustanyalien,ortrivial,codecancountasamoralone.Roughly,moralconsiderationsmustrelatetohumangoodandharm.(Footdoesnotclaimtohavethoroughlyelucidatedthedefinitionalcriteriaatworkinourmoralpredicates.)Alternatively,subjectivistscanoptforarestrictedsubjectivism,byusingproposition(1b).Accordingtothisview,Foot’sconceptuallyrestricteddescriptiveconditions(a)arenecessaryforanapplicationofanevaluativepredicate,and(b)sufficientforamerelydescriptive(or“invertedcomma”)applicationofthem.But,itclaims,theseconditionsdonotsufficeforaproperlyevaluativeuse.Tothinktheydoistomisstheverypointofevaluation,namely,thatthereinagentscontributesomethingoftheirownafterthefactsaresettled,beitacommitmentofwillorintention,feeling,orattitude,etc.Itisthisfurtherelement,ofitsnaturelinkedtoatendencytoact,thatisneededtosecuremorality’spracticality,asin(2).Itissomethingentirelyuptoagents,andcannotbelogicallyrequiredofthembytheworld.In“MoralBeliefs,”partII,Footarguesthatthispositiontooismistaken:itputs“thepracticalimplicationofvaluewordsinthewrongplace.”Herargument,asIreadit,con-tainsacarrotandstick.Thecarrotistheofferofamoreplausibleaccountofmorality’spracticality.Injury,shesuggests,offersahelpfulparallel.Onceweagreethatnotjustanythingcanbecalledaninjury,wecanseethatthereasonforustoavoidinjuryisnotthat“reason-givingness”isbuiltintotheevaluativeuseof“injury.”Rather,itissimplythatcertainkindsofthingscountasinjuries.IhaveareasonnottopokeasharpobjectinmyeyenotbecauseIfindmyselfpreparedtocallthisan“injury”inafullevalu-ativeorfullaction-guiding,sense,butbecauseIwon’tbeabletosee,and,asthingsare,Ineedtosee.Similarlytheconnectionbetweenmoraljudgmentandthewilldoesnotlieinthewill’scommitmentbeingaconditionconstitutiveofevaluativeuse,butinthecontentofmoraljudgment.Itisthefactsaboutwhatthevirtuesare,giventheconditionsofhumanlife,thatsecurethatthereisreasonforeachofustobevirtuous,andactvirtuously.Thestickistoqueryhowanysuchextraelement–beitattitude,dispositiontochoose,self-addressedimperative,orwhatever–couldpossiblyperformthisroleofrationalizingactionsorcharactertraits.Forallsuggestedcandidatesseemobviously352\nPHILIPPAFOOTunnecessary,asfarasreason-givingnessgoes.ImayknowthatcourageisavirtueandthatIhavereasontobecourageous,but,cowardthatIam,havenocommitment,orwhatever,tomendingmyways.Thealternativeaccountofmorality’spracticality,(2),interpretsitasamatterofuni-versalrationality,notofuniversalmotivation.(Footsupposesthisthemoreplausibleversion.)And,byitsverynature,itushersbackonstageanother,mucholder,oppo-nent,theimmoralist,who,likePlato’sThrasymachusagreesthatmorality’srational-ityissettledbyfactsaboutthevirtues,butquerieswhetherthesefavortherecognizedvirtues,andjusticeinparticular;forjustice,onthefaceofit,is“another’sgoodandselfharm.”Therestof“MoralBeliefs”attemptstoanswerthisopponent.Butwhichfactsaboutthevirtuesandvirtuousactionwouldshowthatthereisreasontopursuethem?ThespecificformofFoot’salternativeaccount,andofherreplytotheimmoralist,iscontrolledbyherassumptionofanend-relativeconceptionofpracticalreason,namelythethirdorthodoxy.(3)AconsiderationCisareasonforagentsif,andonlyif,itservessomethingtheydesireorcareabout,thatis,itisareasononlyinrelationtotheirends.Giventhisviewofwhatmakessomethingareason,weallhavereasontobevirtuousandtoactvirtu-ouslyifandonlyifthefactsaboutthevirtuesandvirtuousactionsshowthemtoconnectupwithwhateachpersonhappenstowantorcareabout.Thisconnectioncouldbeinstrumentalorconstitutive.Moralconsiderationsmaybereasonseitherbecausetheyfurtheranagent’snon-moralend,orbecausetheyareconstitutiveofachievingsomemoralendoftheagent’s.Footsupposesthatnoteveryonehasmoralends,andthattheonlyenduniversallysharedisanon-moraloneofself-interest(albeitnotnecessarilyselfish).Thusshefeelsthat,todefendtheuniversalrationalityofmoral-ity(hercommitmentto(2)),shehastodemonstrateakindofmoralinstrumentalism:that,asthingsare,thevirtuesfurtherself-interest.Theresultingpositionhasitsproblems.OnethatimmediatelyoccupiesFootisthedefenceofjustice.Itisnotdifficulttomakeageneralcasethatjusticefurthersanagent’sself-interest;butwhatofthe“tightcorner,”theparticularcasewheretoactjustlyanagenthastolaydownherlife?Howcanjusticeherebemoretotheagent’sadvantageorself-interest?ThesolutionsheofferswasaversionofHume’sbut,byherownaccount,itwasinpartdissatisfactionwiththisthatleadshernexttoabandon(2).Itisnot,however,untilthethirdphasethatFootlocatestherealculprit,theneo-Humeanviewofpracticalreason.Onlythenistherestrictedsubjectivistpresentedwithaproperalternativeaccount,andtheimmoralistwithanadequateresponse.Thesecondphase(1970s):uneaseovermoralityandtherejectionof(2)ThemarkofFoot’ssecondphaseisthesuspicionthatourordinarymoralthoughtandlanguagecontainselementsoffiction,initsassumptionsofcompleteobjectivityandofrationalityorauthority.In“MoralityandArt”(1970)Footstillrejectsthefact/valuedistinction,(1).Thedefinitionalcriteriabuiltintotheconceptofthemoralconstrainwhatcancountasamoralcodeorasmorallygoodandexplainswhysomanymoral353\nGAVINLAWRENCEjudgmentscanbeprovedfromthefacts(e.g.thatHitler’streatmentoftheJewswaswicked).Thecoreofmoralityisobjective,anditstruthnon-relative.Nonethelessthesecriteriaarenotsostringentastoruleoutsubjectivismentirely.Forexample,regardingabortionandeuthanasia,Footsuggeststhatdifferentpeoplecould,subjectively,choosetogobydifferentprinciples,andeachchoicewouldequallycountasmoral.AndatsuchpointstherecouldbethatverykindofbreakdowninmoralargumentthatFootearlierdenied.Footthusembracespartialsubjectivism,(1c).Ananalogouspossibilityispresentedforrelativism,wheredifferentelectivemoralprinciplesmaybepeculiartodifferentgroups.Atthesepointsmoraltruthwouldbeobjectivebutrelativetothestandardsadoptedbyaparticularcommunity.Inthe1978Postscriptto“MoralityandArt”andin“MoralRelativism”(1979),Footislessconfidentaboutthesepointsandholdsthattheycannotbesettledwithoutafirmergriponthenatureofthedefinitionalcriteriaandofcertainkeyconceptssuchashavingavalueandhappiness.Thisisstillunfinishedbusiness(seetheendof“DoesMoralSubjectivismRestonaMistake?”(1995)).Morescandalouslystill,in“MoralityandArt”andthenin“MoralityasaSystemofHypotheticalImperatives,”(both1978)Footchallengestheorthodoxythatevery-oneshouldbemoralandactmorally.Shedistinguishestwousesof“should.”Oneisanon-hypothetical,ordesire-independent,use:itsayswhatisrequiredbyacertainpointofvieworsystem.Andinthissenseitistautologicalthatoneshould,morallyspeak-ing,bemoral.Theotheruseof“should”isreason-giving.Clearlythisistheuseatissuehere.But,Footargues,theclaimthat,whatevertheirdesiresandinterests,everyonehasreasontobemoralandactmorallylacksasense;foritimpliesthatmoralconsid-erationshaveamagicallyautomaticreason-givingforce.Insteadweshouldconcedethatmoral,likeother,considerationsofferreasonsonlyhypothetically,thatis,ontheconditionthattheagenthappenstohavetheappropriateends.(Notethatnowshesup-posesmoralconsiderationsareproperlyreasonsonlyforthosewhohavemoralends,thatis,viaaconstitutive,notaninstrumental,connection.)Ifso,peoplewholackmoralendswillhavenoreasontoactmorally,andtosaythattheyshouldsoact,orshouldhavesuchends,ismerebluff.AndsoFootabandonstheclaimofmorality’suniversalrationality(2),viewingitasanotherpieceofmoralfiction,completewithafictitiouslinguisticuse.Moralitythenturnsouttobeinescapableinonesensebutnotinanother.Theapplicationofthemoralpredicates–just,courageous,mean,cruel,etc.–isanobjectivematter.Butwhethermoralconsiderationsarereasonsturnsonthesubjectivematterofwhattheagenthappenstocareabout.InshortFoot’spositionisthatofmoralobjectivity,rationalsubjectivity.Onceagainthecontrollingassumptionishercommitmenttotheend-relativeconception(3),asbeingtheonlynon-mysteriousviewofreason.Thethirdphase(1980s–1990s):rejecting(3);objectivemorality,objectivityrationality,andthefactsofhumanlifeFoot’smorerecentphaseismostapparentin“RationalityandVirtue”(1994)and“DoesMoralSubjectivismRestonaMistake?”(1995).Centraltoitisthereplacementofthesubjectivetheoryofpracticalreason,(3),byanobjectiveone.ThisallowsFoot354\nPHILIPPAFOOTbothtoreaffirmmorality’srationality,(2),byofferingtherestrictedsubjectivistanalter-nativeaccountofitthatismoreconvincingthanherearlier“instrumentalist”defence;andalsotousetheobjectivityofreasonstogetatwhatisreallywrongwithimmoral-ismandthussuccessfullytoconcludeherlongstrugglewithNietzsche.Footnowclaimsitisamistakeofstrategytostartfromsomepreconceivedtheoryofpracticalreason,suchasthemaximizationofperceivedself-interest,ordesire-satisfaction(cf.(3)),andthentrytoshowthatmoralactionisrationalinitsterms.Insteadtherationalityofmoralactionisonaparwiththatofself-interestedaction:theyarenotrivaltheories,butdifferentparts,ofpracticalrationality.Therearethreemainelementsinhernewposition.First,sheelucidatestheconceptofamoralvirtueasanexcellencethatensuresthatanagentisgoodinrespectofaction(andfeeling).Avirtuedoesthisbybeingadispositioncorrectlytocountcertaincon-siderationsasreasons,andthentoactonthem,thatis,todowellinrespectofactingonreasons.Soamoralvirtueisgoodnessinreason-recognitionandreason-following.Assuchitispartofwhatitisforpracticalrationalitytobeingoodorder.Giventhisformalconnectionbetweenmoralvirtueandrationality,therecanbenoquestionofwhetheravirtuousactisrational.(Ifjusticeisavirtue,thenanunjustactionwillthusbecontrarytopracticalreason.)Howevertheconnectionisonlyformal.Wearestilllefttodeterminewhatactuallyarethemoralexcellencesofactingonreasons(e.g.thatjusticeisone).Footthenarguesthatthisissettled,quiteobjectively,byfactsabouthumannatureandlife.Wereadilygrantthatthereareobjectivefactualevaluationsofwhatcountasexcellencesanddefectsinsuchfacultiesassightormemory–beitinanelephant,owl,orhuman–onthebasissolelyofthenatures,needs,andformsoflifeoftherespectivespecies(e.g.lackofgooddaysightisnotadefectinanowl).Footcallsthis“autonomousspecies-dependentgoodness.”Itappliesequallytobehavioraloperations:nest-building,huntinginpacks,etc.And,allowingforcertaindifferences,Footargues,thesamebasisofevaluationappliestothehumanoperationofactingonreasons,todetermineitsexcellencesanddefects.Becauseofwhatweareandwhatwedo,weneedsuchthingsasbeingabletobindothersbypromises,andmutualhelpfulness.Soweneedtorecognizeandfollowthereasonstheypresent.Thesegeneralfactsofhumannatureandformoflifefixwhatconsiderationsarereasonsforhumans,anddosoquiteobjectively,thatis,regardlessofwhetherornotsomeindividual(e.g.animmoralist)recognizesthem.Finally,Footextendsthesameaccounttotheotherpartofpracticalrationality,prudence.Thatconsiderationsofself-interestarereasonsisvalidatedonceagainbygeneralfactsofhumannatureandlife:thatadulthumansplanandlookoutforthemselvesbetterthanotherscanforthem.Thebasicgroundoftherationalitybothofmoralvirtuesandofprudenceisthesame,allowingustohaveinthisrespectaunifiedtheory.Thegeneralshapeisneo-Aristotelian,albeitwithdistinctiveelements.Muchiscon-troversial:thetreatmentofself-interest,thenormativeviewofhumannature,theswiftnessoftheanswertoimmoralists(whowillcomplaintheyseenoreasontobe“goodhumans”).Moreneedssaying,andFoot’sbook(forthcoming)willsaymore.Itwill,Ibelieve,offerthetwenty-firstcenturyamuchbetterstartthanMoore’sPrincipiaEthica,of1903,didtothetwentieth.355\nGAVINLAWRENCEBibliographyWorksbyFoot1970:“MoralityandArt,”ProceedingsoftheBritishAcademy56,pp.131–44.(ReprintedwithpostscriptinPhilosophyAsItIs,ed.M.BurnyeatandT.Honderich,Harmondsworth:PenguinBooks.)1978:VirtuesandVices,Oxford:BlackwellPublishers.1979:MoralRelativism,LindleyLecture,Kansas:UniversityofKansasPress.1994:“RationalityandVirtue,”inNorms,Value,andSociety,ViennaCircleInstituteYearbook,Amsterdam:Kluwer.1995:“DoesMoralSubjectivismRestonaMistake?,”OxfordJournalofLegalStudies15,pp.1–14.forthcoming:TheGrammarofGoodness,Oxford:ClarendonPress.WorkbyotherauthorsHursthouse,R.,Lawrence,G.,andQuinn,W.(eds.)(1995)VirtuesandReasons:PhilippaFootandMoralTheory,Oxford:ClarendonPress.356\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200128RuthBarcanMarcus(1921–)M.J.CRESSWELLAfterundergraduateworkinmathematicallogicwithJ.C.C.McKinsey,RuthBarcanproceededtoYaletoworkwithF.B.FitchandtoproduceaPh.D.dissertationinmodallogic.Inthemid-1960ssheestablishedthephilosophydepartmentattheUniversityofIllinoisatChicago.LatershemovedtoNorthwesternUniversityandtoYaleUniversity.ShehasbeenPresidentoftheWesternDivisionoftheAmericanPhilosophicalAssociationandoftheAssociationforSymbolicLogicandhasbeentherecipientofmanygrantsandawards,includingtheMedaloftheCollègedeFranceandanhonorarydoctoratefromtheUniversityofIllinois.RuthMarcusistheauthoroftheearliestpublishedworkinmodalpredicatelogic,thatis,modalpropositionallogicextendedwithquantifiersandpredicates.Herfirstarticleappearsinvolume11ofJournalofSymbolicLogic(Barcan1946).(RudolfCarnap’s“ModalityandQuantification”appearslaterinthesamevolume.)ItisperhapssurprisingthatalthoughmodallogicintheforminwhichweknowittodaydatesfromanarticleinMindbyC.I.Lewisin1912,itwasnotuntil1946thatthereisanycon-siderationinprintofwhathappenswhenquantifiersandpredicatesareadded.InherfirstarticleMarcusconsidersthepredicateextensionsofthesystemsofmodallogicknownasS2andS4.S2wasLewis’spreferredmodalsystem,andistheonedevelopedindetailinLewisandLangford1932.Manyofthetheoremsofmodalpredicatelogicaresimplyinstancesofnon-modalpredicatelogicorofmodalpropositionallogic,butsomearenot.OftheseanaxiomthatMarcusadoptedwas‡$xAfi$x‡Awhere‡isthemodalpossibilitysign,andfiisthesignfor“strictimplication,”andwhereAfiBisdefinedas~‡(A&~B).ArthurPrior(1956:60)called“‡$xAfi$x‡A”the“BarcanFormula”andthenamehasstuck.WhetherornottoincludetheBarcanFormulaasatruthofmodallogichasbeenamatterofmuchcontroversy.Prior(1957:26)arguedthatthetemporalversionofthisformulashouldberejectedonthegroundthatasen-tencelike“itwillbethatsomeonewillbealivein2150”doesnotentailthatthereisanyonewhowillbealivethen.(Prior’sexamplewasflyingtothemoon,butthatisnowalittleanachronistic!)Marcusherselfhasenteredthisdiscussionbydefendingtheformula.Sheprotested(inMarcus1962)againstthereadingof$xAas“thereexistsanxsuchthatA,”andproposedinstead(p.252)thatitbereadaseither“somesubstitu-tioninstanceofAistrue”or,alternatively,as“thereisatleastonevalueforxforwhichAistrue.”Suchareadingallowsthequantifiertospeakofthingswhich,inthe357\nM.J.CRESSWELLtemporalcase,nolongerexistordonotyetexist,andinthemodalcasedonotexistbutmighthave.Thealternativesareslightlydifferent.Thesecondcanbetreatedinpossible-worldssemanticsifthedomainofthequantifiersisexpandedtoincludepossibilia(thingswhichexistinotherworldsbutmaynotexistinours).ThefirstalternativeistheoneMarcusherselfhaschampioned;itistoadoptasubstitutionalinterpretationofquan-tification,whereby$xFxistrueiffFaistrueforsomeconstanta.Marcus’sviewsonthishavebeendevelopedinmorerecentwork,mostofwhichhasbeenreprintedinModalities(1993).OnthesubstitutionalinterpretationtheBarcanFormulaisuncon-troversiallytrue,sinceif‡$xFxistruethen$xFxmighthavebeentrue.Butthen,bythesubstitutionalaccountofthequantifier,someinstanceFaofFxmighthavebeentrue,andso‡Fawillbetrue,andso$x‡Fxwillbetrue.Nothingissaidhereaboutwhetherornotaexistsinthisoranyotherpossibleworld(orperhapsmoreaccuratelywhetherthename“a”referstoanything)thoughMarcusisnotunsympathetictoadefenseoftheformulaintermsofafixeddomain,andundertakessuchadefenseherself,inModalities(1993:21f.).ThereisofcourseaquestionofwhatitmeanstosaythatFaistrue.Marcus’sattitudeisthatthisisanissueabouttheinterpretationofsin-gularstatementsinvolvingnames,andnotanissueaboutquantification,andthatitisavirtueofthesubstitutionalinterpretationthatitdivorcesthesetwoquestions.ItisnolongerpossibletofollowQuineandlocatethecommitmentinthequantifiersviathesatisfactionofanopensentenceoftheformFx.InoneofMarcus’searlypapers(Barcan1947)thereisatheoremthatifxandyareidenticalthenthisisnecessarilyso.Marcusderivesthisfromthestandardprincipleinordinarynon-modalpredicatelogicwithidentitythatifx=ythenanytwoformulaethatdifferonlyinthatonehasfreexinsomeplaceswheretheotherhasfreeyareequiva-lent.Yetitseemsthatalthoughnineandthenumberoftheplanetsareidentical,fortherearenineplanets,thisidentityisnotnecessary,fortheremighthavebeenmoreorfewer.ItisnowacommonplacethatthispuzzleiseasilysolvedbyRussell’stheoryofdescriptionswithoutgivingupMarcus’stheoremaboutthenecessityofidentity,aswasshownbyArthurSmullyaninthe1948JournalofSymbolicLogic.Yetin1962,whenMarcuspresentedapaperontheroleofidentityinmodallanguages,itwas(asshesaysinherintroductiontothepaperin1993:3)amistakenassumptiononherpartthatSmullyan’spaperwasfullyappreciated.Theappendixtothispaper,basedonatapeddis-cussionbetweenMarcus,Quine,Kripke,Follesdal,andothers,makesitclearhowdiffi-cultitwastocometogripswiththeseissuesinthedaysbeforethepowerofpossible-worldssemanticsformodallogicwaswidelyunderstood.SomeofQuine’sworriesaboutquantifiedmodallogicareonthegroundsthatitleadsto“Aristotelianessentialism.”Althoughnotingthatmodalpredicatelogicisnotcommittedtoessen-tialismMarcusconcedesthatitiscompatiblewithit,anddefendsaformofessentialisminwhichthemodalitiesareunderstoodcausally(1993:67–70).Marcusisastrongsupporterofthecausaltheoryofnames.Propernames,onthisview,haveastheirmeaningnothingmorethantheobjecttheydenote,andtheyareabletohavethismeaninginvirtueofacausalconnectionbetweenaninitial“dubbing”oftheobjectandsubsequentusesofthename.(Itissomewhatunfortunatethatadisputehasgrownup,towhichneitherMarcusherselfnorKripkeisaparty,aboutthehistoricalprioritybetweenherandSaulKripkeonthetreatmentofnamesasmere358\nRUTHBARCANMARCUS“tags”ratherthanasdescriptions.WhatevermightbesaidaboutthatissuedoesnotintheleastdetractfromthevalueofMarcus’sviewsonthesematters.)Ifquantificationdependsonnaming,andifnamingdemandsacausalconnection,thenitisdifficulttoseehowquantificationcouldeverapplytothingsthatdonotexist.Thecausalviewofcoursegoesveryhappilywiththeviewthatontologyislinkedtoreference.“Actualobjectsaretheretobereferredto.Possibiliaarenot”(1993:205).Manyattemptstodefendpossibiliaappeartobebasedontheviewthatwecanrefertonon-actualobjectslikePegasusandSherlockHolmes.Marcusrightlyrejectssuchattempts.Sheisalsorightlypointsout(1993:194)thatsentencessuchas“thewingedhorsedoesnotexist,”whenanalyzedaccordingtoRussell’stheory,involvenoreferencetoapossiblebutnon-actualwingedhorse(seeRUSSELL).Butsheiswellawarethatthedifficultcasesarenotthese.Theyaresentenceslike“Theremighthavebeenmorethingsthanthereare,”wherethereisnoquestionofreferringtoanyofthem.AphilosopherwithamorerealisticattitudetopossibleworldsthanMarcusmightnodoubtsaythatanypossiblebutnon-actualobjectcanbereferredto,butofcourseonlyinaworldinwhichitexists,notinourworld.Marcus’sattitudetosuchviewsisnotsympatheticandshespeaksofsuchsemanticsforquantifiedmodallogicasprovidingitwith“adifferentsubjectmatterfromthatofnon-modallogic”andasnotbeing“astraightforwardextensionofstandardpredicatelogic”(1993:191).AthirdthemeinMarcus’sworkisconnectedwithcontradictions.Shedefendstheclaimthatmoraldilemmasarereal,butneednotthreatentheconsistencyofamoralcode.Ifwethinkofamoralcodeasasetofsentences,thenacodewillbeinconsistentifandonlyifallitsmemberscannotbesimultaneouslytrue.Inthissenseaconsistentcodemaywellallowthepossibilityofdilemmas.ForsupposethatthecodesaysthatifpthenOq(“Oq”meansthatqisobligatory)andthatifpthenO~q(itisobligatorythatnotq).Ifthisisformalizedas(pÆOq)&(pÆO~q)thenevenifwegrantthatOqandO~qarejointlyinconsistentitstilldoesnotfollowthat(pÆOq)&(pÆO~q)isinconsis-tentsince(pÆOq)&(pÆO~q)willbetrueifpisfalse.Indescribingagame,Marcussays“agamemightbesocomplexthatthelikelihoodofitsbeingdilemmaticunderanycircumstancesisverysmallandmaynotevenbeknowntotheplayers”(1993:134).IfIhaveunderstoodhercorrectly,herpointisthatalthoughitispossiblethatthemorallifemightlandusinsituationswherewecannotdotherightthing,yetmoraldilem-masmayoftenbeavoidedprovidedthattheworldcooperates.Sheendorsesasecond-ordermoralprinciple(pp.139f.)totheeffectthatweshouldsoorderourlivesthatasfaraspossibletheyareinfactavoided.Herattitudetobelievingtheimpossibleisdifferent.Animpossiblepropositionisapropositiontrueinnopossibleworld.Sothereisnopossibleworldwhichwouldbethewaythingsareifabeliefinanimpossibilityweretrue.Yetitwouldseemthatweoftendobelievecontradictions.MarcusdiscussesKripke’swell-knownexampleofPierrewhobelievesthatLondon(thecityheknowsas“Londres”)ispretty,andalsothatLondon(acityhehascometoknowas“London”)isnotpretty.Sincethisconjunctionisacon-tradictionMarcusclaimsthatPierrecannothavethisbelief(1993:158).Marcus’swork(excludingtheearlytechnicalarticles)hasbeencollectedinModalities(1993)andessaysinherhonorappearinModality,MoralityandBelief(Sinnott-Armstrongetal.1995).However,herinfluenceisnotrestrictedtoherwrit-ings,andperhapsdoesnotevenprimarilycomefromherwritings.Attheinstitutions359\nM.J.CRESSWELLatwhichshehastaughtshehasinfluencedseveralgenerationsofstudentswhohavebecomeleadingphilosophers;andherroleintheinternationalphilosophicalcommu-nityhasbeennolesssignificant.BibliographyMarcus’searliestworkislistedunderhermaidenname,Barcan.Ofherlaterarticles,thosethatarereferredtospecificallyarelistedhere;otherscanbefoundinthe1993collection.WorksbyMarcusBarcan,R.C.(1946)“AFunctionalCalculusofFirstOrderBasedonStrictImplication,”JournalofSymbolicLogic11,pp.1–16.——(1947)“TheIdentityofIndividualsinaStrictFunctionalCalculusofSecondOrder,”JournalofSymbolicLogic12,pp.12–15.Marcus,R.B.(1962)“InterpretingQuantification,”Inquiry5,pp.252–9.——(1993)Modalities,NewYork:OxfordUniversityPress.WorksbyotherauthorsLewis,C.I.(1912)“ImplicationandtheAlgebraofLogic,”Mind,newseries21,pp.522–31.Lewis,C.I.andLangford,C.H.(1932)SymbolicLogic,NewYork:DoverPublications.Prior,A.N.(1956)“ModalityandQuantificationinS5,”JournalofSymbolicLogic21,pp.60–2.——(1957)TimeandModality,Oxford:ClarendonPress.Sinnott-Armstrong,W.(1995)Modality,MoralityandBelief,EssaysinHonorofRuthBarcanMarcus,ed.D.RaffmanandN.Asher,Cambridge:CambridgeUniversityPress.(ContainsafullbibliographyofMarcus’swork.)Smullyan,A.F.(1948)“ModalityandDescription,”JournalofSymbolicLogic13,pp.31–7.360\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200129JohnRawls(1921–)NORMANDANIELSJohnBordleyRawls,whodevelopedacontractariandefenseofliberalismthatdomi-natedpoliticalphilosophyduringthelastthreedecadesofthetwentiethcentury,wasborninBaltimore,Maryland.In1939,helefthishometowntoattendPrinceton.HeservedinthePacific(1943–5),andreturnedtoPrincetontoreceivehisPh.D.(“AStudyintheGroundsofEthicalKnowledge”)in1950.HetaughtbrieflyatPrinceton,Cornell,MIT,andthen,forthirtyyears,atHarvard.HemarriedMargaretWarfieldFox,apainter,in1949.TheyraisedtwosonsandtwodaughtersandhavelivedformanyyearsinLexington,Massachusetts.Rawls’senormousinfluenceinphilosophy,law,economics,andpoliticalscienceislargelytraceabletohismajorwork,ATheoryofJustice(1971).Accordingtoonesurvey,itisoneofthefivemostcitedphilosophicalbooksofthetwentiethcentury.Incontrasttothedominantemphasisintwentieth-centuryethicsontheanalysisofmorallan-guageandontopicsinmetaethics,Theoryarguedrigorouslyforsubstantivemoralprin-ciplesanddiscussedtheirimplicationsforthedesignofbasicsocialinstitutions.(See,e.g.ANSCOMBE,AYER,FOOT,HARE,MOORE,STEVENSON;cf.CHOMSKYandPOPPER.)Thisnor-mativestandencouragedotherworkonjustice,aswellasonotherareasofappliedethics,anditexplainstherelevanceofRawls’sworkbeyondphilosophy.Rawls’sinflu-enceisalsotheresultofhisdedicationtoteaching.Hehastrainedmanyoftheleadingphilosophersinethicsandpoliticalphilosophyoverseveralgenerations.Rawls’sfirstpublication,in1951,“OutlineofaDecisionProcedureforEthics,”thoughderivingfromhisPh.D.thesis,expressesalifelongthemeinhiswork.Itpro-posesaprocedureforselectingandjustifyingethicalbeliefsandprinciplesfromamongthediverseviewspeoplehold.Ashestruggledwiththisproblemoverthenexttwodecades,henarrowedthescopeofhisproposedsolution.InATheoryofJustice,Rawlsfocusesonlyonaprocedureforselectingamongcompetingprinciplesofjustice,notmoralprinciplesquitegenerally.InTheory,Rawlsusesahypotheticalsocialcontract(theOriginalPosition)toargueforprinciplesofjusticedifferentfromtheutilitarianismthathaslongdominatedAnglo-Americanphilosophy.Deliberatingbehinda“veilofignorance”thatblindsthemtodis-tinguishingandpotentially-biasingfactsaboutthemselves,rationalcontractorschooseprinciplesthatprotectcertainbasicliberties,includingtheeffectiveexerciseofpoliti-calliberties,guaranteefairequalityofopportunity,andpermitinequalities(measured361\nNORMANDANIELSbyanindexofprimarysocialgoods)onlyiftheinequalitiesworktomakethosewhoareworstoffaswelloffaspossible(theDifferencePrinciple).TogethertheseprinciplesregulatethebasicstructureofsocietyandproduceaformofegalitarianismthatRawlscalls“democraticequality.”Becausetheseprinciplesarechoseninasituationthatisfairtoallcontractors,Rawlslabelshisview“justiceasfairness,”bywhichhemeansproceduralfairness.Inadditiontobeingtherationalchoiceofcontractors,theprinciplesmustalsomeettwootherconditions.Theymustmatchorcoherewith“our”consideredjudgmentsaboutjusticein(wide)“reflectiveequilibrium.”Theymustalsobefeasibleinthesensethatpeopleraisedinasocietygovernedbythemwouldfindthesystemtobemorestable,withlessstrainofcommitment,thanalternatives.FollowingthepublicationofTheory,therewasextensivecriticalresponseinphilos-ophyjournalsandbooks,aswellasinrelatedfields.Rawlsengagedactivelywiththiscriticalliteratureoverthenexttwodecades.InthefirstfewyearsafterTheory,hedefendedhisfocusonthebasicstructureofsociety,hiscoherentistaccountofjustifi-cation,hisuseoftheprimarysocialgoods,hisargumentfortheDifferencePrinciple,thesenseinwhichhisviewwasfairtopeoplewithdifferentconceptionsofwhatisgood,andheclarifiedthe“Kantianinterpretation”ofjusticeasfairness(see1999:chs11–15).In1980Rawlspublished“KantianConstructivisminMoralTheory”(theDeweyLectures),inwhichhecarefullydescribedthedetailsofthecontractsothatitrepresentedtheKantianideaoffreeandequalagentswhoarerationalandreasonable.Onthisconstructivistview,thereisnoclaimthatthemoralprinciplesare“true”orrep-resentapriormoralorder.Beginninginthemid-1980s,Rawlsbecameconcernedthathehadunderestimatedtheimportanceofthedivergenceamongcomprehensivemoralandreligiousviewsthatwouldemergeundertheveryconditionsoflibertypromotedbyhistheory.Couldpeoplewithsuchdivergentviewshaveastableagreementonaconceptionofjustice?Toanswerthisquestionandtoaccommodatethe“reasonablepluralism”hethoughtunavoidable,RawlsrevisedhisaccountofstabilityandpoliticaljustificationinpapersleadinguptoandincludingPoliticalLiberalism(1993).InLiberalism,hereplacedKantianconstructivismwith“politicalconstructivism,”andthesameideasaboutfreeandequalcitizensareusedtoconstructapoliticalconceptionofjustice,againwithnoexplicitclaimsaboutmoraltruth.Rawls’slastmajorworkisthusmotivatedbyacentralquestionaboutjustificationthatevolvedfromhisthesisandfirstpublication:Howcanreasonablepeoplewithdivergentmoralandreligiousviewscometoagreeonandabidebyfairtermsofcooperation?JusticeasfairnessThesocialcontractRawlsrevivesthesocialcontractasawaytospecifyfairtermsofsocialcooperationintheformofahypothetical,notanactualorhistorical,agreement.Theappealtoacon-tractembodiesthreemainideas.First,itisaformofproceduraljustice.Whenwedonothaveapriorprincipledagreementonwhatcountsasfairorjust,forexample,about362\nJOHNRAWLShowtoreconcileconcernsaboutlibertyandequalityandefficiency,wemustrelyonaprocedurethatisfairtoallparties.Wethencancounttheoutcomeofthatprocedureasfairorjust.Second,afairproceduremustembodyfeaturesthatarereasonableinlightofthenatureoftheproblemitaddresses.Inthiscase,theproblemistofindprinciplesofjusticethat“freeandequal”personscanallagreeprovidethebasisfora“well-orderedsociety.”Ascitizensorpersons,wearefreeinthesensethatwecanformandrevisearationalplanoflifethatspecifiesourconceptionofthegood.Weareequalinthatweallnormallyhaveanadequatesenseofjustice,adispositiontoseekandabidebytermsoffaircooperation.Awell-orderedsocietyisoneinwhichcitizensacceptandknowthatothersacceptthesameprinciplesofjustice,andthoseprinciplesgovernitsbasicinstitutions.Third,ifproperlydesigned,thecontractsituationrepresentsanArchimedeanPoint.Itstandsoutsidethebiasedorself-interestedbeliefswehappentohold,aswellastheentrenchedinequalitiesthatmaymotivatethem.Fromthisstandpoint,thecontractorscanleveragenewagreementsonpointsofcontroversybybuildingonrelativelyfixedoruncontroversialpoints.Thoughwemightunderstandthatanactualcontractbindsthosewhomakeit,orthosewhoimplicitlyconsenttoit,whythinkRawls’shypotheticalcontracttellsusany-thingaboutwhatweoughttodo?Theanswermustbethatweshareenoughsubstan-tivemoralagreementaboutournatureasfreeandequalpersons,thegoalofarrivingatawell-orderedsociety,andtheappropriatenessofthedesignofthecontractsitua-tionthatweacceptitasaproceduralsolutiontotheproblemofjustice.TheoriginalpositionThehypotheticalcontractinvolves“reasonable”constraintsoncontractorswhomustmakea“rationalchoice”ofprinciplesofjustice,presentedaspair-wisecomparisons.Thecontractorsarelimitedinbothknowledgeandmotivations,andthustheymustnotbeconfusedwiththe“fullyinformed”bargainersorrationalchooserswhopopu-latestandardrationalchoiceproblems.Theyoperatebehinda“thick”veilofignorancethatblindsthemtoinformationabouttheirage,race,gender,classposition,thesocietytheywillenteranditspositioninhistory.Theyarealsoblindtotheir“rationalplanoflife”ofconceptionofthegood,includingtheirsystemofmoralandreligiousvalues.Thisthickveilassuresthattheirchoiceofprincipleswillnotbeaffectedbytheself-interestthatmightcomefromknowledgeofanyofthesefactsaboutthemselves.Contractorsdo,however,havegeneralsocialknowledgeortheywouldnotbeabletoevaluatethechoiceofprinciplesfortheireffectsonwell-being.Theirmotivationsarealsoconstrained.Theyare“mutuallydisinterested,”meaningtheyareconcernedabouttheirownwell-beingandthewell-beingofthoseinagenerationortwoeitherside,buttheyarenotgenerallybenevolent,malevolent,orenvious.Havingblindedcontractorstotheirowndetailedor“thick”conceptionofthegood,Rawlsmustprovidethemwithabasisfordetermininghowonesetofprincipleswillmakethembetteroffthananother.Otherwisethereisnobasisforarationalchoiceofprinciples.Rawlsintroducesasetof“primarysocialgoods,”containingrightsandlib-erties,powersandopportunity,incomeandwealth,andthesocialbasesofself-respect.363\nNORMANDANIELSAweightedindexoftheseobjectivelymeasurablegoodsisthebasisformeasuringtheeffectsofalternativeprinciplesonwell-beingandformeasuringinequalitiesbetween(representativemembersof)socialgroups.Consequently,therationalchoiceproblemfacingcontractorsistodecidewhichoftwoalternativeprinciplesunderconsiderationleadstothehighestindexscoreforthem.Togetherwiththerequirementthatallcontractorshavevetopoweroverchoices,theresultofthe“reasonable”constraintsistoestablishabaselineofequalitythatelimi-natestheinfluenceofentrenchedsocialinequalities.PrinciplesofjusticeRawlsarguesthathiscontractorswouldchoosethreeprinciplesofjusticeinpreferencetoutilitarianism.HisFirstPrincipleassurescitizenstheywillhaveasetofequalbasicliberties,includingfreedomofthought,expression,andassociation,securityoftheperson,andrightsofpoliticalparticipation.Rationalcontractorsknowthattheymayhavefundamentalmoralandreligiouscommitments,evenifbehindtheveiltheydonotknowexactlywhattheyare.Oncetheyreachsomemodestthresholdofmaterialwell-being,havingthelibertytopursuethosecommitmentsisnotsome-thingtheywouldtradeforincrementsinincomeandwealth.Recognizingothersandbeingrecognizedbythemaspoliticalequalsisanimportantsocialbasisofself-respectaswell.Soimportantaretherecognitionalaspectsoftheeffectiveexerciseofpoliticalpar-ticipationrightsthatRawlsarguesforspecialinstitutionalprotectionsofthem.Theseinstitutionalprotections,suchaspublicfundingofelections,areintendedtomakesurethatpoliticalparticipationrightsarenotmerelyformalbutactuallyeffectivelyexercis-ablebyall,regardlessofotherinequalities.Forthesereasons,contractorswouldassurethemselvesthesebasiclibertiesdirectlythroughtheFirstPrincipleratherthandepend-ingforthemontheoutcomeofanuncertainutilitariancalculation.Rawls’sSecondPrincipleactuallyconsistsoftwootherprinciples.Thefairequalityofopportunityprinciplenotonlyprohibitslegalandquasi-legalbarrierstoopportu-nity,aswouldtheweaker“formal”equalityofopportunityprinciple,butalsorequiresthatpositivestepsbetakentomitigatetheeffectsofsocialandeconomiccontingenciesonthedevelopmentsofthetalentsandskills.Thesestepsminimallyincludemeasuressuchastheprovisionofpubliceducation,butfurtherearlychildhoodinterventionsandfamilysupports,suchasdaycare,mightbenecessarytosupportfairequalityofoppor-tunityforbothchildrenandwomen.BoththeFirstPrincipleandthefairequalityofopportunityprinciplerequirecertainkindsofequality.TheyaregivenpriorityovertheDifferencePrinciple,whichallowscertaininequalities.TheDifferencePrinciplerequiresthatinequalities,forexampleinincomeorwealth,beallowedonlyiftheyworktomaketheworst-offgroups(andthenthenextworstoff,etc.)aswelloffaspossible.Theideaisthatitwouldbeirrationalforcontractorstoinsistonequalsharesofasmallsocialproduceifincentiveswouldcreatealargersocialproductthatcouldbedividedsothatall,eventheworstoff,benefitbygettingmore(accordingtotheindexofprimarygoods)thantheywouldwithoutincentivesandinequalities.Sofar,thisargumentestablishesonlythatitisirrationaltodisallowinequalitiesthatadvantageall.TheDifferencePrinciple,however,isverydemanding364\nJOHNRAWLSinthatitrequiresinequalitiestomaketheworstoffaswelloffaspossible(itismuchmorethan“trickledown”).CrucialtoRawls’sargumentfortheDifferencePrincipleishisclaimthattheveryhighstakes(lifetimeprospectsofwell-being)andthegreatuncertaintyimposedbytheveilofignorancemeanthata“maximin”(maximizetheminimum)principleisrequiredasaprincipleofrationalchoice.Contractorsarenotpermittedtogamblethattheyhaveanequalchanceofbeinginanysocialposition,agamblethatwouldmaketheprincipleofaverageutilitypreferabletothem.Rather,themaximinprinciplerequirestheymaximallyprotecttheworstoffthroughtheDifferencePrinciple.TomaketheDifferencePrincipleseemlessodd,Rawlsalsoarguesintuitivelythatbothitandthefairequalityofopportunityprincipleworktomitigatetheeffectsofmorallyarbitrarysocialcontingencies.Utilitarians,Rawlsnotes,believethatinpursuingtheaggregatewelfare,theadvan-tagesofsomeoutweighthelossesofothers,muchastheexpenditureofeffortinacquir-ingskillsatonestageoflifewillbeoffsetbygreaterrewardsatanother.Rawls’sprinciplestogetherbetterrecognizethe“separatenessofpersons.”Theyaffordstrongerprotectionstoindividualssothattheadvantagesofsocialcooperationworkmoredirectlytothebenefitofall.BasicstructureRawlsintendshisprinciplesofjusticetoregulatethebasicstructureofsociety,thatis,thosemajorsocialinstitutions,suchasthepoliticalconstitutionandtheprincipaleco-nomicandsocialarrangements,thathave“profound”effectsonpeoplebecausetheydistributebasicrightsanddutiesanddeterminethedivisionofadvantagesfromsocialcooperation.Inhislaterwork,heexplicitlyincludesthefamilyinthebasicstructure.Theprinciplesofjusticedonotdirectlyapplytotherelationshipsindividualshavewitheachotherorinprivateassociations.Rawlssuggeststhereisanimportantdivisionofmoralresponsibility:societyassuresthatcitizens’needsaremetthroughtheprinciplesofjustice,whichregulatebasicinstitutions.Individualsareresponsibleforpursuingtheirrationalplansoflifewithintheconstraintsimposedbyjustice.PrimarysocialgoodsRawls’srejectionofutilitarianmeasuresofwell-being,suchaswelfareordesiresatis-faction,infavorofanindexofprimarysocialgoods,waschallengedonseveralgrounds.Theindexseemsincomplete,foritfailstotelluswhoisworseoff,therichbutillpersonorthepoorbutwellone.Moregenerally,individualvariations,suchasthosecausedbydiseaseordisability,wouldmeanthatindividualswiththesameprimarysocialgoodswouldactuallyhavequitedifferentcapabilities.SomeconcludethatRawls’sfocusonthe“resources”includedintheindexofprimarysocialgoodsmeansheisconcernedaboutthewrong“space”;egalitariansshouldfocusmoredirectlyonindividuals’“opportunityforwelfareoradvantage”oronthecapabilitiesorpositivefreedomtheyhave.InTheory,Rawlshadmadethesimplifyingassumptionthatallindividualswerefullyfunctionaloveranormallifespan.Thisassumptioninvitedtheseobjections:thattheindexofprimarygoodswasinsensitivetoimportantindividualdifferences,suchasdiseaseordisability.Byviewingdiseaseanddisabilityasimpairmentsoftherange365\nNORMANDANIELSofopportunitiesopentopeople,itispossibletoextendRawls’stheorytoincludeproblemsofhealthanddisease,andRawlsendorsessuchanextensioninhislaterwrit-ings.Quitesurprisingly,Rawls’sthreeprinciplesthenconstituteafairdistributionofthemajor“socialdeterminants”ofhealth,accordingtocurrentworkinthesocialsciences.Inhislaterwritings,Rawlsalsorepliestotheobjectionthattheprimarysocialgoodsmightnotbevaluedinthesamewaybypeopleholdingquitedifferentconceptionsofthegood.Hereformulatesthemascrucial“allpurposemeans”formeetingtheneedsofcitizens.Thisreformulation,togetherwiththeextensiontohealthcare,makeshisviewconvergemorewiththatofsomeofhiscritics,foritsuggeststhatjusticeasfairnessisaimedatguaranteeingthatallcitizens’“needs”–ascitizens–aremetandthatthereforetheyallhavethecapabilitiestofunction–ascitizens–asfreeandequal.WidereflectiveequilibriumTheprinciplesthatcontractorschoosemustmatchourconsideredjudgmentsaboutwhatisjustin“reflectiveequilibrium.”Toachieve“reflectiveequilibrium,”weworkbackandforthbetweenourconsideredjudgmentsaboutparticularinstancesorcases,theprinciplesthatgovernthem,andthetheoreticalconsiderationsthatbearonaccept-ingtheseconsideredjudgmentsorprinciples,revisinganyoftheseelementswherevernecessaryinordertoachieveanacceptablecoherenceamongthem.ForRawls,thismeansweshouldrevisetheconstraintsonchoiceintheOriginalPositionuntilwearriveatacontractthatyieldsprinciplesthatareinreflectiveequilibriumwithourcon-sideredjudgments.Thus,inhisearlywork,themethodofreflectiveequilibriumplaysaroleinboththeconstructionandjustificationofhistheoryofjustice.Thereseemstobelittlejustificatoryforcetoachievingcoherenceorreflectiveequilibriumsolelyamongprinciplesandjudgmentsaboutparticularcases.Unlesswethinkwehavespecialknowledgeofeithertheprinciplesorjudgments,whichRawlsdoesnot,suchanarrowreflectiveequilibriumcapturesonlywhatwehappentothinkisjust.Itdoesnotshowusthatwearejustifiedinholdingthoseparticularbeliefs.Rawlsbelieveswehavenobettermethodofjustificationthanseekingawidereflectiveequilibrium.Thismethodbroadensthefieldofrelevantmoralandnon-moralbeliefstoincludebothanaccountoftheconditionsunderwhichitwouldbefairforreasonablepeopletochooseamongcompetingprinciples,andevidencethattheresult-ingprinciplesconstituteafeasibleorstableconceptionofjustice,thatis,thatpeoplecouldsustaintheircommitmenttosuchprinciples.Ourbeliefsaboutjusticearejusti-fied(and,byextension,wearejustifiedinholdingthem)iftheycohereinsuchawidereflectiveequilibrium.Centraltothemethodofreflectiveequilibriumistheclaimthatourconsideredmoraljudgmentsaboutparticularcasescarryweight,ifonlyinitialweight,inseekingjusti-fication.Vigorouscriticismofthisclaimcomesfromutilitarians,whodenouncesuch“intuitions”astheresultsofhistoricalaccidentandbias.Since,however,utilitariansallowindividuals’desiresorpreferencestocountincalculatingwhatisgoodandright,andthesedesiresarealsoaffectedbyhistoricalaccidentandbias,Rawls’sopennessaboutexposingmoraljudgmentstocomprehensivecriticisminreflectiveequilibriummaybelessharmfulthantheutilitarianapproach.366\nJOHNRAWLSInATheoryofJustice,Rawlsseemedtothinkthatallpeoplemightconvergeonacommonorsharedwidereflectiveequilibriumthatincluded“justiceasfairness.”Wewouldbeledtothatequilibriumbyphilosophicalargumentaboutthevariousmoralbeliefsthatcontributetothesocialcontractapproach,thedetailsoftheOriginalPosition,andtheargumentsmadewithinit.Inhislaterwork,Rawlsmodifiesthisview(see“Justificationrevisited”below).StabilityandfeasibilityPrinciplesofjusticemustnotonlybechosenbycontractorsandmatchourjudgmentsinreflectiveequilibrium,buttheymustbemorestablethanalternativeviews.Peopleraisedinawell-orderedsocietymustfindthatconformingtotheminvolveslessstrainofcommitmentthanconformitywithalternatives.Forexample,theworstoffarguablywouldfindthestrainofcommitmentlessundertheDifferencePrinciple,whichmakesthemaswelloffaspossible,thantheywouldunderautilitarianprinciplethatsimplymaximizedaggregateoraverageutility,assumingbenefitstootherscompensatedthemfortheirlosses.Becausetheautonomyexercisedenabledbytheprinciplesofjusticewouldbeviewedasagoodbypeople,Rawlsthoughthisviewstable.AgrowingrespectforpluralismledRawlstorevisethisargumentforstability.JusticeaspoliticalBurdensofjudgmentandreasonablepluralismReasonablepeople,especiallyunderconditionsinwhichtheyenjoybasicliberties,willtendtodevelopdivergentcomprehensivephilosophicalandreligiousviewsthroughwhichtheyassesswhatisvaluableinlife.By“reasonable,”Rawlsmeanspeoplewhoareconcernedtolivewithothersonfairterms,assumingthattheothersaresowilling.Reasonablepeoplealsounderstandthattobefairthetermsofcooperationmustbeonesthatotherfreeandequalpersonscanaccept.Reasonablepeoplewillrecognizethatdis-agreementsariseamongthembecauseofthe“burdensofjudgment.”Theseburdensincludetheconflictingandcomplexevidencethatbearsonissues,thedisagreementsabouthowtoweightconsiderations,thevaguenessofsomeofourconcepts,theeffectsofthetotalityofaperson’sexperienceonhowsheweightsconsiderations,themulti-plicityofnormativeconsiderationsthatarerelevantandfromwhichaselectionmustbemadeinanyspecificcase.Wearedriven,Rawlsconcludes,toacceptreasonablepluralismaboutmanymattersofimportance.Thisisabasicfactofpoliticallife,andevenamongreasonablepeoplewewillfinddisagreementsthatthreatentheoriginalsuggestionthatphilosophicalargumentcouldproduceconvergenceonthesamewidereflectiveequilibrium.OverlappingconsensusRawlsaddressestheproblemofproducingstableagreementdespitereasonablepluralismbyrecastingjusticeasfairnessasa“free-standing”politicalconceptionofjustice.Thekeyideasoutofwhichjusticeasfairness(orother,alternativereasonable367\nNORMANDANIELSpoliticalconceptionsofjustice)areconstructed,forexample,theideathatcitizensarefreeandequal,arenowtakentobesharedelementsofourpoliticallife,thatis,ofapublic,democraticculture.Theseideasarealreadyheldoracceptedbymostpeoplewhosharethatculture,whateverotherviewstheydivergeon.Ineffect,itisnotphilosophyalone–aidedbyuniversalreason–thathasledpeopletoconvergeontheseideas,butasharedsetofinstitutionsandhistory.Theappealtoashareddemocraticculture,however,isnotaconcessiontothe“communitarian”criticsofRawls,whohadcom-plainedthatasharedconceptionofthegoodmustunitepeopleandformthebasisforjustice;instead,itisawayforRawlstoseekagreementamongthosewhodisagreeaboutsuchviewsofthegood,amongotherthings.Rawlssuggeststhatwethinkofthepoliticalconceptionofjusticeasfairnessasa“module”withitsowninternalprinciples,reasons,andstandardsofevidence.Forexample,justiceasfairnessincludesthetwoprinciplesofjusticeorderedinaparticu-larway.Togethertheseorderedprinciples,illuminatedbythesharedbackgroundideasandpubliclydefendablestandardsofevidenceandreasoning,specifythecontentof“publicreason”asitisusedtodeliberateaboutmattersofjustice.Thismoduleshouldbecomplete:itshouldgive“reasonable”answerstoabroadrangeofquestionsabout“constitutionalessentialsandbasicquestionsofjustice.”Theseanswersare“reasonable,”however,inlightofthekindsofreasonstowhichthepoliticalconceptionisrestricted.Ineffect,thejustificationfortheseanswersonlygoessofar.Itappealsonlytoreasonscontainedinthepublicview.Rawlscallsthis“protantojustification.”Peoplewithdivergentcomprehensivemoralandreligiousviewscanoverlapintheiracceptanceofaconceptionofjustice,themostreasonableofwhichRawlsthinksisjusticeasfairness.Hedrawsananalogytothesametheorem’sbeingprovablewithindifferentaxiomaticsystems.Nothingthatturnsonthecomprehensiveviewsplaysaroleinthepublicjustificationofthemodule.Noclaimsaboutmoraltruthandnospe-cificmoralorphilosophicalviewsthatarethedistinctivefeaturesofsuchcomprehen-siveviewsplayarole.Overlappingconsensusisnotacompromiseormodusvivendiamongcompetinggroupsthatholddifferentmoralconceptions.Publicjustificationoftheviewmustbeforthe“rightreasons”andturnontheacceptabilityofthemoduleandtheideasitrestsontothosewhoholdthoseviews.PublicreasonAnideathatbecomescentralinRawls’slaterworkisthatof“publicreason.”InPoliticalLiberalism,Rawlsarguesforaratherrestrictiveviewofpublicreason,attemptingtorestricttheintroductionofreligiousandotherviewsintopublicdebate,especiallybypublicofficialsandeveninthethinkingofcitizensastheyvote.Thisviewwaswidelycriticized,andRawlsadoptsamorerelaxed,“wide”viewofpublicreasoninhislastpaperonthetopic,“TheIdeaofPublicReasonRevisited.”Inthewideview,deliberationsaboutjustice,especiallybypublicofficials,aregovernedbya“proviso.”Reasonablecomprehensivemoralorreligiousdoctrinesmaybeintroducedintopublicpoliticaldiscussionatanytime,andtheremaybegoodreasonsfordoingso,provided368\nJOHNRAWLSthatproperpoliticalreasonsarealsoofferedthataresufficienttoestablishthesamecon-clusions.Thisprovisoappliestopublicpoliticalcultureandnottodebateintheback-groundculture,whichhasnosuchrestrictiononit.JustificationrevisitedTosaythataclaimaboutwhatisjustisjustifiedsolelybypublicreason(orprotanto)isnotyettosaythatitisafullyjustifiedbeliefforaparticularperson.Thecriterionforfulljustificationultimatelyremainsacceptabilityinwidereflectiveequilibrium,andprotantojustificationdeliberatelyrefrainsfromseekingsuchdeeperjustification.Bynotseekingoralludingtodeeperjustification,protantojustificationdoesnotalienatethosewhohavedifferentreasonsforacceptingthemodule.Weobtainthegreateststabilitywecanforapoliticalconceptionofjustice,Rawlsargues,answeringhiscentralquestioninLiberalism,whenthereistherighttypeof“overlappingconsensus”onit,thatis,whenthereisoverlappingconsensusfortherightreasons.Peoplewithdifferentcomprehensivemoralviewsmustjustifyforthemselves,bytheirownlights,thatis,intheirownwidereflectiveequilibria,theacceptabilityofthemodule.Theirrationaleswillthusdifferinwaysthatreflecttheirotherphilosoph-ical,moral,andreligiousbeliefs.Somemayinsist,forexample,thatthereis“moraltruth,”othersdenyit.Somemightseetheprinciplesofjusticeasformsofdivinelygivennaturallaw;othersmayseeitasahumanconstruction.Ultimately,peoplearejustifiedinacceptingjusticeasfairnessifitisacceptabletotheminthedifferentwidereflectiveequilibriatheycanachieve.Ifthereisgeneralacceptanceinthiswayofthemodulewithinthedifferent“rea-sonable”comprehensiveviewsinasociety,Rawlssaysthatwehave“general”reflectiveequilibrium.Generalreflectiveequilibriumisnotasharedwidereflectiveequilibrium–exceptfortheoverlaponthemodule.CurrentapplicationsandcontroversiesAtthebeginningofthetwenty-firstcentury,Rawls’sworkcontinuestostimulateextensivediscussioninseveralveryactivefieldsofpoliticalphilosophy,includingthefollowing:ThefamilyandfeminismInhislatestwritings,Rawlsemphasizesthatthefamilyshouldbeincludedinthebasicstructureandthusberegulatedbyprinciplesofjustice.Atthesametime,concernsaboutequalitymustbereconciledwithconcernsaboutthelibertyoffamiliestopursuereligiousormoralviewsthatinvolvegenderroledifferentiationaffectingmothersandchildren.Rawlsimaginesrobustinstitutionalprotectionsofwomen,includingdaycareandotherfamilysupportsystems;atthesametime,heimaginesthedebateaboutgenderrolestobecarriedoninthebackgroundcultureandnotthroughintrusionsintothefamily.Nevertheless,Rawls’semphasisonprinciplesofjusticeshouldbecon-trastedwithvirtue-basedfeministapproachestoethics.369\nNORMANDANIELSEgalitarianismAsubstantialbodyofegalitarianliteraturehasarisenthatchallengesRawlsfromvariousdirections.OneprominentviewsuggeststhatRawls’sintuitiveargumentsfor“democraticequality,”whichappealtothemoralarbitrarinessofsocialandnaturalcontingencies,yieldamoreegalitarianviewthanisembodiedinRawls’sprinciples.Onthisview,weareowedcompensationforanydeficitinopportunityforwelfareoradvan-tagethatarisesthroughnofaultorchoiceofourown.AnotheregalitarianchallengeisthattheleewayRawlsallowstoindividualstopursueincentivesortomakeselfishdomesticchoiceswillunderminethepossibilityofachievingoptimalresults,asjudgedbyRawls’sownprinciples.InternationaljusticeAnearlychallengetoRawlswasthathefailedtodiscusswhatkindsofobligationsofjusticeareowedacrossnationalboundaries;arelatedcriticismisthathistheoryistoomuchwedtotheideaofnationstates.Rawls’slatepublicationson“TheLawofPeoples”extendhiscontractarianviewstodiscusshumanrights.DemocraticdeliberationTheviewofdemocracythatemergesinRawlsemphasizespoliticalparticipationasawayofrealizingourmoralcapabilities.Democracyisnotsimplyaproceduralmethodofachievingagreementthatwemustemploybecausewelacksubstantiveagree-mentonvariousmatters.Instead,Rawlsprovidesafoundationforemphasizingthedeliberativeelementsofdemocratictheory.BibliographyWorksbyRawls1971:ATheoryofJustice,Cambridge,MA:HarvardUniversityPress.1993:PoliticalLiberalism,NewYork:ColumbiaUniversityPress.(Paperbackedition(1996)con-tainsanewintroductionand“ReplytoHabermas,”fromJournalofPhilosophy92/3(March1995).)1999:CollectedPapers,ed.S.Freeman,Cambridge,MA:HarvardUniversityPress.WorksbyotherauthorsDaniels,N.(ed.)(1975)ReadingRawls,NewYork:BasicBooks.Freeman,S.(ed.)(forthcoming)CompaniontoRawls.Reath,A.,Herman,B.,andKorsgaard,C.M.(eds.)(1997)ReclaimingtheHistoryofEthics:EssaysforJohnRawls,NewYork:CambridgeUniversityPress.SymposiumonRawlsianTheoryofJustice:RecentDevelopments,Ethics99/4(July1989).370\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200130ThomasS.Kuhn(1922–1996)RICHARDGRANDYThomasS.Kuhn’ssecondmonograph,TheStructureofScientificRevolutions(1962)isthemostwidelyreadandmostinfluentialbookonthephilosophyofscienceofthetwentiethcentury.Itspawnedtheubiquitoususeoftheterm“paradigm”inpopularculture,includingcartoonsandbusinessmanagementcourses,andamillioncopieshavebeensoldinalmosttwentylanguages.Thecentralthesisofthebookisthatthenatureofscientificdevelopmenthadbeenseriouslymisunderstoodbyphilosophersandscientists,andthat,inthewordsoftheopeningsentence:“History,ifviewedasarepositoryformorethananecdoteorchronology,couldproduceadecisivetransformationintheimageofsciencebywhichwearenowpossessed.”Theimagehesoughttotransformwasoneinwhichscienceiscumulative,varyinginthespeedofitsprogress,butalwaysmovingforward,animageinwhichscientificcontroversiesareasmallandunimportantpartoftheprocess,frictioninthewheelsofprogress.Thecontrastingimagehechampionedportraysmaturesciencesasalternatingbetweentwokindsofchange.Thefirstareperiodsofcumulativeprogressinwhichsci-entistsapplygenerallyacceptedtheoriestotheunresolvedquestionsinadomainaccordingtoasharedunderstandingofwhatconstitutesareasonablescientificques-tionandofwhatcriteriaareusedtojudgeanswers.This“normalscience”isaverysophisticatedformofpuzzle-solvingandcanrequiregreatingenuity,butoccurswithinastableframeworkoftradition.Incontrast,thealternatingperiodsof“revolutionaryscience”consistofconfrontationbetweentwodiverseunderstandingsofwhatconsti-tutesareasonablequestionandwhatcriteriashouldbeusedtoadjudicatedisputes.InTheStructureofScientificRevolutionsKuhnusedtheterm“paradigm”tobothdefineandexplainthedifferencebetweenthetwokindsofscience:normalscienceconsistsoftheelaborationofanacceptedparadigm,whilerevolutionaryscienceconsistsoftheover-throw,orattempttooverthrow,anacceptedparadigm.Inaddition,thenotionofpara-digmplayedanessentialroleindistinguishingprescientificpreludestoascience,forexampleopticsbeforeNewton,becausethecriticalstepinmakingthetransitiontoascienceconsistedofconvergencebyascientificcommunityonaparadigm.Reactionstothebookbyphilosophersandnaturalscientistswerenumerous,vociferous,andalmostallnegative.SomecriticssaidthatKuhnmademostofscience–normalscience–seempedestrianandalmostunnecessary,inspiteofhisclear371\nRICHARDGRANDYinsistencethatitwasonlythepersistentpursuitofpuzzlesbyfirst-ratemindsthatwouldeventuallygeneratetheanomalieswhichwouldleadtorevolutionaryscience.Butthemajorityofcriticsfocusedinsteadontheaccountofrevolutionaryscience;accordingtomany,theprocessesofrevolutionarychangeasdescribedbyKuhncon-stitutedirrationalmobruleandwereantitheticaltotheviewofscienceastheepitomeofreason.Muchoftheattentioncenteredontheclaimthatopposingparadigmsareincommensurable,thatthemeaningsandoftenthereferentsofthetermsofthetheo-riesdiffersothatnodirectsimplecomparisonbetweenthemispossible.Thereactionamongsocialscientistswasmoremixed;someembracedthecentralthemesandbecameobsessedwithwhethertheirfieldhadyetcompleteditspreparadigmaticpreparationforsciencehood.Theanalyticapparatusofthebook,especiallythecentralnotionofaparadigm,cameinforparticularlyseverescrutiny.Onereviewerdiscernedtwenty-twodistin-guishablesensesof“paradigm”inthetext.Kuhnwasastoundedatwhathesawaswidespreadmisunderstandingandmisrepresentationofhisideas,butrecognizedtheneedtoclarifythecentralnotionofparadigmandrelatedapparatus.Thefirstfruitionofthisrethinkingappearedinanumberofpapersduringthelate1960sandina“Postscript,”whichwaspublishedinthesecondeditionofStructurein1970.Beforeelaboratingonthemodificationinthe“Postscript,”itwillbeusefultosketchsomeofthepathbywhichhereachedtheviewsbehindStructure.Kuhn’sstatusasaphilosopherisdifficulttoassessbecausehistraining,career,andindeedthenatureofhisinfluenceareveryunusual.ThomasS.KuhnwasborninCincinnati,Ohioin1922andreceivedhisBachelor’s(1943),Master’s(1946)anddoctoraldegrees(1949)fromHarvardUniversityinphysics.HeonlybegantoreadseriouslyinthehistoryofsciencewhenaskedbyJamesB.Conant,thenPresidentofHarvard,toassistinpreparingahistoricallyorientedundergraduatesciencecoursefornon-sciencestudents.Apivotalmomentoccurredin1947whilehewasreadingAristotle,tryingtoascer-tainhowmuchmechanicsAristotleunderstood.HisconclusionwasthatAristotleunderstoodlittleornomechanicsandindeedseemedtobeapoorobserverandun-systematicscientist.Hewaspuzzledbyhowoneofthegreatestintellectsinthehistoryofwesternthoughtcouldhavebeensoconfused.Then,suddenly,“thefragmentsinmyheadsortedthemselvesoutinanewway,andfellintoplacetogether.Myjawdropped,forallatonceAristotleseemedaverygoodphysicistindeed,butofasortI’dneverdreamedpossible”(Thalheimerlecturep.32).Amongthepiecesthathadsortedthem-selvesoutwastheinsightthatforAristotle,theGreekexpressionthatistranslatedas“motion”meansnotonlyachangeoflocation,butanyofawidevarietyofchanges,ofwhichchangeoflocationisonlyone.Lookingattheworldinthisnewwaywiththistransformedvocabulary,Aristotelianmechanicsmadeverygoodsenseofmanyobser-vations,albeitmanyofthoseobservationswouldnotberegardedasrelevanttomodernmechanics.Aftercompletinghisdissertationinphysics,hespentthreeyearsasamemberoftheHarvardSocietyofFellowsbroadeninghishistoricalandphilosophicalknowledge.ThenensuedanappointmentteachinghistoryofscienceatBerkeleyinwhichmuchofhistime,byhisownobservation,wasspentpreparinglecturesinafieldinwhichhehadnoformaltraining.In1957hepublishedTheCopernicanRevolution,awell-received372\nTHOMASS.KUHNaccountoftheconceptualandtechnicalobstaclestomakingthetransitionfromageo-centrictoaheliocentricuniverse.ThecentralideasofStructurearediscernibleinthisfirstbook,buttheclaimsaremuchnarrowerandgenerallylessphilosophical.ThereisconsiderablefocusontheideathatthetransitionsfromtheAristotelian–Ptolemaicuni-versetotheCopernican–Galilean–Newtonianonearenottransitionsthatcanbearrivedatbysmallincrementalsteps.Toseetheuniverseascenteredonthesun,notontheearth,canonlybeaccomplishedasadramaticchange.Butnowiderclaimsaremadeinthefirstbookabouthowwidespreadthiskindoftransformationhasbeeninthehistoryofscience.Structurerepresentedthegeneralizationofthatideatothelargercanvasofthephysi-calsciencesgenerally.TheprefacetoStructureindicatesinsomedetailtheextenttowhichheisawarethatthereareseriousgapsandshortcomingsinthephilosophicaldevelopmentofkeyconcepts.However,hehadcontractedtoproduceamonographwithinfairlyseveresizelimitsandtheeditorswerepressinghimtocompletethemanuscript.OnelittleknownironicalaspectofthepublicationofthebookisthatalthoughthelogicalpositivistconceptionofscienceisaprimarytargetofKuhn’scriticisms,themonographwasfirstpublishedasVolumeII,no.2,ofTheEncyclopediaofUnifiedScience,thepublishingorganofthelogicalpositivistmovement.Theeditorsresponsibleforsolicitingandencouragingthemanuscript,CharlesMorrisandRudolphCarnap,wereenthusiasticaboutthemonographanditsimportanceforphilosophyofscience.ThesecondeditionofthebookindicateditsstatusaspartoftheEncyclopedialesssaliently,andbythethirdeditionin1996nomentionismadeoftheoriginalimprimatur.In1964KuhnmovedfromBerkeleytoPrinceton,becomingamemberofthehistorydepartmentbutalsojoiningthegraduateprograminthehistoryandphilosophyofscience.Inseminarsthere,aswellasinlecturesandcorrespondence,herevisedandclarifiedtheideasofStructure.Inparticular,theuseof“paradigm”wastobereplacedbyeither“disciplinarymatrix”or“exemplar,”thusrecognizingamajortwo-foldambi-guityintheoriginalterm.Adisciplinarymatrixconsistsofsymbolicgeneralizations,metaphysicalassumptions,models,values,instruments,andexemplars.Thusadis-ciplinarymatrixisaconstellationofelementswhichdefineaworld-viewandcharac-terizeascientificcommunity.Sinceadisciplinarymatrixcontainsmanyelements,therecanbevaryingdegreesofcongruenceamongmembersofacommunity.Thesymbolicgeneralizationsarethemostfamiliarelement;thesewouldbeequa-tionssuchasNewton’sLawsorBoyle’sLaw.Metaphysicalassumptionsconcernthebasicelementsoftheuniverse;exampleswouldbetheassumptionthatavacuumisimpossible,thataction-at-a-distanceisimpossible,thattheuniverseisgovernedbydeterministiclaws,andthatallmatterconsistsofatomsinavoid.Modelsareeasiertoillustratethandescribe:themodeloftheatomasaminiaturesolarsystem,themodelofagasasacollectionofalargenumberofverysmallparticlesinrapidmotion,heatasafluid,andsoon.Valuesincludesimplicity,generalityorscope,accuracy,repro-ducibilityofresults.Exemplars,whicharebothanelementofdisciplinarymatrices,butalsoasignificantsecondsenseofparadigm,areexamplesofnotablescientificaccomplishmentwhichsetastandardforfutureresearchers.Forexample,therigorofEuclid’sgeometrywasanexemplarformanydisciplines,andthenumberoffieldsthathaveproclaimed373\nRICHARDGRANDYCopernicanrevolutionsislegion.Morerecentexamplesmightbethepredicteddiscov-eryofUranusorthediscoveryofthedoublehelicalstructureofDNA.Theexemplarsprovideagluefortheelementsofadisciplinarymatrixbybringingtogethertheexam-plesinconcreteaccomplishments.ItisanessentialpartoftheKuhnianpicturethattheexamplescanbeextendedinvariousways,sothattheexemplarsprovideguidancebutnotrules.AmajoremphasisinKuhn’sdiscussionofscientificchangewasthesuddenandinvoluntarytransformationofperceptionandbelief.ThisclearlyoriginatedinhisownexperienceinunderstandingAristotleandheillustrateditbygivingexamplesofGestaltswitchfiguresinhisbook,forexample,alinedrawingwhichcanbeseeneitherasanoldwomanfromoneperspectiveorayoungwomanfromanother.Theseideasbio-graphicallystemmedbacktoKuhn’sexperiencein1947,butothershadalsobeenstruckbysimilarideas,andanarticulatepresentationofthemcouldbefoundinN.R.Hanson’sPatternsofScientificDiscoveryfouryearsbeforetheappearanceofStructure,whichrefersapprovinglytoHansoninanumberofplaces.Kuhnwasparticularlyperplexedandfrustratedbytheaccusationthathewasunderminingtherationalityofscience.Hestronglybelievedthatscienceisanepitomeofrationality,andthustheprocessesinvolvedinthedevelopmentofscience,includingbothnormalandrevolutionaryscience,mustbeessentialingredientsintherational-ityofscience.Hisgoalinoverthrowingtheacceptedimageofscientificprocesseswastocastasideafalseunderstandingofrationalityandtobegintheprocessofreplacingitwithamoresophisticatedandhistoricallyaccurateapprehension.HemadeonestrategicdecisionincompletingthemanuscriptofStructureandpub-lishinghisideasinthatabbreviatedandhighlyunfinishedform.Hemadeaseconddeci-sioninthelate1960stopublishthe“Postscript”attheendofthesecondeditionofStructure,ratherthanattemptingathoroughrevisionthatwouldsystematicallyreplacetheoccurrencesoftheambiguous“paradigm”withtheappropriatetermfromthevocabularyofdisciplinarymatricesandexemplars.Thismeantthatevenafter1970newreadersofStructurebecameawareoftheextensiveterminologicalandconceptualchangesonlyafterreadingtheoriginal170-pagetextandbeingthoroughlyimmersedinthesweepingandambiguousvocabularyof“paradigms.”ThisdecisionwastheresultofKuhn’srecognitionthatreworkingStructurewasnotaverygoodoptionsincehewasstillinthemidstofchanginghisviews,andsothe“Postscript”strategywasastopgapuntilhecouldreachthestagewhereanewandmorethoroughbookwasprepared.Duringthe1960sand1970shegavefrequentgraduateseminarsonStructureandhisfurtherthoughts,aswellasgivinglecturesandpublishingintermediatehintsofhiselaborations.TwomajorinfluencesonhisthinkingwereconferencesinLondonin1965andChampaign,Illinoisin1969,atwhichStructurewasamajorcriticalfocus.TheproceedingsofthesewereeventuallypublishedasCriticismandtheGrowthofKnowledge(editedbyLakatosandMusgrave)andTheStructureofScientificTheories(editedbySuppe).In1977hepublishedTheEssentialTension,acollectionofhisessaysrangingfromreprintingsofpre-Structurepaperstoitemsthatappearedforthefirsttimeinthatvolume.Theessentialtensionreferredtoisthatbetweenthedesiretoassimilatealldataandobservationswithinthecurrentparadigmandthedesiretofindrevolutionarynewsolutions.374\nTHOMASS.KUHNHecontinuedtobeheavilyinvolvedinhistoryofscience,themainculminationofwhichwasthepublicationin1978ofBlack-BodyTheoryandtheQuantumDiscontinuity,1894–1912.In1979heleftPrincetonfortheMITdepartmentofphilosophyandlin-guistics,wherehebecameaprofessorofphilosophyforthefirsttime.SubsequentlyhisresearchfocusedmoreexclusivelyonrefininghisanswerstothequestionsraisedaboutStructure:aboutthenatureofincommensurability,therelationbetweendisciplinarymatricesandscientificcommunities,theelementsofdisciplinarymatrices,rationality,andtheorychoice.Hisanalytictoolsalsoshifted;the“Postscript”wasphrasedintermsfamiliartoreadersofQuine’sWordandObject,whereashislaterworkinvokedpossibleworlds(seeLEWIS)andrigiddesignators(seeKRIPKE).Onerecurringissuewastheclarificationofhisontologicalviews.ProbablythemostinfamoussentenceofStructureoccursonp.150:“InasenseIamunabletoexplicatefurther,theproponentsofcompetingparadigmspracticetheirtradesindif-ferentworlds.”Welldisposedcriticsurgedthatheprobablydidnotreallymeantosaythattheywereindifferentworlds,justthattheworldlookedverydifferenttothem.Buthewasadamantthattherewasanimportantinsightinthestrongerclaim.Thiswasimportanttohimbecause,forinstance,healsowantedtoclaimthatbeforethemedievalparadigmchangethatintroducedtheconceptofthependulum,therewerenopendulumsbutonlyswingingstones(p.120).Hisattemptstoclarifythisandrelatedlocutionsledhimtofurtherinvestigationsoftheinterrelationsoflanguage,concepts,andperceptionandtoproposethatthesewereatleastpartiallyconstitutiveoftheworld.Hebecameanemeritusprofessorin1989butratherthandiminishinghisefforts,heusedthisasanopportunitytospendmoretimeonhisresearchagenda.Atthetimeofhisdeathin1996thesolutionswerestillnotinhisgraspandtheenvisionedcon-clusivemanuscriptwasstillinanearlystage.AnextensivestudyofhislaterworkisReconstructingScientificRevolutions,a1993translationofPaulHoyningen-Huene’s1989book.Hoyningen-HueneworkedcloselywithKuhninproducingthebookanditisalmostacollaboration.AthoroughevaluationofKuhn’sworkcanbefoundinWorldChanges,editedbyPaulHorwich,whichistherevisedproductofa1990conferenceonKuhn’sworkandincludesresponsestohiscritics.IhavewaxedbiographicaltounderlinethepeculiarityofevaluatingKuhnfromthecontextofanalyticphilosophy.Hehadnoformaltraininginphilosophy,andhismostinfluentialworkwascompletedbeforehewasverythoroughlyconversantwiththeintricaciesoftheanalytictradition.ButhewasalreadysufficientlyfamiliarwithitatthetimeofthewritingofStructuretorecognizethathewouldbeaccusedofconfusingthecontextofscientificdiscoverywiththecontextofscientificjustification,adistinc-tionformalizedbyReichenbachbutwhichwaswidespreadinthetraditionbeforehisarticulation.Kuhn’sresponsetotheaccusationwastoquestionthedistinction:toarguethatonlyahistoricallyinaccurateandoversimplifiedviewofscientificdevelopmentwouldpermitsuchadistinctionandthattomaintainsuchadistinctionwastodoomepistemologytosterility.Otherphilosophersofscience–Hanson,Toulmin,Feyerabend,Hesse,amongothers–publishedbookswithatleastsimilarthemesinthelate1950sand1960s,butnoneofthosehadthesameeffector,possiblyexceptingFeyerabend’sAgainstMethod,producedsostrongareactioninreadersandreviewers.375\nRICHARDGRANDYItwouldbeeasytounderestimatetheinfluenceStructureandKuhn’ssubsequentworkhaddirectlyonphilosophyofscienceandindirectlyonanalyticphilosophygen-erally.Alargepercentageofagenerationofphilosophersofsciencespentacon-siderableportionoftheircareersshowingthatKuhnwaswrong–wrongaboutincom-mensurability,wrongaboutparadigms,wrongabouttheroleofscientificcommuni-ties,wrongaboutrationality,wrongabouttherelevanceofpsychologyforphilosophyofscience,andmostsignificantly,wrongabouttheimportofhistoryofscienceforphi-losophyofscience.However,theresultsoftheinquiriesdemonstratingthedefectsanderrorsofStructurebearafargreaterresemblancetoStructurethantoitspredecessors.Thesitua-tionseemscomparabletotheroleofPiagetindevelopmentalpsychology.Few,ifany,ofPiaget’sspecificclaimsaboutdevelopmentalstagesorevenabouttheabilities(andinabilities)ofchildrenatvariousageshavewithstoodfurthermoresophisticatedresearch.ButPiagetbroughtthefieldintoexistenceandwithouthisimpetusitisnotclearthatanyofthefurtherresearchwouldhavebeendone.OnecouldarguethatIhaveoverstatedtheimpactofKuhn’swork;otherphiloso-phers,includingCarnapandHempel,aswellasthepreviouslymentionedauthors,werecalling,albeitmorequietly,forarethinkingoftheimageofsciencethathadbeendominatingphilosophyofscience(seeCARNAPandHEMPEL).Thereceivedviewofsci-entifictheorieswasunderattackbothfromthosewhoquestionedthepivotaldistinc-tionbetweentheoreticalandobservationalvocabulary,butalsofromthestructuralistapproachtotheorieschampionedbyBraithwaite,Suppes,vanFraassen,andSuppe.Ontheotherhand,themostrefinedversionofstructuralisttheories,thatproducedbyStegmuller,Sneed,andothers,drewstronginspirationfromKuhn.Theimportanceofcarefulhistoricalcasestudies,ofconsiderationofthebroadercontextofscientificdevelopments,ofthecognitiveabilitiesandconstraintsonscien-tists,ofthe“external”influencessuchasmotivationandcompetition,areallnowtakenforgrantedaspartofphilosophyofscience.Debateragesabouttherelativeimportance,interpretations,andsoon,butinthebackgroundtherearesharedassumptionsthatwerenotinplacebeforetheinfluenceofKuhn’swork.ExamplesofimportantrecentbooksthatarenotalwayscognizantoftheirKuhnianheritage,butwhichcanbeseentobefollowinginaKuhniantraditionareLongino’sScienceasSocialKnowledge,Giere’sExplainingScience,andKitcher’sTheAdvancementofScience.Hisworkhasalsoinspiredthedevelopmentofhistoricist,feminist,andsociologicalmovementsinthephilosophyofsciencewithwhosedoctrinesheoftendisagreed.BibliographyWorksbyKuhn1957:TheCopernicanRevolution,Cambridge,MA:HarvardUniversityPress;7thedn.,1985.1962:TheStructureofScientificRevolutions,Chicago:UniversityofChicagoPress.(2ndedn.,with“Postscript,”1970.)1977:TheEssentialTension,Chicago:UniversityofChicagoPress.1978:Black-BodyTheoryandtheQuantumDiscontinuity,1894–1912,NewYork:OxfordUniversityPress.376\nTHOMASS.KUHNWorksbyotherauthorsHoyningen-Huene,P.(1993)ReconstructingScientificRevolutions,trans.AlexanderLevine,Chicago:UniversityofChicagoPress.Horwich,P.(ed.)(1993)WorldChanges:ThomasKuhnandtheNatureofScience,Cambridge,MA:MITPress.Lakatos,I.andMusgrave,A.(eds.)(1974)CriticismandtheGrowthofKnowledge,Cambridge:CambridgeUniversityPress;3rdedn.,corrected,1977.Suppe,F.(ed.)(1974)TheStructureofScientificTheories,Urbana,IL:UniversityofIllinois;2ndedn.,1977.377\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200131MichaelDummett(1925–)ALEXANDERMILLERMichaelDummett(WykehamProfessorofLogicatOxford1979–92)isoneofthemostimportantandinfluentialBritishphilosophersofthesecondhalfofthetwentiethcentury.InadditiontomakingseminalcontributionstotheexpositionandstudyofthephilosophyofFrege,Dummettstartedadebate–concerninghowissuesinmetaphysicsmightbestbeprosecutedviaargumentsinthephilosophyoflan-guageandtheoryofmeaning–whichcontinuestobeoneofthecentralissuesincontemporaryanalyticphilosophy.Thetwoareintimatelyrelated.Wearegivenalarge-scaleexpositionandpartialdefenseofabroadlyFregeantheoryofmeaning.Itisthenarguedthattherealistpositioninmetaphysicaldebatesaboutadisputedsubjectmatterisbestcastasasemanticalthesisaboutthemeaningofsentencesconcerningthatsubjectmatter.OnceWittgensteinianinsightsaboutlinguisticunderstandingandlan-guagemasteryareincorporatedintotheFregeantheoryofmeaning,itemergesthatthesemanticalthesisinwhichtherealistviewisbestcastturnsouttofaceveryseriouschallenges.Ananti-realistalternativeisexplored,drawingonthetheoryofmeaningproposedbyintuitionismformathematicalstatements,anditisarguedthatoneconsequenceofthisistherejectionofcertaintheoremsofclassicallogic,suchasthelawofexcludedmiddle.FregeDummett’sexpositionandpartialdefenceofaFregeantheoryofmeaningiswritlargethroughouthiswork,butthekeytextsareFrege:PhilosophyofLanguage,TruthandOtherEnigmas(essays7–9),TheInterpretationofFrege’sPhilosophy,FregeandOtherPhilosophers,andOriginsofAnalyticPhilosophy.ForFrege,whetherornotasentenceisgrammaticallywellformedisdeterminedbysyntacticalrules.Thesearetheprovinceofsyntax,andtellushowexpressionsfromdifferentsyntacticcategoriesmaybecombinedtoformgrammaticalsentences.Intheprovinceofsemantics,theBedeutungofanyexpressionisthatfeatureofitthatdetermineswhethersentencesinwhichitoccursaretrueorfalse.TheBedeutungofasentenceisitstruth-value(trueorfalse).WhetherasentenceistrueisdeterminedbytheBedeutungenofitsconstituents.ExpressionsfromdifferentsyntacticcategoriesareassigneddifferenttypesofBedeutung:theBedeutungofapropernameistheobject378\nMICHAELDUMMETTitstandsfor,theBedeutungofapredicateisafirst-levelfunctionfromobjectstotruth-values,theBedeutungofasententialconnectiveisafirst-levelfunctionfromtruth-valuestotruth-values,andtheBedeutungofaquantifierisasecond-levelfunctionfromconcepts(first-levelfunctions)totruth-values.TheBedeutungofaconstituentofasentenceisdeterminedbyitsSinnorsense(thatingredientofitsmeaningthatdeterminesitscontributiontothetruthorfalsityofsentencesinwhichitmayappear),whichinturndetermines,inconjunctionwiththesensesoftheotherconstituents,theSinnofthesentence.TheSinnofasentenceisthethoughtwhichitexpresses,conceivednotassomepsychologicalepisodeorentity,butasatruth-condition:theconditionwhichmustobtainifthesentenceistobetrue.TheSinnofanexpressioniswhatsomeonewhounderstandsanexpressiongrasps:ourunder-standingofwholesentencesthereforeconsistsinpartinourgraspoftheirtruth-conditions(seeFREGE).MuchofDummett’sworkconsistsinasophisticatedelaborationofthetheorythuscrudelysummarized,andanexaminationoftheothernotions–suchasforce(thatingredientinmeaningwhichdistinguishes,e.g.,assertionsfromquestionsandcommands)andtone(thatingredientinthemeaningof,e.g.,“and”whichdistinguishesitfrom“but”eventhoughithasthesameSinn)–thatneedtobeaddedtothenotionsofSinnandBedeutunginordertoobtainacomprehensivetheoryofmeaning.Interalia,DummettdefendstheFregeantheoryinthefaceofattacksfromthecausaltheoryofreferenceadvocatedbyKripke(1973:appendixtoch.5)andtheholisticpictureoflanguageadvancedbyQuine(1978:essays9and22,1973:ch.17).OnDummett’sinterpretationofFrege,itispossibleforanexpression,suchasthepropername“Vulcan,”tohaveaSinn,eventhoughithasnoBedeutung,sincethereisnoobjectforwhichitstands:sentencescontainingexpressionswhichhavenoBedeutungthemselvesfailtohaveaBedeutung,thatis,failtopossessatruth-value(1973:ch.6).ThisfacetofDummett’sinterpretationischallengedinimportantworkbyGarethEvans(1982)andJohnMcDowell(1998a:essays8–12).WhereasforDummett,thesenseofanexpressionisamethodorprocedurefordeterminingitsBedeutung,sothatthesenseofasentence,forexample,isamethodorprocedurefordeterminingitstruth-value,amethodwhichcanexistevenifthesentenceinquestionhasnotruth-value,forEvansandMcDowellthesenseofanexpres-sionis“awayofthinkingaboutitsBedeutung.”Forthem,lackofBedeutungnecessarilyinvolvesacorrespondinglackofSinn.Foradiscussion,seeDummett1993a:ch.7.DummetthimselfcriticizesandqualifiestheFregeantheoryinvariouswaysinFrege:PhilosophyofLanguage(e.g.Frege’sexplanationofhowexpressionscandifferintoneisrejectedinchapter5;Frege’sassimilationofsentencestocomplexpropernamesandtheassociatedclaimthattruth-valuesareakindofobjectisquestionedinchapter6;andhisclaimthatexpressionsotherthanpropernames,suchaspredicatesandquantifiers,alsohaveBedeutungen,isqualifiedinimportantwaysinchapter7).However,thesuggestionofhisthathasexcitedthegreatestinterestamongcontempo-raryphilosophersisthat,inmanyimportantcases,debatesinmetaphysicsbetweenrealistsandtheiropponentsmayandperhapsmustbecastasdebatesinthetheoryofmeaning,betweenrivalaccountsofthenatureofourgraspofsentences’Sinne,ortruth-conditions.379\nALEXANDERMILLERRealismTheremarksgermanetothissuggestionarescatteredthroughoutDummett’swritings,butthekeytextsareTruthandOtherEnigmas(1978)(Prefaceandessays1,10,14,and21),TheSeasofLanguage(1993b)(essays1–7,11,and20),andTheLogicalBasisofMetaphysics(1991c).ThisapproachtometaphysicalquestionsisindicativeofDummett’sviewthatthephilosophyoflanguage–thetheoryofmeaning–hasafoundationalroletoplaywithinphilosophy.Indeed,Dummettseesthisviewaboutthepriorityofphilosophyoflanguageasthedefiningcharacteristicofanalyticphilosophy:Whatdistinguishesanalyticalphilosophy,initsdiversemanifestations,fromotherschoolsisthebelief,first,thataphilosophicalaccountofthoughtcanbeattainedthroughaphilo-sophicalaccountoflanguage,and,secondly,thatacomprehensiveaccountcanonlybesoattained.(1993a:4)Dummettbelievesthatoneofthereasonswhyphilosophicalspeculationaboutmeta-physicalissueshasmadelittleprogressoverthecenturiesisthattheopposingpositionsinvariousmetaphysicaldisputeshaveonlybeenexplainedinpictorial,ormetaphori-calterms:Eventoattempttoevaluatethedirectmetaphysicalarguments,wehavetotreattheoppos-ingthesesasthoughtheircontentwerequiteclearanditweresolelyamatterofdecidingwhichistrue;whereas...theprincipaldifficultyisthat,whileoneoranotherofthecom-petingpicturesmayappearcompelling,wehavenowaytoexplaininnon-pictorialtermswhatacceptingitamountsto.(1991c:12)Dummett’sapproachisintendedtoremedythis:themetaphysicaldisputesarerecastasdisputesabout“thecorrectmodelofmeaningforstatementsofthedisputedclass,”thusgivingthedebatessomenon-metaphoricalcontent,andenablingthedisputestoberesolvedwithinthetheoryofmeaning.Whatdoesitmeantosaythatthetruth-conditionsofarangeofsentencesare“realist”?Inshort,Dummett’sanswerisasfollows:tosaythatarangeofsentenceshaverealisttruth-conditionsistosaythatthosetruth-conditionsarepotentiallyverification-transcendent.Tosaythatatruth-conditionispotentiallyverification-transcendentistosaythatwemaybeincapable,eveninprinciple,ofdeterminingwhetherornotitobtains.Thus,considerdiscourseaboutthepast:intuitively,thesentence“JamesIIsufferedamigrainein1665,ontheafternoonofhis32ndbirth-day”hasatruth-condition–James’ssufferingamigraineontheafternooninquestion–andwecansaythatthisconditioneitherobtainedoritdidnot,eventhoughwemayhavenoway,eveninprinciple(becausealltheevidenceappearstohavevanishedandtime-travelisimpossible)ofdeterminingwhichofthesewasthecase(cf.ANSCOMBE).Thus,“JamesIIhadamigraineontheafternoonofhis32ndbirthday”hasatruth-condition,andwemaybeincapableofdetermining,eveninprinciple,whetherthatconditionobtainedornot:itispotentiallyverification-transcendent.Likewise,considerarithmeticaldiscourse.Goldbach’sconjecture,thateveryevennumbergreaterthan380\nMICHAELDUMMETTtwoisthesumoftwoprimes,hasapotentiallyverification-transcendenttruth-condition:ithasadeterminatetruth-valueeventhoughweareincapableofdeter-miningwhatthistruth-valueis,sincewehavenoguaranteeeitherthataproofoftheconjecturewillbeconstructedorthatacounterexample–anevennumberwhichisnotthesumoftwoprimes–willbefound.Thus,sentencesaboutthepastandaboutarithmetichavepotentiallyverification-transcendenttruth-conditions:inthissense,Dummettwillclaim,theirtruth-conditionsarerealist.Now,whyistheclaimthatthesentencesofadiscoursearepoten-tiallyverification-transcendentawayofcashingoutrealismaboutthesubjectmatterofthatdiscourse?Inordertoseethiswehavetorecall–fromthefirstsection–thattheSinn,orsense,ofasentenceisgivenbyitstruth-conditions,andthatunderstand-ingasentenceconsistsingraspingitssense.Thus,understandingasentenceconsistsingraspingitstruth-conditions.Anythesisaboutthetruth-conditionsofasetofsen-tencesisinteraliaathesisaboutwhatourunderstandingofthosesentencesconsistsin.Inaslogan,atheoryofmeaningisalsoatheoryofunderstanding.Thus,someonewhoacceptsthatthetruth-conditionsofaregionofdiscoursearepotentiallyverification-transcendentalsoacceptsthatourunderstandingofthesentencesofthatdiscourseconsistsinourgraspofpotentiallyverification-transcendenttruth-conditions.Andnowtheconnectionwithrealismaboutthatdiscourseisrelativelyeasytosee.AsCrispinWright,anotherimportantBritishphilosopherwhohasdonemorethananyonetofurtherDummett’sagenda,putsit:Toconceivethatourunderstandingofstatementsinacertaindiscourseisfixed...byassigningthemconditionsofpotentially[verification]-transcendenttruthistograntthat,iftheworldco-operates,thetruthorfalsityofanysuchstatementmaybesettledbeyondourken.Soweareforcedtorecogniseadistinctionbetweenthekindofstateofaffairswhichmakessuchastatementacceptable,inthelightofwhateverstandardsinformourpracticeofthediscoursetowhichitbelongs,andwhatmakesitactuallytrue.Thetruthofsuchastatementisbestowedonitindependentlyofanystandardwedoorcanapply;acceptabilitybyourstandardsis,forsuchstatements,atbestmerelycongruentwithtruth.RealisminDummett’ssenseisthusonewayoflayingtheessentialsemanticgroundworkfortheideathatourthoughtaspirestoreflectarealitywhosecharacterisentirelyinde-pendentofusandourcognitiveoperations.(1992:4)Inalargeclassofcases,wecanthusconceiveofthemetaphysicaldebatebetweenreal-istsandtheiropponents–anti-realists–inaparticularregionofdiscourseDascon-cerningwhetherthesentencesofDcanplausiblybeviewedaspossessingpotentiallyverification-transcendenttruth-conditions.Realism:thesentencesofDhavetruth-conditionsandthesetruth-conditionsarepoten-tiallyverification-transcendent.Anti-Realism:thesentencesofDhavetruth-conditionsbutthosetruth-conditionsarenotpotentiallyverification-transcendent.Note1:Theabovecannotbeusedtocharacterizealltypesofdebatebetweenreal-istsandtheiropponents.OnecharacteristicformofoppositiontorealismaboutaregionofdiscourseDisthedenial(bynon-cognitivists,expressivists,ornon-factualists)thatthesentencesinquestionhavetruth-conditions,whetherpotentiallyverification-381\nALEXANDERMILLERtranscendentornot.Dummetthasadmittedthatthisstyleofdebatebetweenrealismanditsopponentsispriortothatwhichturnsonthepossibilityofverification-transcendenttruth(1993b:467).Inaddition,therearealsosomeseriousquestionsaboutwhethertheabovecharacterizationcanadequatelycapturewhatisatissueinontologicaldisputesbetweenrealistsandtheir(e.g.nominalist)opponents.(Forillumi-natingdiscussion,seeHale1997,esp.sect.3.SeealsoMichaelDevitt1993.)Note2:Itisfarfromclearwhatitmeanstosaythatthetruthofasentenceisnotpotentiallyverification-transcendent.Doesthismeanthatthesentencecanbeverifiedbyusasweactuallyare?Bysomeone,somewhere,astheyactuallyare?Bysomeone,somewhere,givensomesuitableidealizationoftheirpresentcognitivepowers?Andwhatispermissibleasa“suitableidealization”?Andhowcanthenotionofeffectivedecidability(seebelow)beextendedfromthemathematicalcasetotheempiricaldomain?Thesequestionsmustbeansweredifanti-realismistohaveanydeterminatecontent.(SeeWright1986:32.)Note3:ThischaracterizationofrealismmakesessentialuseofFrege’sideathatunderstandingasentenceconsistsingraspingitstruth-conditions.ItisplausiblethatFregehimselfwasarealistinthesensethuscharacterized.Hewrites:“Athinkerdoesnotcreate[thoughts]butmusttakethemastheyare.Theycanbetruewithoutbeinggraspedbyathinker”(1967:30).Ifathought–thesenseofasentence–canbedeterminatelytrueorfalseeventhoughthatthoughtisnotevengraspedbyathinker,thenthesentenceinquestioncanbetrueorfalseeventhoughthinkersareincapable,eveninprinciple,ofdeter-miningitstruth-value.DummettnowsuggeststhatFrege’srealismisseriouslychallengedwhenweaddtohistheoryofmeaningtheinsightsaboutlinguisticunderstandingtobefoundinLudwigWittgenstein’sPhilosophicalInvestigations.FregeandWittgensteinontheobjectivityofsenseTheinterpretationofWittgenstein’sviewsonmeaningandunderstandingisacomplexandsubtlematter:herewecangivebutthebriefestsketchofoneofthefacetsofWittgenstein’spositionthatDummettreliesoninchallengingourintuitivelyrealistpictureoftheworld.(Foramorecomprehensivetreatment,seeMiller1998:chs5and6.)Infact,theWittgensteinianviewaboutlinguisticunderstandingthatfeaturesinDummett’schallengetorealismisadevelopmentofanotherinsightofFrege’s:thatsenseis,inawaytobeexplained,objective.AccordingtoDummett,ifFregehadfollowedthisinsightthroughtoitslogicalconclusion,hewouldhaveseenthatitchallengesseriouslyhiscommitmenttorealism(1993b:essay2,§6).Recallthatthesenseofasentenceisathought,andthataccordingtoFregethoughtsareinsomesenseobjective,asopposedtosubjectiveorpsychological.ThisisanextremelyimportantpartofFrege’sposition.Indeed,intheintroductiontoTheFoundationsofArithmetic,hestatesthefollowingasthefirstofhisthree“fundamentalprinciples”:“Alwaystoseparatesharplythepsychologicalfromthelogical,thesubjec-tivefromtheobjective”(1953:x).Thisappliesnotonlytothesensesofsentences,buttothesensesofexpressionsgenerally.Butwhatexactlydoesitmeantosaythatsenseisobjectiveandnotsubjec-382\nMICHAELDUMMETTtive?Onethingthatitmeansisthatgraspingasense–understandinganexpression–isnotamatterofassociatingthatexpressionwithsomesubjectiveitemlikeamentalimage,picture,oridea.Fregeisquiteexplicitabouttheneedtodistinguishsenses,whichareobjective,fromideas,whicharesubjective:Thereference[Bedeutung]andsenseofasignaretobedistinguishedfromtheassociatedidea...Thereferenceofapropernameistheobjectitselfwhichwedesignatebyusingit;theideawhichwehaveinthatcaseiswhollysubjective;inbetweenliesthesense,whichisindeednolongersubjectiveliketheidea,butisyetnottheobjectitself.(Frege1960:60–1)Theviewthatunderstandinganexpressionconsistsinthepossessionofsomeassoci-atedideaorimageisonethathasalonglistofadherentsinthehistoryofphilosophy.Indistinguishingthesenseofanexpressionfromanyassociatedidea,Fregewasdirectlyattackingthistradition.TheclassicexampleofthisviewofsensecanbefoundinBookIIIofJohnLocke’sAnEssayConcerningHumanUnderstanding.Somecreatureswhoutter,forexample,theword“cube”understandthatword,andsomedon’t.Aparrotcansaytheword,butunlikeanormalhumanspeakerofEnglish,theparrotpossessesnounderstandingofwhatissaid.InFregeanterminology,thehumanspeakergraspsthesenseof“cube,”whereastheparrotdoesnot.Butwhatdoesthisdifferenceconsistin?Locke’ssuggestionisthattheword“cube”is,inthecaseofthecompetenthumanspeaker,associatedwithanideaofacubeinthatspeaker’smind,whileinthecaseoftheparrotthereisnosuchideaandsonosuchassociation.Lockeisthusledtotheviewthatunderstandinganexpressionconsistsinassociatingitwithsomeidea:“Words,intheirprimaryorimmediatesignification,standfornothingbuttheideasinthemindofhimthatusesthem”(1975:III,ii,2).Locketakesideastobementalimagesorpictures:anideaofacubeistakentobeamentalimageorinnerpictureofacube.ThisisclearfromthewayLockespeaksthroughouttheEssay.Forexample,inhisaccountofmemorythetalkofideasisexplicitlycashedoutintermsofpicturingandimagery:Theideas,aswellaschildrenofouryouthoftendiebeforeus.Andourmindsrepresenttousthosetombstowhichweareapproaching;wherethoughthebrassandmarbleremain,yettheinscriptionsareeffacedbytime,andtheimagerymouldersaway.Thepicturesdrawninourmindsarelaidinfadingcolours.(1975:II,x,5.1)WecouldthussumupLocke’sviewofsenseasfollows(wherethesenseof“cube”determinesthatitrefersto,precisely,cubes):aspeakergraspsthesenseof“cube”ifandonlyifheisdisposedtohaveamentalimageofacubewheneverhehearsorutterstheword.WhydoesFregeobjecttothisaccountofsense?Locke’saccountleadstoatensionbetweenthepublicnatureofmeaningfullanguage,andtheprivatenatureofideasandmentalimages.Ontheonehand,languageispublicinthatdifferentspeakerscanattachthesamesensetotheirwords,andonespeakercanknowwhatanotherspeakermeansbyhiswords.Differentspeakerscancommunicatewitheachotherinvirtueofthecommonsensesthattheyhaveattachedtotheirwords.On383\nALEXANDERMILLERtheotherhand,ideasareprivate.AsLockehimselfputsit,aman’sideasare“allwithinhisownbreast,invisible,andhiddenfromothers,norcanofthemselvesbemadetoappear.”Also,myideas,my“internalconceptions,”arevisibleonlytomyconscious-ness,andlikewiseyourideas,your“internalconceptions,”arevisibleonlytoyourconsciousness.Butweareattemptingtogiveanaccountofsense,anaccountthatshouldhelpexplainhowweareabletocommunicatewitheachotherviatheuseoflanguage;andhowcanatheorywhichconstruesgraspofsenseintermsofthepos-sessionofprivateinneritemshelpexplainourabilitytouselanguageinsuccessfulpubliccommunication?DummettseesFrege’santi-Lockeanargumentfortheobjectivityofsenseasvitiatedbyanerroneousconstrualofmentalimagesasnecessarilyprivate,butneverthelessagreeswithitsupshot(1973:157–9).But,Dummettsuggests,inordertoallowfortheobjectivityofsense,weneedtogofurtherthanmerelydenyingthatgraspofsenseissubjectiveinthemannerjustoutlined:weneed,inaddition,toconstruegraspofsenseintermsofuse.Frege’sthesisthatsenseisobjectiveis...implicitlyananticipation(inrespectofthataspectofmeaningwhichconstitutessense)ofWittgenstein’sdoctrinethatmeaningisuse...yetFregeneverdrewtheconsequencesofthisfortheformwhichthesenseofawordmaytake.(1993b:91)Inordertoallowfortheobjectivityandcommunicabilityofthesenseofanexpression,graspofitssensehastobeconstruedintermsofpossessionofanabilitytouseitincertainpublicandobservablecircumstances.Itfollowsthatifspeakerspossessapieceofknowledgewhichisconstitutiveoflinguisticunderstanding,thenthatknowledgeshouldbemanifestedinspeakers’useofthelanguage,thatis,intheirexerciseofthepracticalabilitiesthatconstitutelinguisticunderstanding.We’llnowseehowDummettattemptstochallengerealism,byincorporatingtheWittgensteinianinsightaboutunderstandingwithintheFregeantheoryofmeaning.Dummett’schallengestorealismAccordingtoDummett,thedebatebetweenrealismandanti-realismaboutaregionofdiscourseisadebateaboutthenatureofthetruth-conditionspossessedbythesen-tencesofthatdiscourse.Anyaccountofthetruth-conditionsofarangeofsentenceswillbeunacceptableifitcannotcoherewithaplausibleaccountofwhatourunder-standingofthosesentencesconsistsin.Dummett’sstrategyistoarguethattheaccountoflinguisticunderstandingwhichrealismleadstofacesseriousproblems.Themeta-physicaldebateconcerningtheplausibilityofrealismboilsdowntoadebatewithinthephilosophyoflanguage.Why,then,doesDummettthinkthatthereareproblemswiththerealisticconstrualoflinguisticunderstandingasgraspofpotentiallyverification-transcendenttruth-conditions?Therearetwomainchallenges:theacquisitionchallenge,andthemani-festationchallenge.(Forthecanonicalstatementoftheformer,see1978:essay1;forthelatter,seeessay14;forastate-of-the-artexpositionofboth,andofotherchallengestorealismasconceivedbyDummett,seetheIntroductiontoWright1986).384\nMICHAELDUMMETTDummett’sacquisitionchallengeSupposethatweareconsideringsomeregionofdiscourseD,thesentencesofwhichweintuitivelyunderstand.Suppose,forreductio,thatthesentencesofDhavepoten-tiallyverification-transcendenttruth-conditions.Thus,1WeunderstandthesentencesofD.2ThesentencesofDhaveverification-transcendenttruth-conditions.Now,from(1)togetherwiththeFregeanthesisthattounderstandasentenceistograspitssenseorknowitstruth-conditions,wehave3WegraspthesensesofthesentencesofD:i.e.weknowtheirtruth-conditions.Wenowaddtheapparentlyreasonableconstraintonascriptionsofknowledge:4Ifapieceofknowledgeisascribedtoaspeaker,thenitmustbeatleastinprinciplepossibleforthatspeakertohaveacquiredthatknowledge.So5Itmustbeatleastinprinciplepossibleforustohaveacquiredknowledgeoftheverification-transcendenttruth-conditionsofD.But6Thereisnoplausiblestorytobetoldabouthowwecouldhaveacquiredknowledgeofverification-transcendenttruth-conditions.So,byreductio,wereject(2)toget:7ThesentencesofDdonothaveverification-transcendenttruth-conditions,sorealismaboutthesubjectmatterofDmustberejected.Thecrucialpremisehereisobviously(6).Wrightputsthepointasfollows:Howarewesupposedtobeabletoformanyunderstandingofwhatitisforaparticularstate-menttobetrueifthekindofstateofaffairswhichitwouldtaketomakeittrueisconceived,exhypothesi,assomethingbeyondourexperience,somethingwhichwecannotconfirmandwhichisinsulatedfromanydistinctiveimpactonourconsciousness?(1986:13)However,asWrightnotes,thisargumentisatbestinconclusive.Itreallyonlypresentstherealistwithachallenge:Inordertobemorethanachallenge,[it]wouldneedthebackingofaproventheoryofconcept-formationofabroadlyempiricistsort.[And]thetraditionaltheoriesofthatsorthavelongbeenrecognizedtobeinadequate.(1986:15)Thechallengetotherealististhus:givesomeplausibleaccountofhowtheknowledgeofverification-transcendenttruth-conditionswhichyouimputetospeakerscouldhavebeenacquired.Whetherornotthischallengecanbemetbytherealistisverymuchanopenquestion,intheabsenceofaproventheoryofconceptacquisition.385\nALEXANDERMILLERDummett’smanifestationargumentSupposethatweareconsideringregionofdiscourseDasbefore.Then:1WeunderstandthesentencesofD.Suppose,forreductio,that2ThesentencesofDhaveverification-transcendenttruth-conditions.From(1)andtheFregeanthesisthattounderstandasentenceistograspitssenseorknowitstruth-conditions,wehave:3WegraspthesensesofthesentencesofD;thatis,weknowtheirtruth-conditions.Wethenaddthefollowingpremise,whichstemsfromtheWittgensteinianinsightthatunderstandingdoesnotconsistinthepossessionofaninnerstate,butratherinthepossessionofsomepracticalability(seethesection“FregeandWittgensteinontheObjectivityofSense,”above):4Ifspeakerspossessapieceofknowledgewhichisconstitutiveoflinguisticunder-standing,thenthatknowledgeshouldbemanifestedinspeakers’useofthelan-guage,thatis,intheirexerciseofthepracticalabilitiesthatconstitutelinguisticunderstanding.Itnowfollowsfrom(1),(2)and(3)that:5Ourknowledgeoftheverification-transcendenttruth-conditionsofthesentencesofDshouldbemanifestedinouruseofthosesentences,thatis,inourexerciseofthepracticalabilitieswhichconstituteourunderstandingofD.Since6SuchknowledgeisnevermanifestedintheexerciseofthepracticalabilitieswhichconstituteourunderstandingofD,itfollowsthat7Wedonotpossessknowledgeofthetruth-conditionsofD.(7)and(3)togethergiveusacontradiction,whence,byreductio,wereject(2)toobtain:8ThesentencesofDdonothaveverification-transcendenttruth-conditions,sorealismaboutthesubjectmatterofDmustberejected.Thebasicpointisthat,sofarasanaccountofspeakers’understandinggoes,theascriptionofknowledgeofverification-transcendenttruth-conditionsissimplyredun-dant:thereisnogoodreasonforascribingit.Consideroneofthesentencesweconsid-eredearlierascandidatesforpossessingverification-transcendenttruth-conditions,“JamesIIhadamigraineontheafternoonofhis32ndbirthday”or“Everyevennumbergreaterthantwoisthesumoftwoprimes.”Therealistaccountviewsourunder-standingofthesesentencesasconsistinginourknowledgeofapotentiallyverification-transcendenttruth-condition.But,inWright’swords:386\nMICHAELDUMMETTHowcanthataccountbeviewedasadescriptionofanypracticalabilityofuse?Nodoubtsomeonewhounderstandssuchastatementcanbeexpectedtohavemanyrelevantprac-ticalabilities.Hewillbeabletoappraiseevidencefororagainstit,shouldanybeavailable,ortorecognizethatnoinformationinhispossessionbearsonit.Hewillbeabletorecog-nizeatleastsomeofitslogicalconsequences,andtoidentifybeliefsfromwhichcommit-menttoitwouldfollow.Andhewill,presumably,showhimselfsensitivetoconditionsunderwhichitisappropriatetoascribepropositionalattitudesembeddingthestatementtohimselfandtoothers,andsensitivetotheexplanatorysignificanceofsuchascriptions.Inshort:intheseandperhapsotherimportantrespects,hewillshowhimselfcompetenttousethesentence.Buttheheadingsunderwhichhispracticalabilitiesfallsofarinvolvenomentionofevidence-transcendenttruth-conditions.(Wright1986:17)Thisestablishes(6),andtheconclusionfollowsswiftly.Adetailedassessmentoftheplausibilityofthisargumentisimpossiblehere:butweshouldnotethatpremise(4)dependsuponaninterpretationofWittgenstein’sworkonrule-followingandunder-standing(seeWITTGENSTEIN),andthatthisisanextremelycontroversialmatter(seeMiller1998:chs5and6).Inparticular,oneissuethatneedstobeaddressediswhethertheinterpretationofpremise(4)requiredbyDummettfortheanti-realistargumentisleftintactbyJohnMcDowell’sinterpretationofWittgenstein,accordingtowhichunderstandingcanharmlesslybeconstruedasastateofmind(seeMcDowell1998b:essays11–14).OnDummett’santi-realistargumentsgenerally,seeMcDowell1998a:essays1,4,5,14,15,16.Foranexcellentsurveyofpossiblerealistresponsestoboththeacquisitionandmanifestationchallenges,seeHale1997.Anti-realism(1)Asentenceissaidtobeeffectivelydecidableifthereissomeprocedurewhichwecaninprincipleapplyandwhichwillguaranteeananswertothequestionwhetherornotthesentenceistrue.Thus,“2+2=4”and“TheQueenhadcornflakesforbreakfastyesterday”arebotheffectivelydecidable:wecancarryoutanelementaryarithmeticalcalculationinthefirstcase,andwecangathertheobvioussortsofevi-denceinthesecondcase,inordertodeterminethetruth-valuesoftherespectivesen-tences.But“JamesIIhadamigraineontheafternoonofhis32ndbirthday”and“Everyevennumbergreaterthantwoisthesumoftwoprimes”arenotknowntobedecidable:inneithercasedoweknowaprocedurewhichwecanapplytodeterminewhetherornottheyaretrue.Nowintuitively,wethinkthateventhoughthesesentencesarenotknowntobedecidable,wecanneverthelessstillassertthattheyareeithertrueorfalse:“Everyevennumbergreaterthantwoisthesumoftwoprimes”hasadeterminatetruth-value,it’sjustthatwecannotworkoutwhatthistruth-valueis.Inotherwords,eventhoughthesentenceisnotknowntobedecidable,westillthinkthattheprincipleofbivalence,thatevery(non-vague)sentenceisdeterminatelyeithertrueorfalse,appliestoit.Nowthisisanideathatisputunderpressurebytheconclusionoftheanti-realistargumentsofDummett’swehavebeenconsidering.Iftruthisnotverification-transcendent,itisepistemicallyconstrained.Onewaytospelloutwhatitmeanstosaythattruthisepistemicallyconstrainedistosaythatitmustbeconstruedintermsofsomenotionlikecorrectorwarrantedassertability:tosaythatasentenceistrueistosaythatthereisawarranttoassertit,orthatitpossessessomeotherprop-387\nALEXANDERMILLERertythatisconstructedoutofwarrantedassertability.(Thisisgreatlyoversimplified:formoredetail,seeWright1986:§vand1992:ch.2,wherehesuggeststhatforcertaindiscourses,truthmaybemodelledon“superassertibility.”Foranotherattempttocon-struetruthasessentiallyepistemicallyconstrained,seePutnam1981(seePUTNAM).SeealsoTennant1987,1997.)Nowgiventhattruthisthusepistemicallyconstrained,whatcanwesayabout“Everyevennumbergreaterthantwoisthesumoftwoprimes”?Wedonothaveawarranttoassertthis–sincenoonehasyetbeenabletoconstructamathematicalproofofit–nordowehaveawarranttoassertitsnegation–sincenoonehasyetproducedacounterexampletoit,orestablishedthatsuchacounterexam-plemustexist.Giventhis,andgiventhattruthistobetoconstruedintermsofwar-rantedassertability,wecannotassertthatthesentence“Everyevennumberisthesumoftwoprimes”iseithertrueorfalse.Thatistosay,wecannotassertaprioritheprincipleofbivalenceforsentencesthatarenotknowntobedecidable:wecannotassert,apriori,thattheyareeithertrueorfalse.(2)Notethatwehaveherecharacterizedrealismastheviewthattruthisnotessen-tiallyepistemicallyconstrained,andderivedtherealist’sattitudetotheprincipleofbivalenceasaconsequence.Dummetthimselfpreferstocharacterizerealismdirectlyintermsofadherencetotheunrestrictedprincipleofbivalence(seee.g.1993b:230),sothatanydenialofthatprinciplemustbeseenasincliningoneinthedirectionofanti-realism.Butthisseemstobeamistake.Therearemanyreasonswhytheprincipleofbivalencemightfailforaparticularregionofdiscourse:becausetherelevantsentencescontainemptynames,havefalsepresuppositions,containvaguepredicatesappliedtoborderlinecases;noneoftheseseemtoconcerntheissueofrealismversusanti-realism.Sotherejectionofbivalenceisasymptomofanti-realism,whichmayormaynotsignaltherejectionofrealismdependingonwhetherornotthefailureoftheprincipleofbiva-lencestemsfromtherejectionoftruthasessentiallyepistemicallyunconstrained.Soitisbettertocharacterizerealismdirectlyintermsofepistemicconstraintsontruth,andviewissuesaboutbivalenceashavingonlysecondary,derivativesignificance.Foranexcellentdiscussionofthispoint,seeRosen1995.(3)Notethatifwecharacterizemeaningintermsofanepistemicallyconstrainednotionoftruth–perhapsintermsofconditionsofwarrantedassertibility–wetherebyavoidtheproblemsraisedbythemanifestationchallengefortherealistconceptionoflinguisticunderstanding.Becausetheconditionswhosegraspconstitutesunderstand-ingareconditionswhich,bytheirverynature,areinprinciplecapableofbeingrecog-nizedwhenevertheyobtain,wecanidentifygraspofasentencewithapracticalability.Thisistheabilitytodiscriminatebetweenthoserecognizablecircumstancesinwhichthesentenceistrueandthosewhichitisnot.Sothatthemanifestationchallengehasasimpleanswerwhendirectedattheanti-realistconceptionofunderstanding.LikewiseforDummett’sacquisitionchallenge.(4)Dummett’santi-realistclaimsthatwecannotassertaprioritheprincipleofbivalence,atleastasappliedtosentencesthatarenotknowntobedecidable.Nowtheprincipleofbivalence–thatevery(non-vague)sentenceisdeterminatelyeithertrueorfalse–iscloselyassociatedwiththeprincipleofclassicallogicknownasthelawof388\nMICHAELDUMMETTexcludedmiddle:|P~P.Refusingtoassertaprioritheprincipleofbivalence,astheanti-realistproposes,thusappearstothreatenthelawofexcludedmiddle,andtheclas-sicalsystemoflogicwhichisfoundeduponit.Thereismuchdebateamonganti-realistsaboutwhetheranti-realismimpliesrevisionismaboutclassicallogic:Dummetthasarguedthatanti-realismimpliesthatclassicallogicmustbegivenupinfavorofsomeformofintuitionisticlogicwhichdoesnothavethelawofexcludedmiddleasatheorem.(Theissuesherearecomplex.ForDummett’sownexaminationofintuition-ismseehisElementsofIntuitionism;fordiscussionoftheallegedrevisionaryaspectsofDummettiananti-realism,seeWright1986,and1992,ch.2).(5)Itisimportanttobeclearthatalthoughtheanti-realistclaimsthatwecannotassertthatsentencesnotknowntobedecidableareeithertrueorfalse,heisnotclaim-ingthatwecanassertthattheyareneithertruenorfalse.Dummettisexplicitthatalthoughtheanti-realistdoesnotwishtoassertaprioritheprincipleofbivalence,hedoesnotrejecttheprincipleoftertiumnondatur,thatthereisnothirdtruth-value(“neithertruenorfalse”)standingbetweentruthandfalsity.Thismightseempuzzling.Supposethattheprincipleofbivalencecorrespondstothelawofexcludedmiddle:|P~P,andthattheprincipleoftertiumnondaturcorrespondsto|--(P~P)(isnotthecasethatneitherPnornot-P).Sinceitisalogicallyvalidsequentthat--P|P,doesn’titfollowthatrejecting|P~Pentailstherejectionof|--(P~P),thatrejec-tionoftheprincipleofbivalenceentailsrejectionoftheprincipleoftertiumnondatur?Thecrucialpointisthatthesequent--P|Pthatlicensesthisentailmentisvalidinclassicallogicbutnotvalidinintuitionisticlogic.Rejectionofbivalenceentailsrejectionoftertiumnondaturonlygivenclassicallogic;buttheDummett-styleanti-realistrejectsclassicallogic,andsocanrejectbivalencewhilstholdingontotertiumnondatur.(SeeDummett1978).(6)Finally,notethattheanti-realistattitudetosentenceswhicharenotknowntobedecidableiscompletelydifferentfromthelogicalpositivistattitudetosentenceswhicharenotinprincipleverifiable.WhereasalogicalpositivistsuchasA.J.AyerinLanguage,Truth,andLogicclaimsthatsuchsentences–becausetheyareinprincipleunverifiable–areliterallymeaningless,Dummett’santi-realistclaimisthatsentencesthatarenotknowntobedecidablearemeaningfulbuttheirmeaningshavetobeconstruedintermsofanepistemicallyconstrainednotionoftruth(seeAYER).LimitationsandprospectsAccordingtoDummett,toopposerealism,toespouseanti-realism,istodenythattruthispotentiallyverification-transcendentandtoarguethattruthmustbeviewedasepis-temicallyconstrained.Butisthisthebestwayofcashingoutthemetaphysicaldebatesbetweenrealistsandtheiropponents?Wehavealreadymentionedsomeofthelimita-tionsofthisapproachinthesection‘Realism.”Butevenwaivingtheseproblems,thereareothers.Forexample,althoughDummett’swayofcharacterizingthemetaphysicaldebateseemstobeappropriateinsomecases(e.g.mathematics,statementsaboutthepast,statementsabouttheexternalworld)thereareothercaseswhereitsimplyseemsbesidesthepoint.Considerdiscourseabout,forinstance,moralsorcomedy.Itseems389\nALEXANDERMILLERthatinthesecasesamoralrealistwouldnothavetoclaimthatthetruth-conditionsoftherelevantsentencesarepotentiallyverification-transcendentandthatboththemoralrealistandthemoralanti-realistcanagreethatstatementsaboutcomedyormoralvaluedonothaveverification-transcendenttruth-conditions.AsWrightputsit:Thereare,nodoubt,kindsofmoralrealism[orrealismaboutcomedy]whichdohavetheconsequencethatmoral[orcomic]realitymaytranscendallpossibilityofdetection.Butitissurelynotessentialtoanyviewworthregardingasrealistaboutmorals[orcomedy]thatitincorporateacommitmenttothatidea.(1992:9)Intuitively,asensibleversionofrealismabout“Thatremarkwasfunny”or“Thatdeedwaswrong”doesnothavetoviewfactsaboutfunninessorwrongnessaspotentiallyverification-transcendent.Soalthoughconstruingrealisttruth-conditionsasverification-transcendenttruth-conditionsmaybeusefulforcharacterizingrealismaboutsomeareasofdiscourse,thereareotherareasforwhichthisisnotausefulchar-acterization.Theupshotofthisisthatweneedotherwaysoffleshingoutthenotionofarealisttruth-condition.ThemostimportantrecentworkonthephilosophicalagendainitiatedbyDummetthasconsistedofattemptstodojustthis.InTruthandObjectivity,CrispinWrightarguesthatnon-cognitivism–thedenialthatthesentencesofadiscoursearetruth-aptorevenpossesstruth-conditions–doesnotprovideausefulwayofformulatingoppositiontorealism.Thedebatebetweenrealismandanti-realismaboutadiscoursetakesplaceonlyafterithasbeengrantedthatthesentencesofthatdiscoursearetruth-apt.TherearetwomainpartstoWright’ssketchoftheshapeofthedebates.First,hedevelopsaversionofminimalismabouttruth-aptness,accordingtowhichallofthediscourses,includingmorals,comedy,theexter-nalworld,mathematics,thepast,andsoon,doturnouttobetruth-apt.Wright’sapproachisthussuperiortoDummett’s,insofarasheisnotcontentsimplytoignorethedebatebetweencognitivismandnon-cognitivismaboutaregionofdiscourse.Second,hedevelopsanumberofwaysofcharacterizingrealismandanti-realismaboutdiscourseswhosetruth-aptnesshasalreadybeengranted–thatis,anumberofdiffer-entwaysinwhichtruth-conditionscanbemoreorlessrealist.Itturnsoutthatviewingthesentencesofadiscourseashavingpotentiallyverification-transcendenttruth-conditionsisonlyoneofanumberofwaysofcharacterizingrealism.Wright’sapproachisthussuperiortoDummett’sinsofarasitdoesnotinvolvesaddlingthemoralrealistwithclaimsaboutpotentialverification-transcendencethatanysensiblemoralrealistwouldbaulkat.BothpartsofWright’sprogramhavebeenwidelydiscussedintheliterature,andarethepointofdepartureforphilosopherswishingtofollowupthedebatestartedbyDummett.Formore,seeHale1997andMiller1998:ch.9.OtherworkIntheabove,IhavetouchedbrieflyonlyonthoseaspectsofDummett’sworkthatItaketobehismostimportantcontributiontoanalyticphilosophy.Therearemanyotheraspectswhichlackofspacehaspreventedmefromdiscussing,andIcanmentiononlyafewofthesehere.Frege:PhilosophyofMathematics,isDummett’sstudyofFrege’sPlatonistandlogicistviewsonarithmetic.Muchofthebookiscriticaloftheattempts390\nMICHAELDUMMETTofWrightandHaletodevelopa“neo-Fregean”viewofarithmeticin,respectively,Frege’sConceptionofNumbersasObjects,andAbstractObjects.Forsomeneo-FregeanresponsestoDummett’scriticisms,seeHale1994andWright1994.Dummett’swrit-ingsonthephilosophyofmathematicscannoteasilybedisentangledfromhiswritingsonthephilosophyoflanguage,butkeypapersare1978:essays11and12and1993b:essay18.ForDummett’sintroductorysurveyofthearea,seehis1998paper“PhilosophyofMathematics.”Dummetthasalsodoneimportantworkonvagueness(1978:essay15)andcausation(1978:essays18and19,1993b:essay15).BibliographyWorksbyDummett1973:Frege:PhilosophyofLanguage,London:Duckworth.1977:ElementsofIntuitionism,Oxford:ClarendonPress.1978:TruthandOtherEnigmas,London:Duckworth.1981:TheInterpretationofFrege’sPhilosophy,London:Duckworth.1991a:FregeandOtherPhilosophers,Oxford:ClarendonPress.1991b:Frege:PhilosophyofMathematics,Cambridge,MA:HarvardUniversityPress.1991c:TheLogicalBasisofMetaphysics,Cambridge,MA:HarvardUniversityPress.1993a:OriginsofAnalyticalPhilosophy,Cambridge,MA:HarvardUniversityPress.1993b:TheSeasofLanguage,Oxford:ClarendonPress.1998:“PhilosophyofMathematics,”inPhilosophy2,ed.A.Grayling,Oxford:OxfordUniversityPress.WorksbyotherauthorsAyer,A.J.(1946)Language,Truth,andLogic,NewYork:DoverPress.Devitt,M.(1993)RealismandTruth,Princeton,NJ:PrincetonUniversityPress.Evans,G.(1982)TheVarietiesofReference,Oxford:OxfordUniversityPress.Frege,G.(1953)TheFoundationsofArithmetic,Evanston,IL:NorthwesternUniversityPress.——(1960)“OnSenseandMeaning,”inTranslationsfromthePhilosophicalWorksofGottlobFrege,ed.P.GeachandM.Black,Oxford:OxfordUniversityPress,pp.56–78.——(1967)“TheThought,”inPhilosophicalLogic,ed.P.Strawson,Oxford:OxfordUniversityPress.Hale,B.(1987)AbstractObjects,Oxford:BlackwellPublishers.——(1994)“Dummett’sCritiqueofWright’sAttempttoResuscitateFrege,”PhilosophiaMathematica3/2,pp.122–47.——(1997)“RealismanditsOppositions,”inTheBlackwellCompaniontothePhilosophyofLanguage,ed.B.HaleandC.Wright,Oxford:BlackwellPublishers,pp.271–308.Locke,J.(1975)AnEssayConcerningHumanUnderstanding,ed.P.Nidditch,Oxford:OxfordUniversityPress.McDowell,J.(1998a)Meaning,KnowledgeandReality,Cambridge,MA:HarvardUniversityPress.——(1998b)Mind,Value,andReality,Cambridge,MA:HarvardUniversityPress.Miller,A.(1998)PhilosophyofLanguage,London:UCLPress.Putnam,H.(1981)Realism,Truth,andHistory,Cambridge:CambridgeUniversityPress.Rosen,G.(1995)“TheShoalsofLanguage,”Mind104,pp.599–609.Tennant,N.(1987)Anti-RealismandLogic,Oxford:ClarendonPress.——(1997)TheTamingoftheTrue,Oxford:ClarendonPress.391\nALEXANDERMILLERWittgenstein,L.(1974)PhilosophicalInvestigations,Oxford:BlackwellPublishers.Wright,C.(1983)Frege’sConceptionofNumbersasObjects,Aberdeen:AberdeenUniversityPress.——(1986)Realism,Meaning,andTruth,Oxford:BlackwellPublishers.——(1992)TruthandObjectivity,Cambridge,MA:HarvardUniversityPress.——(1994)“CriticalNoticeofDummett’sFrege:PhilosophyofMathematics,”PhilosophicalBooks35,pp.89–102.392\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200132HilaryPutnam(1926–)JOHNHEILThePhilosophicalLexiconcontainsthefollowingentryfor“hilary”:hilary,n.(fromhilaryterm)Averybriefbutsignificantperiodintheintellectualcareerofadistinguishedphilosopher.“Oh,that’swhatIthoughtthreeorfourhilariesago.”(Dennett1987:11)TheentrymakesreferencetoHilaryPutnam’spenchantforchanginghisviews,evencompletelyreversinghimselfoncentralthemes.Whatarewetomakeofthisinconstancy?Emerson,inafamousbutwidelymis-quotedpassagewrote:Afoolishconsistencyisthehobgoblinoflittleminds,adoredbylittlestatesmenandphilosophersanddivines.Withconsistencyagreatsoulhassimplynothingtodo.(Emerson1940:152)Philosophers,especiallytechnicallyadroitphilosophers,oftenadoptapositionthenstickwithitnomatterwhat.Ifyouareclever,youcanfindwaysaroundalmostanyobjection.Defendingacherishedthesiscanbelikedefendinganoldfriendagainstachargeofdishonesty:yourownhonoraswellasyourfriend’sisatstake.Inphilosophy,however,candortrumpsconstancy.ItistoPutnam’screditthathehasbeenwillingtoallowhisviewstoevolveastheywillevenwhenthisleadshiminsurprisingdirections.Commendableasitis,thiskindofintellectualforthrightnessputspressureonanyonesettingouttosummarizePutnam’sviews.NotonlyhasPutnamwrittenonawidevarietyofissues,buthistakeonthoseissueshasshifted,oftendramatically.ThereisnotonePutnam,butmany.InwhatfollowsIhaveselectedfromamongtheavailablePutnamsthosethatseemtometohavehadthemostimmediatephilosophicalimpact.Inevitably,Ihavehadtoleaveoutmuchthatisinteresting.IshallnotdiscussanyofPutnam’simportanttechnicalworkinthephilosophyofmathematics,logic,andthephilosophyofscience.(HisviewsontheseandmanyothertopicscanbefoundinPutnam1975a,1975b,and1983.)Ishallfocus–selectively–onthreedomainsinwhichPutnamhashadconsiderableinfluenceoverphilosophyasitisnowpracticed:philosophyoflanguage,philosophyofmind,andmetaphysics.393\nJOHNHEILPhilosophyoflanguageSupposeyouutterthesentence“That’sabanana”meaningtoindicateabanana.Whatisitaboutyouthatmakesitthecasethatyourutterance,accompaniedperhapsbyagesture,indicatesabanana?Onenaturalresponseisthatthepertinentfeatureofyouthatmakeitthecasethatyourutteranceconcernsabananaisyourownstateofmind.Whenyoudelivertheutterance,yourmindisfocusedinacertainway(onbananas!),andthismentalfocusingiswhatgivesyourlinguisticutteranceitssignificance.Iunder-standyourutteranceasindicatingabananawhentheutterancetriggersinmeacom-parablestateofmind.Thisstateofmindconstitutesmygraspofyourutterance’ssignificance.Proponentsofthetraditionalideationalmodelofmeaningappealtoideasormentalimages.Asyouutterthesentence,youentertainabananaimage.Theimagesecuresaconnectionbetweenyourwordsandtheworld.Imageryisnotessential,however.Ameaningmightbeakindofdefinitionyoucarryaroundinsideyourhead,arecipethattellsyouwhentoapplyaparticularterm.(Letusignorearegressproblemthatseemstoundercutanysuchview:ifweunderstandtermsonlybypossessingadefinition,whatenablesustounderstandthetermsconstitutingthatdefinition?)Thinkofimagesandmentalrecipesasmechanismsforfixingtheextensionofterms.(Theextensionofatermisthesetofobjectsitdesignates.TheEnglishword“water,”forinstance,desig-natesakindofstuff.Whenyouusethisword,youdesignatethatstuff.)Approachingthetopicfromthisdirectioninvitesustodistinguishsharplybetweenmeaningandreference.Themeaningof“water”includesonlyelementsthatyouthespeakercangrasp.Competentspeakersneedknownothingofthechemicalcom-positionofwater;theymaybeignorantofthenatureofthestuffdesignatedbytheterm“water.”Eventhosewithanintimateknowledgeoftheconstitutionofwaterarerarelyinapositiontoapplythisknowledgetodeterminewhenaliquidsubstanceisorisnotwater.Ifweeliminatespecializedknowledgeasarequirementforknowingthemeaningof“water,”weareleftwiththeideathatthetermmeansaclear,colorless,tastelessliquidfoundinoceans,ponds,andrainpuddles.Yourgraspofthismeaningiswhatenablesyoutousetheterm“water”correctly.Themeaning,under-stoodasakindofrecipeorrulegraspedbyspeakers,iswhatconnectstheword“water”tothestuff,water.Wittgensteininauguratedasustainedattackonviewsofthiskindintheperiodbetweenthetwoworldwars(seeWittgenstein1953).AccordingtoWittgenstein,meaningisdeterminedbysocialandcontextualfactors.Yourutterancemeanswhatitdoes,notbecauseofastateofmindthatliesbehindtheutterance,butbecauseyouproducetheutteranceasamemberofaparticularcommunityoflanguageusers.Inthiscommunitywordsareseamlesslyintegratedwithactionsandwithinteractionswithnon-linguisticstatesofaffairs.Tounderstandthemeaningofawordistounder-standhowitfiguresinthepractices–linguisticandotherwise–ofmembersofyourcommunity(seeWITTGENSTEIN).Thisunderstandingisultimatelygroundedinyourcapacitytoengageinthepertinentpractices.Putnam’sapproachtomeaningcanbeseenasinvolvinganarticulationandexten-sionofWittgenstein’sinsight.Putnambeginswithanattackonthefamiliardistinc-tionbetweenmeaningandreference.Welearntousetheterm“water”atanearlyage.394\nHILARYPUTNAMLaterwelearnthatthestufftowhich“water”refersisH2O.Inthiswaywediscoverempiricallywhatwateris.Thisdiscovery,however,shedslightaswellonwhatourterm“water”means.Inspeakingofwaterwemeantobespeakingaboutstufflikethisstuff(herewepointtosomewater).Wemayknownothingofthestuff’shiddennature,butweuse“water”todesignatethestuffwiththenatureofthisstuff,whateverthatmightbe.Thestuffinquestionis,aswenowknow,H2O.Sowater’sbeingH2Oispartofthemeaningof“water.”TwinEarthToreinforcethispoint,PutnaminvitesustoimagineadistantplanetthatpreciselyresemblesEarthinalmosteveryrespect(seePutnam1975b:ch.12).Itscontinentsarearrangedexactlyasourcontinentsarearranged,itsinhabitantsspeaklanguagespre-ciselyresemblinglanguagesspokenonEarth.Wereyouinstantaneouslytransportedtothisplanet,youwouldnoticenodifferencesatall.Inhabitantsoftheplanetwholiveinaplacetheycall“America”refertotheirplanetas“Earth,”buttoavoidconfusionletusdubitTwinEarth.TwinEarthdiffersfromEarthinjustonerespect:thecolorless,tasteless,odorlessliquidstuffthatfillsTwinEarthoceans,rivers,icetrays,andfishtanksisnotH2O,butadifferentsubstance,XYZ.AlthoughXYZsuperficiallyresemblesH2O,itpossessesaverydifferentchemicalconstitution.Inotherrespects,however,TwinEarthpreciselyresemblesEarthdowntothelastdetail.Supposethatyouutterthesentence,“I’llhaveaglassofwater,please.”Yourutter-anceconcernswaterand,assumingitoccursinappropriatecircumstances(youarenotrehearsingforaplay,forinstance,ormakingaphilosophicalpoint),youareissuingarequestforaglassofwater.Imagine,now,thatyourtwinonTwinEarthproducesanexactlyresemblingutterance.Yourtwin’sutterancedoesnotconcernwater,nordoesyourtwinrequestaglassofwater.WaterisH2O,andthestuffcalled“water”onTwinEarthisnotH2O,butXYZ.Wemightsaythatyourtwin’sutteranceof“water”con-cernstwinwater;yourtwinisrequestingaglassoftwinwater.YouandyourTwinEarthcounterpartmaybeasalikeasyouplease(leavingasidethefactthatthechemicalconstitutionofyourrespectivebodieswillbeimportantlydiffer-ent!);theimagesrunningthroughyourmindcouldpreciselyresembletheimagesrunningthroughyourtwin’smind;yourfeelingscouldbethesame.Yetyourutteranceandyourtwin’sappeartohavedifferentmeanings.Giventheintrinsicsimilaritiesbetweenyouandyourtwin,themeaningsofwordsyouuttermustbedeterminedbysomethingotherthanyourintrinsicmakeup.Youandyourtwinmaybeentirelyigno-rantofthechemicalconstitutionofwhatyoueachcall“water.”Indeed,Englishspeak-ers(andtheircounterpartsonTwinEarth)usedthetermforgenerationsbeforeanyonewasinapositiontoappreciatethatwaterwasaparticularsortofchemicalcompound.Earlier,weregardedthisasagoodreasontosupposethatthemeaningofwordsmustbelimitedtowhatspeakerscanindividuallygrasp.ButthisconceptionofmeaningasstronglydistinguishedfromreferenceispreciselywhatPutnam’sTwinEarththoughtexperimentchallenges.Ifwethinkofthemeaningsofourtermsaswhatfixestheexten-sionofthoseterms–wheretheextensionofatermisjustthestufforsetofobjectsdes-ignatedbytheterm–thenwemustgiveuptheideathatmeaningsarelikepicturesorrecipeswecarryaroundinsideourheadsandconsultwhenweapplywordstoobjects.395\nJOHNHEILThedivisionoflinguisticlaborEnglishspeakersusetheterms“beech”and“elm”todesignatespeciesoftree.Ifyouarelikeme,however,youwouldbehardpressedtosayhowbeechesandelmsdifferandutterlyunabletodistinguishabeechfromanelminthewild.Doesthismeanthat,forEnglishspeakerswholackacapacitytotellbeechesandelmsapart,thewords“beech”and“elm”aresynonymous?Thatseemsimplausible.Putnamsuggeststhat,whenitcomestosuchnaturalkindterms,werelyonadivisionoflinguisticlabor.(Naturalkindterms–“gold,”“water,”“planet,”“tiger”–designatestuffsandobjectsthoughttooccurnaturally,andaredistinguishedfromartifactualkindterms:“table,”“senator,”“dollarbill.”)Weuse“elm”and“beech,”forinstance,todesignatespeciesoftreethatwouldbesolabeledbyexperts.TwinEarthcasesandthephenomenonofthedivisionoflinguisticlabormakeitclearthatagentswhoareindiscernibleinallrelevantintrinsicrespectscouldnever-thelessdifferinwhattheymeanbytheirutterances.Ifthisisright,accountsofmeaningthatfocussolelyonagentsconsideredinisolationareboundtofail.Anade-quateaccountofmeaningapparentlybringswithitabatteryofsocialandcontextualelements.Putnamputsitsuccinctly:“Cutthepieanywayyoulike,‘meanings’justain’tinthehead!”(1975b:227).Thephilosophicalimpactofthisthesis–Ishallcallitexter-nalism–wouldbehardtooverstate.Asweshallsee,whatgoesformeaninggoesforthoughtaswell.IfPutnamisright,thetraditionalconceptionofthemindasaspec-tatoronthe“externalworld”mustbeabandoned,replacedbyaconceptionofthemindasconstituting–andconstitutedby–theworld.Thisistogetaheadofourstory,however.LetuslookfirstatPutnam’sarticulationanddefenseoffunctionalism,aconceptionofthemindaccordingtowhichmindscom-prisesystemsofrelationsamongelementsthatresemblethestatesofacomputingmachine.PhilosophyofmindInthe1950s,English-speakingphilosophersunderthespellofWittgensteincametoregardphilosophicalquestionsasexpressionsoflinguisticbefuddlement.Weask,forinstance,“Whatistruth?”andinterpretthisasasubstantivequestion,onethatcallsfortheinvestigationofsomeindependentlyexistingreality.Wittgensteinarguedthatsuchquestionsoccurtousonlywhenwedistanceourselvesfromlinguis-ticpracticesthatgiveformtotalkoftruth(oranyotherphilosophicallychallengingconcept).Augustine(Confessions,XI,xiv)remarkedabouttime:“Whatthenistime?IfnooneasksofmeIknow;ifIwishtoexplaintohimwhoasks,Iknownot.”Thesentiment(thoughnotAugustine’ssubsequenttreatmentofit)isprofoundlyWittgensteinian.Solongasweuselanguageinthepursuitofordinaryhumanends,weremaininnocentofphilosophy.Wearemovedtophilosophicalquestion-ingwhenwestepoutsidethelinguisticpracticesthatgroundouruseofwords.Weask“Whatistime?”andseekananswerinawaythatignorestheway“time”anditscognatesareactuallydeployedinourlinguisticcommunity.Onceweloseourmooringswithinlanguage,ourtheorizingiscoloredbyamisapprehensionof396\nHILARYPUTNAMtherolesoftermsthatgeneratefamiliarphilosophicalpuzzles.Thismisapprehen-sionissystematic:thesamekindsoftheoryariseoverandoverinthehistoryofphilosophy.Wittgenstein’spositiveproposalisdeflationary.Philosophicalpuzzlementrequirestreatment.Whenaphilosopherre-immerseshimselfinthelinguisticpracticesandformsoflifethatgivesensetothetermshefindsbewildering,thebewildermentebbs.Philosophicalquestionsarenotansweredbutlaidtorest.Thetemptationtoposesuchquestions,althoughperfectlynatural,requiresakindoftherapy(seeWITTGENSTEIN).Thephilosopherwhorespondstothistherapyisnolongerimpelledtophilosophize;theendofphilosophyistheendofphilosophy.Wittgenstein’sapproachtophilosophicalissuesconcerningthemindledtotherejectionofthetraditionalideaofmindsasmentalorgansthatreceiveinputsviathesensesandyieldoutputsintheformofutterancesandbodilymotions.“Mind”isasubstantivenoun,buttalkofmindsisnottalkofasubstanceorentityassociatedwith,butsomehowdistinctfrom,thebody.Onthecontrary,inregardingyouaspossessingamind,Iregardyouasengaginginintelligentactivities,respondingtotheworldinintelligibleways,andsoon.Thoughtsliketheseled,inturn,tophilosophicalbehav-iorism(seeRYLE):possessingamindisexclusivelyamatterofbehaving,orbeingdis-posedtobehave,inparticularways(see,e.g.,Ryle1949;foraresponse,seePutnam1975b,chs14,15,16).Behavioristshopedtoanalyzeortranslatetalkofmentalgoings-on(feelings,thoughts,intentions)intotalkofbehaviorandbehavioraldisposi-tions.Tobedepressed,forinstance,isnottobeinaparticularkindofinnerstate,buttomopeabout,complain,orbedisposedtocomplain,andthelike.Behavioristsneednotdenythatinnerstatesaccompanyboutsofdepression,onlythattheseinnerstatesarethedepression.Onedifficultyforthebehavioristprogramstemmedfromthefactthatbehavioristanalysesofmentalconceptstypicallyincludedreferencetoothermentalconcepts.Someonewhoisdepressed,forinstance,isdisposedtoformthoughtsofcertainsorts,andtoacquire(orlose)certainmotivesanddesires.Whenweattempttoanalyzethesementalconceptsbehavioristically,wefindwemustappealtoothermentalconcepts;analysesoftheseconceptsrequirereferencetostillothermentalconcepts;andsoon.(Asweshallsee,theinterconnectednessofmentalconceptscomestotheforewiththedevelopmentofbehaviorism’sintellectualsuccessor,functionalism.)Theanalyticalprogramofbehaviorismwaschallenged,firstbytheadventofthemind–brainidentitytheory(seePlace1956,Smart1959)andthenbyfunc-tionalism.Mind–brainidentitytheoristsdefendedthethesisthatconsciousstateswereatbottomstatesofbrains.Theyarguedthatthekindsofcorrelationknowntoholdbetweensubjects’reportsofstatesofconsciousnessandstatesofthosesubjects’brainsarebestconstruedasevidencefortheidentificationofstatesofconscious-nesswithbrainstates.Imaginethat,whileshopping,youdropacanoftomatosouponyourfootand,asaresult,youexperienceathrobbingpaininyourbigtoe.Neuroscientiststellusthat,whenyouexperienceapainofthissort,certainkindsofeventoccurinyourbrain.(LetuspretendthatpainsareassociatedwiththefiringofC-fibersinthespinalcord.)IdentitytheoristsarguedthatthebestexplanationofthecorrelationbetweenC-fiberfiringsandreportsofpainwasthatbeinginpainjustisthefiringofC-fibers.397\nJOHNHEILFunctionalismNowadaysmanyscientificallymindedtheoristsregarditasclosetoobviousthatstatesofmindarebrainstates,mentaleventsareneurologicalevents.Professionalphiloso-phers,however,havebyandlargeresistedthisconclusion.Thisisnotbecausephiloso-phershaveapreferenceformind–bodydualism,butbecausemostphilosophershavebeenconvincedbyargumentspioneeredbyPutnamthatthemind–brainidentitytheorysuffersafundamentaldefect(seePutnam1975b,chs18,19,20).Considerthefactthatweunhesitatinglyascribestatesofmindtocreaturesotherthanhumanbeings.Thinkofbeinginpain,andsupposeforthesakeofargumentthemind–brainidentitytheorywerecorrect:painsarebrainstates;yourbeinginpainisyourbeinginaparticularkindofbrainstate:yourC-fibersarefiring.Sofar,sogood.Butnowconsider:cananoctopusfeelpain?Itsurelyseemsso.Theneurologicalmakeupofanoctopusisverydifferentfromtheneurologicalmakeupofahumanbeing,however.Thisseemstoimplythatoctopodes,sportingadifferentphysiology,lackacapacityforpain:ifpainsareC-fiberfirings,andoctopodes’painresponsesaretrig-geredbydifferentmechanisms(astheysurelyare),thenoctopodesdonotfeelpain!Supposeweencounteredintelligentcreaturesfromdistantplanetswhowerelikeusinmanywaysbutwhosebiologywassilicon-based.Wemighthaveexcellentgroundsforregardingsuchcreaturesasexperiencingpain,yet,ifthemind–brainidentitytheoryweretrue,thiswouldbeimpossibleifsuchcreatureslackedC-fibers(astheyalmostcer-tainlywould).Ifhavingapainisamatterofbeinginaparticularkindofneuralstate,nocreaturelackingsuchstatescouldexperiencepain.Ifhumanbeings,octopodes,andAlphaCentaurianscanallexperiencepain,thenitishardtoseehowpaincouldbeidentifiedwithkindsofbrainstatefoundonlyinhumanbeings(andtheirnearrelations).Suppose,however,wethinkofpainstates,notasneu-rologicalstates,butonthemodelofcomputationalstates.Reflectonanordinarydesktopcomputer.Thedevice’soperationisgovernedbyprogramsthatitruns.Whenyouelecttoprintadocumentyouhavebeenworkingon,yourdesktopcomputerrunsasimpleprogramthatsendssignalstoaprinter,whichthenprintsthedocument.Nowsupposewedistinguishbetweentheprogramyourdesktopcomputerisrunningandaparticu-larphysicalimplementationofthatprogram.Themachine’srunningtheprogramisamatterofitsgoingintoasequenceofphysicalstates.Thesestates,wemightsay,realizetheprogram.Note,however,thatadifferentmachinecouldruntheverysameprogrambygoingintoasequenceofverydifferentkindsofphysicalstate.Inthe1950s,com-putingmachinesconsistedofungainlyarraysofvacuumtubes;moderncomputersmakeuseoftinytransistors;inthenineteenthcentury,CharlesBabbageconstructedasophisticatedcomputingmachineusingbrassgearsandcylinders;andtodaythereistalkofmolecularcomputers.Itispossibleforallofthesedevicestoruntheverysameprogram,toengageintheverysamesequenceofcomputations,andsotoencompasstheverysamecomputationalstates.Distinctmachinescanbeinthesamecomputationalstate,then,eveniftheyaremadeofverydifferentphysicalingredients.Allthatisrequiredisanisomorphism–aone–onecorrespondence–betweensequencesofoperationsperformedbythemachines(andsamenessofinputsandoutputs).Youfeedintoasimplecalculator“7,”“+,”and“5,”andthecalculatordisplays“12.”Thecausalchainleadingthecalculatorthrough398\nHILARYPUTNAMthiscomputationhasacertainphysicalcharacter.WhenyoutypethissamesequenceintoyourdesktopcomputerordialitintoaBabbagemachine,thesedevicesgothroughvastlydifferentkindsofcausalsequencetoarriveatthesameoutput:“12.”Atanyrate,thesequencesarevastlydifferentconsideredsolelyasphysicalevents.Theyexhibit,however,acommonstructure,acorrespondingsetofrelations.Youmightputthisbysayingthat,consideredconcretely,theeventsareverydifferentbut,consideredatahigherlevelofabstraction,thesequencestheyembodyarethesame.Whathasanyofthistodowiththemind?Insuggestingthatstatesofmindarecom-putationalstates,Putnamisnotimaginingthatcreatureswithminds–humanbeings,forinstance–are“mererobots,”creatureswhoseactionsareinflexibleand“mindless.”Theidea,rather,isthatstatesofmindowetheiridentity,nottotheirphysicalmakeup,buttotheirplacewithinastructuredsystem.Toavoidmisleadingassociations,Ishallspeakhenceforth,notofcomputationalstates,butoffunctionalstates.(Arefunctionalstatesandcomputationalstatesco-extensive?Thisiscontroversial.Acomputationalstatecanbegivenaparticularsortofformalcharacterization.Ifeveryfunctionalstateorprocessischaracterizableinthisformalway,theneveryfunctionalstateisacom-putationalstate.)Functionalstatesarepickedoutbyreferencetorolestheyoccupy–functionstheyperform–withinasystem.Ananalogymayhelp.Wayneisavice-presidentoftheGargantuanCorporation.WhatexactlydoesWayne’sbeingavice-presidentamountto?Wayneis175cmtall,balding,andoverweight.TheseintrinsicpropertiesofWayneseemnottobearonhisbeingavice-president.Waynecouldbe“re-orged,”andreplacedbyBecky,apetitbrunette,byOscar,arobot,byHans,achimpanzeefluentinsignlan-guage,orevenbyRenée,animmaterialangel.Wayneisavice-president,notinvirtueofhisintrinsicproperties,butinvirtueofrelationshebearstoothersintheorganiza-tioninwhichheoccupiesthisoffice.Anyone(oranything!)bearingtheserelationswouldbeavice-president.Functionalistscontendthatwhatgoesforvice-presidentsgoesforstatesofmind.Beinginpainisnotamatterofbeinginaparticularkindofneurologicalstate,butamatterofbeinginastatethatbearstherightsortsofrelationtoothercomponentsofthesystemtowhichitbelongs.Inyourcase,aparticularneurologicalstateoccupiesthepainrole;inthecaseofanoctopusoranAlphaCentaurian,verydifferentkindsofphysicalstatefillthepainrole.Acreatureisinastateofpainwhenitisinastatethatistypicallycausedbytissuedamage,andthatcausescertaincharacteristicbeliefsanddesires(thebeliefthatthishurts,forinstance,andadesireforthepaintostop),andcertaincharacteristicactions(ifyou’vesteppedonatackyouwillquicklymoveyourfoot).Thismakesfunctionalismsoundlikedressedupbehaviorism.Functionalism,however,unlikebehaviorism,doesnotrequirethatstatesofmindbecharacterizablesolelyintermsofstimuliandresponses.Howyourespondtopain–yourbehavior–candependpartlyonwhatyoubelieveanddesire.Ifyouaretryingtoimpressacom-panionwithyourtoughness,youmayshrugoffapainthatyouwouldreacttoverydifferentlywereyoualone.Youmayworrythatthiswayofcharacterizingmentalstatesisultimatelycircular.Wedesignateamentalstatebynotingitsrelationstoothermentalstates.These,inturn,arecharacterizedbyreferencetoothermentalstates.Eventuallywecomebacktotheoriginalstates.399\nJOHNHEILThethreatofcircularityiswardedoffbymeansofatechniqueintroducedinadif-ferentcontextbyFrankRamseyandrefinedbyDavidLewis(seeLewis1972).Theissuesaretechnical,buttheguidingideaisstraightforward.Imaginethatyoudefinestatesofmindbylocatingthemasnodesinanetworkofnodes,eachofwhichrepresentsadis-tinctivekindofmentalstate.Thesystemofnodesisanchoredatoneendbyrelationstoincomingstimuli,andattheotherendbybehavioraloutputs.Nowthepainnodewillhaveacertainuniquestructuralrelationtoothernodesinthesystem;afeelingofpleasurewillhaveanotherkindofstructuralrelation;andabeliefordesirewillexhibitotherkindsofstructuralrelation.Wecanthensaythatbeinginpainisamatterofbeinginastateexhibitingthesekindsofrelationtoelementsinasystemwiththiskindofstructure.Putnamsummarizesthislineofreasoningbydescribingstatesofmind(indeedcom-putationalorfunctionalstatesgenerally)asmultiplyrealizable.Thismeansthattheverysamestateofmindcanberealizedbymanydifferentkindsofphysical(orperhapsnon-physical,ectoplasmicorangelic)state.Ifonestatecanhavemanyrealizers,thatstatecannotbeidentifiedwithorreducedtoanyofitsrealizers.Thus,althoughstatesofmindarepossessedbyordinaryconsciousagentsbyvirtueofthoseagents’beinginsomephysicalrealizingstate,mentalstatesarenotreducibletothephysicalstatesthatrealizethem–orsofunctionalistscontend.Despiteitsimmensepopularity,functionalismhasbeenwidelycriticized.Putnamhimselfhasbeenamongitsmostvocalcritics.Evenso,itisfairtosaythatfunctional-ismremainshugelyinfluential,bothinsideandoutsidephilosophy.Functionalismpro-videsawayofunderstandinghowmentalitycouldbehousedinthebrainsofhumanbeings(andinthenervoussystemsofotherintelligentspecies).Inaddition,function-alismleavesroomfordistinctivelevelsofexplanation.Wemightcometounderstandthebehaviorofacomputingmachinebyinvestigatingitsphysicalmakeuporbystudy-ingitsprogram.Inthesameway,youmightexplainmybehaviorbycitingcomplexprocessesinmycentralnervoussystemorbyreferencetomybeliefs,desires,andinten-tions.Ofcourse,thephysicalmakeupofacomputingmachinemightbeextremelycomplicated,andthephysicalmakeupofahumanbeingmorecomplicatedstill.Inmostcases,apurelyphysicalexplanationofthebehaviorofeitherwouldbeapracticalimpossibility.Nevertheless,functionalismprovidesawayofseeinghowwecouldbewarrantedinoffering“higher-level,”functionalexplanationsofthebehaviorofcomplexsystems,anddoingsoinawaythatdoesnotcompetewithlower-level,purelyphysical,explanations.FunctionalismspurnedRecallPutnam’slineonmeaning:themeaningofyourutterancesdepends,notmerelyonyourintrinsicfeatures,butonrelationsyoubeartoyoursurroundings.Whenyouutter“that’swater,”forinstance,yourutteranceconcernswater(H2O)onlyifyoustandinanappropriaterelationtowater.WhenyourTwinEarthcounterpartproducesanindistinguishableutterance,thatcounterpartsayssomethingdifferent.Whenyouspeakofelmsandbeecheswhatyoumeanisdeterminedinpartbyexpertsinyourlin-guisticcommunitywhoareinapositiontoidentifyanddistinguishelmsandbeeches.Inthisregard,meaningsarecommunityaffairs.Youandtheexpertmeanthesame400\nHILARYPUTNAMwhenyouspeakofelms,eventhoughyoulacktheexpert’sknowledgeofthedistin-guishingmarksofelms.Yourbeliefsaboutelmsmightbelargelyfalse,stillyoumeanby“elm”whatothersinyourlinguisticcommunitymean:yourtalkofelmsistalkofelms.Sofar,thisisathesisaboutthemeaningofutterances.Thethesisiseasilyextended,however,tothecontentsofourthoughts:whatthosethoughtsconcern.Imaginethat,onEarth,Debbieisanticipatingacooldrinkofwateronahotday.Debbieentertainsathoughtshewouldexpressbysaying“that’swater.”AtthesametimeDebbie’scoun-terpartonTwinEarth,TwinDebbie,isentertainingathoughtshewouldexpressbysaying“that’swater.”Debbie’sutteranceandthoughtconcernwater.TwinDebbie’sutteranceandthought,incontrast,arenotaboutwater;water,afterall,isH2O,andDebbie’sutteranceandthoughtarenotaboutH2O;theyareaboutXYZ,twinwater!Ofcourse,TwinDebbiecallstwinwater“water,”butthatisanothermatter.DebbieandTwinDebbie’susesof“water”resembletheuseof“burro”byaSpanishspeakerandanItalian.InthemouthofaSpanishspeaker,“burro”meansdonkey;utteredbyanItalian,“burro”meansbutter.Putnamholdsthatcaseslikethesemakeitclearthatwhatourthoughtsconcern,aswellaswhatourwordsmean,isfixed,notsolelybywhatisinsideourheadsbutbyrela-tionswebeartotheworldaroundus.“Water”inDebbie’s(butnotTwinDebbie’s)mouthmeanswaterinpartbecauseDebbie(butnotTwinDebbie)standsinanappro-priatecausalrelationtowater–H2O.Similarly,thoughtsDebbie(butnotTwinDebbie)wouldexpressbyutterancesfeaturingtheword“water”concernwater–H2O–inpartbecauseDebbie(butnotTwinDebbie)standsinanappropriatecausalrelationtowater.TwinDebbie’s“water”utterancesandthoughtsshewouldexpressbymeansoftheseutterancesconcern,notwater,buttwinwater,XYZ.Tobesure,thereisnorelevantinternaldifferencebetweenDebbieandTwinDebbie(ignoringthefactthatDebbie’scon-stitutionincludesH2O,TwinDebbie’s,XYZ).Thecontentsofourthoughts,likethemeaningsofourwords,dependonourcontext,mostparticularlyoncausalrelationswebeartoobjectsintheworldandsocialrelationswebeartoothersinourlinguisticcommunity.Theseexternalrelationsarepartlyconstitutiveofthemeaningsofwordsandthecontentsofthoughts.Itisnaturaltoextendtheexternalistthesisthatmeaningsarenot“inthehead”tothemeaningsorcontentsofstatesofmind.Thisseemstoimplythatstatesofmind,oratanyratethecontentsofstatesofmind,arenotinthehead!(Ifyouthinkthatthecontentofastateofmindisessentialtoit–thebeliefthatsnowiswhiteisessentiallythebeliefthatsnowiswhite–thenanexternalismaboutcontentstraightforwardlyyieldsanexternalismaboutstatesofmindwithcontent:beliefs,desires,intentions,andthelike.)Surelythisisridiculous!Orisit?Beforetryingtoanswerthisquestion,letusreflectontheimplicationsofallthisforfunctionalism.Afunctionalstateisastateofasystem,astatedefinablewhollybyrelationsitbearstootherstatesofthesystem(andtoinputsandoutputs).TwinEarthcases,however,appeartoshowthatdistinctagents(DebbieandTwinDebbie,forinstance)couldbefunctionallyidenticalyetdiffermentally:oneisthinkingofwater,anotheroftwinwater.Youmightregardfunctionalismasframingamoreorlesstraditional“internalist”conceptionofthemindanditscontents.Ifyouareattractedtoexternalism–abroadlycontextualaccountofthe401\nJOHNHEILcontentsofstatesofmind–youtherebyabandonthetraditionalviewofmindsasself-containedentitiesradiatingthoughtsontoan“externalworld.”Theboundarybetweenmindandworldbecomesblurred:“themindandtheworldjointlymakeupthemindandtheworld”(Putnam1981:xi).Iffunctionalstatesarepurelyinternalstatesofasystem(statesdefinablebytheirrelationstootherstatesofthesystem),however,andifstatesofmindarenot,thenstatesofmindarenotfunctionalstates.Weareinthiswayledbacktotheideathatmakingsenseofthemindanditscontentsisamatterofseeingintelligentagentsincontext.Theattempttolocatemindsinsideheadsisanalogoustoanattempttochar-acterizechesspiecessolelybyreferencetotheirintrinsicfeatures.MetaphysicsAllthisbringsustoPutnam’sattackon“metaphysicalrealism,”adoctrine,rarelyarticu-latedbutwidelytakenforgranted,accordingtowhichthemindandtheworldareseparatedbyanepistemologicalchasm.Mindsrespondperceptuallyto,andrepresenthowthingsstandin,theworld.Neverthelessmindsaredepictedasoccupyingastand-pointoutsidetheworld.Althoughmostofusarebyandlargecommittedtotheviewthatmindsarephysi-calconstituentsoftheworld,itcanstillseemperfectlynaturaltorepresenttherelationmindsbeartotheworldinawaythatsituatesthoughts“inhere”andtheirworldlyobjects“outthere.”Wearespectatorsontheworld.Perceivingisamatteroftheworldaffectingthemind(theincomingarrowinthefigure),andthinkingabouttheworldisamatterofaimingthoughtsattheworld(theoutgoingarrow).MINDPerceptionWORLDRepresentationsStates(thoughts,ofRepresentationtheories)affairsDescartes’smind–bodydualismisonlyoneextremeformofmetaphysicalrealism.Indeed,PutnamholdsthatthemetaphysicalforceofCartesiandualismliesnotinthecontentionthatmindsareimmaterialentities,butintheideathatmindsstandapartfromtheworldonwhichtheirthoughtsaredirected.Thispicture,thecoreofmeta-physicalrealism,survivesthetransitiontomaterialistconceptionsofthemind.Theresultisakindofinternalinstability.Ontheonehand,modernscienceencouragesustoregardmindsasmaterialobjectsalongsideothermaterialobjects.Ontheotherhand,metaphysicalrealismdepictsthemindasaspectatorontheworld,seeminglylocatingmindsoutsidetheworldtheyrepresent.Ifweinsistonsituatingmindsintheworld,however,wemustabandonmetaphysicalrealism.402\nHILARYPUTNAMOnewaytogetatallthisistoconsidertheimplicationsofmeaningexternalism.Externalismunderminestheconvictionthatmeaningsandmentalcontentsare“inthehead.”WhatyoumeanwhenyouutteranEnglishsentenceandwhatyourthoughtsaredirectedon,dependsinpartonyourcircumstances(andnotmerelyonyourcon-stitutionorinternalorganization).Thepointcanbeillustratedbymeansofasimpleanalogy(borrowedfromWittgenstein).Imagineapictureofasmilingface.Imaginethefaceappearingintheforegroundofadepictionofachild’sbirthdayparty.Herethefaceexpressesbenevolenthappiness.Nowimaginethefacesituatedagainstabackgroundofhorriblesuffering.Inthiscontext,thefaceexpressesevil.Perhapsourthoughtsarelikethis.Thesameformofthoughtinonecontextexpressesonecontent,and,inadif-ferentcontextexpressessomethingentirelydifferent.(“Form”herereferstothe“shape”or“intrinsiccharacter”ofathought.“Burro”inSpanishencompassesdonkeys,inItalianitdesignatesbutter.Thesamelinguisticformcanhavedifferentmeaningsindifferentlinguisticsettings.)ThisisonecomponentofPutnam’sattackonmetaphysicalrealism.Asecondcom-ponentisepistemological.Insharplydistinguishingrepresentationsoftheworld–beliefsandtheories–andtheworldthoserepresentationsconcern,wecreateanunbridgeablechasm.Ourrepresentationspurportto“match”reality,butweareinnopositionevertoeffectacomparison.Atbestwecanmeasurerepresentationsagainstotherrepresentations.Supposeyoubelievetheiceisthin.Youdecidetocheckyourbeliefbyexaminingtheice.Youarenotmeasuringyourbeliefagainsttheice,butmeasur-ingyourbeliefagainstotherperceptuallyinducedbeliefsabouttheice.Ifyouregardthisasanaturalandunavoidablefeatureofthehumanpredicament,youareatleastaclosetmetaphysicalrealist.Oneconsequenceofsuchaviewisthatitopensthedoorto“externalworldskepticism”:whatgroundscouldwehaveforthink-ingthatourbeliefsaboutthe“externalworld”aretrue?Ifwehaveaccessonlytoourownrepresentations,thenwecouldhavenoassurancethatthoserepresentations“match”therealitytheypurporttorepresent,orevenwhetherthereisanyexternalrealitybeyondtherepresentationsthemselves!ThesituationisoneDescartesdramatizedbyimagininganevildemon.Theevildemonhasthepowertomakeourbeliefsabouttheworldfalse.Descartes’sattempttoreconcilemetaphysicalrealismwiththeconvictionthat,properlypursued,knowledgeoftheexternalworldwasattainable,involvedappealtoabenevolentGod:Godissuchthathewouldnotletuserrconcerningtruthswefindindubitable.ThisgivesDescartesafoundationonwhichtoerectanaccountofknowledgeaccordingtowhichweareentitledtobeconfidentthatourbeliefsaretrueprovidedthosebeliefsarenotbasedonunprovenassumptions.Theargument’sappealtoabenevolentGod,however,strikesmostreadersasunconvincing.Notethattheskepticalchallenge–whatentitlesustosupposethatourrepresenta-tionsoftheworldmatchtheworld?–presupposesmetaphysicalrealism.Skepticismismetaphysicalrealismseeninthemirrorofepistemology(Heil1998).Metaphysicalrealismmakestheskepticalquestioninevitable,andtheskepticalquestionmakessenseonlygiventhekindofmind–worldseparationthatmakesupthecoreofmetaphysicalrealism.Arefutationofrealism,then,canbeseenasarefutationof“externalworld”skepticism–andviceversa.ThisispreciselyPutnam’sstrategy.403\nJOHNHEILBrainsinvatsAppealstoabenevolentGodaside,letusfollowPutnaminupdatingtheskepticalchal-lengebyaskingwhatgroundswehaveforbelievingthatwearenotbrainsinvats.Imaginethatyouhavebeenkidnappedbyanevilscientist,drugged,andyourbrainremovedfromyourbodyandkeptaliveinavatofnutrients.Nerveendingspreviouslyattachedtobodilyorgansareattachednowtoasupercomputer.Thecomputerpre-ciselysimulatesincomingnerveimpulses.Nervousstimulationthattwodaysagowouldhavecomefromyourretina,forinstance,nowissuesfromthecomputer.Asfarasyoucantell,theworldisunchanged.Yourvisualandauditoryexperiences,eventhekines-theticfeedbackyoureceivewhenyouseemtomoveyourbodyarefedtoyoubythecomputer.Allthis,althoughundoubtedlyfanciful,seemsatleastphysicallypossible.But,theskepticinsists,ifitispossible,howcouldweeverbeinapositiontoknow(orevenreasonablybelieve)thatwearenotbrainsinvats?Note,first,thatthebrain-in-a-vatpossibilityisapossibilityonlysolongasweacceptmetaphysicalrealismanditsattendantgapbetweenhowtheworldisandhowwerep-resentitasbeing.Doesthisgiveusareasontoabandonrealism?Ithardlyseemsso.Ifrealismimpliesthatwemightbebrainsinvats,thenthisissomethingweshallhavetolivewith.(Ifyouareinclinedtodismissthepossibilityasidle,askyourselfwhatgroundsyouhavefordismissingit.)HerePutnamgoesontheoffensive.Supposemetaphysicalrealismimpliesthatwemightbebrainsinvats.Ifwecouldestablishthatitisnotpossiblethatwearebrainsinvats,thenwewillhaveestablishedthatmetaphysicalrealismisfalse.Buthowcouldanyonehopetoprovethatitisnotpossiblethatwearebrainsinvats?Wehave,afterall,grantedthattheenvisagedenvattingofabrainlieswithintherealmofphysicalpossibility.RecallPutnam’stakeonmeaning.Themeaningofanutterance(orthecontentofathoughtyoumightexpresswiththatutterance)dependsoncontext,mostespeciallyonrelationsspeakersandthinkersbeartotheirsurroundings.Inthesimplestcase,yourthoughtofatreeconcernsthistreebecausethistree(andnoother)iscausallyrespon-sibleforit.Supposewegeneralizethisobservation.TheEnglishwords“brain”and“vat”meanwhattheydoinpartbecausemembersoftheEnglish-speakinglinguisticcommunitystandinappropriatecausalrelationstobrainsandvats.Similarly,yourthoughtsofbrainsandvats,thoughtsyoumightexpressusingtheterms“brain”and“vat,”concernbrainsandvatsbecauseyouare,asanEnglishspeaker,anagentstand-inginappropriatecausalrelationstobrainsandvats.Thecausalrelationsinquestionarenodoubtcomplex,anditwouldbedifficulttospellthemoutindetail.But,astheTwinEarthcasesseemtoshow,thepresenceofanappropriatecausalconnectionisatleastanecessaryconditionforourwordsandthoughtsconnectingtotheworld.Debbie’sutteranceof“water”designateswater,andthoughtsshewouldexpressusingthistermconcernwater,inpartbecauseDebbie’siscausallyrelatedtowater.TwinDebbie’sutterancesof“water”differinmeaningandhercorrespondingthoughtsdifferintheircontentbecauseTwinDebbiestandsincomparablecausalrelations,nottowater,buttoXYZ,twinwater.Letusallowthat“vat”and“brain”meanwhattheydoinpartbecausethoseofuswhodeploythesetermsstandincertaincausalrelationstovatsandbrains.Now404\nHILARYPUTNAMconsideranenvattedbrain,Evan.Consider,inparticular,Evan’scausallinkstotheworldoutsidethevatandtheirbearingonthesignificanceofEvan’s“utterances”of“brain”and“vat”(andthoughtsEvanmight“express”usingtheseterms).Sourcesofstimulationfortheseutterancesandthoughtsarenotbrainsandvats,butelectricaleventsinsidethesupercomputertowhichEvanisattached.Theseelectricalevents–stand-insforrealbrainsandvats–producesensoryexperiencesinEvanthatpreciselyresemblethesensoryexperiencesyoumighthavewhenyouencounteravatorabrain.Evan’ssituationobligesustoreconstrueEvan’sutterancesandthoughts,justaswedidintheTwinEarthcase.WhenEvanhasathoughtthathemightexpressbyutteringthesentence“That’sabrain,”weshouldhavetointerprethisutteranceasmeaningsomethinglike“That’selectricalstates1.”Similarly,whenEvanharborsathoughthewouldexpressbyuttering“That’savat,”wemustinterpretthisutteranceasexpress-ingsomethinglikewhatwewouldexpressinEnglishas“That’selectricalstates2.”Evan’sutteranceswillneedtobesystematicallyreinterpreted.Hisutterancesfor-mallyresembleEnglishutterances,buttheydiffersignificantlyinwhattheymean.WecanmarkthissystematicdifferencebydescribingEvanas“speaking,”notEnglish,butVat-English,justasyourtwinonTwinEarthspeaksTwinEnglish,notEnglish.Ofcourse,justasyourtwincallsTwinEnglish“English,”soEvancallsVat-English“English.”WhenEvan“says”“IspeakEnglish,”his“utterance,”translatedintoEnglish(ourlanguage),means“IspeakVat-English,”andthisutteranceistrue.Withallthisasbackground,weareinapositiontoappreciatePutnam’santi-skepticalargument.Supposeyouentertainathoughtthatyouwouldexpressbyutter-ingthesentence“Iamabraininavat.”IfyouareanEnglishspeaker,thenthissen-tenceisfalse.Why?IfyouareanEnglishspeaker,youareconnectedinanormalwaywithbrains,vats,trees,andthelikeandnotpluggedintoasupercomputer.IfyouareanEnglishspeaker,then,youarenotabraininavat.Astutereaderswillbequicktopointoutthatitishardtotakemuchcomfortfromthisfact.True,ifwegrantmeaningexternalism,wearenotbrainsinvatsifweareEnglishspeakers.ButwhatgivesustherighttoassumethatweareEnglishspeakers?Afterall,ifwewerebrainsinvats,wewouldnotbeEnglishspeakers!Itlooksasthough,inordertoknowthatwespeakEnglish(andnotVat-English)weshouldfirsthavetoknowthatwearenotbrainsinvats.Wecannot,then,underpainofcircularity,appealtothisfact(orallegedfact!)toestablishthatwearenotbrainsinvats.Letusthinkalittleharderaboutthis.ConsiderEvan,andhis“utterance”of“Iamabraininavat.”Evan,asweknow,isenvattedand“speaks,”notEnglish,butVat-English.Evan’s“utterance”of“Iamabraininavat”translatedfromVat-EnglishintoEnglishwouldmeansomethinglike“Iamacomputerstateoftypesn.”Butthisutter-anceismanifestlyfalse:Evanisabraininavat,notacomputerstate!Itappearsthatyoucangeneralizethispointanduseitinsimpleargumenttotheconclusionthatyouarenotabraininavat:1IfIamabraininavat,Iexpressafalsehoodinutteringthesentence“Iamabraininavat.”2IfIamnotabraininavat,Iexpressafalsehoodinutteringthesentence“Iamabraininavat.”3IamabraininavatorIamnotabraininavat.405\nJOHNHEIL4Inutteringthesentence“Iamabraininavat,”Iexpressafalsehood.(From(1),(2),and(3))5Inutteringthesentence“Iamabraininavat,”IutterasentencemeaningthatIamabraininavat.6Iamnotabraininavat.(From(4)and(5))Thisargumentisvalid:itspremiseslogicallyimplytheconclusion.Butistheargumentsound?Doesitestablishwhatitpurportstoestablish?Doestheargumentenableyoutoprovethatyouarenotabraininavat?Beforetakingupthisquestion,letusremindourselveswhytheissueisimportantforPutnam.Metaphysicalrealism,inbifurcatingmindandworld,impliesthatitispos-siblethatwearemassivelydeluded:itispossiblethatwearebrainsinvats.Ifwecanexcludethisskepticalpossibility,weshallhavetherebyestablishedtheinadequacy–thefalsehoodorperhapsincoherence–ofmetaphysicalrealism.DoesPutnam’sargumentwork?Supposethat,afterstudyingtheargumentyoucon-cludethatitfailstoshowthatyouarenotabraininavat:youmightbeabraininavatanyway.Supposeyouexpressthispossibilityviathesentence“Iamabraininavat.”Aswehaveseen,yourutteranceofthissentenceisfalseifyouareanEnglishspeaker,anditisfalseifyouspeakVat-English.Ineithercaseitisfalse.Assuming(forsimplic-ity)thesearetheonlypossibilities,thesentencemustbefalse.Yes,youmightreply,butitisvitaltoknowwhatthefalsesentencemeans.Premise(5)oftheargumenttellsusthatthesentencemeansthatyouareabraininavat.Ifyouknewthis,andknewaswell,thatthesentencewasfalse,youwouldknowthatyouarenotabraininavat.Butwhyshouldweacceptpremise(5)?Remember,wearesupposingthatyouarerunningthroughtheargumentinanefforttoestablishthatyouarenotabraininavat.Imaginethatyouhavejustrunthroughpremise(5).Howcouldpremise(5)expressafalsehood?Toseethedifficultyindenying(5),pretendthatweareeavesdroppingonEvanasherunsthroughtheargu-ment(thecomputerconnectedtoEvanincludesaloudspeakersothatwecaneaves-droponEvan’sruminations).WhenEvanreachespremise(5),heconcludes:“Inutteringthesentence‘Iamabraininavat,’IutterasentencemeaningthatIamabraininavat.”Evanis“speaking”Vat-English,soweshallneedtotranslatethisutter-anceintoEnglish.Whenwedosoweobtainsomethinglike:“Inutteringthesentence‘Iamabraininavat,’IutterasentencemeaningthatIamacomputerstateoftypesn.”Thisutteranceiscertainlytrue.Moregenerally,utterancesoftheform“inutteringthesentence‘P’,IutterasentencemeaningP,”areboundtobetrue.IfPutnamisright,thenthemeaningsofthesesentencesdependonallsortsofcausalandcontextualfactors.Weneedknownothingofthese,however,forthemtodeterminethemeaningofwhatwesayandthecontentofwhatwethink.Debbie’sutterancesof“water,”andthoughtsshewouldexpressbysuchutterances,concernwater,notbecauseDebbiehasfiguredoutthatsheiscausallyconnectedtoH2O(andnotXYZ),butbecauseDebbieiscausallyconnectedtoH2O(andnotXYZ).Imagine,again,thatweareeavesdroppingonEvanrunningthroughPutnam’sargument.PretendthatEvanfindstheargumentconvincing,concludinginanexcitedtoneofvoice:“Iamnotabraininavat!”Surely,youthink,Evanisdeluded.Aswecanplainlysee,Evanisabraininavat.Thisistooquick,however.Beforewecanevaluate406\nHILARYPUTNAMEvan’sconclusion,wemusttranslateitfromVat-EnglishintoEnglish.Whenwedoso,weobtainsomethinglike:“Iamnotacomputerstateoftypesn!”This,ofcourse,istrue;Evanisnotdeluded.Wecanapplythislessontoourowncase:givenmeaningexter-nalism,wecouldnotbedeludedinconcludingfromPutnam’sargumentthatwearenotbrainsinvats.ImplicationsformetaphysicalrealismEvaniscertainlynotdeludedinthesenseofbelievingafalsehood.Hebelievesthatheisnot(asweshouldputit)acomputerstate,andheiscorrect.Indeed,mostofEvan’sbeliefsabouthisactualsituationarecorrect.(TheargumentisnotaffectedbysupposingthatEvanhasfalsebeliefs;allofushaveourshareoffalsebeliefs.)OfcourseEvancannotappreciatethatheisabraininavathookedtoacomputerprogrammedbyanevilscientist.ThethoughtisnotoneEvanisinapositiontoentertain.PerhapsthisisallPutnamneedstoshowthatmetaphysicalrealismisuntenable.Metaphysicalrealismpresumesanepistemologicalgapbetweenwhatwetaketobethecaseandwhatisthecase.Onewaytoexpressthisisintermsofthepossibilitythatourbeliefsaboutwhatisthecasearemassivelyfalse.Ifweacceptexternalismaboutthemeaningsofourwordsandthecontentsofourthoughts,ifwesupposethatwhatwemeanandwhatourthoughtsconcernisfixedbyourcircumstances,wetherebyexcludethepossibilityofmassiveerror.ThisiswhatPutnam’sargumentshows.IfwereturntoEvan,however,wecanseethat,althoughEvanmaynotbemassivelydeceived,agapremainsbetweenwhathetakestobethecaseandwhatisthecase.Thefact,ifitisafact,thatEvanisinnopositiontoentertainthoughtsconcerningwhatisthecase–thoughtsaboutbrainsandvats–providesscantcomfortwhenweconsiderourowncircumstances.Wecanstillenvisionagap,evenifwecannotenvisionwhatmightlieonthefarsideofthegap.Andthisevidentlyleavesrealismstanding.Isthisunfair?Weareimaginingthatourcircumstancesmightbewildlydifferentfromwhatwetakethemtobeeventhoughourbeliefsaboutthosecircumstancesare,onthewholecorrect:thedeeptruthisnotthinkablebyus.Butwhatsensecouldbemadeofthesuggestionthatthingsmightbesomeway,althoughwecannotsomuchasconsiderwhatthatwaymightbe?Thissoundslikenonsense;andanonsensicalpossibilityisnopossibilityatall.ReadersfamiliarwithBerkeleywillrecognizethislineofreasoning.Berkeleydis-missesthepossibilityofamaterialworld,aworldofobjectsexistingmind-independently,onthegroundsthatwecannotsomuchasentertainthoughtscon-cerningsuchaworld.IfwecannotentertainthoughtsconcerningX,thenplainlywecouldhavenoreasontothinkXexistsormightexist:talkofXisempty.Thesituationwehavebeenenvisaging,however,isnotoneinwhichweendeavortothinktheunthinkable,butoneinwhichweacknowledgeourfallibility,recognizingthatwecouldbewrongaboutalmostanythingwithoutbeinginapositiontoentertainthoughtsastohowthingsactuallyare.Perhapsthisisallarealistneeds:itispossiblethatreality(orasignificantportionofreality)isnotjustunknown,butunknowablebyusowingtoourcircumstances.Orperhapsarealismofthissortleavesbehindthetraditionalimpetusforrealism.Ineithercase,Putnam’sreflectionspushrealists–and407\nJOHNHEILtheirbedfellows,theskeptics–toexaminetheirfundamentalassumptions.Forthat,realistsandanti-realistsalikeshouldbegrateful.OntologicalpluralismWehavebeenfocusingonPutnam’scontentionthattheworldisnotmind-independent:“themindandtheworldjointlymakeupthemindandtheworld.”ButthereisanotherdimensiontoPutnam’sdissectionofmetaphysicalrealism(seePutnam1987,lecture1).Ametaphysicalrealistregardstheworldaspossessingadefinitecharacterquiteindependentlyofourwaysofthinkingaboutit.Werepresentthischaracterinvariousways:trulyorfalsely,subtlyorclumsily.Oureverydaybeliefsabouttheworldrepresentitasbeingoneway,forinstance;thesciencesrepresentitdiffer-ently.The“scientificimage”andtheeveryday,“manifestimage”oftheworldappearinvariouswaystobeatodds.(TalkofscientificandmanifestimagesoriginatedwithWilfridSellars(seeSELLARS);seeSellars1963,ch.1.)ThesurfaceofthedeskatwhichIamsittingappearssmoothandcontinuous.Physics,however,tellsusthatthedeskisacloudofparticles,widelyspacedandinconstantmotion.Whichdescriptionofthedeskisthecorrectone?Perhapstheapparentdesk,thedeskofthemanifestimage,isamereappearance.Thisisthereactionofthemetaphysicalrealist.Therealiststartswiththeideathattheworldisasingledefiniteway.Wecandescribetheworldinmanydifferentways;somekindsofdescriptioncapturetheworldbetterthanothers,however.Myordinarydescriptionofmydesk,forinstance,isatbestacrudeapproximation.Takenliterallyitisfalse.Wecanedgeclosertothetruthbyturningtophysics.Whenwedo,welearnthattheworldcontainsnodesks,onlycloudsofinvisibleparticles.Canwemakesenseofthispicture?Returntomydesk.Howmanythingsarestackedonit?Ananswertothisquestionwilldependonhowwedecidetocount.Wecould,forinstance,countpencils,pens,books,andmemos.Wecouldjustaseasilycountpagesofbooks,partsofpensandpencils.Orwecouldcountparticlesofwhichallthesethingsarecomposed.Whatisthecorrectwaytocount?Whatisthecorrectanswertotheques-tion,howmanythingsareonmydesk?Suchquestionsarewrong-headed.Therearemanywaystosortobjectsonmydesk,manycorrectanswerstotheoriginalquestion.Thecontentsofmydeskcanbe“carvedup”indifferentways.Howwedosodependsonus:ouraimsorpurposes.Wedeploysystemsofconceptsinrepresentingtheworld.Themetaphysicalrealistseestheseconceptsasmatchingwellorbadlywhatis“outthere.”Butthisisthewrongmodel.Theconceptsweusedetermine,ratherthanmerelyreflect,whatisoutthere,atleastinthesensethattheydetermineobjects’boundaries,hencewhatistocountasanobject.Itisnotthattherearenodivisionsinnature,butthattherearetoomany.Ourconcepts,orrathersystemsofconcepts,makesomeofthesedivisionssalient.Insayinghowtheworldis,weinvokeoneoranotherconceptualsystemorscheme.Whichschemeweinvokedependsinpartonfeaturesofus,ourneedsandinterests.Aswelearnmoreandasourneedsandinterestschange,ourconceptualschemesevolve.Supposethisisright.Itisthenhardtoseehowwecouldmakesenseoftalkofaworldindependentofanyconceptualscheme(Kant’snoumenalworld,the“thinginitself”).Wecouldhavenowayofdescribingthatworld,nowayofthinkingit:describ-408\nHILARYPUTNAMingandthinkinginvolverepresentingintermsofaconceptualscheme.ConsideramapofthesurfaceoftheEarth.WecandepicttheEarth’ssurfacebymeansofaMercatorprojection,aPetersonprojection,asphericalprojection.ImaginesomeonedissatisfiedwiththeseinsistingthattheEarthbedepictedusingnoprojectionatall!Thisiswhatthemetaphysicalrealistdemandsforrepresentationsofrealityingeneral:arepresen-tationalsystemorconceptualschemethatisutterlytransparent.Butatransparentschemeisnoschemeatall.Onceyouacceptthattherecanbenosenseintalkofascheme-independentworld,youmaybemovedtoaskhowcompetingschemescouldbeevaluated.Again,thisisthewrongquestion.JustasMercatorprojectionsandPetersonprojectionsdonotcompete,soourmodernscientificschemedoesnotcompetewithoureverydaycon-ceptionoftheworld.Bothareentirelysatisfactoryontheirownterms,bothprovideperfectlyadequatedepictionsofourworld.Aneverydaydescriptioncanbewrong;Imayfalselybelievethatthereisadeskinmyoffice.Butifthisbeliefisfalseitisnotbecausesciencetellsusthattherearenodesks(onlycloudsofparticles).Ontology–whatthereis–isrelativetoaconceptualscheme.Desksdonotfigureintheontologyofthephysicist,butthisdoesnotmeanthatdesksaremereappearances.Insofaraswefindituseful,orunavoidable,todeployaconceptualschemeinwhichdeskshavearole,theontologicallegitimacyofdesksisassured.Themetaphysicalrealisthankersafterasingleontology:theontologyoftheworld.Instead,Putnaminsists,weshouldembraceontologicalpluralism:whatthereisdependsinpartonschemesorsystemsofconceptswefinditconvenienttodeploy.ExternalismagainInassessingPutnam’sattackonmetaphysicalrealism,wehavebeengrantingexter-nalism,theviewthatwhatwemeanandwhatourthoughtsconcerndependsinpartonourcircumstances.Putnam’sdefenseofexternalismreliesheavilyonTwinEarthcases:weimagineagentswhoareintrinsicallyindiscernibleandyetwhoseutterances,andthoughtsthoseutterancesexpress,differinsignificance.Partoftheideahereisthattheprojectivecharacterofthoughts–whatisoftencalledtheirintentionality–isdue,nottointrinsicfeaturesofthosethoughts,buttomattersexternaltothinkers.AnotherofPutnam’sexamplesnicelyillustratesthepoint.Supposeyouformamentalimageofaparticulartree,oneinanearbypark,forinstance.NowimagineanAlphaCentaurian,Fred,wholivesonaplanetbarrenofveg-etationandsoknowsnothingoftrees.OnedayFredspillssomepaintsthatpurelybychanceformadesignthatyouwouldregardasperfectlyrealisticrepresentationofthetreeinthepark.Later,reflectingonthespilledpaint,Fredformsamentalimageindis-tinguishableintrinsicallyfromyourtreeimage.IsFredimaginingthetreeinthepark?Thatseemsunlikely.IfFredisimagininganything,heisimaginingadesignproducedbyspilledpaint.Whataccountsforthedifference?Theimagesareintrinsicallyalike,sothedifferencemustlieelsewhere.Perhapsthedifferencestemsfromyourbeingcausallyrelatedtothetreeinthepark,inawayFredisnot.Yourimageofatreeprojectstothattreebecauseyoustandinanappropriatecausalrelationtothetree;Fred’simagepro-jects,nottothetreebuttothespilledpaint,becausethespilledpaint,notthetree,playstherequiredcausalrole.409\nJOHNHEILAviewofthissortreversesthemetaphorofprojection.Thoughtdoesnotprojectfromthe“insideout,”butfrom“outsidein.”Mustwegoalong?Perhapsnot.Perhapswecanaccountfortheprojectivecharacterofthoughtbyreferencetointrinsicfeaturesofagents;perhapsprojectionis“insideout.”How,then,couldweaccommodateTwinEarthcases?Suspenddoubtforamomentandpretendthattheprojectivityofathoughtislikethebeamofaflashlightradiatingoutward.Whatthebeamoftheflashlightillu-minatesdependsbothonthenatureofthebeam,asdeterminedbyintrinsicfeaturesoftheflashlight,andonwhathappenstobe“outthere”tobeilluminated.Justasflash-lightsonEarthilluminatewater(H2O),andflashlightsonTwinEarthilluminatetwinwater(XYZ),sothoughtsonEarthprojecttowater,andthoughtsonTwinEarthprojecttotwinwater.Themoral?TwinEarthcasesdonotestablishthattheprojectivecharacterofthoughtisduetoincomingcausalchains,then,onlythatwhatthoughtsdesignatedependsinpartonthecircumstancesofthinkers.Granteditissillytocomparetheprojectivityofathoughttothebeamofaflash-light.Neverthelessitmaybepossibletobaseanaccountoftheprojectiveaspectofthoughtonintrinsicfeaturesofagents.Agentsandtheirstatesofmindpossessdispo-sitionality,anddispositionsareinherentlyprojective.Locke’sexampleofalockandkeyillustratetheidea.Thekeyisforlocksofacertainsort,andnotforothers.Thisissoevenifnosuchlockhasbeenmanufactured(orifthelockthekeyfitsisdestroyed).Thekeyisdisposedtoopenonelock,butnotanother:thekeyprojectstoonelock,butnottoanother.Thekey’ssoprojectingdoesnotdependonthekey’shavingbeenincausalcontactwiththelock,butsolelyonintrinsicfeaturesofthekey(andintrinsicfeaturesofthelock).Imaginenowthatanagent’sstatesofmindincorporatefine-tuneddispositions(MartinandHeil1998;MartinandPfeiffer1986).Yourthoughtsoftrees,forinstance,projecttotrees,notperhapsbecausetheyarecausedbytrees,butbecausetheydisposeyoutointeractinappropriatewayswithtrees.Tobesure,intrinsicallyindiscerniblethoughtsmightdisposeaninhabitantofTwinEarthtointeractwithtwintrees.Thisdoesnotshowthatyourthought’sprojectivecharactercomesfromtheoutside,however.Theprojectivecharacterofthosethoughtsmightbe“builtin”evenifobjectsthosethoughts“illuminate”dependsonwhatobjectsareavailabletobeilluminated.Consideranothermuch-discussedcase(Davidson1987).Doniswadingthroughaswampinathunderstorm.Suddenly,aboltoflightningreducesDontoapileofashesandsimultaneouslyreconstitutesanearbytreestumpintoa“molecularduplicate”ofDon.Supposethatthemolecularduplicate,Swampman,functionsjustasDondidpriortohissuddendemise,andinparticularSwampmanhasthoughts,images,andmemo-riesintrinsicallyindiscerniblefromDon’s.Swampmanhasmanyfalsememories.Heseemstorememberhistwelfthbirthdayparty,butheisinfactonlyafewminutesold.WhatofSwampman’sotherthoughtsandimages,however?SwampmanentertainsthoughtsandformsimagesintrinsicallyindiscerniblefromDon’sthoughtsandimagesoftrees,stars,water,andsoon.OughtwetosaythatthesethoughtsandimagesareemptyofsignificanceuntilSwampmancomesintocausalcontactwithtrees,stars,andwater?Onlysomeonewithapriorcommitmenttoan“outside-in”conceptionofthoughtwouldsayso.Swampman’smentalconditionincludesfinely-tuneddisposi-tionsthatundergirdtheprojectiveaspectofhisthoughts.Ofcourse,whatthosethoughtsprojecttodependsinsomemeasureonwhatis“outthere.”Thisisthelesson410\nHILARYPUTNAMofTwinEarth.Butthisneednotleadustoimaginethattheprojectivityofthoughtmustbeexplainedbyincomingcausalchains.Putnam’ssignificanceThisentryprovidesonlythebriefestintroductiontoonecornerofPutnam’sphilo-sophicalwork.Ihaveexpressedreservationsconcerningtwothemesthathaveprovedespeciallyinfluentialinrecentphilosophy.Thesecriticalcommentsafford,atbest,onlyhintsastowhereareadermightdisagreewiththosedoctrines.Putnam’sworkiswide-ranging,rich,andinterconnectedinawaythatundercutspiecemealcriticism.InattackingonePutnamthesis,acriticrisksassumingpositionsthatPutnamelsewhererejects.PerhapsIhavesaidenoughtomakeitclearthatPutnam’sworkisdeeplyinsightful,penetrating,andsynoptic.EvenwhenPutnamself-confessedlygoesupablindalley,itisworthfollowinghimforthesakeofobservingsometopicinanewandrevealinglight.(Besides,Putnam’sblindalleysaremoreinterestingthanthewell-troddenpathsofotherphilosophers.)HilaryPutnamisoneofahandfulofphiloso-pherswhohaveindividuallyshapedthefundamentalcharacterofcontemporaryphilosophy.BibliographyWorksbyPutnam1971:PhilosophyofLogic,NewYork:HarperandRow.1975a:Mathematics,Matter,andMethod,PhilosophicalPapersvol.1,Cambridge:CambridgeUniversityPress.1975b:Mind,Language,andReality,PhilosophicalPapersvol.2,Cambridge:CambridgeUniversityPress.1978:MeaningandtheMoralSciences,London:RoutledgeandKeganPaul.1981:Reason,Truth,andHistory,Cambridge:CambridgeUniversityPress.1983:RealismandReason,PhilosophicalPapersvol.3,Cambridge:CambridgeUniversityPress.1987:TheManyFacesofRealism,1985PaulCarusLectures,LaSalle,IL:OpenCourt.1988:RepresentationandReality,Cambridge,MA:MITPress.1990:RealismwithaHumanFace,ed.J.Conant,Cambridge,MA:HarvardUniversityPress.1992:RenewingPhilosophy,Cambridge,MA:HarvardUniversityPress.1994:“Sense,Nonsense,andtheSenses:AnInquiryintothePowersoftheHumanMind,”JournalofPhilosophy91,pp.445–517.1995:Pragmatism:AnOpenQuestion,Oxford:BlackwellPublishers.WorksbyotherauthorsBoolos,G.(ed.)(1990)MeaningandMethod:EssaysinHonorofHilaryPutnam,Cambridge:CambridgeUniversityPress.Clark,P.andHale,B.(eds.)(1994)ReadingPutnam,Oxford:BlackwellPublishers.Davidson,D.(1987)“KnowingOne’sOwnMind,”ProceedingsandAddressesoftheAmericanPhilosophicalAssociation60,pp.441–58.Dennett,D.(ed.)(1987)ThePhilosophicalLexicon,Newark,DE:AmericanPhilosophicalAssociation.411\nJOHNHEILEmerson,R.W.(1940)“Self-reliance,”inTheCompleteEssaysandOtherWritingsofRalphWaldoEmerson,ed.B.Atkinson,NewYork:TheModernLibrary.Heil,J.(1998)“SkepticismandRealism,”AmericanPhilosophicalQuarterly35,pp.57–72.Lewis,D.(1972)“PsychophysicalandTheoreticalIdentifications,”AustralasianJournalofPhilosophy50,pp.249–58.Martin,C.B.andHeil,J.(1998)“RulesandPowers,”PhilosophicalPerspectives12,pp.283–312.Martin,C.B.andPfeiffer,K.(1986)“IntentionalityandtheNon-psychological,”PhilosophyandPhenomenologicalResearch46,pp.531–54.Place,U.T.(1956)“IsConsciousnessaBrainProcess?,”BritishJournalofPsychology47,pp.44–50.Ryle,G.(1949)TheConceptofMind,London:Hutchinson.Sellars,W.(1963)Science,Perception,andReality,London:RoutledgeandKeganPaul.Smart,J.J.C.(1959)“SensationsandBrainProcesses,”PhilosophicalReview68,pp.141–56.Wittgenstein,L.(1953)PhilosophicalInvestigations,trans.G.E.M.Anscombe,Oxford:BlackwellPublishers.412\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200133DavidM.Armstrong(1926–)FRANKJACKSONDavidArmstrong’smanymajorcontributionsarefocusedintraditionalepistemologyandmetaphysics.Heofferscomprehensiveaccountsofwhatthereis,itsnature,andhowweknowaboutit.Heisa“systembuilder.”Hisworkisinformedbythecon-victionthatphilosophersmusttakeveryseriouslytheteachingsofscience.Heisarealist:aboutmentalstates,aboutproperties,aboutlaws,andaboutsingularcausa-tion.Indeed,onalmostanyphilosophicaltopic,ifthereisarealistpositionavailable,Armstrongwilloccupyit.Also,heseekswhatisnowoftencalled“theviewfromnowhere.”Heisopposedtotheideathattheremaybedifferent,equallylegitimatebut,toonedegreeoranother,incommensurate,viewsofhowthingsarefromoneoranotherperspective.Or,asitissometimesput,hedeniesthattherearedifferentkindsofbeingoroftruth.MaterialismaboutthemindArmstrongisprobablybestknownforAMaterialistTheoryoftheMind.Histheoryiscommonlyknownascentralstatematerialismorasthecausaltheoryofmind.Armstrongstartedhisphilosophicallifeasabehavioristbut,partlyasaresultoftheinfluenceofJ.J.C.(Jack)Smart,movedtotheviewthatmentalstatesarestatesofthecentralnervoussystem,andmoreespeciallythebrain.ArmstrongdevelopshiscentralstateversionoftheidentitytheorybyfirstarguingthattheconceptofamentalstateMistheconceptofastatethatplaysadistinctivecausalrolethatconnectsstimulus,behavioralresponse,andothermentalstates.Thus,togivetheroughidea,painisthestatetypicallycausedbybodilydamage,andtypicallycausingadesirethatititselfceaseandabehavioralresponsethattendsto,orisbelievedto,minimizethedamage.Obviously,anaccountofthiskindisexactlywhatevolutionaryconsiderationswouldsuggest.Insimilarfashion,beliefisastateinducedbysubjects’environmentsthattendstomakethembehaveinwaysthatrealizewhattheydesireifwhattheybelieveistrue.Armstrongseestwomajoradvantagesofthiskindofviewoverbehaviorism.First,itallowsmentalstatestobecausesofbehavior.Secondly,bybringinginreferencetoothermentalstates,itallowsforsuitablycomplexaccountsoftheconnectionsbetweenmentalstatesandbehavior.Itisnotoriousthatthereisnosimpleone-to-onematchingofmentalstatesandbehavior.Whatyoudowhenyouthinkitisabouttoraindepends,413\nFRANKJACKSONinteralia,onwhetheryouwanttostaydry,onwhereyouthinktheumbrellais,onwhetheryouthinkyouareGeneKelly,andonhowcoldyoufeel.Onthecentralstatetheory,toaskaftertheidentityofagivenmentalstateMistoaskwhatstateplaysthedistinctive,causallyintermediateroleassignedbytheconceptofM.Armstrongarguesthat,foreachmentalstate,itwillturnouttobesomestateorotherofthebrainthatplaystheroleinquestion.Heconcludes,therefore,that,asanempiricalmatteroffact,mentalstatesareidenticalwithbrainstates.Theseidentitieswillbecontingentbecausewhichbrainstatesplaywhichrolesisacontingentmatter.Somehaveobjectedthattherearenocontingentidentities:everythingisnecessarilyidenticaltoitself,andthatwhatArmstrong(andSmart)shouldsayisthattheidentitiesareaposteriori.Infact,theyholdthattheidentitiesarebothcontingentandaposteriori,buttheobjectiontothecontingentidentitypartoftheirtheoryisamisunderstanding.Alltheymeanisthatsentencesoftheform“MisB”arecontingent,inthesamewaythat“Redisthecolorofbullfighters’capes”clearlyis.AsecondmisunderstandingisoverArmstrong’sstanceonthepossibilitythatquitedifferentstatesmightplaythecausalroledistinctiveofpainindifferentspeciesand,maybe,indifferentmembersoftheonespecies.Itisoftenobjectedthatidentitytheo-ristsarecommitted(implausibly)topainbeingthesamestateineverythingthatexpe-riencespain.Butconsiderthefollowingparallel.Themostdangerousvirusfordogsisdifferentfromthemostdangerousvirusforpeople,andthereasonforthisisthatthekindthatplaystherelevantroleindogsdiffersfromthekindthatplaystherelevantroleinpeople.Nevertheless,wecan,anddo,identifythemostdangerousvirusfordogsandforpeople–oranywaytheexpertsdoitforus.AmorepressingquestioniswhetherArmstrong(andSmart)shouldhavesaidthatmentalstatesareconstitutedby,ratherthanidenticalwith,brainstates.Therelationbetweenatableandthepartsthatmakeitupisoneofconstitution,notidentity.Becausethelifehistoriesofthetableanditspartsdiffer–forexample,thepartstypi-callycomeintoexistencesomewhatearlierthanthetable–Leibniz’sLawmeansthattherelationcannotbeoneofidentity;itmustbeconstitution.(Aseparatequestioniswhetherthisrelationofconstitutioncanbeanalyzedintermsofidentitybetweentem-poralpartsofthetableandtemporalpartsofvariousaggregationsofparts.)Similarly,itmaywellbethatArmstrong(andSmart)should,strictly,saythatmentalstatesareconstitutedbybrainstates,notthattheyareidenticaltothem.Armstrong’scentralstateviewofmindissometimescontrastedwiththekindoffunctionalisttheoryofmindassociatedwiththeearlyHilaryPutnam(seePUTNAM).Theybothagreeingivingfunctionalrolesacentralroleinthetheoryofmind.ThisisbecauseArmstrong’scausalrolescanequallybedescribedasfunctionalroles.ThestimulithatArmstrongtalksofareinputs,asfunctionalistssayit,andthebehavioralresponsesareoutputs,asfunctionalistssayit.Therearetwobigdifferences,though.Armstrongthinksofthementalstatesastheoccupantsofthefunctionalroles,asthestatesthataresuitablyinterconnectedtoinputs,outputs,andother,internalmentalstates.Putnamthinksofthem(orthoughtofthemwhenhewasafunctionalist)asthefunctionalrolesthemselves.And,secondly,thefunctionalrolesinArmstrong’stheoryarethosesometimescalled“commonsense.”Theirinputsandoutputsaredescribedintermsfamiliartousall:rain,umbrellas,movementsthatleadtobeerinsidethemouth,414\nDAVIDM.ARMSTRONGetc.InPutnam’sversionoffunctionalism,thoughnotinallversionsoffunctionalism,theinputsandoutputsarethoughtofasinternalones.Perception,sensations,belief,knowledgeArmstrong’sPerceptionandthePhysicalWorldisanargumentfordirectrealisminper-ception.Hearguesthatwearedirectlyacquaintedwithindependentlyexistingphysi-calobjectsinperception.Thedistinctivefeatureofhisargumentisthewayitisfoundedonananalysisofperceptionandperceptualexperienceintermsoftheacquisitionofbeliefthroughtheoperationofone’ssenseorgans.Thismakesgoodsenseofthecentralbiologicalfunctionofperception,whichistheacquisitionofbeliefaboutwhatisgoingonaroundandinsideone.AnobviousquestionforArmstrong’saccountisraisedbythefactthattheverysamebelief,say,thatitisrainingoutside,canbeacquiredinverydifferentwaysthroughverydifferentperceptualexperiences.Youmight,forinstance,seethatitisraining,betoldthatitisraining,readonacomputerscreenthatitisraining,orhearthatitisraining.Perhapsthemostplausiblewayofapproachingthisproblemisintermsofthedistinctivelydifferentclustersofbeliefsineachcase.Innoneofthesecases,doesthebeliefthatitisrainingcome“byitself”;rather,itcomesasanintegralpartofawholeclusterofbeliefs,andtheclustersaredifferentin,anddistinc-tiveof,eachcase.InBodilySensations,Armstronggivesanaccountofsomaticsensationsintermsofperceptionofone’sownbody.Asensationisanexperienceofperceivingthatone’sbodyisinsuchandsuchastate,anexperiencewhichmayormaynotbeveridical.Forexample,afeelingofwarmthistheputativeperceptionthatapartofone’sbodyiswarm.Inthecaseofcertainsensations,theputativeperceptionisaccompaniedbyacharacteristicattitude.Pain,forexample,istheputativeperceptionthatthereissome-thingamisswithpartofone’sbody,accompaniedbyanimmediatedislikeofthisputa-tiveperception.Armstrong’streatmentofbelieffollowsasuggestionofF.P.Ramsey’sthatbeliefislikeamapbywhichwesteer.Insideourheadsisamastermapthatmovesusthroughtheworldinsuchawaythatwhatwedesireisachievedtotheextentthatthemapiscorrect,andindividualbeliefsarethoughtofassub-mapsofthemastermap.Thisapproachtobeliefisnowastandardalternativetotheinternalsentencetheoryofbeliefsupportedbylanguageofthoughttheorists.Hisaccountofknowledgeisareliabilistone.Knowledgenecessarilyinvolvestruebelief:ifSknowsthatP,thenStrulybelievesthatP.Butnotalltruebeliefisknowl-edge;thetruthofabeliefmaybeanaccident,andhowcangettingitrightbyaccidentbeknowledge?Armstrong’ssuggestion,roughly,isthatS’struebeliefthatPisknowl-edgeifitisareliablesignthatP.Herehediffersfromthetraditionthatrequiresthatone’sbeliefbejustifiedinordertocountasknowledge.TimeandactionArmstrongholdsatemporalpart,orstage,metaphysics.Identityovertimeisamatterofhavingpartsorstagesatthetimesinquestion.IwasattheMelbourneTestwhen“Typhoon”Tysontook7for27becauseacertainperson-stagewiththeright415\nFRANKJACKSONconnectionstotheperson-stagewritingthesewordswaspresentatthattestmatch.Armstrong’smaincontributiontothedebateisoneoftheveryfirstdiscussionsofthefamousrotatinghomogeneouscylinder/disk/sphereexample.Hearguesthattheexampleshowstheconceivabilityofaconceptionofidentitythroughtimenotframedintermsoftemporalstages,butthat,nevertheless,thetemporalstageaccountofiden-titythroughtimeisinfactcorrect.Whatmakesittrue,onhisview,thatsuchanobjectisrotatingarethedependenciesbetweendifferentstages.Armstrongwasalsooneofthefirst,withBrianO’Shaughnessy,toarguethatifoneacts,onemusthavetriedtoact,andthatthisistheessenceoftruthintheoldvolitionaltheoryofaction.Universals,laws,causation,possibility,andstatesofaffairsTruth-makersplayacrucialroleinArmstrong’slaterphilosophizing.Thebasicideaisthatifsomesentenceorpropositionistrue,theremustbesomethingthatmakesittrue;similarly,ifsomepredicateappliestosomething,theremustbesomethingthatmakesittruethatthepredicateapplies.Youcannotsaythattheword“square”appliestoA,andthatthatisallthereistosay.TheremustbesomethingaboutAthatmakesittruethatthewordappliestoit,thatAsatisfiesit.InArmstrong’shands,thetruth-makerprinciple,ashecallsit,ismorethanthewidelyacceptedsupervenienceoftruthandsatisfactiononnature.Superveniencesaysthatifasentenceistrueinonesituationandfalseinanother,andifapredicateissatisfiedbyonethingbutnotbyanother,thesituationsandthingsmustdifferinnature.Thetruth-makerprinciplegoesfurther.Itsaysthattheremustbesomethingthatmakes–necessarilymakes–thetruesentencetrueandthesatisfiedpredicatesatisfied.Armstrongholdsthatwhatmakesittruethatpredicatesapplytoparticularsarethepropertiesoruniversalsthattheparticularspossess.Inkeepingwithhisrealistleanings,theseuniversalsexistindependentlyoftheclassificationsthatwefindnatural.Theyareinnature.Secondly,theyarenottobereducedtosets,ortoresemblancesbetweenpar-ticulars.Armstrongisnotanominalist.Heargues,inparticular,thatnominalistscannothandlethefamous“oneovermany”problem,theproblemofwhatunifiesthingsthatshareaproperty.Thirdly,therearenouninstantiateduniversals;everyuniversalispossessedbyatleastonething.Inthissense,heiswithAristotleandnotPlato.HeregardsthePlatonicviewthatthereareuninstantiatedpropertiesoruniversalsasanunmotivatedontologicalextravagance.Fourthly,thereisnotaone–onerelationbetweenpropertiesandpredicates:oneandthesameuniversalorpropertymaybethetruth-makerforanumberofdifferentpredicates.Toillustrate:supposethatUisauni-versalandthat“A”isapredicatethatsaysthatsomethingisU,and“B”issomequitedifferentpredicate.Surely,“AorB”mightbetrueofsomethingwhichisUsimplybecauseitisU.Wearenotrequiredtopostulateanextrapropertyjustbecause“AorB”isadistinctpredicatefrom“A.”Also,theremaybepropertiesforwhichthereisnopred-icate.Finally,whichuniversalsorpropertiesthereareisanaposteriorimattertobesettledbytotalscience.Philosophytellsusthattheremustbetruth-makersfortruepredications,butwhattheyareisultimatelyamatterforsciencebroadlyconceived.ArmstrongarguesstronglyagainstHumeanandneo-Humeanaccountsoflaws.Forhim,nofactsaboutregularities,howevertrickedup,caneveradduptolawfulness416\nDAVIDM.ARMSTRONGproper.Whatthenmustbeaddedtoaregularitytogetalaw?Hisansweristhatwhatdistinguishestheuniversalstatementsoftheform“EveryFisG”thatexpresslawsofnature–thatarenomicornomological–fromthosethatexpressaccidentalregulari-tiesisthat,roughly,thelawscorrespondtorelationsofnomicnecessitationbetweenuniversals.Initssimplestversion,theideaisthat“EveryFisaG”isalawifandonlyifFnessnecessitatesGness.Butmoredetailedaccountswouldneedtoadverttohismetaphysicsofstatesofaffairs,mentionedbrieflybelow,andtohistreatmentoflawsthatdonotfallobviouslyintothe“EveryFisG”mold,derivedlaws,andlawsthathavenoinstances(forexample,concerningmotionintheabsenceofgravity).Thisaccountoflawsis,obviously,stronglyanti-Humean.Armstrong’saccountofcausationisequallycountertothetraditionthatcomestousfromHume,andinthreerespects.First,Armstronginsiststhatcausationissingularinthatitisanon-relationalpropertyofasequence(seeANSCOMBE).Secondly,heholdsthattheconnec-tionbetweencausationandlawisaposteriori.Hedenies,thatis,thatitisapriorithatanysingularcausalsequencefallsundersomelaw.Hedoes,though,allowthatitmaywellbethatsomeorallcausalsequencesareidentical,asanaposteriorimatter,withtheinstantiationofalaw.Finally,heholdsthatwesometimesdirectlyperceivesingularcausalconnections.Hereheisgoingagainstawidelyheldview,evenamongthosewhowouldnotdescribethemselvesasHumeans.Manywhoagreewithhimthatcausationismorethansequencesuitablyconstrainedthink,nevertheless,thatsequenceisallweliterallyperceive.WedonotseethatXcausedY;weinferit.Sometimestheirargumentforthisviewisthatanon-causalsequencecanseemascausalascanbe,asPiaget’sfamousexperimentstellus.Armstrongrightlypointsoutthatthisonlyshowsthatillusionispossible,andthepossibilityofillusionconcerningafeaturedoesnotshowthatwedonotliterallyperceiveitwhenallgoeswell.However,thereisastrongerargument.Itishardtoidentifythecausalrolethatsingularcausa-tionplaysinitsallegedperception.WhenIseethatsomethingissquare,itssquarenessplaysaroleininducingmyperceptualexperience.ThisseemscrucialtoitsbeingcorrecttosaythatIperceiveitssquareness.Butwhatroledoessingularcausationplaythatmightmirrortherolesquarenessplays?Allthecausalworkseemstobebeingdonebythesequenceperse.Armstrong’saccountofpossibilityisacombinatorialone,drawingonhisrealismaboutuniversals.Wecanthinkofhowthingsareasavast,complexarrangementofparticularsanduniversals.Thevariouspossibilitiescanthenbethoughtofasallthecombinationsandrecombinationsoftheseparticularsanduniversalsaccordingtovariousrulesforcombiningparticularsanduniversals.Thus,togivethebarestbonesoftheidea,supposethatthereisinfactchargeXatpointy,andchargeUatpointv.WhatmakesitpossiblethattherebechargeXatv,andchargeUaty?HisansweristhefactthatputtingXwithv,andputtingUwithy,doesnotviolatetherulesofcombination.Inhismostrecentbook,AWorldofStatesofAffairs,Armstrongarguesthatthebestwaytobringhisideasonuniversals,laws,truth-making,andpossibilitytogetherisbyadoptingametaphysicsofstatesofaffairs.Forexample,universals–thekeytohisaccountoflaws–turnouttobetypesofstatesofaffairs.Inanycase,forArmstrong,theworldisnottheaggregationofallthethingsthereare.Itistheaggregationofallthestatesofaffairsthereare,wherestatesofaffairsarethings-having-properties.His417\nFRANKJACKSONviewisessentiallythesameasWittgenstein’sintheTractatus,namely,thattheworldisthetotalityoffacts,notofthings.BibliographyofworksbyArmstrong1960:Berkeley’sTheoryofVision,Melbourne:MelbourneUniversityPress.1961:PerceptionandthePhysicalWorld,London:RoutledgeandKeganPaul.1962:BodilySensations,London:RoutledgeandKeganPaul.1968:AMaterialistTheoryoftheMind,London:RoutledgeandKeganPaul.1973:Belief,TruthandKnowledge,Cambridge:CambridgeUniversityPress.1978:UniversalsandScientificRealism,2vols.,Cambridge:CambridgeUniversityPress.1983:WhatisaLawofNature?,Cambridge:CambridgeUniversityPress.1984(withNormanMalcolm):ConsciousnessandCausality:ADebateontheNatureofMind,Oxford:BlackwellPublishers.1989a:ACombinatorialTheoryofPossibility,Cambridge:CambridgeUniversityPress.1989b:Universals:AnOpinionatedIntroduction,Boulder,CO:WestviewPress.1997:AWorldofStatesofAffairs,Cambridge:CambridgeUniversityPress.418\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200134NoamChomsky(1928–)PETERLUDLOWNoamAvramChomsky,borninPhiladelphia,Pennsylvania,receivedhisPh.D.inlin-guisticsfromtheUniversityofPennsylvaniain1955.Since1955hehastaughtatMIT,wherehecurrentlyholdsthepositionofInstituteProfessor.Chomskygainedtheatten-tionofphilosophersearlyoninhiscareerbytheintroductionofmathematical/logicaltoolsforthedescriptionoflinguisticphenomena.InthisrespecthisearlyworkwasinfluencedbyfiguressuchasNelsonGoodmanandW.V.Quine,bothofwhomarethankedintheintroductiontohisSyntacticStructures(1957).Nevertheless,Chomsky’sprincipalphilosophicalsignificancerelatestohisrejectionoftheapproachtolanguageandmindtakenbyQuineandmanyotheranalyticphilosophers.Indeed,Chomskyhasbeenadirectparticipantinseveralkeyphilosophicaldebatesinthelasthalfcentury,takingissuewithinterlocutorssuchasQuine,DonaldDavidson,HilaryPutnam,SaulKripke,andJohnSearleonthenatureoflanguageandmind.Intheviewofmanyanalyticphilosophers,languageisasocialobjectthathasbeenestablishedbyconventionforpurposesofcommunication.Chomsky’stakeisdifferent:theconceptionoflanguageasanexternalsocialobjectisunfruitful(ifnotincoherent),andtheonlyplausiblestrategyfortheempiricalscientististoviewlanguage,orrather,1thelanguagefaculty,asanaturalobjectthatispartofourbiologicalendowment.Theexactnatureofthispicturehasevolvedsincethe1960s,andithastakentheformofa“principlesandparametersmodel,”whichcanbeviewedinthefollowingway:thinkofthelanguagefacultyasbeingalargelypre-wiredmechanismwithasetofswitches(parametricsettings)whichcanbesetinvariouswaysdependingupontheenviron-mentalsettingintowhichthelanguage-learnerisborn.Thetaskofthelinguististostudythismechanism,todeduceitsinitialstate,andtounderstandwhatthepossibleparametricsettingsare,thatis,todeterminepreciselywhatvariationisallowedbythelanguagefaculty.Chomsky(1986a)introducestheterms“I-language”and“E-language”todistin-guishhisgeneralthesisaboutthelanguagefacultyfromtheloosecollectionoftheo-riesaboutlanguagethatholdthatitissocialorexternal.TheI-language/E-languagedistinctionisuseful,sinceithighlightstheideathattheobjectofstudyinlinguisticsis“internal”inasense,andisnotdirectlyconcernedwith“external”phenomenalikewrittencorporaofdata.ChomskyisthusopposedtotheconceptionreflectedinthedefinitiongivenbytheAmericanlinguistLeonardBloomfield:languageis“the419\nPETERLUDLOWtotalityofutterancesthatcanbemadeinaspeechcommunity.”ForChomsky,I-languagesare“inthemind,ultimatelythebrain.”Tohighlightthedifferencebetweenthesetwoapproaches,considerthetwodiffer-entpicturesoflinguisticrulesthatemerge.Traditionalgrammarians(citingconven-tionsandcommonpracticeforwrittenEnglish)giveussuperficialrulessuchas“Donotendasentencewithapreposition”or“Use‘whom’,not‘who’whenthepronounhasaccusativeordativecase.”Ontheotherhand,generativegrammarianslikeChomskynotethattherearemoresubtleandinterestinglinguisticruleswhichgounnoticedbythetraditionalgrammarianbutwhichseemtobeemployedbyabroadclassofspeak-ers.Forexample,nonativespeakerofEnglishwouldrecognize(1)(below)asawell-formedquestioninEnglish,eventhoughthereareseeminglysimilarstructureslike(2)thatarequiteacceptabletolanguageusers:1*WhodidJohnseetheboythatBillhit?2WhodidJohnsaythatBillhit?Theaccountforthedifferenceinthesecasesissubtle,andthedetailsoftheexplana-2tionhavechangedasgenerativegrammarhasevolved.Thefactremains,however,thatnativespeakersofEnglishknowthat(2)isacceptableandthat(1)isnot,andfurtheritisclearthatnooneistaughttohavethispreference.Whateverrulesaccountforthejudgmentsabout(1)and(2)theyarefarmoresubtlethantheusualprescriptiverules.Similarconsiderationsapplytothefollowingexamples,discussedinChomsky(1986b;seealso1982,1986a).3Johnfiledeveryletterwithoutreadingit.4WhatletterdidJohnfilewithoutreadingit?SomehowspeakersofEnglishknowthatifwedeletethepronoun“it”inthesetwosen-tences(asin(3¢)and(4¢))theeffectsonmeaningaredifferent.3¢Johnfiledeveryletterwithoutreading.4¢WhatletterdidJohnfilewithoutreading?(4¢)isambiguousinawaythat(3¢)isnot.Both(3¢)and(4¢)havethemeaninginwhichthefilingwasdonewithoutsome(unspecified)readingtakingplace,but(4¢)alsopre-servesthemostsalientpossiblemeaningof(4):itcanstillbeunderstoodasaskingwhatletterJohnfiledwithoutreadingit–thefiledletter.Clearly,noonetaughtusthis,thereisnoconvention(tacitorotherwise)touselanguageinthisway,andnopre-scriptivegrammarianeverstipulatedthatweshouldinterpretthesesentencesinthismanner.But,justasclearly,thesefactsdescribethelinguisticcompetenceofalargeclassofindividuals.Thisisjustoneoftheproblemsfortraditionalgrammarsandforthemoregeneralassumptionthatlanguagesareobjectsthatareestablishedbyconvention.Ourbestattemptstostipulatetherules–ortomakeexplicittheconventions–justscratchthesurfaceaboutourlinguisticcompetence.InChomsky’swords,Traditionalgrammarsdonotdescribethefactsoflanguage;rather,theyprovidehintstothereaderwhoalreadyhas,somehow,therequisite“notionofstructure”andgeneralcon-ceptualresources,andcanusethehintstodeterminetheexpressionsofthelanguageand420\nNOAMCHOMSKYwhattheymean.Thesameistrueofdictionaries....Traditionalgrammarsanddictio-naries,inshort,presuppose“theintelligenceofthereader”;theytacitlyassumethatthebasicresourcesarealreadyinplace.(1994b:160)Likewise,institutionssuchastheAcadémieFrançaisedonotstipulateasmuchastheythinktheydo.AtbesttheygivesomesuperficialrulesofthumbforproperlinguisticbehaviorregardingFrench.Theycannotevenbegintocovertherangeoffactsofinter-esttopracticinglinguists.Indeed,foragenerativelinguist,thetraditionalnotionofalanguagelikeFrenchorGermanissuspectatbest(seeChomsky1980b:ch.6).Inwhatsenseisa“speakerofGerman”fromtheDutchborderofGermanyanda“speakerofGerman”fromBavariaspeakingthesamelanguage?(Especiallygiventhattheirlanguagesarenotmutuallyintelligible?)Thefactthatwesaytheseindividualsspeakthesamelanguageismoreofapoliticaldecisionthananythingelse,andindeedanindividualraisedinnorthernGermanyandanindividualraisedinTheNetherlandsmayfindthattheirlanguagesaremoremutuallyintelligiblethanthetwoaforementionedGermancitizensdo.SayingthatthetwoGermancitizensspeakthesamelanguageisatbestaloosewayoftalkingaboutsomecontextuallyrelevant(andcertainlypoliticalratherthanlinguistic)simi-larities.Chomsky(1994b)comparesittosayingthattwocitiesare“near”eachother;whethertwocitiesareneardependsonourinterestsandourmodeoftransportationandvirtuallynotatallonbrutefactsofgeography.Thenotionof“samelanguage”isnomorerespectableanotioninthestudyoflanguagethan“nearness”isingeogra-phy.Informallywemightgrouptogetherwaysofspeakingthatseemtobesimilar(relativetoourinterests),butsuchgroupingshavenorealscientificmerit.Asasubjectofnaturalinquiry,thekeyobjectofstudyhastobethelanguagefacultyanditssetofpossibleparametricvariations.OnemightthinkitpossibletoretreatslightlybygivingupontheideaofanE-languageandendorsinganotionofE-dialectorE-idiolect,buteventhisretreatwillnotsavethelanguage-as-external-objectposition,accordingtoChomsky.Considerationsthatmakeitarbitrarywhentosaythattwoindividualsspeakthe“samelanguage”alsoapplytosayingwhentheyspeakthe“samedialect.”Furthermore,wehavenowayofidentifyingthelinguisticformsthatwouldbepartofagivenindividualA’sE-idiolect.Inthefirstplace,Aspeaksindifferentwayswithdifferentgroupsofindividuals(sayAusesadifferentvocabularyamongphilosophersthanamongfamilymembers)andindeedatdifferentstagesoflife(contrastA’suseoflanguageatage3andage30).Doallofthesewaysofspeakingcountasbeingpartofthesameidiolect?WhatunifiesthemotherthanthattheyarewaysinwhichAhappenstohavespoken?Stillworse,wecertainlycan’tidentifyA’sE-idiolectwithsomecorporaofutterancesandinscriptions,fortheseintuitivelyincludespeechandspellingerrors.OnthebasisofwhatcanwesaythatagivenhiccupisanerrorandnotpartofthespokencorpusofA’sE-idiolect?IfwetryandidentifyerrorsbyappealingtoA’slan-guagecommunity,thatlandsusbackintheproblemofindividuatingE-languagesandE-dialects;thereissimplynofactofthematteraboutwhichlanguagecommunityAbelongsto.OntheI-languageapproach,however,thisproblemtakestheformofawell-definedresearchproject.Theidiolect(I-idiolect)isdeterminedbytheparametricstateofA’s421\nPETERLUDLOWlanguagefaculty;thelanguagefacultythusdeterminesA’slinguisticcompetence.Speechproductionthatdivergesfromthiscompetencecanbeattributedtoperformanceerrors.Thus,thecompetence/performancedistinctionisintroducedtoilluminatethedistinctionbetweensoundsthatarepartofA’sgrammarandthosethataresimplymis-takes.TheE-languageperspectivehasnosimilarrecourse.ForChomsky,theseareamongthemyriadreasonswehaveforabandoningtheideathatlanguage(asstudiedbythelinguist)isasocialobject,andadoptingtheperspec-tivethatitisanaturalobject.Butwhatkindofnaturalobject?Sincechildrenacquiretheirlinguisticcompetencewithoutseriousformaltraining(certainlynonethatwouldcoverthefactsin(1)–(4))andindeedwithimpoverisheddata,Chomskyhypothesizesthattheremustbeaninnatelanguageacquisitiondevicewhichaccountsforthiscom-petence.Thetaskofthelinguististolearntheinitialstateofthisdevice,andtodeter-minethepossibleparametricvariationsofthedevicethatarebroughtaboutbyexposuretolinguisticdata.Thisthesishasledtocontroversy;indeed,ithascometobeatthecenterofrecentinnatenessdebatesbetweenChomskyandPiaget,andQuineamongothers(seeQUINE).Thedebateshaveturnedonwhetherlanguageacquisitionrequiresadedicatedlan-guagefacultyorwhether“generalintelligence”isenoughtoaccountforourlinguisticcompetence.Chomskyconsidersthe“generalintelligence”thesishopelesslyvague,andarguesthatgeneralizedinductivelearningmechanismsmakethewrongpredictionsaboutwhichhypotheseschildrenwouldselectinanumberofcases.Considerthefol-lowingtwoexamplesfromChomsky(1975,1980a).5Themanistall.6Isthemantall?Chomskyobservesthatconfrontedwithevidenceofquestionformationlikethatin(5)and(6)andgivenachoicebetweenhypothesis(H1)and(H2),thegeneralizedinduc-tivelearningmechanismwillselect(H1).(H1)Movethefirst“is”tothefrontofthesentence.(H2)Movethefirst“is”followingthefirstNPtothefrontofthesentence.Butchildrenapparentlyselect(H2),sinceinformingaquestionfrom(7)theynevermaketheerrorofproducing(8),butalwaysoptfor(9).7Themanwhoishereistall.8*Isthemanwhohereistall?9Isthemanwhoisheretall?Notethatthisistruedespitethefactthattheonlydatatheyhavebeenconfrontedwithbeforeencountering(7)issimpledatalike(5)and(6).Chomsky’sconclusionisthatwhateveraccountsforchildren’sacquisitionoflanguageitcannotbegeneralizedinductivelearningmechanisms,butrathermustbeasystemwithstructure-dependentprinciples/rules.Ineffect,onehastothinkofthelanguagefacultyasbeingadomain-3specificacquisitionmodule.ObviouslythedistinctionbetweenI-languageandE-languageputsChomskyatoddswithanumberofphilosophersonthenatureoflanguage,butitalsoleadstoanumberofsubsidiaryphilosophicaldisputes,notleastofwhicharethosedisputesthatare422\nNOAMCHOMSKYdrivenbyquestionsaboutthenatureofrulesandrepresentations(orprinciplesandparameters)incognitivescience.ForexampleQuineextendshis“gavagai”argumentandattendantskepticismaboutmeaningstosimilarskepticismaboutgrammaticalrules.AtthecoreofQuine’sworryistheideathatifseveralrulesystemsareconsistentwiththelinguisticbehaviorofanindividual,thentherecanbenofactofthematteraboutwhatsetofrulesisactuallybeingemployed(seeQUINE).Chomsky(1969,1975,1980)hasmadeseveralresponsestothisargument.Inthefirstplace,ChomskytakesQuine’sargumenttobearehashofthestandardscientificproblemoftheunderdeterminationoftheorybyevidence.So,forexample,evenifthereareseveralgrammarsthatareconsistentwiththeavailablelinguisticfacts(notlinguisticbehavior,forChomsky,butintuitionsaboutacceptabilityandpossibleinterpretation)westillhavetheadditionalconstraintofwhichtheorybestaccountsfortheproblemoflanguageacquisition,acquiredlinguisticdeficits(e.g.frombraindamage),linguisticprocessing,etc.Inotherwords,sincegrammaticaltheoryisembeddedwithincognitivepsychology,thechoicebetweencandidatetheoriescan,inprinciple,beradicallyconstrained.Butfurther,evenifwehadtwodescriptivelyadequategrammars,eachofwhichcouldbenaturallyembeddedwithincognitivepsychology,thereremainstandardbesttheorycriteria(simplicity,etc.)whichcanhelpustoadjudicatebetweenthetheories.AmorerecentassaultonrulesandrepresentationshascomefromKripke’srecon-structionofWittgenstein’sprivatelanguageargument(seeKRIPKE).Accordingtothatargument,therecanbenofactofthematteraboutwhatrulesandrepresentationsasystemofunknownoriginmaybefollowing.Kripkeconcludes:ifstatementsattributingrule-followingareneithertoberegardedasstatingfacts,nortobethoughtofasexplainingourbehavior...itwouldseemthattheuseoftheideaofrulesandofcompetenceinlinguisticsneedsseriousreconsideration,evenifthesenotionsarenotrenderedmeaningless.(1982:31n.22)Chomsky’sinitial(1986b:ch.4)responsetotheKripke/Wittgensteinargumentappearstobethatthereisafactofthematteraboutwhatrulesacomputationalsystemisoperatingon,butinmorerecentarticles(1993,1994b,1995a)hehasarguedthattheKripke/Wittgensteinargumentappliesonlytoartifactsandnottonaturalobjects.Thatis,computersareartifacts–theproductsofhumanintentions–andhencethereisnofactabouttheirdesignthatexistsapartfromthoseintentions.Theprinciplesandparametersofthelanguagefaculty,ontheotherhand,areembeddedwithincognitivepsychologyandultimatelyfactsabouthumanbiology.Thereforethestructureofthelanguagefacultyisnolessgroundedthan,forexample,thehumangenome.ChomskyhasalsoclashedwithSearleoverthepossibilityofrulesincognitivesciencethatare“inprinciple”inaccessibletoconsciousness.Cantherebeaspectsofthementalwhicharenot“inprinciple”accessibletoconsciousness?Searlearguesthattherecannotbe(seeSEARLE).Chomsky(1990,1994a)arguesthatthenotionof“inprinciple”inSearle’sargumentisvacuous.Forexample,whatevidenceistherethatthegrammaticalprinciplesgoverningourjudgmentsaboutexamples(1)–(9)can’tbeaccessibletoconsciousness?Isitalawoflogicthattherecouldnotbeaspecieswithalanguagefacultyjustlikeoursbutwithfullconsciousaccesstoitsprinciplesand423\nPETERLUDLOWparameters?ChomskyalsonotesthatSearlemustintroducethenotionof“blockage”tocoverthosecasesinwhichanindividual,perhapsthroughbraindamage,isabletocorrectlysolveaproblem,butbeunabletosayhowitwassolved.OnSearle’stheory,suchapersonhas“inprinciple”accessbutsuffersfrom“blockage.”ButChomskyobservesthatitisentirelyarbitraryastowhatcountsasblockageandwhatcountsasinprincipleinaccessibility(e.g.perhapsanunfortunatemutationblockedouraccesstothelanguagefaculty).Accordingly,Chomskyarguesthatsuchnotionshavenoroleinnaturalisticinquiryintothenatureofthemental(andindeed,cognitivesciencerightlyignoressuchnotions).ForChomsky,itisnotenoughtodefendtheideathatrulesandrepresentations(principlesandparameters)beapartofournaturalisticinvestigationintothemind,theircharactermustalsobeindividualisticallydetermined.Thatis,thereisabrutefactaboutthestateofanindividual’slanguagefacultyandthatfactisdeterminedinturnbyfactsabouttheindividualinisolation,notbytheenvironmentinwhichtheindi-vidualisembedded.Thethoughtisthatifthelanguagefacultyispartofourbiologi-calendowment,thenthenatureoftherepresentationsutilizedbythelanguagefacultyarefixedbyourbiologyandarenotsensitivetoenvironmentalissuessuchaswhetherwearemovingaboutonEarthorTwinEarth.ThisappearstoputChomskyonacollisioncoursewithfiguressuchasTylerBurge,whoarguesin“IndividualismandPsychology”thatthecontentoftherepresentationspositedinpsychologyaredeterminedatleastinpartbyenvironmentalfactors.Ifthenotionofcontentinvolvesexternalistorenvironmentalnotions,thenChomskyisdubiousthatitcanplayaninterestingroleinnaturalisticinquiryincognitivepsy-chology.Furthermore,sinceforChomsky“themental”issimplyanaspectofthenaturalworldthatisinvestigatedbysciencessuchascognitivepsychology(seeChomsky1994a),thenatureofthementalitselfmustbeindividualisticallyandnotenvironmentallydetermined.ThusChomsky(1993,1995a)rejectsthecontentionsoffiguressuchasPutnam(“TheMeaningof‘Meaning’”),Burge(“IndividualismandtheMental”),andDavidson(“KnowingOne’sOwnMind”)thatthecontentsofourmentalstatesareenvironmentallydetermined.Moreaccurately,hedismissesthetalkof“con-tents”asill-defined.Indeed,heisdismissiveofthethoughtexperiments(Putnam’s“water/twater,”Burge’s“tharthritis,”andDavidson’s“Swampman”)thatpurporttosupportexternalism,andsuggeststhattheyreflectphilosophicalprejudicemorethananygenuinefactsaboutthemind/brain.Ifenvironmentalismistoberejectedinpsychology,thenitnaturallymustberejectedinsemanticsaswell.Thatis,ifthetaskofthelinguististoinvestigatethenatureofI-language,andifthenatureofI-languageisachapterofcognitivepsychology,andifcognitivepsychologyisanindividualisticratherthanarelationalscience,semanticswillwanttoeschewrelationalpropertieslikereference(wherereferenceisconstruedasarelationbetweenalinguisticformandsomeobjectintheexternalenvironment).ThusChomsky(1981,1995b)rejectsthenotionofreferencethathasbeencentraltothephilosophyoflanguagesinceabout1970,characterizingitasanill-definedtech-nicalterm(certainlyonewithnoempiricalapplications),andsuggestingthatintheinformalusageof“refer,”individualsreferbutlinguisticobjectsdonot.Italsofollowsthatsemantictheoriesthatemploythetechnicalnotionofreferenceshouldberejectedinfavorofsemanticaltheorieswhichdonotpurporttostatelan-424\nNOAMCHOMSKYguage/worldrelationsor,followingChomsky(1975a),infavorofaWittgensteinianapproachinwhichthereisnosemanticsperse,butratheroneinwhichthelingusticformsareusedincertainways.Withthisrejectionofreferentialsemanticsalsocomesarejectionofanyattempttousethesemanticsofnaturallanguagetogaininsightsintoontology.Itisnogoodtoarguefromthestructureoflanguagetotheexistenceofevents,orpluralobjects,ortimes,etc.AsChomskyhasargued,thereareanumberofconstructionswherethestructureoflanguageandthestructureoftheexternalworlddiverge.Forexample,somenounphrasesintuitivelyhavecounterpartsintheworld(forexample,thenounphrase“coatsinthecloset”)whileothersdonot(“flawsintheargument”):IfIsay“theflawintheargumentisobvious,butitescapedJohn’sattention,”Iamnotcommittedtotheabsurdviewthatamongthingsintheworldareflaws,oneofthemintheargumentinquestion.Nevertheless,theNPtheflawintheargumentbehavesinallrel-evantrespectsinthemannerofthetrulyreferentialexpressionthecoatinthecloset.(1981:324)Stillmore,Chomskyholdsthatthereisadeepreasonwhyourontologycannotbereflectedinnaturallanguage:ontologyisdeterminedbyhumanintentions,whiletherepresentationsinthelanguagefacultyarenaturalisiticallydetermined.Wedonotregardaherdofcattleasaphysicalobject,butratherasacollection,thoughtherewouldbenologicalincoherenceinthenotionofascatteredobject,asQuine,Goodman,andothershavemadeclear.Butevenspatiotemporalcontiguitydoesnotsufficeasageneralcondition.Onewingofanairplaneisanobject,butitslefthalf,thoughequallycontinuous,isnot....Furthermore,scatteredentitiescanbetakentobesinglephysicalobjectsundersomeconditions:considerapicketfencewithbreaks,oraCaldermobile.Thelatterisa“thing,”whereasacollectionofleavesonatreeisnot.Thereason,apparently,isthatthemobileiscreatedbyanactofhumanwill.Ifthisiscorrect,thenbeliefsabouthumanwillandactionandintentionplayacrucialroleindeterminingeventhemostsimpleandelementaryofconcepts.(1975b:204)Theupshotisthatpursuingmetaphysicalquestionsbyappealtonaturallanguage(I-language)isadeadend.PerhapslesscleararetheprospectsforrecentphilosophicalattemptstoemploytheresourcesofgenerativegrammarincarryingoutDavidson’sprogramofdefiningtruthinnaturallanguage(seeDAVIDSON).Chomsky’sviewappearstobethateverythingdependsuponhowtheseenterprisesareinterpreted.Iftheyaretakentobewaysofexe-cutingareferentialsemantics,thentheyaremisguided.Ifthe“semanticvalues”ofthesetheoriesaretakeninanon-referentialwaythenthereispresumablyroomforinterestingtheorizing.InthisentryI’veonlyscratchedthesurfaceofworkbyChomskythatispotentiallyofinteresttophilosophers(analyticorotherwise).Oneglaringomissionishiswritingonsocialissues(see,forexample,Chomsky1987)andonthemedia(seeHermanandChomsky1988).I’vealsopassedoverhiscontributionsforformallanguagetheory(Chomsky1956,1959)andgeneralissuesinepistemology(Chomsky1981).Finally,thereismuchthatcouldhavebeensaidabouthisearliersyntacticwork(1957,1965,425\nPETERLUDLOW1975a)andtheinfluencethatithadinthephilosophyoflanguageinthe1960sand1970s.Ihope,however,thattheforgoingdiscussionhashelpedtoilluminatesomeofChomsky’sworkandplaceditinthecontextofthedebatesthathavetakenplaceinanalyticphilosophysincethemid-twentiethcentury,debateswhichremainopen4largelyduetohisefforts.Notes1Chomskyoftensuggeststhatifonedigsbeneaththesurface,onefindsthatthesephilosophers(eventhebehaviorists)arealsobelieversinalanguagefacultywhichispartofourbiologicalendowment.SeehisdiscussionofQuineinChomsky1975b:198ff.OnChomsky’sview,ofcourse,nocoherentstorycanbetoldwithoutthisassumption.2InChomsky1975b,1977,forexample,theideaisthat(1)representsasubjacencyviolation;formationofthequestionwouldrequirethewh-element“who”tomoveoutofbothanNP(nounphrase)andanS(clause)withoutasafeintermediatelandingsite.FormorecurrentaccountsoftheseconstructionsseeChomsky1986a,1995b.3However,ChomskyobservesthatitisnotmodularinthesenseofFodor’sModularityofMind,butisanacquisitionmodulemoreinthesenseofGallistelinTheOrganizationofLearning.4IamindebtedtoNoamChomsky,RichardLarson,andA.P.Martinichforcommentsonanearlierdraftofthisarticle.BibliographyWorksbyChomsky1956:“ThreeModelsfortheDescriptionofLanguage,”I.R.E.TransactionsofInformationTheory,IT-2,pp.113–24.1957:SyntacticStructures,TheHague:Mouton.1959:“OnCertainFormalPropertiesofGrammars,”InformationandControl2,pp.137–67.1965:AspectsoftheTheoryofSyntax,Cambridge,MA:MITPress.1966:CartesianLinguistics,NewYork:HarperandRow.1969:“Quine’sEmpiricalAssumptions,”inWordsandObjections:EssaysontheWorkofW.V.Quine,ed.D.DavidsonandJ.Hintikka,Dordrecht:D.Reidel.1971:ProblemsofKnowledgeandFreedom:TheRussellLectures,NewYork:VintageBooks.1975a:TheLogicalStructureofLinguisticTheory,Chicago:UniversityofChicagoPress.(Originallyappearedinunpublishedmanuscriptformin1955.)1975b:ReflectionsonLanguage,NewYork:Pantheon.1977:“ConditionsonRulesofGrammar,”inEssaysonFormandInterpretation,Amsterdam:ElsevierNorthHolland,pp.163–210.1980a:“OnCognitiveStructuresandtheirDevelopment:AReplytoPiaget,”inLanguageandLearning:TheDebatebetweenJeanPiagetandNoamChomsky,ed.M.Piatelli-Palmerini,Cambridge,MA:HarvardUniversityPress,pp.35–54.1980b:RulesandRepresentations,NewYork:ColumbiaUniversityPress.1981:LecturesonGovernmentandBinding,Dordrecht:ForisPublications.1982:SomeConceptsandConsequencesoftheTheoryofGovernmentandBinding,Cambridge,MA:MITPress.1986a:Barriers,Cambridge,MA:MITPress.1986b:KnowledgeofLanguage,NewYork:Praeger.1987:TheChomskyReader,NewYork:PantheonBooks.426\nNOAMCHOMSKY1990:“Accessibility‘inPrinciple’,”BehavioralandBrainSciences13,pp.600–1.1992:“ExplainingLanguageUse,”PhilosophicalTopics20(Spring),pp.205–31.1994a:“NaturalismandDualismintheStudyofLanguageandMind,”InternationalJournalofPhilosophicalStudies2,pp.181–209.1994b:“NoamChomsky,”inACompaniontothePhilosophyofMind,ed.S.Guttenplan,Oxford:BlackwellPublishers,pp.153–67.1995a:“LanguageandNature,”Mind104,pp.1–61.1995b:TheMinimalistProgram,Cambridge,MA:MITPress.1988(withHerman,E.):ManufacturingConsent:ThePoliticalEconomyoftheMassMedia,NewYork:PantheonBooks.WorksbyotherauthorsGeorge,A.(ed.)(1989)ReflectionsonChomsky,Oxford:BlackwellPublishers.Harman,G.(ed.)(1974)OnNoamChomsky:CriticalEssays,GardenCity:AnchorBooks.Hornstein,N.andAntony,L.(eds.)(forthcoming)ChomskyandHisCritics,Oxford:BlackwellPublishers.Kripke,S.(1982)WittgensteinonRulesandPrivateLanguage,Cambridge,MA:HarvardUniversityPress.427\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200135RichardRorty(1931–)MICHAELWILLIAMSRichardRortyhastaughtatWellesley,Princeton,andtheUniversityofVirginia.SinceretiringfromVirginia,hehasbeenamemberoftheDepartmentofComparativeLiteratureatStanford.Earlyinhiscareer,Rortywroteextensivelyontopicsinthephilosophyofmind,emergingasaninfluentialdefenderofeliminativematerialism.Buthewasalsocon-cernedwithmetaphilosophicalquestions.Hisintroductiontohisanthology,TheLinguisticTurn,surveysthehistoryoftheanalyticmovementwiththeaimofcastingdoubtontheviewthat,bycenteringphilosophyonquestionsoflanguageandmeaning,analyticphilosophyprovidesphilosopherswithnewandmore“scientific”methodsforsolvingtraditionalphilosophicalproblems.Thisargumentforeshadowstheradicalturntakenbyhismaturework.Themainthemesofthisworkemergeinaseriesofessayspublishedinthe1970sandcollectedinConsequencesofPragmatism(1982).However,itwashisbook,PhilosophyandtheMirrorofNature(1979),thatmadehimtheobjectofintenseandoftenoutragedcriticalscrutiny.Inthisbook,hearguesthatphilosophy,aspracticedinmainstreamAnglo-Americanphilosophydepartments,hasexhausteditstheoreticalresourcesandoutlivedwhateverusefulnessitmayoncehavehad.Itthereforedeservestocometoanend.Likeother“therapeutic”philosophers,Rortyholdsthatourcanonical“problemsofphilosophy”aretobeavoidedratherthansolved.However,thisisnotbecauseheseesthemaspseudo-problems,rootedinmisunderstandingsormisusesoflanguage.Rorty’sapproachishistoricist.Hedeniesthatphilosophydealswithperennialproblems,intel-ligibletoanyreflectivepersonbecausepartofthehumancondition.Ourcanonicalproblems,then,aregenuineenough,butonlyinthecontextofahistoricallycontin-gent,hencepotentiallyoptional,configurationofideas.Althoughtheymayoncehavepromisedgreatthings,theseideascannowresponsiblybedropped.InRorty’snarrative,modernphilosophytakestheformofepistemologyorthetheoryofknowledge.PhilosophyofthiskindoriginatesintheseventeenthcenturyandachievesitsdefinitiveforminthewritingsofKant.Descartesinauguratesmodernphilosophy’sepistemologicalturnbymakingtwomoves:introducingmethodologicalskepticismastheprincipaltoolforinvestigatingthefoundationsofknowledge,andredefining“mind”asthattowhicheachofushasprivilegedaccess.Giventhis428\nRICHARDRORTYconceptionofmind,skepticismitselfacquiresanewandmoreradicalform.Fortheancients,skepticismraisedthequestionofwhetherwecanattaincertaintyaboutthe“realnature”ofthings.AfterDescartes,itraisesthequestionoftowhatextent,ifany,our“ideas”areaccuraterepresentationsof“external”reality.Theveryexistenceoftheexternalworldissubjecttodoubt.Descartes’sphilosophicalprojectisfoundationalintwosenses.Itaimsatidentifyingbothepistemologicalfoundations(certaintiesthatresistskepticalchallenge)andmeta-physicalfoundations(themostbasicexplanatorycommitmentsoftheNewScience).AsKantsaw,themetaphysicalaspirationsofDescartesandhisrationalistsuccessorsareproblematic.Metaphysicianswanttodetermineapriori,onthebasisofourideasalone,fundamentalfactsabouttheworld.Thiscannotbedone:Descartes’sskepticalproblemthwartshismetaphysicalambitions.Rationalistmetaphysicsisthusmeredogmatism.Locketakesanimportantsteptowardsamorepurelyepistemologicalconceptionofthephilosopher’staskbysuggestingthat,byinvestigatingthepowersoftheCartesianmind,wecandeterminethescopeandlimitsofhumanknowledge.Locke,however,isinsensitivetothepowerfulandgeneralskepticalproblemformulatedbyDescartes.Lockeclaimstoinvestigatethelimitsofhumanknowledge.But,asKantcharges,inadoptinga“historical”–i.e.empirical-psychological–approachtotheoriginsofourbeliefs,Lockefailstoaddresstheepistemologicalquestionofourrighttoholdthem.Moreover,wherethemetaphysiciansatleastattempttojustifythebasicpresuppositionsofmodernscience,Lockesimplytakesforgrantedthecorpuscular-mechanicalpictureoftheworld.Kantpresentshistranscendentalidealismasthewaybeyondrationalistdogmatismandempiricistnaturalism.Histhoughtisthat,sinceallempiricallyknowableobjects,“outer”aswellas“inner,”aresubjecttoconditionsinherentinourcognitiveconstitu-tion,wecanhaveaprioriknowledgeoffeaturesnecessarilycharacteristicoftheworldasweareabletoknowit.However,notallmattersofhumanconcernanswertotheseconditionsofobjectiveknowability.Thosethatdonotremainmattersofjudgmentorfaith.Kantthuspresentsuswiththeideaofepistemologyasanon-empiricaldisciplinethatdeterminesthecognitivestatusofallothersubjectsaccordingtohowfartheyarecontrolledbyreasonandevidence,hencewhethertheyaimatobjectivetruth.Thusinmodernphilosophy,“refutingtheskeptic,”nowconceivedasestablishingourrighttoclaimknowledgeofanobjective,causallyorderedworld,ceasestobe“thelanguidaca-demicexerciseofcomposingareplytoSextusEmpiricus”(Rorty1979:223),becom-inginsteadthekeytodistinguishingbetweenformsofdiscoursethatare“rational,”“scientific,”or“cognitivelysignificant”andthosethatare“emotive”or“merelyexpres-sive.”Philosophy-as-epistemologybecomescentraltoculture.MichaelDummettarguesthatFrege,thefounderofanalyticphilosophy,isasmucharevolutionaryasDescartes(seeDUMMETT).InDummett’sview,Frege’srevolutionreplacesepistemology,asthefoundationofphilosophy,withphilosophyoflanguageor“thetheoryofmeaning,”withtheresultthatanalyticphilosophyissharplydiscontin-uouswithphilosophy-as-epistemology.Rortyseesnosuchdiscontinuity.Fregeisa(notablyoriginal)memberofthe“backtoKant”movement.Histurntologicandlan-guageisanattempttoeliminatetheKantiantradition’slastvestigesofpsychologism,therebyrescuingphilosophyfromthescientificnaturalismthatwasthreateningto429\nMICHAELWILLIAMSoverwhelmit.Analyticphilosophythuscontinuestopursue,intheidiomof“lan-guage,”theepistemologicalquestionsthatKantandhispredecessorspursuedintheidiomof“ideas”:segregatingthecognitivelysignificantfromthemerelyexpressive,drawinglinesbetweentheaprioriandtheempirical,showingwhereweshouldandshouldnotbe“realists”abouttruth,andsoon.AdistinctionthatisabsolutelyessentialtothisKantianstyleofphilosophizingisthatbetweenschemeandcontent.Acceptingthisdistinction,wewillseeempiricalknowl-edgeasinvolvingtwoclearlydistinguishablecomponents,conceptsandintuitions,orasresultingfromthecooperationoftwofaculties,understandingandsensibility.Onthismodel,“mind”or“language”ordersorinterpretsthefactualelements“given”toconsciousness.Takentogether,Rortyargues,Sellars’sattackon“themythofthegiven,”Quine’sskepticismabouttheanalytic/syntheticdistinction,Wittgenstein’scri-tiqueofostensivedefinitionand“privatelanguage,”andAustin’ssarcasmabout“theontologyofthesensiblemanifold”leavethisfundamentalcommitmentnolongercred-ible(seeAUSTIN,QUINE,SELLARS,andWITTGENSTEIN).RortyseesthesecriticsoftheKantiantraditionasunitedbyakindofmethodo-logicalbehaviorism.Intheirdifferentways,theyinviteus,first,tolookathowweactuallyusewords,revisebeliefs,evaluatetheories,orconductinquiriesand,second,toaskwhetherthereisanypayoff,theoreticalorpractical,inpartitioningourbeliefsorstatementsinto“true-by-virtue-ofmeaning-aloneversustrue-by-virtue-of-fact”or“purelyobservationalversustheory-laden.”Theansweris“No.”Theadvantageoftakingthelinguisticturn,then,isnotthatitoffersnewwaysofsolvingoldproblemsbutthatitmakesthismethodologicalorientationplausible,therebyallow-ingustosettheoldproblemsaside.Inthisway,analyticphilosophytranscendsandcancelsitself.Thepictureofinquiryandjustificationthatresultsfromabandoningthedualismofschemeandcontentisholistic,coherentist,andpragmatic.Inquiryisaprocessofcon-stantlyreweavingourwebofbeliefundertheimpactofobservationandinthelightofmultipleinterestsandcriteria,theoreticalandpractical.Rortythinksthatthisholisticpictureblursallthemethodologicaldistinctions–betweentheaprioriandtheapos-teriori,thenecessaryandthecontingent,factandvalue,thesciencesandthehuman-ities,andsoon–thatphilosophersbentonprojectsofepistemologicalormetaphysicaldemarcationwanttokeepalive.PresentinPhilosophyandtheMirrorofNature,butmuchmorestronglyemphasizedinsubsequentwritings,istheclaimthatthemostfundamentalerrorofourphilo-sophicaltraditionisthenotionthattruthiscorrespondencewithrealityoraccuracyofrepresentation.Thequestfortruth-as-correspondencereflectsanurgetobeguidedbysomethinggreaterthanourselves:theWorld,theTrue,ortheGood.(Rortythinksoftoday’shard-headedscientificrealismasevincinganessentiallyreligiousattitude.)Thisquest(whichisasoldasphilosophyitself,philosophy-as-epistemologybeingsimplyitsmodernincarnation)isalwaysassociatedwithdemarcationalprojectsdividingmattersofhumanconcernintoanupperandlowerdivision:knowledgeversusopinion,natureversusconvention,philosophyversuspoetry.However,inadditiontounderminingmethodologicalgroundsforsuchdemarcations,theholistic,broadlycoherentistandpragmaticconceptionofinquirycommontoQuine,Sellars,andWittgensteinmakesitdifficulttoseeindividualsentencesorbeliefsas“corresponding”toanything.Whether430\nRICHARDRORTYwelookatinquiryfromthestandpointofmethodorthatoftruth,wefindnoroomforphilosophy.Rorty’sfocusontruthreflectsanincreasingself-identificationwithpragmatism.Havingadoptedabroadlycoherentistpictureofjustificationandinquiry,RortyflirtedbrieflywiththePeirceansuggestionthattruthisidealjustification.However,hissettledoutlook–whichheidentifieswiththepragmatismofJamesandDewey–isaradicalanti-essentialismwithrespecttothetraditionalobjectsofphilosophicalconcern.Rorty’sPragmatistdoesnotreplaceacorrespondencetheoryoftruthwithanepistemicaccountbutratherholdsthattruth(orrationalityorgoodness)isnotthesortofthingthatwecanusefullytheorizeabout.Rortythinksthat,amongcontemporaryphilosophers,DonaldDavidsonhasdonemosttoadvancethepragmatistcause.AccordingtoRorty,Davidson’sworknotonlyreinforcesSellars’srejectionof“given”factsandQuine’srepudiationoftheanalytic/syntheticdistinction,ittracestheconnectionsbetweenbelief,truth,andmeaninginawaythatdeprivesthesenotionsofalldemarcationalsignificance.Foralltheircriticismsoftraditionalepistemology,SellarsandQuinearepronetobackslidingbecausetheyremaincommittedtotheviewthatthenaturalsciences,especiallyphysics,getat“hardfacts”or“theultimatenatureofreality”inawaythatthesofterdisciplinesdonot.DavidsonisabletogobeyondSellarsandQuinebecauseheiswhollyfreeofthislingeringscientism.Perhapsbecauseneitherapproachtotruthmakesourunderstandingoftruththekeytotraditionalepistemologicalormetaphysicalproblems,RortypaysscantattentiontothedistinctionbetweenDavidson’sviewthattheconceptoftruth,whileofconsid-erableexplanatorysignificanceinthetheoryofmeaning,mustbetakenasprimitive,andthe“deflationary”viewthattruth-talkisonlyanexpressiveconvenience.Indeed,heoftentreatsDavidson’sviewasaformofdeflationism,asuggestionthatDavidsonemphatically(thoughperhapsnotentirelyconvincingly)repudiates.AnothernotableinfluenceonRorty’sversionofpragmatismisThomasKuhn.RortythinksthatKuhn’sdistinctionbetween“normal”and“revolutionary”scienceinviteswideapplication.Inallareasofdiscourse,therearetimeswheninquiryproceedsmoreorlessnormally,solvinginagreed-uponwayscommonlyrecognizedproblems,formu-latedinafamiliarvocabulary.Butsometimeswecanmakeprogressonlybydroppingoldquestionsinfavorofnewones,orbychangingthebasicvocabularyintermsofwhichourproblemsandprojectsaredescribed.Rortythinksthathisownpragmatistattackontraditionalphilosophyisaninstanceofjustsuchanattemptatrevolution-arychange.Rorty’srejectionofthecorrespondenceor“realist”conceptionoftruthisoftenthoughttoamounttoanextremeformoflinguisticidealism.Ifourbeliefsdonotanswertotheworld,truthissomethingwemakeup:theideaofobjectivetruthgoesbytheboard.Rortythinksthattheideaof“answeringtotheworld”confusescausa-tionwithjustification.Becausewearetrainedinobservation-reportingpracticesinvolvingthecausaltriggeringofreportingdispositionsbyexternalcircumstances,theworldplaysacausalroleinregulatingourbeliefs.Butitdoesnotplayajustifyingrole.Thesituationsthatprovokesuchreportsdonotdemandtobedescribedinanyparticularvocabularyanddonotdeterminetheinferentialortheoreticalsignificanceofthereportstheyprovoke.431\nMICHAELWILLIAMSCriticssometimeschargethatgivinguponasubstantivenotionoftruth,whetherrealistorPeircean,preventsRortyfromseeinginquiryasprogressing.Rortymeetsthischargebysayingthatimprovementsaremeasuredretrospectivelyandcomparatively–byreferencetoproblemssolved,improvementsmade,oralternativesforegone–ratherthanbytheirshorteningthedistancebetweenourselvesandtheEndofInquiry.Wehavenoconceptionofwhatitwouldbeforinquirytohaveanend,noideaof“theTruth”astheIdealTheoryofEverythingorthewaythatNatureitselfwouldliketobedescribed.Rortyhasalsobeenwidelycriticizedforpreachingirrationalismandrelativism.Herejectsbothcharges.Heagreesthathisrelaxedversionofcoherentismentailsthatjus-tificationislessalgorithmicthanmanyepistemologistshavewantedittobebutdeniesthatthisisequivalenttotheclaimthatanyonecan(rationally)thinkwhateverhelikesorthatanysystemofbeliefsisasgoodasanyother.Oursettledbeliefs,involuntaryobservations,andtheoreticalandpracticalinterestsprovidealltheconstraintweneed(andcanpossiblyhave).Hisposition,heconcedes,is“ethnocentric”inthefollowingsense:atanystageofinquiry,wecanonlyworkwithwhateverbeliefsandtheoriesandcriteriawehaveonhand.Thatis,wehavetoaccepttheirreduciblecontingencyofourinvestigativeandargumentativeresources.Giventhiscontingency,therearelikelytobeissueswithrespecttowhich,atanygiventime,notallpeoplecanfindcommonground.Butthisdoesnotmeanthatsome(orany)disputesreflectcommitmentsthatareinprinciple“incommensurable.”Wecannotpredictthefutureofinquiryandneverknowhowthedialecticalsituationwillevolve.Rortythinksthatonlydisappointedfoundationalistswillequatehisthoroughgoingfallibilismwithskepticism,relativism,orirrationalism.Inrecentyears,Rorty’swritingshavetakenapoliticalturn.Hedefendsapositionhesometimescalls“postmodern,bourgeoisliberalism”:“bourgeoisliberalism”becauseitfullyendorsestherightsandfreedomstypicallyguaranteedbytherich,industrialdemocracies;and“postmodern”becauseiteschewstheneedforprovidingthoserightsandfreedomswithaphilosophicaljustification.Rortyrecognizesthatmanyphiloso-phersthinkthat,ifwegiveuponsuchEnlightenmentconceptionsasuniversalreasonandtheRightsofMan–thekindsofthingphilosophyisinvokedtounderwrite–weleaveourselveswithnowayofshowingwhatiswrongwithoppressive,discriminatory,ortribalistformsofpoliticallife.Indeed,hethinksthatconcernsaboutrelativismandirrationalismgrowoutofjustthisfear.Inreply,headvocatesfacinguptothe“priorityofdemocracytophilosophy.”Democraticconstitutionsandtheruleoflawareappeal-ingtopeoplewithourhistoryandculturalbackground,butoftentootherpeopletoo,iftheygetthechancetoenjoythem.Thosewhowantphilosophicalfoundationsforliberal-democraticinstitutionsshouldrecallthatsuchinstitutionsdidnotappearovernight.Extendingpoliticalrightsandlegalprotectionstoallcitizens,withoutregardtoreligion,race,orgendertooktime;and,inRorty’sview,thisincreasinginclusivenessowesmoretoanenlargementofsympathiesthandiscoveriestotheeffectthatratio-nalityormoralconsiderabilityismorewidespreadthanusedtobethought.Imaginativeliteratureandinvestigativejournalismhavedonemorefortheoppressedandexcludedthaninquiriesintothe“foundations”ofmoralsandpolitics.UnusuallyforanAmericanphilosopher,Rortyhaswrittenextensivelyaboutsuch“continental”figuresasHusserl,Heidegger,Foucault,andDerrida.Heseescontinen-432\nRICHARDRORTYtalandanalyticphilosophyashavingfollowedparallelcourses.LikeFrege,Husserlwantedphilosophytoberigorousandscientific,yetdeeperthanandpriortothespecialsciences.AlsolikeFrege,hesoughtthisdepthandpriorityinageneralaccountofrep-resentation.UnlikeFrege,whoturnedtologicandlanguage,Husserllookedforatheoryoftheinvariantstructuresofconsciousness.Buthetooprovokedapragmatistreac-tion.Roughlyspeaking,Heidegger(especiallytheHeideggerofBeingandTime)standstoHusserlasthelaterWittgensteinstandstoFregeandRussell.TheconsequencesofthisreactionarefurtherworkedoutinDerrida’sdeconstructivereadingsofseminalphilosophicaltextsandFoucault’shistoricistreconstructionsofvanishedconceptionsofscientificknowledge.Whileheisgenerallyregardedasarguingforthedeathofphilosophy,thisisadescriptionRortyrepudiates.FollowingSellars,hesuggeststhat“philosophy”canbeunderstoodtwoways.Ontheonehand,thereisphilosophy(littlep):theattempt“toseehowthings,inthebroadestpossiblesenseoftheterm,hangtogether,inthebroad-estpossiblesenseoftheterm.”ThisHegelianprojectofgraspingone’stimeinthoughtcouldonlycometoanendifinquiryitself(inabroadsensethatencompassesscience,thehumanities,literature,politics,andthearts)groundtoahalt.Ontheotherhand,thereisPhilosophy(bigp):thePlatonic–Kantianprojectofdetermininghowtoseektruth(orconductoneselfrationallytodomoregood)throughdiscoveringthenatureoftruth(orrationalityorgoodness).Wherephilosophyseeksreflectiveself-understanding,andperhapsself-transformation,butalwaysataparticularstageofinquiry,Philosophytriestodiscernthepermanentframeworkwithinwhichallinquiryproceeds.IntryingtokilloffPhilosophy,Rortylooksforwardtoa“post-Philosophicalculture,”inwhichsuchaquestwilllookasquaintasmedievaltheologicaldisputeslooktosecularintellectualstoday.LearningtodowithoutPhilosophy,asmostintellectualshavelearnedtodowithoutreligion,meanscomingfinallytotakefullresponsibilityforouropinionsandvalues.Rorty’sphilosophyisthusaversionof“humanism,”inSartre’ssense.BibliographyWorksbyRorty1967:TheLinguisticTurn,Chicago:UniversityofChicagoPress.1979:PhilosophyandtheMirrorofNature,Princeton,NJ:PrincetonUniversityPress.1982:ConsequencesofPragmatism,Minneapolis:UniversityofMinnesotaPress.1988:Contingency,Irony,andSolidarity,Cambridge:CambridgeUniversityPress.WorkbyotherauthorsMalachowski,A.(ed.)(1990)ReadingRorty,Oxford:BlackwellPublishers.433\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200136JohnR.Searle(1932–)A.P.MARTINICHJ.L.AustinwasbeingrecruitedbytheUniversityofCaliforniaatBerkeleyinthelate1950s.Hedeclined,saying,“IthinkIshouldbedeadbythen”andthereuponadded,“Sinceyoucan’tgetme,getSearle.”Searlebecameanassistantprofessortherein1959,thesameyearhereceivedhisD.Phil.fromOxford.Asidefromvisitingappointmentsandleaves,hehasspenthisentirecareeratBerkeley,whereheisMillsProfessorofPhilosophy.InadditiontoAustin,SearlehadbeenastudentofP.F.Strawson,PeterGeach(whodirectedhisdissertation),andotherdistinguishedOxfordphilosophersduringtheheydayofordinary-languagephilosophy.Adeptatcriticism,Searleisevenmoreimpres-siveasaconstructivephilosopher.Eveninsomeofhisearlyarticles,whichareosten-siblycriticismsofothers,hisownpositivetheoryisnotfarbelowthesurface.LanguageIn“AustinonLocutionaryandIllocutionaryActs,”SearleshowshowAustin’soriginallinguisticdistinctionsshouldberecast.Themostimportantresultisthatparadigmaticcasesofillocutionaryactsshouldbeunderstoodasconsistingofaforceandapropo-sitionalcontent(seeAUSTIN).Consider,forexample,thesesentences:IstatethatJoneswillbeattheparty.IpromisethatJoneswillbeattheparty.IquestionwhetherJoneswillbeattheparty.Itisobviousthatallofthesesentenceshavesomethingincommon.EachwouldbeappropriatelyusedtoexpressthesamecontentorpropositionthatJonesisattheparty.(Forsimplicity’ssake,thetemporalelementwillbeignored.)Inthesesentences,thepropositionalcontentisexpressedwitha“that”clause;butsomesentencesexpresstheircontentwithgerundivephrasesorinfinitives:IcongratulateJonesforbeingattheparty.IorderJonestobeattheparty.Itisequallyobviousthatastandarduseofeachoftheabovesentencesexpressestheirpropositionalcontentwithadifferent“force,”theforceofastatement,promise,ques-434\nJOHNR.SEARLEtion,congratulation,andorder,respectively.SothestructureofillocutionaryactscanberepresentedasF(p).Thepcorrespondstothepropositionalcontent,andtheFindi-catesthe“force”attachedtotheproposition.Ineffect,SearlemovedAustin’srheticactsfromthecategoryoflocutionaryacttoasubpartofanillocutionaryact.In“WhatisaSpeechAct?”(1965)andSpeechActs(1969),Searleusedpromisingtoillustratetheappropriateformofanalysisforillocutionaryacts.Withslightmodifi-cation,Searle’sanalysiswasthis:InutteringasentenceT,aspeakerSpromisesanaddresseeHtodoanactionAifandonlyif1Normalinputandoutputconditionsapply.2SexpressesthepropositionthatpintheutteranceofT.3Inexpressingthatp,SpredicatesafutureactAofS.4HwouldpreferS’sdoingAtoS’snotdoingA,andSbelievesHwouldpreferS’sdoingAtoS’snotdoingA.5ItisnotobvioustobothSandHthatSwilldoAinthenormalcourseofevents.6SintendstodoA.7SintendsthattheutteranceofTwillplacehimunderanobligationtodoA.8Sintends[i-1]toproduceinHtheknowledgeKthattheutteranceofTistocountasplacingSunderanobligationtodoA.SintendstoproduceKbymeansoftherecognitionofi-1,andheintendsi-1toberecognizedinvirtueof(bymeansof)S’sknowledgeofthemeaningofT.9ThesemanticalrulesofthedialectspokenbySandHaresuchthatTcorrectlyandsincerelyutteredTifandonlyifconditions(1)–(8)obtain.CertainaspectsofSearle’sanalysismightbefine-tuned.Abetteranalysandumis:“InutteringasentenceT,aspeakerSexplicitlyandnon-defectivelypromisesanaddresseeHtodoanactionA.”Perhapsanotherpreparatoryconditionisappropriate:“SisabletodoA.”Notwithstandingtheseandotherpossibleadjustments,theformofSearle’sanalysisispowerfulandeasilyadaptedtoanalyzethefullspectrumofillocutionaryacts.Anothermeritofthisformofanalysisisthatcategoriesofconditionseasilyemergefromthem.Mostimportantly,conditions(2)and(3)expressrequirementsforthepropo-sitionalcontentoftheact.Conditions(4)and(5)expresspreparatoryconditions.(6)expressesasinceritycondition.(7)expressesan“essentialcondition,”whichconveystheaimorgoaloftheact.InExpressionandMeaning,Searledevelopedataxonomyofspeechacts:assertives(forexample,statementsandasseverations);directives(forexample,commandsandsuggestions);commissives(forexample,promisesandvows);expressives(forexample,apologiesandcongratulations);anddeclarationsandassertivedeclaratives(forexample,declarationsandverdicts,respectively).UnlikeAustin’staxonomy,whichwasfoundedonnoprinciplesandZenoVendler’s,whichwasfoundedonsyntacticprinci-ples,Searle’staxonomyissemanticallybased(andalsocloselyrelatedtothetypesofconditionsalreadydescribed).Firstofall,illocutionaryactsarecategorizedonthebasisoftheirpointorpurpose,asexpressedbytheiressentialconditions.Assertivesaimatcommittingspeakerstobeliefs.Directivesandcommissivesaimatcommittingsomeonetoacourseofaction.Expressivesaimatexpressingamentalstate,suchashappiness435\nA.P.MARTINICHorsadness.Declarationsandassertivedeclarationsaimatbringingaboutsomefactabouttheworld;whentheChairmanoftheOlympicCommitteesays,“IherebyopentheGames,”thegamesaretherebyopened.Anotherdimensionofcategorizationisword/worldfit.Thus,toassert“Thedoorisopen”istoaimatgettingthewordstofitorcorrespondtothewaytheworldis.Ifthedoorisnotopen,thedeficiencyiswiththewords.Incontrast,tocommand,“Youwillopenthedoor,”istoaimatgettingtheworldtofitthewords.Iftheaddresseedoesnotopenthedoor,the“deficiency”iswiththeworld.Ofcoursethisworldlydeficiencytendstomaketheaddressee,nottheworld,culpable.Takingourleadfromthesetwoexam-ples,wecansaythatassertivesaimathavingtheirwordsfitthewaytheworldis,whiledirectivesandcommissivesaimatgettingtheworldtofitthewaythewordssayitistobe.Expressivesdonothaveadirectionoffitbutpresupposesomefactabouttheworld.Theuseof“Icongratulateyouonwinning”and“Iapologizeforsteppingonyourfoot”presupposeavictoryandanoffense.Athirddimensionofcategorization,relatedtosincerityconditions,isthepsycho-logicalstateexpressedintheillocutionaryact.Assertivesandassertivedeclarationsexpressthespeakers’beliefs.Directivesexpressthespeakers’wantsanddesires.Commissivesexpressthespeakers’intentions.Declarationsdonotexpressanypsycho-logicalstate;warrantedbyinstitutions,declarationsdonotneedsincerity.Afourthdimensionofcategorization,relatedtopropositionalcontentconditions,concernswhatillocutionaryactscanbeabout.Assertiveshavevirtuallynorestrictiononpropositionalcontent.Directivesandcommissivesmustbeaboutfutureactions.Expressivesmustbeaboutpresentorpastactionsorconditions.Declarationscanbevirtuallyaboutanything,althoughtherearelimits.Onecannotfryaneggbyuttering,“Iherebyfrythisegg.”Otherdimensions,suchastheintensityofillocutionarypoint(suggestingversusinsisting)andthewaytheactrelatestotherestofthediscourse(objectingversusreply-ing),whileinformative,arenotcrucialtothebasictaxonomy.Asrevealingasthetax-onomyis,itwillacquireevengreaterimportancebecauseofitsconnectionwithintentionality.Sofar,ourdiscussionhasfocusedonthenatureofillocutionaryforce.Thisfocussug-geststhenoveltyofspeechacttheory,forphilosophershadconcentratedonpropositionsfortwoandahalfcenturies.Let’snowconsiderhowSearletreatsthishoarymatter.Paradigmaticpropositionsconsistofareferenceandapredication.Traditionally,ref-erenceisconsideredthebasicwayinwhichwordsrelatetotheworld.ThespiritofthattraditioniscapturedbySearle’sformulationoftheaxiomofexistencenamely,thateverythingreferredtomustexist.Thisinvitesthequestionofhowthewordorphraseusedtorefergetshookeduptotheworld.Thecompletestoryrequiresatheoryofinten-tionality(tobepresentedlater)sincereferencedependsonphysicalexpressionssuchaswordsbeingbackedupbyinherentlyrepresentationalmentalstates.Atthispoint,however,Searleisabletosaythatreferencedependson“theaxiomofidentification,”namely,thataspeakermustbeabletoidentifytheintendedreferentforhisaudienceinanontrivialway.Theobjectisidentifiedeitherbythedescriptivecontentoftherefer-ringexpression(“thefirsthumantosetfootonthemoon”),bythereferringexpressionplusthecontext(“thisonehere”),orbyacombinationofthetwo(“thisredshoehere”)(Searle1969:80).436\nJOHNR.SEARLETraditionally,propernameshavebeentheparadigmatickindofreferringexpression;andtheproperunderstandingofthemhasoccupiedphilosophersfortwomillennia.Torestrictourdiscussiontothelasthalf-century,twotheorieshavedominated:thedescriptivetheoryandthecausaltheory.Accordingtothecausaltheory,whichSearlerejects,apropernamegetsconnectedwithitsreferentinvirtueofacausalconnectionbetweenthatname,thespeaker’sintentiontohavetheaddresseeidentifyaparticularobjectthroughheruseofthatname,andthatparticularobject.Incontrast,accordingtothedescriptivetheory,namesrefertoreferentsinvirtueoftheirdescriptivecontent.TakingofffromFrege’sviews,Searlemaintainsthatpropernameshavebothsense(Sinn)andreference(Bedeutung),andthereferenceisafunc-tionofthesense(seeFREGE).Consideranamelike“Aristotle,”whichreferstoAristotle.Thatreferenceoccursinvirtueofacertaindescriptivecontent.Unlikegeneralwordslike“red”or“human,”whichseemtobeobviouslytiedsemanticallytorednessandhumanity,respectively,propernamesseemtohavealooserbutnonethelessindispens-ablesense.Themeaningof“Aristotle”is“thelogicalsum[inclusivedisjunction]ofthepropertiescommonlyattributedtohim”(Searle1969:173).AlthoughSearlehasnevertakenbackanypartofthisviewofthemeaningofpropernames,hisfocusseemstochangeinhislaterwork.Whileinsistingthatsomerepre-sentationalcontentmustaccompanytheuseofapropername,hedoesnotassertthatthename’smeaningisthatrepresentationalcontent.Further,likethecausaltheorist,Searlemaintainsthattherepresentationalcontentiscausallyrelatedtotheexternalworld(Searle1983:238).IthinkthatSearleoughttoabandontheclaimthatpropernameshaveSinne.Fromthefactthateveryuseofapropernamemustbeaccompaniedbysomerepresentationalcontent,itdoesnotfollowthatthatcontentisthemeaningofthatname.Itfurtherdoesnotfollowthatthereisanystablecontent,sharedbythepeoplewhousethatname.SearlemaybeconflatingFrege’smodeofpresentationwiththepropositionalcontentofanutterance(seeSearle1983:249,251).WhatismostimportantinSearle’stheory,itseemstome,istheroleofintentionalityinreference.Hethinksthatthiscommitshimtodescriptivismbecausehetakeshismainopponents,KripkeandDonnellan,todiscountintentionality.IbelievethatSearle’stheorycapturestheintuitionsofbothdescriptivismandthecausaltheory,withoutfallingintotheerrorsofeither,andthathegivesthewrongimpressioninclaimingtobeadescriptivist.Concerningtheotherpartofaproposition,predication,SearleacceptspartoftheasymmetrythesisofGottlobFregeandP.F.Strawson.Whilereferenceistheactofpickingoutoridentifyinganobjectforthepurposeofclassifyingorcategorizingit,predi-cationistheactofassigningapropertytothereferent.Asubject-predicateproposi-tionistrueifandonlyifthereferenthasthepropertyascribedtoit.ForSearle,thedis-tinctionbetweensubjectandpredicateorreferenceandpredicationisprimarilyoneoffunction,thatis,ofhowsomethingisoperatingandonlysecondarilyoneofontol-ogy.Asanontologicalissue,thereweretwobasicchoices:nominalism(thedenialthatproperties,oruniversals,exist)orrealism(theassertionthatuniversalsexistasmuchasindividualmaterialobjectsdo).Searle’sbasicpositionisthatthereisnosubstantiveissuehere.Universalsexistsolelybecausepredicatesallownominalization.Becauseofsentenceslike“Socratesiswise,”wecanformsentenceslike“Wisdomisavirtue.”Consequently,althoughthereareuniversals,theydepend“merelyonthemeaning437\nA.P.MARTINICHofwords”(Searle1969:105).Predicatingauniversalmeansusing“apredicateexpres-sionintheperformanceofasuccessfulillocutionaryact”(Searle1969:121;seealso124).Accordingtothetraditionaltaxonomy,Searlewouldcountasaconceptualist,anominalistofaliberalexpression;wordsarepredicatedoftheobjectsreferredto(Searle1969:124)anduniversalsexist.ConsciousnessSinceaspeechactisakindofhumanactionthatrequiresamentalrepresentationoftheworld,acompletetheoryofspeechactswillbepartofatheoryofmind.Formostphilosophersthemainprobleminthisareais“themind–bodyproblem.”Whatisthemind;whatisthebody;andhowdotheyinteract?ForSearle,thesequestionsarenotproblematic:Thesolutionhasbeenavailabletoanyeducatedpersonsinceseriousworkbeganonthebrainnearlyacenturyago,and,inasense,weallknowittobetrue.Hereitis:Mentalphenomenaarecausedbyneurophysiologicalprocessesinthebrainandarethemselvesfeaturesofthebrain.(Searle1992:1;seealso1984:14–15)Searlecallshisview“biologicalnaturalism,”andsays,“Mentaleventsandprocessesareasmuchpartofourbiologicalnaturalhistoryasdigestion,mitosis,meiosis,orenzymesecretion....[Further,]intentionalstatesstandincausalrelationstotheneu-rophysiological(aswellas,ofcourse,standingincausalrelationstootherIntentionalstates),and...thatIntentionalstatesarerealizedintheneurophysiologyofthebrain”(Searle1983:1,15;seealsop.90and1984:21–2).Heiswillingtoadmitthatmentalstatescausallysuperveneonbrainstates,butheisuneasyabouttheadmissionbecausesuperveniencewasoriginallyalogi-calrelationandheworriesthathiscausalformmaybeconflatedwiththelogicalone(Searle1992:124–6).Theoriesofsuperveniencearetypicallyreductionistic.IsSearle’s?Yes,butinonlyoneofthevarioussensesofreduction.Theoreticalreductionism,forexample,triestoexpressorreduceallthelawsofonetheoryT1tothoseofanotherT2.Oncethisisaccomplished,itisstandardtoclaimthattheenti-tiesreferredtoinT1donotreallyexistandthatonlytheentitiesinT2do.Sotheoreti-calreductions,likemostreductions,ultimatelyaimatsomeontologicalreduction.ButSearle’sdoesnot.Rather,heexplainsthatheespousesaformofcausalreductionismbecauseheholdsthatconsciousnessiscausallyreducibletothebrainprocesses(Searle1992:116).Thiscausalreduction,unlikemost,doesnothaveacorrespondingontol-ogicalreductionbecauseoftheirreduciblefirst-personontologyofconsciousness.Whileaperson’spainisnodoubtcausedbyacertainpatternofneuronalfiringsinthethalamusandotherpartsofthebrain,acompletespecificationofthispatternwouldstillleaveout“essentialfeaturesofthepain”(Searle1992:117),namely,howthepainfeels.Whenheatwasreducedtomeankineticenergyandcolorswerereducedtotherefractionofphotons,thesetwoentitieswereredefinedinordertoeliminatethesub-jectiveelementintheperceptionofthem.Butasimilarkindofredefinitionofcon-sciousnessisnotpossiblebecausethereisnothingtoconsciousnessexceptthesubjectivity(Searle1992:121–3).438\nJOHNR.SEARLESearle’spositionisdesignedtoavoidbothmaterialismanddualism.Unlikeamateri-alist,heaffirmstheexistenceofirreduciblementalphenomenaanddeniesthattheyareidenticalwithbrainstates.Unlikeadualist,heassertsthatmentalpropertiesarephysicalproperties.SincemanycriticsthinkthatSearleisapropertydualistmalgrélui,somethingshouldbesaidaboutthis.Apropertydualistholds(1)thatthereareonlytwokindsofproperties,and(2)thatallpropertiesarementalorphysical.Searledeniesboth(1)and(2).Eitherthereisonlyonekindofproperties,physicalones,ortherearemanykinds.And,iftherearemanykinds,noneofthemarementalorphysicalinthesenseintendedin(2).Searlethinksthatdualismandmaterialismsharetheassump-tionsthatgiverisetothe“mind–bodyproblem.”Butthereisnosuchproblemanymorethanthatthereisa“stomach-digestionproblem”(Searle1992:15;seealso1984:14).Theseriousissueinthephilosophyofmind,inSearle’sopinion,isthenatureandstructureofconsciousness.HedealtwithitsnatureinTheRediscoveryoftheMind(1992),andwithitslogicalstructureinIntentionality(1983).Searlecametoappreci-atehowthestructuresofintentionalityhavetobeunderstoodaswaysthatcon-sciousnessexercisesitselfafterhewroteIntentionality.Soitissensibleforustobeginwithhisviewsaboutconsciousness.Concerningthenatureofconsciousness,Searlehasanegativeandapositiveproject.Thenegativeprojectistoshowthatcurrentworkonthenatureofthemental,espe-ciallyincognitivescience,isconceptuallyconfused.Theconfusiondoesnotjustinter-ferewithconstructingacorrecttheory,butalsomotivatesmisconceivedresearchstrategiesandcoversoverthefactthatconsciousnessisthecentralphenomenonofthemind.Hisrefutationcentersonthe“Chineseroom”thoughtexperiment.ApersonwithnoknowledgeofChineseislockedinaroom.HeisgivenabatchofChinesewriting(input);hehasarulebook,writteninhisownlanguage,thatcorrelatestheinputwithotherChinesewritingandexplainshowtoselectorproduceChinesewriting(output)thatundercertainconditionswouldbeanappropriatefollow-uptotheinput.Toanobserveroutsidetheroomandignorantoftheprocessincamera,itmayappearthatthepersonormechanisminsidetheroomknowsChinese.Ofcourse,neitherthatpersonnoranyconjunctionofthatpersonandanythingelserelevanttotheoutputtingknowsChinese.HemaynotevenknowthatheisdealingwithChineseoranylanguageatall.Heisperformingtotallyformaloperations,thatis,thepersontreatsthewritingsolelyinvirtueofphysicalshape(orotherphysicalproperties).Thewritingmeansnothingtotheperson.Sincethepersonintheroomisdoingjustwhatacomputerdoes,Searleconcludesthatcomputersdonothavemindsnorcognitivestates(Searle1992:45).Searlehasvariouslydescribedwhatthepersonintheroom(andsimilarlyanyexistingcomputer)lacks:asemantics,oranappropriatecausalconnectionwiththeinputandoutput,orunderstanding(seeSearle1992:69).However,whatseemstobemostbasicforSearleisthefactthatneitherthemanintheroomnoranyexistingcomputerhasanyunderstandingofChinese.Thisiscrucialbecauseunderstandingrequiresconsciousnessandconsciousnessemploysauniquekindofcausation.Thistype,explainedinIntentionality,maybedescribedtentativelyhereasintentionalcausation(seeSearle1992:107–9).TheoperationofintentionalcausationisimportantbecausesomecognitivescientistshavetriedtocircumventtheconsequenceofSearle’sscenariobyconstructinga“room”orrobotthatcanperformcomplex439\nA.P.MARTINICHoperations.Supposearobotisoutfittedwithtelevisioncameras,connectedtoleversandpulleys,poweredbyamotorthatdrivestherobotonwheelstovariouslocationstoarrangeandrearrangeboxesorotherobjects.Thesescientiststhinkthatbyincreasingthecomplexityoftheinternalmechanismsandbyhavingtherobotbeaffectedbyandtoaffectitsenvironment,theyhaveunderminedtheChineseroomargument(Searle1984:34–5,40–1).Thescientistsaremistakeninthinkingthatcomplexityorgenericcausalityistheissue.Rather,itisthenatureofthecontrollingmechanism(consciousness)andthetypeofcausation(intentional)thatarecrucialforSearle.InRediscoveryoftheMind,hearguesthatthefactthatcognitivescientiststhinkthatthebrainisadigitalcomputerandthatadigitalcomputercanbeconstructedoutofaninfinityofmaterials,includ-ingpaperormagnetictapes,cogsandlevers,“ahydraulicsystemthroughwhichwaterflows...anelaboratesystemofcatsandmiceandcheese...[and]pigeonstrainedtopeckasaTuringmachine”–alloftheseareactualexamplesfromtheorists–provesthatthetheoryisbankruptandirrelevanttothenatureofthebrainandmind(Searle1992:206).Ashesays,“wewantedtoknowhowthebrainworks,”anditisnoanswertosaythatthebrainworkslikeanydigitalcomputermightwork,where“digitalcom-puter”isdefinedsobroadlythat“stomach,liver,heart,solarsystem,andthestateofKansas”countasdigitalcomputers(1992:208,and1984:36).Anotherwayofgettingatthesamegeneralpointistosaythatessentialtotheopera-tionofthebrainis“intrinsicintentionality.”Whilethisconceptwillbeexplicatedlater,itcanbeunderstoodprovisionallyasmeaningthatthebraincausesstatesthatareinherentlyrepresentational.Incontrast,nocurrentcomputerandnocomputertobedesignedintheforseeablefuturehasintrinsicintentionality.Forthesakeofsimplicity,let’ssaythattheoperationofeverycomputercanbeunderstoodasinvolvingthepro-ductionofsequencesof“1”sand“0”s.Itisalsocommontothinkof“1”and“0”assyntactic(purelyformal)entities,andoneandzeroassemanticentities,thatis,themeaningsofthesyntacticentities.Evenwhen“1”and“0”arenotunderstoodtobesemanticentities,itiscommontothinkthatthecomputeritselftakesthosemarkstorefertooneandzero.Searleobjectstothisview.Computersunderstandnothingbecausetheydonotrep-resentanythingtothemselves.Rather,thecomputerdesignersandthecomputerusersinterpretthemarks,“1”sand“0”s,insomewaythatisusefultothem,somecompu-tationalway.Buttheinterpretationandtheunderstandingareinthedesignersandusersonlyandnotatallinthecomputer.Thishastheconsequencethatthecomputer“1”sand“0”sarenotevenintrinsicallysyntacticentities,sincesyntaxandsemanticsarecorrelativeideas(Searle1992:209–10).Syntaxisnotinphysics.Itisnogoodtourgethatcomputerscanordohavephysicalstatesotherthan“1”sand“0”s.Searle’spointstillholds:notionssuchascomputation,algorithm,andprogramdonotnameintrinsicphysicalfeaturesof[computer]systems.Computationalstatesarenotdiscoveredwithinthephysics,theyareassignedtothephysics.(Searle1992:210)Searle’sscenariooftheChineseroomispowerfulbecausetheonlyplausiblelocusoftheintentionalcausationrequiredforunderstandingChineseisintheconsciousness440\nJOHNR.SEARLEofthepersonintheroom,anditisobviousthatthatpersonisnotconsciousofanddoesnotknowChinese.OneconsequenceofSearle’scriticismofstandardcognitivescienceisthatanentireleveloftheorydisappears;theideaof“anunconsciousmentalprocess”andhencetheideaoftheirprincipleslosesalljustification(Searle1992:239–40).Therearethenonlytwolegitimateobjectsofcognitivestudies,thebrainandconsciousness.Thestudyofthebrainbelongstoneurophysiology;thestudyofconsciousnessbelongstophiloso-phyandastillmisguidedcognitivescience.ThisbringsustoSearle’spositiveprojectaboutthemind,namely,togiveanon-reductionisticaccountofconsciousnessthatisconsonantwithneuroscience.WhatisontologicallyessentialtoconsciousnessaccordingtoSearleissubjectivity:“thementalisessentiallyafirst-personontology”(Searle1992:70;seealsopp.77,94–5).Itisthisfactthatmakesthe“first-person”understandingofitfundamental.Third-personaccesstoconsciousnessviaobservationofbehaviorisinherentlyin-completeinsofarasithasnoaccesstotheexperienceofconsciousness.Itisalsoderivativebecausewhileconsciousnessisessentialtocausingbehavior,consciousnessislogicallyindependentofbehavior(1992:69).Therecanbeconsciousnesswithnobehavior.Searleidentifiesanumberoffeaturesofhumanconsciousness.Weshalldividethemintosevencategories.(1)Consciousnessmanifestsitself“inastrictlylimitednumberofmodalities”(Searle1992:128).Inadditiontothetraditionalfive–seeing,touching,tasting,smelling,andhearing–thereisthesenseofbalance,bodilysensations,whichincludes“proprioception,”thatis,thefeelingofhowone’sbodyandpartsofone’sbodyisoriented,andthestreamofthought.(2)Consciousnessisunifiedwithrespecttobothtemporalcontinuityofimpres-sionsandthespatialunityofvariousimpressions.Yesterday,today,andtomorrowareallpartofthesametemporalsystem;here,there,andtheotherplacearepartofthesamespatialsystem;andthetwoformaspatiotemporalsystem.(3)Consciousnessisanecessaryconditionforintentionalityandtypicallyisintentional,thatis,directedatobjects.Allintentionalityisaspectual.Thisiseasiesttoseeinvisualperception;thingsarealwaysperceivedfromapointofviewandasbeingthingsofacertainkind(1992:133).ArelatedaspectofintentionalityisthefactthatconsciousnesshasafocusandthisinturngivesrisetothedifferencebetweenfigureandgroundinGestaltpsychology.Also,attentionisdirectedtosomecontentsofconsciousnessmorethantoothers.Thedriverofacarmaybepayingmoreattentiontohisvacationplansthantohisdriving;yetbotharesimultaneouslyconscious.(4)Consciousnesshasa“subjectivefeeling.”Thereisadifferencebetweenhumanconsciousnessandwhatitisliketobeabatoraporpoise.(5)Althoughitisnotaspecialfeeling,thereisanairoffamiliarityabouttheobjectsthatapersonisconsciousof.Eventheunfamiliarisfamiliarinthesenseintendedhere.Apersonwalksintoanofficebuildingandexpectsittohaveelevators;theelevatorsarefoundinafairlypredictablelocation;theyareeasytooperate;andthedooropenstoafloor,which,thoughneverseenbefore,hasenoughfamiliarityaboutitthattheappropriateroomisdiscovered.Thereisasenseinwhichpeoplehaveknowl-441\nA.P.MARTINICHedgeoftheworldinawaythatismoregeneralthananyparticularbitofknowledgeaboutit.Theworldisnotstrangeandmysterious.Thisfactishighlightedbysurrealistartistswiththeirmeltingwatchesandever-ascending-and-descendingstaircases.Thereisanotherkindoffamiliaritywiththeworld:peopleknowgenerallywheretheyareandwhattimeitis,inrelationtomanyotherplacesandtimes.Searlecallsthisgeneralspatialandtemporalfamiliaritywiththeworld“situatedness”(1992:141)(cf.CHOMSKY).Relatedto(5)is(6).(6)ThisiswhatSearlecalls“overflow,”thefeaturethathassomespecificpercep-tionorbeliefconnecttootherbeliefsseeminglywithoutendinsomeelaborate,notfullyarticulatableweb:theseeastTexastreesarepines,likethepinesofCalifornia,butnotexactly;theyflourishinwetareasnotquitemarsh,etc.PerhapsthisfeatureiscloselyrelatedtoSearle’sconceptsoftheNetworkandtheBackground.(7)Clearlymanystatesofconsciousnessaresuffusedwithamood(elation,depres-sion,cheer)eventhoughamood“neverconstitutesthewholecontentofaconsciousstate”(1992:140).Construedbroadlyenough,everystateofconsciousnesshassomemoodorother.Formostpeople,thereisapermanentlowlevelofpleasureconnectedwithconsciousness,andforsome,thereisapermanentlowlevelofdispleasure.(Seealso1998:73–80.)Atheoryofconsciousnesswouldnotbecompletewithoutatheoryoftheuncon-scious.Searle’smainthesisisthat“everyunconsciousintentionalstateisatleastpoten-tiallyconscious”(1992:132;seealsop.152,and1984:43–4).Inholdingthis,heisopposingtheidea,standardamongcognitivescientists,ofadeepunconsciousthatcanneverbemadeconscious.Theypurporttosubtractconsciousnessfromconsciousmentalstatesandtofindthe“computationalmind”astheremainder.Thisisanattempttoeffectaseparationofintentionalityfromconsciousness,afterwhichamistakenaccountofintentionalityascomputationispresented.Searle’sproofofhisthesisisroughlythis:onlymentalstatesareintrinsicallyintentionalinthesensedescribedabove.Unconsciousmentalstatesareintrinsicallyintentional.Allintrinsicallyinten-tionalstatesareaspectual.Unconsciousmentalstates,asunconscious,existonlyasneurophysiologicalevents.Therefore,inordertobementalstatesatall,theuncon-sciousonesmustbecapableofbeingbroughtintoconsciousnessbytheunderlyingneurophysiologicalevents(1992:155–9,172).Searle’stheoryoftheunconsciouspermitsaneatexplanationfortheexistenceofunconsciouspains.Forexample,peoplewithchronicbackpainoftenawakefromsleepwithpain.Itisplausiblethatthecharacteristicneurophysiologicaleventsthatunder-lietheconsciouspainarepresentduringsleep,absentwhateveradditionaleventsthatbringthepaintoconsciousness.Whentheseadditionaleventsgettriggered,thepainbecomesconsciousandthesleeperawakes(1992:164–7).Further,consciousnessistransparenttoitself.WhenDavidHumelookedintohimself,inadditiontonotfindingthe“mind”hewaslookingfor,hedidnotfindcon-sciousnesseitherbecauseitisalwaysdirectedatotherthingsandcannotbeanobjectofdirectstudy(Searle1992:97).Thatiswhynineteenth-century“introspectionism”inpsychologywasdoomedtobeafailureandwhyitisnaturaltothinkthatconsciousnessisnotpartofthephysicalworld.Thetransparencyofconsciousnessalsoexplainswhyitmakesnogoodsensetosaythateachpersonhas“privilegedaccess”442\nJOHNR.SEARLEtoherown.Thatfigureofspeechrequiresonetothinkthatapersonentersaspaceseparatefromherselfinwhichshealonecanstand.Butthereisnothinginconscious-nessanalogoustothatspace(1992:97,104–5,170–1).IntentionalityThecashvalueofSearle’stheoryofthenatureofconsciousnessdependsuponhisabilitytoexplicatethestructuresofconsciousness.Themostsalientstructureshaveintentionality;thatis,beingdirectedatsomethinginthesenseofrepresentingsome-thing(Searle1983:3,11–12).Whilenotallconsciousstatesareintentional–undi-rectedanxietyandnervousnessarenotdirectedatanything,nor,Ithink,ispurejoy–themostimportantformsare.Inordertoemphasizethatintendingtodosomethingisonlyonekindofintentionality,Searlecapitalizesthelatterterm,butIshallnotfollowthatconvention.Becausethoughtsandfeelings(prominentformsofintentionality)areexpressedinlanguage,Searlethinksthatmuchofthestructureofintentionalitycanbereadofffromthestructureofspeechacts.So,justasthebasicformofanillocutionaryactisF(p),thebasicformofanintentionalStateisS(r),apsychologicalmode,suchasbeliev-ing,hoping,orloving,andarepresentationalcontent.Likespeechacts,thecontentisoftenaproposition,asinbelievingthatGeorgeWashingtonwasthefirstpresident.ButitcanalsobeanobjectaswhenJoneslovesSmith.Inthislattercase,adistinctionneedstobedrawnbetweenJones’srepresentationalcontent,andtherealpersonSmith,whichistheintentionalobjectofthatcontent.Fourfeaturessharedbyillocutionaryactsandintentionalphenomenaareespeciallyrevealing.(1)Analogoustotheforceandpropositionalcontentofspeechacts,intentionalstateshaveapsychologicalmodeandacontent.“Istatethatyoubrokethevase”and“Iorderyoutowashthisfloor”arestructurallyidenticalwith“Ibelievethatyoubrokethevase”and“Iwantyoutowashthisfloor,”respectively.(2)Directionoffit:likestatements,beliefshaveaword-to-worlddirectionoffit,andlikeorders,desiresandintentionshaveaworld-to-worddirectionoffit.Likeapologies,sorrowhasnodirectionoffitbutpresupposessomethingabouttheworld,aninjurycausedbytheagent.(3)Forillocutionaryactsthathaveasinceritycondition,theexpressedproposi-tionalcontentpurportstobethecontentofanintentionalstateofthespeaker.Forexample,statingthatsnowiswhiteisrepresentingthatonebelievesthatsnowiswhite.Andpromisingtogotothepartyisrepresentingthatoneintendstogototheparty.Further,therelevantintentionalstateisthesincerityconditionoftheillocutionaryact.Boththesincerityconditionandtheintentionalstateofassertingthatsnowiswhiteisthebeliefthatsnowiswhite.(4)Bothillocutionaryactsandintentionalstateshaveconditionsofsatisfaction.Astatementissatisfiedifandonlyifitistrue;anorderissatisfiedifandonlyifitisobeyed;apromiseissatisfiedifandonlyifitiskept.Correspondingly,abeliefissatisfiedifandonlyifitcorrespondstothewaythingsare;adesireissatisfiedifand443\nA.P.MARTINICHonlyifitisfulfilled(idiomatically,desiresaresaidtobe“satisfied”),andanintentionissatisfiedifandonlyifitiscarriedout.Sofar,nothinghasbeensaidaboutatraditionalpartoftheconceptofintentional-ity,namely,thatwhatisintentionalisalwaysdirectedatanobject.WhileSearleholdsthatsomestateshaveintentionalobjects,hedeniesthatallofthemdo,becausehedoesnotholdthestandardviewaboutwhatanintentionalobjectis.Forhim,itisalwaysanexistentobject.TheintentionalobjectofbelievingthatJonesishappyisJones,theverypersonherself.Theintentionalobjectseemsanalogoustothereferentofareference.Bothreferentsandintentionalobjectsareidiomaticallyspokenofasbeing“in”mind,althoughneither,ofcourse,couldliterallybespatiallyinthemind.Andthereferentisnomoreapartofthepropositionexpressedthantheintentionalobjectispartoftherepresentationalcontentofamentalstateorevent.Giventheanalogybetweeninten-tionalobjectsandreferents,onemightthinkthatitformsthebasisforafifthsharedfeature.ButSearledoesnotgothisroute;andingeneralhementionsintentionalobjectsonlytoputthemaside.Inadditiontothesefourparallelsbetweenillocutionaryactsandintentionalstates,thereisanimportantdisanalogy.Intentionalstatesareintrinsicallyinten-tional,whileillocutionaryactsarenot.Illocutionaryactshavetoberealizedinutterances(sounds,marks,orgestures)thatareintrinsicallyphysicalandonlyderivetheirintentionalityfromthefactthathumansintendthemtorepresentstatesthatareintrinsicallyintentional(Searle1983:27;1992:78–80).Inotherwords,whenaspeakeruttersasentence,he(i)intendsthatsentencetohavecondi-tionsofsatisfactionthataregivenby(ii)theintentionalstateexpressed.TheparentheticalromannumeralsindicatewhatSearlecallsa“doublelevelofIntentionalityinthespeechact,”namely,(i)theintentionofgivingtheutteranceconditionsofsatisfactionthatare(ii)theintentionalstates.Theconditionofsatis-factionof“snowiswhite”isthatsnowiswhite.Toaskforthemeaningofasentenceistoaskfortheintentionalstatethatistheconditionofitssatisfaction(1983:28,164).Whatdeterminesthatsomeintentionalstateoreventisonewithsomeparticularcontentratherthansomeother?InIntentionality,Searle’sanswerwasthatanessen-tialelementisitsplacewithinasystemofintentionalstates.Forexample,theinten-tionalstateofrunningforthepresidencyoftheUnitedStatesispossibleonlyaspartofaNetworkofotherintentionalstatesinvolvingsuchbeliefsasthatthereisaUnitedStates,constitutedintheeighteenthcentury,separateandindependentofeveryothercountry,boundedinlargepartbyCanadaonthenorthandMexicointhesouth,andsoon.SoSearleendorsesakindofholism.Further,theNetworkofintentionalstatesispossibleonlyagainstabackgroundofnonrepresentational“mentalcapacities”(Searle1983:21).Thebackgroundisgener-alizedknow-how.Itisaworldlycompetencethatmakespossibleparticularkindsofknow-how,suchasknowinghowtoopenadoororknowinghowtowritealetter.But,becauseparticularknow-howscanbeexplicatedasessentiallyinvolvingpropositionalcontents,suchasthatsomethingisthecaseorthatsomethingistobedone,theseinstancesofknow-howarenotpartofthebackgrounditself.Openingadoorinvolvesrepresentations,but“theabilitytorecognizethedoorandtheabilitytoopenitarenotthemselvesfurtherrepresentations”(1983:143).Itisdifficulttobeexplicitandprecise444\nJOHNR.SEARLEaboutthebackgroundjustbecausebackgroundcapacitiesarenotpropositionalandexplicitwhileexplanationsare.SearlemaybediscussingthesamephenomenonasMartinHeideggerwhenhetalkedabouthumanbeingsasbeings-in-the-worldwiththingsbeingready-at-handandasLudwigWittgensteinwhenhetalkedaboutcertainthingsbeingsobasicthatthey“standfast”andhenceareobjectsneitherofcertaintynordoubt.Inthisspirit,Searleclaimsthatmetaphysicalrealismisnota“hypothesis,belief,orphilosophicalthesis...buttheprecondition”ofanyofthesethings(1983:158–9).Consciousnessisnotahypothesiseither(1992:79;seealso1995:178,195).Searle’sviewsabouttheNetworkandtheBackgroundchangedinTheRediscoveryoftheMindalthoughthecoreremained.“Intentionalstatesdonotfunctionautonomously,”hesays.TheirconditionsofsatisfactiondependuponasetofBackgroundcapacities.Someofthesecapacitiesarecapableofgeneratingotherconsciousstates.Ofcoursethesenewlygeneratedconsciousstatesarejustlikethefirstonesmentioned;theydonotfunctionautonomouslyandtheydependuponasetofBackgroundcapacities.TheBackgroundroleissoimportantthatthe“sametypeofintentionalcontentcandeterminedifferentconditionsofsatisfaction”whentheBackgroundcapacitiesaresignificantlydifferent(1992:190).Consciousnessthenconsistsofanoccurrentrepresentationalcontent(item(a)),whichdependsuponaneurophysiologicalbasewith“thecapacitytogeneratealotofotherconsciousthoughts.”Thiscapacity(item(b))itselfispartofaneurophysiologicalsystemthatisnonrepresentationalbutnecessaryforboththerepresentationalcontentandtheindividualconsciousness.Item(a)replacesSearle’searlierideaofaNetwork;item(b)replacestheearlierideaoftheBackground(1992:190–1;seealso1995:181–2,184–9).Havinglaidoutthegeneralstructureofintentionality,wecanlookatSearle’streat-mentofthe“primaryformsofIntentionality,perceptionandaction.”Concerningper-ception,heisarealist;peopleseecars,tables,trees,andsoon.Theydonotseetheirperceptualexperiences;theyhavethem(1983:36,38).Experiences,likebeliefsanddesires,areintentional,beingdirectedatobjectsandhavingconditionsofsatisfaction.Theexperienceofperceivingayellowcarisdirectedatthecarandissatisfiedonlyifthereisayellowcarwhereitisperceivedtobe.Whilethecarisyellowandcar-shaped,theexperienceisneither.Further,thecaritselfhastocausethevisualexperience.Howshouldthisinformationberepresented?Searleproposesthefollowing:“Ihaveavisualexperience(thatthereisayellowstationwagonthereandthatthereisayellowstationwagonthatiscausingthisvisualexperience)”(1983:48).Thecrucialpointaboutthisanalysisoftheconditionsofsatisfactionofvisualper-ceptionandrelatedphenomenaistheelementofself-referentiality:thevisualexperi-encethatistobesatisfiedismentionedintheconditionsofsatisfaction.Thisdoesnotmeanthattheself-referentialcausalrelationisitselfseen(Searle1983:48–9).Searlehassomesympathywiththesuggestionthataclearerwaytorepresenttheconditionsofsatisfactionisthis:VEc(thatthereisayellowcar),wherethesubscript“c”indicatesthecausalself-referentiality.Thevisualexperienceisactualseeingonlyiftheappropriateobjectinthenon-mentalworld,inthiscase,therightyellowstationwagon,iscausingtheexperience.ThisallowsSearletosaythatonhisaccount445\nA.P.MARTINICHperceptionisanIntentionalandcausaltransactionbetweenmindandtheworld.Thedirectionoffitismind-to-world,thedirectionofcausationisworld-to-mind;andtheyarenotindependent,forfitisachievedonlyifthefitiscausedbytheothertermoftherela-tionoffitting,namely,thestateofaffairsperceived.(Searle1983:49;seealsopp.61–2)DoesSearle’saccountsolveafamiliarphilosophicalpuzzle?Supposethattwoiden-ticaltwinshavetype-identicalvisualexperienceswhilelookingattwodifferentbuttype-identicalstationwagons.Whatmakesonetwin’sperceptionaperceptionofcarAandtheothertwin’sperceptionaperceptionofcarB?Searlesaysthattheremustbesomethingintherepresentationalcontentitselfthatspecifiesthepropercarasacon-ditionofsatisfaction.Thespecificationisachievedbecause“eachexperienceisself-referential”(Searle1983:50).Thevisualexperienceofonetwinincludesasacondi-tionofsatisfactionthefactthatthatveryexperienceisbeingcausedbyayellowstationwagon.Inotherwords,perceptionsasamodeofconsciousnessareessentiallyfirst-personphenomena.Incontrast,acausaltheoryofperceptionaccordingtoSearletakesathird-personperspective:carAcausesoneperceptionandcarBcausestheother.Thetheoryisinadequate,however,asSearleobserves,becauseithasnoaccountofhowtheperceiver’sintentionalitymatterstotheperception(1983:64).WhileSearleisrightinhiscriticism,hedoesnotseemtoconcedeasmuchtothecausaltheoristasheshould.Thefactthatanyperceiverhasherownindividualvisualexperiencewithaself-referentialcontentdependsonthefactthatsomeindividualnon-intentionalobjectiscausingthatuniqueexperience.Thatis,theobjectofperceptionhelpsindividuatethevisualexperience.Searle’soppositiontothecausaltheoryinclineshimtoassertthathumanbeingsare“brainsinvats,”thevatsconsistingoftheirskulls(Searle1983:230).Althoughhemayhaveprimarilyintendedthethenfashionablephrase“brainsinvats”metaphorically,itishardtoseehowthispositionisconsistentwithcentralfeaturesofhistheory.Abraininavatdoesnotreceiveitsintentionalcontentsintherightkindofway;itspriorintentionsdonotcauseactionsbecausethebrainisnothookeduptotheappropriatebiologicalorgans,andsoabraininavatdoesnotperformanynon-mentalphysicalaction.Adifferent,and,Ithink,ineffectiveobjectiontoSearle’stheoryofperceptionisthatitleadstoafamiliarformofskepticism.(1)Sincethecarcausesthevisualexperience,thevisualexperienceisthebasisforbelievingthatthecarisseen;and(2)sincetheper-ceiverinfersfromtheperceptualexperiencethatthecarexists,itmightbethecasethattheexperienceexistswithoutthecarexisting.Searlerejectsbothpartsofthislineofreasoning.Theperceptualexperienceisnotevidenceforthebeliefthatthecarisseen;andtheperceiverdoesnotinferthatthecarexists.Onesimplyseesthecar:“Theknowl-edgethatthecarcausedmyvisualexperiencederivesfromtheknowledgethatIseethecar,andnotconversely....[W]eperceiveonlyonethingandinsodoinghaveaper-ceptualexperience”(1983:73,74).Let’snowconsiderSearle’stheoryofaction.Intentionalactionsareconditionsofsatisfactionofintentionstoact;butnoteveryintentionalactionisprecededbyapriorintentiontoact.Peopleoftenactwithoutplanning(1983:84–5,107).Thatis,“priorintentions”shouldnotbeconfusedwith“intentionsinaction.”Priorintentionsarelikeplans;theytemporallyprecedeanaction.Theymaybeexpressedbysentence-formslike,“IintendtoA”or“IwillA.”Incontrast,anintentioninactionispartofthewarpand446\nJOHNR.SEARLEwoofofanaction.Allactionsareintentional.So-calledunintentionalactions,suchasOedipusmarryinghismother,isrelatedtosomethingintentional,namely,OedipusmarryingJocasta.Breathing,snoring,andsneezingarebodilymovementsbutnotactions,becausetheyarenotintentional.Likeperception(andmemory),both“priorintentionsandintentionsinactionarecausallyself-referential”(Searle1983:85).Intrinsictotheirconditionsofsatisfactionisacausalrelationshipbetweentheintentionalstateandthethingdone.Toraiseone’sarmistohaveone’sintentionofraisingitcauseittogoup(p.86).Searlesaysthat“thepriorintentioncausestheintentioninaction”(p.94).Thisistruewheneverthereisapriorintentiontodosomethingimmediately;suchintentionscauseatleastattemptsorefforts.Totrytodosomethingistodosomething.However,priorintentionstodosomeactAarenotsufficienttocausethepersontodoA.Thereisagapbetweentheinten-tionandeventhedecisiontoactandtheactingitself.Perceptionandactionarenicelycontrastedbytwofacts:(1)Whileperceptionhasamind-to-worlddirectionoffit,actionhasaworld-to-minddirectionoffit.(2)Whileanobjectcausesaperceptualexperience,anexperienceofactingcausesanevent.Butper-ceptionandactionarethesameinsofarasbothpresent,ratherthanrepresent,theirexperiences,incontrastwith,say,memoryorimagination,whichdore-presentthings.Searle’sresearchesintothelogicalstructureofintentionalitycausedhimtoreworktheanalysisofmeaningthathegaveinSpeechActs,wherehismaingoalhadbeentogiveananalysisthatwouldnotincludeintendingtocauseaneffect,asH.P.Gricehadinhis.InIntentionality,Searleclaimsthattheessenceofmeaningisrepresenting;ineffect,itisimposingconditionsofsatisfactiononsomethingthatisnotinherentlyrep-resentational.Forexample,inraisinghisarm,apersonmeansthatthearmyisretreat-ingifandonlyifhisintentiontoraisehisarmcauseshisarmtogoupandhisarmraisinghasasaconditionofsatisfactionthatthearmyisretreating.Obviously,thefactthatthearmraisinghasaconditionofsatisfactionisduetothemindimposingthatconditiononit.Allmeaningfulgesturesorutteranceshaveanintentionalitythatisderivedfrommentalstatesthatareinherentlyintentional.Althoughrepresentingistheheartofmeaning,accordingtoSearle,sincemeaningisstandardlyusedtocommunicate,acompleteaccounthastosayhowcommunica-tionoccurs.Whatneedstobeaddedtomeaningasrepresentationisanintentiontogettheaudiencetorecognizewhattheconditionofsatisfactionoftheperson’sgestureorutteranceisandtohavetheaudiencerecognizeitinvirtueofthegestureorutteranceitself(Searle1983:168).Onemeritofthisanalysisisthatitgivesapreciseandinformativeanswertothequestion“Whatisthedifferencebetweensayingsomethingandmeaningitversussayingsomethingandnotmeaningit?”Theansweristhattheformerhasconditionsofsatisfactionandthelatterdoesnot.When“Esregnet”issaidandmeant,aconditionofsatisfactionisthatitisraining,whereaswhenitissaidmerelyinthecourseofprac-ticingGerman,theweatherisirrelevant.SocialrealityAttheendofTheRediscoveryoftheMind,Searleofferssomeguidelinesfortheproperstudyofmind;thelastofthese,thelastsentenceofthebook,is:“weneedtorediscover447\nA.P.MARTINICHthesocialcharacterofthemind”(1992:248).Thiswasanadvertisementforhisnextbook,TheConstructionofSocialReality(1995).Thetitleisadirectchallengetoadia-metricallyopposedidea,thesocialconstructionofreality.Thispopularalternative,whichmaintainsthattheworldisaconstructofthehumanmind,hasnoappealforSearle,whosaysitisa“preposterous”viewthatrestson“anarrayofweakorevennonexistentarguments”(1995:160).Heisarealistabouttherealworld.Thenaturalworldconsistsoftwobasickindsofthings:non-mentalandmental.Thenon-mentalthingsatarelativelyfundamentallevelareatomsandatthemostfundamentallevelarespace-timepoints,accordingtocurrentscience.Thementalthingsevolvedoutofthenon-mentalafterbillionsofyears.Thenon-mentalthingsareontologicallyobjective;theyexistindependentlyofminds.Thementalthingsareontologicallysubjective;theydependfortheirexistenceonminds.Forexample,apainisapartofthenaturalworld,butsubjective.Thesocialworld,whichincludesmoney,government,andmarriage,arisesoutofmentalreality,largelybecausethemindcanrepresentthingsasotherthings.Totakethemostpowerfulinstrumentofrepresentation,languageusessounds,marks,orges-turestorepresentotherthings;andthisispossiblebecausepeoplearewillingandabletotakethemasrepresentingotherthings.Inshort,thesocialworldisobserver-anduser-dependent.Moneyismoneybecausepeopletakepiecesofpaperormetaltohaveexchangevalue.Citizensarecitizensbecausepeopletreatthemashavingcertainrightsandresponsibilities.Incontrast,thenaturalworldisobserver-independent.Aroseisaroseisarose,whetheranyoneviewsitornot.Giventhisontologyasbackground,Searleusesthreeelementstoexplainsocialreality.Firstistheideathatpeoplecanimposefunctionsonobjectsthatdonothavethatfunctionbeforehand.Never“intrinsic”tothethingitself,functionsarealwaysobserver-dependentandintroduceanormativedimension(Searle1995:19).Originally,someobjectbecameahammerwhenitwasusedtohammerandnotbefore;and,inbecomingahammer,itbecamepossibletojudgegoodandbadhammers,dependinguponhowwelltheyfunctioned.Thesecondelementiscollectiveintentionality.Alinemanblocksanopposingline-backeronlyaspartofhisteam’splay,andamusicianplaysfirstviolinonlyaspartoftheorchestra’ssymphony.Footballplayersplayasmuchinconcertasmusiciansdo(Searle1995:22).Collectiveintentionalityisnotreducibletoindividualintentionality.Indoingsomethingtogether,eachparticipanthasherownindividualintentions,butthesederivefromthecollectiveintentionalityofthegroup(pp.24–5).Thethirdelementisthedistinction,introducedbySearleinSpeechActs,betweenregulativeandconstitutiverules.Regulativerulesdirectorcontrolpre-existingbehav-ior,astherulesofetiquettecontrolhowpeopleshouldeat.Theytypicallytaketheformofimperatives:Useanapkin,notyoursleeve,towipeyourmouth;eatyourpeaswithafork,notyourknife.Constitutiverulescreatenewformsofaction.Therulesoffoot-ballcreatethegameoffootball.Moreimportantly,thebasicrulesofagovernment,for-mulatedinconstitutions,creategovernments.Whenonegovernmentfallsandanotherarises,anewconstitutionisformulated.Theseconstitutiverulesoftentaketheformofindicatives:“TheSupremeCourtisthehighestcourtofthejudicialbranchoftheUnitedStates.”Suchsentencesmayseemtobestatementsoffactbuttheyaremoreproperlyseenasdeclarations(Searle1995:55,74).Searlebelievesthatthedeepformofcon-448\nJOHNR.SEARLEstitutiverulesis“XcountsasY(incontextC),”forexample,“ApersonbornonAmericansoilcountsasanAmericancitizen”(p.28).InkeepingwiththetheoryofSpeechActs,onemightsuggestthatadeeperformis,“WedeclarethatXisY(incontextC)”(cf.1995:104–11).Thesuggestionistemperedbythefactthatsomanyinstitu-tionalactsevolveslowly,haltingly,andunreflectively.The“XcountsasY”formulacanbeiterated.SomethingthatonceoccurredasaYterm,forexample,“citizen,”canoccupytheplaceoftheXterminanotherformula.“Acitizencountsasthepresidentwhendulyelected,etc.”Roughly,themorecomplexthesociety,themorenumerousanditeratedthe“counts-as”formulae(Searle1995:80).Whenconstitutiverulesareenacted,institutionalfactsarecreated.Thesefactsareself-referential:somethingisafivedollarbillbecauseitisacceptedasafivedollarbill.Atadeeperlevel,alevelthatexploitsthethreeelementsoffunction,collectiveinten-tionality,andconstitutiverules,thelogicalstructureofinstitutionalfactsisthis:Wecollectivelyaccept(Sisenabled/required(SdoesA)).Applyingthisstructuretoafivedollarbill,callitX,weget,“Weaccept(S,thebearerofX,isenabled(SbuyswithXuptothevalueoffivedollars))”(Searle1995:97–8,104–12).Contrarytoappearances,thisanalysisisnotviciouslycircular,because“fivedollarbill,”and,moregenerally,“money,”occupyonlyacoupleofnodesin“awholenetworkofpractices,thepracticesofowning,buying,selling,earning,payingforser-vices,payingoffdebts,etc.”(p.52).Therangeofinstitutionalfactsisenormous,from“wivestowarfare,andfromcocktailpartiestoCongress”(p.96).They,andthepowersthatgowiththem,comeintoexistencewhenpeopleacceptthemasfactsandcontinuetoexistaslongaspeopleacceptthemasfacts.ConclusionTheclaimthatSearlecountsasaphilosopherofthefirstrankturnsonthispoint:heusesasmallnumberofinterlockingelementstoexplainabroadspectrumofrealityinanilluminatingway.Themostimportantelementsaretheideasofrepresentation,directionoffit,self-referentialintentionalcausation,andthedistinctionbetweencon-stitutiveandregulativerules.Thespectrumincludesthenatureoflanguage,mind,andthesocialworld,allpresentedwithinanaturalisticbutnotmaterialistworld-view.BibliographyWorksbySearle1969:SpeechActs,Cambridge:CambridgeUniversityPress.1979:ExpressionandMeaning,Cambridge:CambridgeUniversityPress.1983:Intentionality:AnEssayinthePhilosophyofMind,Cambridge:CambridgeUniversityPress.1984:Minds,BrainsandScience,Cambridge,MA:HarvardUniversityPress.1985(withVanderveken,Daniel):FoundationsofIllocutionaryLogic,Cambridge:CambridgeUniversityPress.449\nA.P.MARTINICH1992:TheRediscoveryoftheMind,Cambridge:MITPress.1995:TheConstructionofSocialReality,NewYork:FreePress.1998:Mind,LanguageandSociety,NewYork:BasicBooks.WorkbyotherauthorsLePore,ErnestandGulick,Robertvan(1991)JohnSearleandHisCritics,Oxford:BlackwellPublishers.450\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200137JerryFodor(1935–)GEORGESREYJerryFodoriswidelyregardedasthemostsignificantphilosopherofmindinrecenttimes.WithNoamChomskyatMITinthe1960shemountedadecisiveattackonthebehaviorismthatthendominatedpsychologyandmostphilosophyofmind,andhastriedtopresentinitsplaceanaturalisticandrealistaccountofmentalprocessesthatrendersthemamenabletoseriousscientificstudy.Indeed,heisoneofthefewphilosopherswhohascombinedphilosophicalandempiricalpsychologicalresearch,publishingworkinbothdomains,developingatleasttwotheoriesthathavebecomehighlyinfluentialinbothofthem:thecomputational/representationaltheoryofthoughtprocesses(seethesection“CRTT:Computation”)andthemodularitytheoryofperception(“ModularityandtheLimitsofCRTT”).Thesetheoriesare,however,bestappreciatedagainstthebackdropofanumberofotherthemesinFodor’swork,whichprovidethebestoverviewofhiswork,asfollows:(1)IntentionalRealism;(2)NomicExplanation;(3)TheProblemsofMind;(4)CRTT:Computation;(5)CRTT:Representation;(6)SolipsismandNarrowContent;(7)Nativism;(8)ModularityandtheLimitsofCRTT.IntentionalrealismFodor’sprimaryconcernistodefendthefamiliar“belief/desire,”or“propositionalattitude”psychologywithwhichthefolkroutinelyexplaineachother’sbehavior;forexample,someone’sheadingsouthisexplainedbytheirwantingwaterandthinkingthere’ssomethere.Whatwemightcall“scientificintentionalrealism”issimplyawayoftakingfolkpsychologyseriouslyasthebeginningofaseriousscientificpsychology.Notthatthefolkarealwaysrightaboutthemind.Indeed,manyoftheclaimsandeventhespecifictermstheyemploy:“learning,”“memory,”perhapseven“belief”and“desire,”maywellturnouttobetheoreticallyinadequate.Fodoronlypresumesthat,whatevertheparticularkindsofphenomenainvokedbyanultimatepsychology,theywilldisplaycertaincrucialproperties:1assomeorotherspeciesofpropositionalattitude,theywillbeintentional(being“about”things)andsemanticallyvaluable(capableofbeingtrueorfalse);2asphenomenainvolvedinrationalthought,theyneedtobelogicallystructured;451\nGEORGESREY3asultimatelyexplanatoryofaction,theyneedtobecausallyefficacious;and4giventhefundamentalroleofphysicsinourunderstandingoftheworld,theyhadbetterbecomposedoforidenticaltophysicalphenomena.Fodor’spostulationofstatesdisplayingthesepropertiesmayseemrathertruisticuntilonenotesthequitesubstantialeffortsofagoodnumberofphilosophersandpsychologistsinthetwentiethcenturytoargueotherwise:therearenotonlytheusualCartesiandualists,whoresistthematerialityofthemind(andhence(4)),butmorerecentlytherehavebeenthelogicalpositivists,behaviorists,Wittgenstein,Quine,Gibson,Davidson,Dennett,theChurchlands,manyconnectionists,allofwhomhavetriedinonewayoranothereithertodenythecausalrealityofthemindaltogether,ortorelegatementalisticwaysofdescribingtheworldtosomesortof“secondgradestatus,”insomewaylessobjectivethanphysics(denying(3)).MuchofFodor’sworkconsistsindefendingintentionalrealismagainsttheseattacks,notonlyastheyariseinphilosophy,butparticularlyinrelationtopsychology,wherewhatisatstakeareentireresearchprogramscommittedtobehaviorism,connectionism,neurophy-siology–or,ashewouldrecommendinstead,tointentionalrealism(1968a,1998b:chs1,8–10).Fodorisspecificallyconcernedwiththekindsofchallengestoamaterialistmental-ismthatwereraisedbyDescartes’sconcernwithrationality,andBrentano’swithinten-tionality,andsoisconsequentlymostconcernedtobearealistbothaboutattitudestates,suchasbeliefordesire,andabouttheircontents(thebeliefthatsnowiswhiteorGodisdead).Indeed,heisadamantthatthelatteraredecidedlynottobeexplainedawayasmattersof“interpretation”ormere“similarityrelations”(1998a:30–4).Heisalsoasrealistasanyoneabout“qualia”andconsciousness,buthasrelativelylittletosayaboutthem.HeisconvincedthatempiricalpsychologyatleastsinceFreudhasgivenusreasontosupposethatrationalandintentionalphenomenaneedn’tbeconscious(1968a,1998b),andthathecanthereforeaddresstheformidabledifficultiesoftheformerwithoutworryingaboutwhatheregardsasthecurrentlyimpenetrableprob-lemsposedbythelatter(1994:121,and1998b;butsee1972and1998b:73forsomestraysubstantiveremarks).ExplanationasnomicsubsumptionFodoralsotakesitforgrantedthatexplanationingeneralissubsumptionunderlaws,andthattherealmofthementalisnoexception(1994:3,1998a:7).Exceptpossiblyinultimatephysics,heassumestheseareceterisparibuslaws,orlawsthatarenot“strict”and“exceptionless,”butholdinabstractionfromvariousinterferencesor“com-pleters”thatafullertheoryoftheworldmightinclude(1991c).MuchofhisviewhereisofapiecewiththekindofidealizationthatChomskynotedistypicalofanyscience(seeCHOMSKY).ButwhereasChomskyislargelyconcernedwithonlyaspecificsetofidealizations–thosecapturinglinguistic“competence”inabstractionfromits“perfor-mance”–Fodorisconcernedwithwhatheregardsasthenecessaryvarietyofthemthatareenlistedintheexplanationofpsychologicalprocesses.Towardsaddressingthisconcern,Fodor(1968a)presentedoneofthefirstlengthydefensesoffunctionalism,accordingtowhichpsychologicalstatesare452\nJERRYFODORindividuatedbytheircausalrelations.Sincedifferentphysicalphenomenacansatisfytheserelations,functionalismnaturallygivesrisetocross-classificatorylayersofexplanation:onelevelofcausalrelationsmaybe“multiplyrealized,”orvariously“implemented,”bydifferentmechanismsatalowerlevel(1968a,1998b:ch.2).Specifically,theintentionallevelofacognitivepsychologymaybeimplementedatalowerlevelbyvariouscomputational/syntacticprocesses(§4),whichinturnmaybeimplementedbydifferentphysicalmechanisms–brainsinthecaseofpeople,transistorsinthecaseofmachines.(Note,though,thatFodorisneverthelessskepticalonecanprovideanydefinitionsofmentalstates,functionalistorotherwise;see§5.1).EspeciallyinFodor’swork,thisfunctionalistconceptionisresponsibleforaconsid-erable“autonomy”ofcognitivepsychologyfromdetailsofitsimplementation,analo-goustothewayacomputerprogramcanbespecifiedinabstractionfromtheelectronicdetailsofthecomputersthatrunit.Becausetheyaregenuinelaws,involving,forexample,“projectible”predicates,thekindstheymentionarenotreducibletomerefinitedisjunctionsofthekindsatthelowerlevels(1998b:ch.2),althoughFodorpre-sumesthatthey“supervene”onthem.ThedemandformindIfweliveinapurelyphysicaluniverse,however,itmightbewonderedwhatseriousexplanatoryrolementalphenomenahavetoplay.Whydoesn’tphysicsalonesuffice?Theshortansweris,ofcourse,thatpurelyphysicalprocessescancometoexhibitallmannerofspecialstructuresandorganizations–molecules,crystals,cells,livingorganisms,andsometimesminds–thatitisthebusinessofspecial“macro”-sciencestodescribe.Fodorisparticularlyimpressedbythesensitivityofhumanbeingstoindefinitelymanynon-local,non-physicalproperties:notonly,asChomskyhasemphasized,tohighlyabstractgrammaticalproperties,likebeingamorphemeoranounphrase,butalsotoarbitrarynon-physicalornon-localproperties,suchasbeingacrumpledshirt,agrievingwidow,oracollapsingstar(1986).Thesesensitivitiesareparticularlyimpressivegiventhattheyseemtobe(1)productiveand(2)systematic.(1)Peopleseemcapableofdiscriminatingstimuliofindefinitelogicalcomplexity,suchasbeingacrumpledshirtthatwaswornbythethiefwhostolethecatthatchasedtherat...(1975a:ch.1);and(2)anyonecapableofdiscriminatingonelogicalformiscapableofdiscriminatinglogicalpermutationsofit;forexample,onecandiscriminateJohn’slovingMaryifandonlyifonecandiscriminateMary’slovingJohn(1987b:147ff.).AgooddealofFodor’sworkhasbeendevotedtoshowingthatnonon-mentalisticaccountcanexplainthesephenomena.Thushehasarguedatlengththatpurelyphysicalist,behaviorist,Gibsonian,syntactic,andeliminativeconnectionistaccountsofbehaviorareeithervacuousorempiricallyinadequate(1968a,1981b,1987b:161–3,1988a,and1991a).Itisdifficulttoseehowanyphysicalmechanismcouldbesensitivetosuchanextraordinaryrangeofarbitrarypropertiesoftheworldwithoutexploitinginternalprocessesoflogicalcombination,inference,andhypothesisconfirmationthatessentiallyinvolvephenomenasatisfyingthefourdemandslistedonpp.451–2.453\nGEORGESREYCRTT:ComputationFodor’smainproposalformeetingthosefourdemandsisthecomputational/representa-tionaltheoryofthought(CRTT).Indeed,muchofhisworkcanberegardedasanefforttoincorporateintopsychologyAlanTuring’scrucialworkonmechanicalcomputation,accordingtowhichcertainrationalprocessescouldberealizedmechanically:forexample,eachoftherulesoflogiccanbeshowntoinvolvemechanicaloperationsontheformallyspecifiedsentencesofaformallanguage.Fodorregardsthisaspromisingforpsychology,since,heargues,peopleatleastsometimesengageinthesortoftruth-preservinginferentialprocessescapturedbylogic(1994:9).Thisalreadymarksanimportantbreakwithtraditionalpsychology,whichtendedtorelyonmereassociationsamongideas(Hume)orstimuli(Skinner).Theseseemincapableofcapturingtherela-tionbetween,forexample,thepremisesandtheconclusionofavalidargument;mereassociations,likethatbetween“salt”and“pepper,”areneithernecessarynorsufficientforunderstandingvalidarguments.Thisconcernwithlogicandtruthalsocommitspsychologytovehiclescapableoftherequisiterepresentationalrichness.Traditionalempiricistpsychology(asinLockeandHume)tendedtobenotonlyassociationist,butalsotoregardmentalrepresentations(or“ideas”)asimages.Butalthoughimagesmayhavearoletoplayinthought(1975a:174–94),itisdoubtfulthattheyareremotelyadequatefortheexpressionofitingeneral.Whatimagecouldexpresstheconditionals,quantifiers,negations,andmodalsinsuchathoughtasIfeveryonedrinksthennooneshoulddrive?Merelyapictureofalotofnon-drivingdrinkerswon’tquitedo.Theonlyvehiclesthatseemremotelycapableofexpressingsuchthoughtsarethelogico-syntacticvehiclesofalanguage,naturalorartificial,withpreciselytheresourcesofoperators(quantifiers,connectives)andreferentialdevices(predicates,variables,names)thatweordinarilyusetoexpressthosethoughts.Thatis,theremustbesomesortoflanguageinwhichathinkerthinks,a“languageofthought”(an“LOT”or“mentalese”).Talkof“sentences”inthebrainmustn’tbetakenonthemodelofsentencesastheyareinscribedonpagesofbooks.Sentencesarehighlyabstractobjectsthatcanbeentokenedinanendlessvarietyofways:aswaveforms(inspeech),assequencesofdotsanddashes(Morsecode),assequencesofelectricallychargedparticles(onrecordingtape).Itispresumablyinsomethinglikethelatterformthatsentenceswouldbeentokenedinthehead.Indeed,CRTTisbestviewedassimplytheclaimthatthebrainhaslogicallystructured,causallyefficaciousstates,athesisthat,whateveritsmerits,isn’tpatentlyabsurd.(Notealsothatthisisnotathesisthatissupposedtobeintrospectiblyplausible:CRTTdoesnotentailthatpeople’smentallivesshouldappear“introspectively”toinvolvesentences,muchlesssentencesofanaturallanguage.)AnextremelysimpleversionofCRTTcouldbetrueofanintelligentsysteminthefollowingway:therearesensorymodules(e.g.visualandauditorysystems,see“ModularityandtheLimitsofCRTT,”below)thattransduceambientenergyformsintoelectricalsignalsthatinturnproducestructuredsentencesasinputtoacentralcognitivesystem(perception).Thiscentralsystemselectscertainsentencesfromapre-established(perhapsinnate)set,teststheirdeductiveconsequencesagainstthisinput454\nJERRYFODORfora“bestfit,”andproducesasoutputthosesentencesthatpassthattestabovethresh-old(belief).Thesesentencesinturnmaybetheinputtoadecision-makingsysteminwhich,onthebasisofthatinput,innatepreferences,andutilityfunctions,acourseofactionisdetermined,thatis,abasicact-descriptionisselected(intention)thatcausesabasicactsatisfyingthatdescriptiontobeperformed(action).Acommonobjectiontosuchanaccountisthatitrequires“homunculi”inthebrain,possessingpreciselythesameintelligencethatissupposedtobeexplained,inorderto“read”thesesentences.However,whatTuring’stheoryofcomputationhastaughtusisthatcomplexcomputationalprocesses,suchasoperationsuponsymbolsinalan-guage,canbebrokendownintosimpleroperationsthatareeventuallysosimplethat(it’sobviousthat)amachinecanexecutethem(1975a:73)withoutatthatpointexploitinganyintelligenceatall.Besidesthestraightforwardargumentfromtheexpressivepowerofalanguage,FodoradvancesanumberofotherreasonsforCRTT:1Itispresupposedbystandardtheoriesofperception,hypothesisconfirmation,anddecision-making,allofwhichinvolvetheagentrepresentingtheworldandassign-ingutilitiestotheconsequencesofvariouscoursesofaction(1975a:ch.2).2Itisabletoexplainthetruth-preservingtransitionsinthoughtofwhichrationalcreaturesareatleastsometimescapable:forexample,theabilitytodeduce“Ratsdie”from“Catslive”and“Ifcatslivethenratsdie”(1994:9).3ItiseasytoconceiveofcomputationalarchitecturesexploitinganLOTthatwouldexplaintheproductivityandsystematicityofourmindsmentionedearlier,forexample,amachinethatwassensitivetosyntacticstructurescouldinstandardrecursivewaysproduceindefinitelycomplexrepresentations(productivity),andcouldaccessonerepresentationifandonlyifitcouldaccessalogico-syntacticper-mutationofit(systematicity)(1987b).4Itoffersaperspicuousaccountoftheintensionalityofthought:Oedipus’thinkinghe’llmarryJocastaisdistinctfromhisthinkinghe’llmarryhismothersincethevehiclesofthetwothoughts(theLOTequivalentsof“Jocasta”and“mymom”)aredifferent.ItisatleastthesefourargumentsthatFodordeploysbothagainsttraditionalassociationisminpsychology,aswellasagainstwhatheregardsasitscontem-porarymanifestation,radicalconnectionism(1988a,1991a).(AgainstconnectionistproposalsthatwouldmerelyproposeanovelimplementationofCRTT,Fodorhasnoobjection.)DespitehiscommitmenttoaCRTT,Fodorhasdoubtsaboutitseventualscope,towhichwe’llreturnbelow(in“ModularityandtheLimitsofCRTT”).CRTT:RepresentationSofar,we’vediscussedFodor’sviewsaboutmentalprocesses,ascomputationsoverlogico-syntacticrepresentations.Fodorwouldbethefirsttorecognizethatthisisatbestonlyhalfanaccountofmentality:computationsmaywellpreservetruthbetterthanassociationsdo,butwheredotheserepresentationsacquireanysemanticpropertiesliketruthinthefirstplace?455\nGEORGESREYInferentialroletheoriesInearlywork(1963),Fodorwasdrawntowhatis,broadlyspeaking,an“inferentialrolesemantics”(IRS).Thisisafamilyofviewsaccordingtowhichthemeaningofanexpressionhastodowithitsinferentialrelationstootherexpressions,asinthecaseof“bachelor”entailing“unmarried.”Theserelationsmightinvolvedefinitions(asintraditional“analyses”and“meaningpostulates”inphilosophy),“semanticdecompo-sition”inlinguistics,“proceduralsemantics”inartificialintelligence,or“prototypes”and“wholetheories”inpsychology.Bythelate1970sFodorbecameconvincedthatthestandardargumentsforIRSsufferedfromseriousempiricalandphilosophicaldifficulties:proposedlinguisticdecompositionswereseriouslyinadequate;therewasanembarrassingpaucityofpsychologicalevidenceforanythinglikedefinitions(1975b),whichthehistoryofanalyticphilosophyhadshownwerenotoriouslydifficulttoprovideinthefirstplace(1970);andQuinehadcastseriousdoubtonwhethertherecouldeverbeanytheoreticallysatisfactorywayofdistinguishingconstitutiveinferentialrelations(the“analytic”)frommerelycommonbeliefs(the“synthetic”)(1998a)(seeQUINE).Indeed,althoughFodorhasnopatiencewithQuine’sbehaviorism,hewholeheartedlyendorseshisrejectionofdefinitionsashavinganyseriousexplanatorystatusinanysciencewhatsoever(theintuitiveappearanceofan“analytic/synthetic”distinction,Fodorargues,isdueeithertothe“centrality”ofaclaimtoone’sthought,ortoitsinvolving“one-criterion”concepts(1998a:80–6)).UndertheinfluenceofQuine,andespeciallyofhisdictum,“Theunitofmeaningisthewholeofscience,”manyIRStheoristshavethemselvestendedtoforgotheanalytic/syntheticdistinctionandregardallofanexpression’sinferentialrelationsasconstitutive.This“meaningholism”hasthedisturbingconsequencethatitwouldbevirtuallyimpossiblefortwopeopleevertomeanexactlythesamething,indeed,forevenonepersontomeanthesamethingoveranychangeofbelief–renderingmemoryimpossible!Fodorthinksthatthisrendersanyseriouspsychologicalgeneralizationsimpossibleaswell,andsoisatpainstoblockthemanyargumentsforit(1992a),andforanyIRS,which,hethinks,inevitablyinvitesit.InFodor’sview,theoriginalsinendemictoIRStheoriesconsistsinconflatingsemantics(oratheoryofthecontentofconcepts)withepistemology(oratheoryofhowweapplyconcepts).ThisconflationnotonlyburdenssemanticswiththenotoriousproblemsofaverifiabilitytheoryofmeaningthatlurksinmostoftheaboveIRSproposals,butalsopresentssubstantialproblemsforaccountingfortheaforementionedproductivityandsystematicityofthought.Fodorarguesthattheselatterphenomenarequireacompositionalsemantics(i.e.oneinwhichthemeaningofacomplexexpressionisafunctionofthemeaningofitsparts),andepistemologicalcapacitiesarenotingeneralcompositional:onecouldknowalotaboutpetsandalotaboutfishwithoutknowingmuchatallaboutpetfish,forexample,whatonetypicallylookslike(1990b,1998b:chs4–5).Unconfoundingepistemologyandsemantics,Fodorinsteadforgoesany“molecular”accountofmeaningthatdependsuponrelationsamongsymbols,andinsteadembracesan“atomistic”theorythatrequiresonlythatasymbolstandinaspecificrelationtotheexternalworld.456\nJERRYFODORInformationtheoriesFodortakesashispointofdeparturethe“information”theoreticsemanticsdevelopedbyFredDretske,whichtreatssemanticmeaningasaspeciesof“natural”meaning(wherebydarkclouds“mean”rain,orsmokefire).Thisideaoftenappearsinpsycho-logicaldiscussionsundertheguiseofdiscriminationabilities:forexample,somethingisa“shapereceptor”ifandonlyifitreliablydiscriminatesshapes.Theideaisnaturallyspeltoutintermsofcertaincounterfactualdispositionalpropertiestoco-varywithspe-cificphenomenaintheworld.Sostated,informationsemanticsisopentoseveralimmediateobjections.1pan-semanticism:somethingneedstobesaidaboutwhat’sspecialaboutsemanticorpsychologicalmeaning,sinceeverythingiscausallyrelated(andso“carriesinformation”)aboutsomething;2transitivity:“information”istransitive,butmeaningisn’t:if“smoke”co-varieswithsmoke,andsmoke,itself,withfire,then“smoke”co-varieswithfire;but“smoke”doesn’tmeanfire(1990b:93);3robustness:mosttokeningsofsentencesareproducedintheabsenceofthecondi-tionsthattheyneverthelessmean.“That’sahorse”canbeutteredonadarknightinthepresenceofacow,orjustidlyinthepresenceofanything.Fodor(1987b)callstheselatterusages“wild,”andthepropertywherebytokensofsymbolsmeanthingsthataren’tonoccasiontheiractualcause,“robustness.”4Inaccountingforrobustness,asemantictheoryneedstosaywhatdistinguishesthe“wild”fromthemeaning-constitutivecauses,aproblemmadevividbythe“dis-junction”problem:whatmakesittruethatsomesymbol“F”meansHORSEandnot23HORSEORCOWONADARKNIGHT,orHORSEORCOWONADARKNIGHTORWORWORi...(whereeachWisoneofthepurportedly“wild”causes)(1987b).Afifthproblemcouldberaisedregardingthecontentsoflogicalandmathematicalsymbols,whichdonotobviouslyenterintocausalrelationswithanyworldlyphe-nomenon.Fodorsetsasidethisproblemforthenonce,althoughsuspectingthattheyaretheonlysymbolsforwhichanIRSisplausible.TeleologicalviewsAnaturalsuggestionregardingthemeaning-constitutiveconditionsisthattheyareinsomesense“optimal”conditionsthatobtainwhennothing(e.g.poorvision,limitedspatiotemporalaccess)is“interfering”withbelieffixation,anditisfunc-tioningasitwas“designedto.”Fodor(1987b)callssuchtheories“teleological”andhehimselfproposedaversionofoneinthewidelycirculatedpapercalled“Psychosemantics”(1990a)(tobedistinguishedfromthebook(1987b),ofthesametitleinwhichherejectsanysuchtheory!).Theattractionofsuchatheoryliesinitscapturingtheideathattwoindividualsmeaningthesamethingbysomesymbolconsistsintheiragreeingaboutwhatitwouldapplyto,weretheytoagreeaboutevery-thingelse.Theirdisagreementsaretobeexplainedasduetotheirdifferingepistemicpositionsandreasoningcapacities.AlthoughFodornowheresuggestssuchtheoriesarefalse,hedoesthinktheyaresubjecttoanumberofdifficulties,thechiefoneconsistingofthecircularitythatseems457\nGEORGESREYunavoidableinspecifyingtheoptimalconditions:itwouldappearthatthoseconditionscannotbespecifiedwithoutemployingtheveryintentionalidiomthetheoryissup-posedtoexplain(1987b:104–6).InordertoavoidthisandotherproblemsFodor(1987b,1990b)wentontoproposehis“asymmetricdependency”theory.Althoughitmakesnoexplicitappealtoidealepistemicconditions,muchofitsmotivationcanbeappreciatedbythinkingoftheidealco-variationaltheoryinthebackground.AsymmetricdependencyAccordingtotheidealco-variationaltheory,tokensofanexpressionmaybe“wild,”thatis,producedbyapropertyitdoesn’texpress.Now,onewaytounderstandtheasymmetricdependencytheoryisfirsttonoticethat,plausibly,allsuchwildcasesdependupontheidealcase,butnotviceversa:thewildtokeningsdependupontheidealones,buttheidealonesdon’tdependuponthewildones.Gettingthingswrongdependsupongettingthingsrightinawaythatgettingthingsrightdoesn’tdependupongettingthingswrong.Thus,thepropertyHORSEcauses“cow”becausesomehorses,forexample,thoseatthefarendofthemeadow,looklikecowsand,underidealconditions,COWcauses“cow.”Soformulated,ofcourse,theaccountstillmentionsidealconditions,andtheseFodorhasconcededcannotbespecifiedwithoutcircularity.Hisfurtherinterestingsug-gestionisthatmentionoftheidealconditionshereisentirelyinessential:thestructureofasymmetriccausaldependencyalone,abstractedfromanyspecificconditionsorcausalchains,willdoalltherequiredwork(1990b:99,1998a:156ff).Tosimplifythediscussion,wecandefineapredicate,“xislockedontoy,”tocapturethisasymmetriccausalstructure:Asymbol“S”islockedontopropertyFjustincase:1there’sa(ceterisparibus)lawthatFcausestokeningsof“S”;2tokeningsof“S”arerobust:i.e.aresometimescausedbyapropertyGotherthanF;3whenGs(otherthanFs)causetokeningsof“S,”thentheirdoingsoasymmetricallydependson(1)i.e.onthelawthatFcauses“S”s,whereX’scausingYs“asymmetricallydepends”onalaw,L,ifandonlyifX’scausingYwouldn’tholdbutforL’sholding,butnotviceversa:LcouldholdwithoutX’scausingY.Thus,smoking’scausingcancer,dependinguponmanylaws,asymmetricallydependsuponNewton’s,sinceNewton’sdoesn’tdependuponsmoking’scausingcancer.Fodor’sproposalaboutcontentthenis:(M)if“S”islockedontoF,then“S”expressesF.Thus,apredicate“C”expressesCOWif(a)itwerealawthatthepropertyCOWcauses“C”tokenings,and(b)othercausalrelationsbetweenproperties(e.g.HORSE,MILK,etc.)and“C”tokeningsasymmetricallydependuponthislaw.Notethat(M)suppliesonlyasufficient,notanecessaryphysicalisticconditionforpredicateexpression.Fodorbelievesthisisallthatheisrequiredtodo,givenhismerely“supervenient”physicalismmentionedonp.453.Fodorarguesthat,ifthereareno458\nJERRYFODORcounterexamplesto(M),thenhehasdoneallthatheneedstodotoshowthat,con-trarytodualism,certainphysicalarrangementsaresufficientforintentionality.NotealsothatFodoravailshimselfoftheconvenientlargesse(somemightregarditasaprofligacy)ofpropertiesintheworld.Forhim,asformanyphilosophers,there’svirtuallyapropertyforeveryprimitivepredicate,whetherornotthepropertyhappenstobeinstantiated.Thus,therearepropertiesofbeingaunicornandbeingphlogiston,despitethelackofanyinstantiationsofthemintheactualworld.AndsotheconceptsUNICORNandPHLOGISTONaredistinguishableinthiswaybytherespectivelockings.Fodor(1987b,1990b,1991b)defends(M)withconsiderableingenuity.Whetherornotthisatomisticaccountofmeaningcansucceed,itiscrucialtounderstandingFodor’slaterworkwhereitissimplytakenforgranted.(Fodorhaswrittenalmostnothingfurtheron(M)since1991.)SolipsismandnarrowcontentDespitetheattractionsofan“externalist”theorylike(M),itishardtoresisttheideathatthereissomethingsemanticpurely“inthehead.”Therearetwostandardwaysofpressingthispoint:so-called“Frege”cases,and“Twin”cases.FregecasesThereseemtobeplentyofexpressionswiththesameworldlyreferencethatneverthe-lesshavepatentlydifferentmeanings.Frege’sexamplewas“themorningstar”and“theeveningstar,”butthesearedistinguishableindifferentpossibleworlds,andsoinvolvedifferentproperties.Butwhataboutpredicatesthatarenecessarilyco-extensive,notonlyinthisbutallpossibleworlds,like“equilateral”versus“equiangular”triangle?HereFodoravailshimselfoftheresourcesoftheLOT:“equi-angu-lar”and“equi-later-al”aresyntacticallycomplex,andsothoughtsinvolvingthemcanbedistinguishedthereby(1998a:15–21,163–5).Inhisterminology,theyaredifferentconceptswiththesamecontent.Canallcasesbehandledintheseways?Arealldifferencesinthoughteitherdiffer-encesinthedenotedpropertiesorstructuraldifferencesinthewaythepropertiesarerepresented?Propernamespresentonekindofproblem;termsforkinds(“lawyer,”“attorney”)another;necessarilyco-instantiatedterms,suchasQuine’snotorious“rabbit/undetached-rabbit-parts”example,stillanother.Fodorclaimsthat,solongascoreferentialnamesandsimplekindtermsaretreatedastokensofdifferenttypesinter-nallybyanyagent,interagentcomparisonsaremerelypragmaticallydifferent(1994:109–12),aviewthatmarksachangefromhisearlierviewandwhichheshareswithmany“directreference”theoristssuchasDavidKaplan.HealsoprovidesadetailedresponsetotheQuinianchallenge,exploitingadistinctioninthelogicalrolethedifferentco-instantiatedthoughtsplay(1994:55–79).TwincasesTwincasesaretheconverseofFregecases:insteadoftwoexpressionswiththesamereferencebutdifferentsenses,herewehaveexpressionswiththesamesensebutdif-ferentreferences.HilaryPutnaminvitedustoimaginetherewasafarawayplanet,459\nGEORGESREY“TwinEarth,”exactlylikeEarthineveryway(includinghistory)exceptforhavinginplaceofH2Oasuperficiallysimilar,butatomicallydifferentchemicalXYZ(seePUTNAM).OscaronEarththinksaboutwater,i.e.H2O,wherehistwinonTwinEarthdoesn’tthinkaboutwateratall,butabouttwater,i.e.XYZ.ThequestionnowiswhetherpsychologyshouldcareaboutdistinguishingOscarfromhistwin:afterall,aren’ttheirinternalmentallivesindistinguishable?Fodor’sviewsabouttwincaseshavechangedovertheyears.Inamuchdiscussedpaper(1980a)hearguedthatpsychologycouldn’twaituponafulltheoryofallanagent’senvironment,tellinguswhichwaswaterandwhichtwater.AndsoithadbetteradoptwhatPutnamcalledapolicyof“methodologicalsolipsism,”theorizingonlyaboutwhatgoesoninsideanagent’shead.Hetakesthistocomportwellwithwhathecallsa“formalitycondition”thatfollowsfromCRTT:mentalstateshavetheirefficacyasaconsequenceoftheformalcharacteroftheirtokens.However,healsorecognizesthatintentionalpropertiesofmentalrepresentationsareessentialtotheirroleinpsychologicalexplanation,andso,ifpsychologyissolip-sistic,theremustbesomesolipsistic,or“narrow”kindofcontentthatsupervenesontheinternalstatesofathinker.Forawhile,Fodorsettledonthefollowingconception(originallyproposedbyStephenWhite,developingtheseminalworkofKaplan):thenarrowcontentofaMentaleseexpressionisafunction(inthesettheoreticsense)thatmapsaperson’slifecontextontoabroadcontent.Forexample,thenarrowcontentofOscarandTwinOscar’s“water”isthefunctionthatmapsOscar’scontextontoH2Oandhistwin’scontextontoXYZ.WhenOscarutters“Wateriswet,”hetherebyexpressesthecontent“H2Oiswet,”whilewhenTwinOscaruttersitheexpressesthecontent“XYZiswet.”Twosymbolshavethesamenarrowcontentjustincasetheyservetocomputethesamesuchfunc-tion:itisthisthatissharedbyOscarandhistwin.Whetherthisfunctioncanactuallybespecifiedbypsychologyis,however,notaltogetherclear:howdoesonecontinuetospecifyitbeyondEarthandthefancifulcaseofTwinEarth?OnlywidecontentafterallMorerecentlyFodorhasmovedawayfromanyrelianceonnarrowcontentatall.HearguesthatbothFregecasesandTwincasesdon’thavetobetakenseriouslybypsy-chology:theyareviolationsoftheceterisparibusconditionsunderwhichseriouspsy-chologicallawsaresatisfied(1994:ch.2).WithrespecttotheTwincases,heclaimsthatit’simportanttorememberwhatheearlierforgot,that,althoughtheyareconcep-tuallypossible,theyarenotnomologicallyso,and“empiricaltheoriesareresponsibleonlytogeneralizationsthatholdinnomologicallypossibleworlds”(1994:29).WithrespecttotheFregecases,hispositionismorecomplex.Citingthecaseofbelief,hearguesforwhathecallsthe“PrincipleofInformationalEquilibrium”(PIE),Agentsarenormallyinepistemicequilibriuminrespectofthefactsonwhichtheyact.Havingalltheinformation–havingalltheinformationthatGodhas–wouldnotnormallycauseanagenttoactotherwisethanashedoes.(1994:42)Heclaimsthat,sincethesuccessofouractionisnoaccidentandtendstodependuponthetruthofourbeliefs,“nobelief/desirepsychologycanfailtoacceptPIE”(1994:42).460\nJERRYFODORConsequently,Fregecases,inwhichanactiondependsuponbeingignorantofaniden-titystatementare,fromthepointofviewofpsychology,“aberrations.”Theydooccur;and,followingtheearlierdiscussion,theycanbedescribedbyinvokingsyntacticallydifferentLOTexpressionsasthedifferent“modesofpresentation”Fregethoughtwereneeded.But,paceFregeandhismanyfollowers,no“thirdrealm”of“senses,”ornarrowcontents,isneeded.Whetherasimilarargumentcanbemadeforattitudesotherthanbeliefisaquestionthatisunfortunatelynotaddressed.NativismInthesamebookinwhichFodorpresentedtheCRTThypothesishealsodefendedahighlycontroversialthesisabouttheinnatenessofallconcepts.ThatthesisoriginallystirredmorecontroversythandidCRTTitself,muchofwhichwasaddressedinpubli-cation(1981c);butitisindependentofCRTT,dependinguponfurtherclaimsaboutthenatureofconcepts,definitions,andlearning,issueswhich,moreover,arenotlikelytobesettledbycommonsensethoughtonthematter(1998a:28).In(1975a:ch.2)Fodorarguedthat,sincestandardmodelsoflearninginvolvedhypothesisconfirmation–onelearnsthat“cat”inEnglishmeanscatbyconfirminghypothesesaboutEnglishusage–thesemodelsarecommittedtotheconstituentcon-ceptsofthesehypothesesbeinginnate(1975:ch.2):afterall,onecan’tformhypo-thesessuchas“‘red’meansred,”withoutusingandthereforealreadypossessingthoseveryconcepts!Laterhedeepensthediscussionbyconsideringthetraditionalempiricistsuggestionthatoneacquiresconceptsbyconstructingcomplexideasoutofsensoryprimitivesthatarevariouslyassociatedinexperience(1981b).Thepersistentfailureoverthecenturiestosetouttheseconstructions,whichiscloselyrelatedtothefailuresofanyIRS(seesection“Inferentialroletheories”),suggeststhattheyareprobablynottobehad.AsPlatoobservedearlyon(andChomskymorerecently),mostofourcognitivecapacitiesseemtotranscendourspecificsensoryhistories.However,Fodor(1998a:chs6–7)hasrecentlyraisedwhatheregardsasaseriousobjectiontohisearlierviewsthatallconceptsareinnate,whathecallsthe“DOORKNOB/‘doorknob’”problem:Whyisitsooftenexperiencesofdoorknobs,andsorarelyexperiencewithwhippedcreamorgiraffes,thatleadsonetolockontodoorknobhood?...assumingthatprimitiveconceptsaretriggered,orthatthey’re“caught,”won’taccountfortheircontentrelationtotheircauses;apparentlyonlyinductionwill.Butprimitiveconceptscan’tbeinduced;tosupposetheyareiscircular.(1998a:127–32)Thatis,ashearguedin1975a:ch.2,youcan’tperforminductionstoconceptsthatyoucan’talreadyrepresent.Hissolutiontothisproblemis“ontological”:thepropertiesthatcorrespondtoourprimitiveconceptsarejustthepropertiestowhichwegeneralize,fromphenomenallyspecifiedstereotypes.Forexample,itisconstitutiveofbeingadoorknobthatitistheprop-ertyontowhichpeoplelockasaresultofexposuretostereotypicaldoorknobs.Fodorrelieshereonwhatheregardsasthepsychologicalevidencethatstereotypicalinstances461\nGEORGESREYofprimitiveconceptscanbespecifiedindependentlyofthoseconcepts,e.g.byenumerat-ingtheshapes,colors,functionsthattypicalinstancesshare(1998a:137–45).Fodorthinkswenowhaveanacceptableaccountofthenon-arbitraryrelationbetweentheacquisitionofmanyofourconceptsandtheexperienceoftypicalinstancesofthem.Moreover,ithastheinterestingconsequencethatsomeonenotexposedtotypicaldoorknobsmightwellnothavetheconcept“doorknob,”since,withoutthatexperience,theremightwellbenosuchlocking.Hespeculatesthatthisisthecaseformostcommonsenseconcepts.So,heconcludes,“maybetherearen’tanyinnateideasafterall”(1998a:143).Allthereareareinnatedispositionstolockontopropertieswhenexposedtotheirstereotypicalinstances.Thereareanumberofproblemsraisedbythisview.Itmightappeartounderminetherealityofthereferentsofourcommonsenseconcepts,makingthem“dependentuponus.”Butthisisanillusion.Adoorknobmaybeidentifiedbyitstendencytohaveacertaineffectonus;butitcanexistevenif,ifwedidn’texist,itdidn’thavethateffect.But,inanycase,mindsareinfactapartoftheworld,anddoorknobsdoinfacthavetheireffectsuponthem(1998a:148–9).Amoreseriousproblemisraisedbywhatseemtobegenuine“naturalkind”con-cepts,whosereferencepatentlydoesnotinvolveanyrelationtowhatpeoplemightdo:beinggenuinewaterisamatterofbeingH2O,whetherornotpeoplewouldgeneralizetoitonthebasisofstereotypicalsamples.Theseconcepts,Fodorclaims,arelatecom-ersinourcognitivedevelopment,dependentuponthesocialinstitutionofscience,thedevelopmentofsophisticatedtheoriesandtheconsequentdeferencetoexperts(1998a:150–62).Hishopeisthatthepeculiarsemanticeffectsoftheselatedevelopmentscanbeentirelyspeltoutintermsofdetailsofthecounterfactualsinvolvedinhisasymmetricdependencetheory(see“Teleologicalviews”).ContraselectionismFodor’sinterestinnativisthypothesesmightleadonetothinkthathebelievesthatourcognitivecapacitiesaretheresultofnaturalselection.Nothingcouldbefurtherfromthetruth.Althoughhedoesn’tdoubtforamomentthatthehumanmind/brainevolved,heseesnoreasonwhatevertothinkthatitsconsiderablecognitivecapacitieswerespecificallyselected.Inparthisoppositiontoselectionismislikehisoppositiontoempiricisttheoriesoflearning:heseesnoreasontothinkthatthesecapacitiesreflectsomesortofregularitiesinourhistories.Manyofthem–suchasourgrammaticalormathematicalabilities–seemtofarexceedanythingthateitherourupbringingorourevolutionaryhistorycouldplausiblyhavesupported.Additionally,hethinksthatselectioniststoriesabouttheevolutionofcognitivecapacities,besidebeingflagrantlyspeculative,seriouslyunderestimatethecomplexityoftherelationbetweenmindandbrain:“makeanape’sbrainjustalittlebitbigger(ordenser,ormorefolded...)andit’sanyone’sguesswhathappenstothecreature’sbehavioralrepertoire”(1998b:209).It’saslikelyasnotthatsomesmallchangeinourancestors’brainsmadethemtremendouslysmarter,likethemodestchangerequiredtotransformafinitestatemachineintoaTuringmachine.Doubtlessthisprovideduswithsomeselectionaladvantages.Butthatisnoreasontosupposethatanythinglikethemajorityofmentalabilitieswedisplay–fromacquisitionofgrammarstothegrasp462\nJERRYFODORofhighermathematics,physics,orfolkpsychology–werethemselvesindividuallyselected.ModularityandthelimitsofCRTTIndiscussingIRStheoriesinthesection“Inferentialroletheories,”wenotedtheirten-dencytobecome“holistic,”atendencyFodorseesasinimicaltotheinterestofseriouspsychology.Thistendencyhasalsobeenprevalentinmuchcontemporary“NewLook”theoriesofperception:theworkofJeromeBrunerinpsychology,ThomasKuhninthehistoryofscience(seeKUHN),andNelsonGoodmaninphilosophy(seeGOODMAN),emphasizeshowmuchpeople’stheoreticalexpectationscancolortheirperceptions,tothepointthattheyandothersinsistthatweoughttoabandon“themythofthegiven”(seealsoSELLARS).Fodordeploresthisholistictendencyhereaswell,andinthiscasemarshalsinterestingpsychologicalevidenceagainstit.Fodorfirstofallcallsattentiontothesurprisinglylittlenoticedfactthattheveryper-ceptualillusionsthatNewLooktheoristsofteninvokedtomaketheirpointactuallytellagainstit:formanyoftheseillusionsdonotdisappearwhenweknowbetter.NomatterhowsurewearethattheMuller–Lyerillusionlinesareequal,theupperstilllookslongerthanthelower.InaphraseFodortakesfromthepsychologist,ZenonPylyshyn,per-ceptionseemstobe“cognitivelyimpenetrable.”Fodorcitesfactslikethese,andconsiderabledataaboutlanguagecomprehension,toarguethatthereareanumberofdedicatedmental“modules”thatare“informa-tionallyencapsulated”fromthe“central”systemwherebywereasongenerallyandfixourbeliefs.Theseincludethestandardsensorysystems,certainlevelsoflanguagepro-cessing,andperhapsotherdedicatedsystemssuchasfaceandmusicalperception.Amongotherthings,suchsystemsare,furthermore:extremelyrapid(ontheorderofaquarterofasecond,1983:61–4);shallow(theiroutputsarelimitedto“basicpercep-tualcategories,”suchaschairordog,1983:86–99);associatedwithacharacteristicdevelopment(visionandlanguageseemtodevelopinspecificwaysthatareindependentofothermentalcapacities,1983:100–1);andtheyaredomainspecific(confinedto,forexample,processingoflight,faces,orgrammar,1983:47–52).Inthislastrespect,Fodor’sconjectureoverlaps,ofcourse,withChomsky’spostulationofaninnategram-maticalcompetence,but,further,involvesthemodularityofnotonlythedomain,butofthestandardprocessingofinformationfromthatdomain(1989,1998b:ch.4).Allofthesepropertiescontrastwithpropertiesofthecentralsystem,whichisvoluntary(youcanchoosewhattothinkabout),slow(youcanpotterformonths),sometimesdeep(youcanthinkaboutnon-perceptualcategories),non-localized(“thereis,toputitcrudely,noknownbraincenterformodusponens,”1983:98),anddomainindependent(youcanthinkaboutalmostanything).Philosophically,whatisperhapsmostintriguingaboutthepostulationofsensorymodulesisthewayinwhichitprovidesanewbasisforthecontroversialobserva-tion/theoreticdistinction,althoughabasisthatmayonlypartiallyoverlapthetradi-tionalintrospectiveone(1984b,1988b).463\nGEORGESREYItmightbethoughtthat,giventheirencapsulation,thesemodulesarenotreallycog-nitive,andsodon’tinvolvealltheissuesaboutcomputationandrepresentationthatarethefocusofmostofFodor’swork.Butpreciselythepointofpostulatingmodulesasopposedtostandardtransducersisthattheydoseemtoinvolvecomputation.Indeed,Fodorregardsthemas“compiledtransducers....‘compiled’toindicatethattheyhaveaninternalcomputationalstructure,and‘transducer’...toindicate...informationencapsulation”(1983:41).Indeed,inhisrecentwork(2000),Fodorarguesthatmodulesmaybetheonlyappro-priatedomainforaCRTT.Inbeingunencapsulated,thecentralbelieffixationsystemexhibitsanumberofpropertiesthatpresentseriousprima-faciedifficultiesforanystan-dardcomputationaltreatment.RelyingonwhatheregardsasQuine’sastuteviewsabouttheoryconfirmation,Fodorclaimsthatcentralsystemsare:1“Quinian,”i.e.computedoverthetotalityofabeliefset,aswhenwesettleonatheorythatis,forexample,simplestandmostconservativeoverall;2isotropic(everybeliefispotentiallyrelevanttotheconfirmationofeveryother,aswhenapatternoflightonapieceofpaperconfirmsatheoryabouttheageoftheuniverse)(1983:105ff).Fodor(1999)arguesthatthesefeaturesrenderbelieffixationholisticandcontext-dependent,inafashionthatisnotclearlyamenabletotheTuringcomputabilityinvokedbyCRTT.Thislatterdependsuponexploitingarepresentation’slocalsyntacticfeatures:anargument’sdeductivevaliditycanbecheckedbylookingatitslocalspelling.However,abductivecogencyseemstobeascertainableonlybylookingataclaim’srela-tiontoother,indefinitelyremoterepresentations,anditseffectonthebeliefsystemasawhole.Fodorseesthisastheproblemunderlyingtheso-called“frameproblem”encounteredinartificialintelligence(1987a),andisconsequentlypessimisticabouttheprospectsofitbeingultimatelysolvedbyCRTT.AlthoughCRTTisnecessaryforanadequatetheoryofmind,itseemstobefarfromsufficient.BibliographyofworksbyFodorForafullbibliographyofFodor’sworkupuntil1991,seeB.LoewerandG.Rey(1991)MeaninginMind:FodorandhisCritics(Oxford:BlackwellPublishers),whichalsocontainscriticalessaysbyanumberofprominentphilosophers,andanintroductiononportionsofwhichthepresententryrelied.1963(withKatz,Jerrold):“TheStructureofaSemanticTheory,”Language39,pp.170–210.1968a:PsychologicalExplanation,NewYork:RandomHouse.1968b:“TheAppealtoTacitKnowledgeinPsychologicalExplanation,”JournalofPhilosophy65,pp.627–40.1970:“ThreeReasonsfornotDeriving‘Kill’from‘CausetoDie’,”LinguisticInquiry1,pp.429–38.1972(withBlock,N.):“WhatPsychologicalStatesareNot,”PhilosophicalReview81,pp.159–81.1974(withBever,T.andGarrett,M.):ThePsychologyofLanguage,NewYork:McGrawHill.1975a:TheLanguageofThought,NewYork:ThomasY.Crowell.1975b(withFodor,J.D.andGarrett,M.):“ThePsychologicalUnrealityofSemanticRepresentations,”LinguisticInquiry6,pp.515–31.464\nJERRYFODOR1978:“TomSwiftandhisProceduralGrandmother,”Cognition6,pp.229–47.1980a:“MethodologicalSolipsismConsideredasaResearchStrategyinCognitiveScience,”BehavioralandBrainSciences3,pp.63–109(withrepliestocommentators).1980b(withGarrett,M.,Walker,E.andParkes,C.):“AgainstDefinitions,”Cognition8,pp.263–367.1981a(withPylyshyn,Z.):“HowDirectisVisualPerception?SomeReflectionsonGibson’s‘EcologicalApproach’,”Cognition9,pp.139–96.1981b:“ThePresentStatusoftheInnatenessControversy,”inFodor1981c,pp.257–316.1981c:Representations:EssaysontheFoundationsofCognitiveScience,Cambridge,MA:MITPress.1983:TheModularityofMind:AnEssayonFacultyPsychology,Cambridge,MA:MITPress.1984a:“Semantics,WisconsinStyle,”Synthese59,pp.231–50.1984b:“ObservationReconsidered,”PhilosophyofScience51,pp.23–43.1986:“WhyParameciaDon’tHaveMentalRepresentations,”inMidwestStudiesinPhilosophy,vol.X,ed.P.French,T.Uehling,Jr.,andH.Wettstein,UniversityofMinnesotaPress.1987a:“Frames,Fridgeons,SleepingDogsandtheMusicoftheSpheres,”inZ.Pylyshyn(ed.)TheRobot’sDilemma;TheFrameProbleminArtificialIntelligence,Norwood,NJ:Ablex.1987b:Psychosemantics:TheProblemofMeaninginPhilosophyofMind,Cambridge,MA:MITPress.1988a(withPylyshyn,Z.):“ConnectionismandCognitiveArchitecture,”Cognition28/1–2,pp.3–71.1988b:“AReplytoChurchland’s‘PerceptualPlasticityandTheoreticalNeutrality’,”PhilosophyofScience55,pp.188–98.1989:“WhyShouldtheMindbeModular?,”inA.George(ed.)ReflectionsonChomsky,Oxford:BlackwellPublishers.1990a:“Psychosemantics,orWheredoTruthConditionsComeFrom,”inW.Lycan(ed.)MindandCognition,Oxford:BlackwellPublishers.1990b:ATheoryofContentandotherEssays,Cambridge,MA:MITPress.1991a(withMcLaughlin,B.):“ConnectionismandtheProblemofSystematicity:WhySmolensky’sSolutionDoesn’tWork,”Cognition35/2,pp.185–204.1991b:“Replies,”inMeaninginMind:FodorandhisCritics,ed.B.LoewerandG.Rey,Oxford:BlackwellPublishers,pp.255–319.1991c:“YouCanFoolAllofthePeopleSomeoftheTime,EverythingElseBeingEqual:HedgedLawsinPsychologicalExplanations,”Mind100/1:pp.19–34.1992a(withLePore,E.):Holism:AShopper’sGuide,Oxford:BlackwellPublishers.1992b:“SubstitutionArgumentsandtheIndividuationofBeliefs,”inG.Boolos(ed.)EssaysforHilaryPutnam,Oxford:BlackwellPublishers.1994:TheElmandtheExpert,Cambridge,MA:MITPress.1998a:Concepts:WhereCognitivePsychologyWentWrong,Oxford:OxfordUniversityPress.1998b:InCriticalCondition,Cambridge,MA:MITPress.2000:TheMindDoesn’tWorkThatWay:TheScopeandLimitsofComputationalPsychology,Cambridge,MA:MITPress.465\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200138SaulKripke(1940–)DAVIDSOSALifeKripkeoncesaid,“Peopleusedtotalkaboutconceptsmore,andnowtheytalkaboutwordsmore....SometimesIthinkit’sbettertotalkaboutconcepts.”Infact,Kripkehimselfhassaidimportantthings,anddevelopedanddeployedsignificantconceptualresources,aboutbothwordsandconcepts.SaulAaronKripkewasborninBayShore,NewYork.HismotherDorothywasateacherandfatherMyerarabbi.ThefamilysoonmovedtoOmaha,NebraskawhereKripkespentmostofhischildhood.Hewasachildprodigy,learningHebrewonhisownattheageof6andreadingallofShakespeareinthefourthgrade.Butitwasinmath-ematicsthatheexhibitedthegreatestprecocity:hederivedresultsinalgebra–intui-tively,withoutthebenefitofalgebraicnotation–infourthgradeandtaughthimselfgeometryandcalculusbytheendofelementaryschool.Bythetimehewasinhighschool,Kripke’sworkinmathematicallogicwassoadvancedthathepresentedsomeofitataprofessionalmathematicsconference.Aroundthetimehepublishedhisfirstarticle,“ACompletenessTheoreminModalLogic,”KripkewasonhiswaytoHarvard,fromwhichhegraduatedwithabachelor’sdegreeinmathematicsin1962.ButduringhisyearsatHarvard,Kripke’sinterestsalreadybegantoshifttophilosophy.In1963KripkewasappointedtotheHarvardSocietyofFellowsandlatertoposi-tionsaslectureratPrincetonUniversity(1965,1966)andbackatHarvard(1966–8).Finally,hewasappointedAssociateProfessoratRockefellerUniversityin1968andpro-motedtoProfessorin1972.ButtheoutstandingphilosophydepartmentatRockefellerwasdisbanded(bytheUniversity’sPresident,FrederickSeitz)inthemid-1970sandKripkewasappointedMcCoshProfessorofPhilosophyatPrincetonin1977,theposi-tionfromwhichheretiredin1999.ModallogicEarlyinhiscareer,Kripkemadeessentialandseminalcontributionstomodallogic.Modallogicis,ineffect,thelogicofnecessityandpossibilityanditshistorycanbetracedtoatleastAristotle.Inthefirsthalfofthetwentiethcentury,C.I.Lewis,C.H.Langford,andthenCarnaprevivedanddevelopedmodallogic.Lewiscriticizedthelogicalsystem466\nSAULKRIPKERussellandWhiteheadhadproposedinPrincipiaMathematica(whichcouldnotdistin-guishwhatissimplyfalsefromwhatisnecessarilyfalse–thatis,whatisimpossible).WithLangford,Lewisdescribedfivedifferentaxiomsystemsthatcouldrepresentanewconceptoflogicalentailment:strictimplication.Unlikethenotionofimplicationfor-malizedinPrincipia,pdoesnotgettostrictlyimplyqsimplyinvirtueofbeingfalse:ithastobeimpossibleforptobetrueandqfalse.Carnaplatercharacterizedthesortoflogicalnecessityinvolvedinstrictimplicationintermsoftruthinall“statedescrip-tions.”ButKripke,withhis“Kripkemodels,”madethisideaofnecessityprecise,refinedit,andgeneralizedit.Kripkemodelsinvolveasetof“possible”worldsand,foreachworld,anassignmentoftruth-valuestosimple(“atomic”)sentences.AsdevelopedbyKripke,thissystemenablesustocharacterizethenotionoflogicalnecessitythatCarnapdiscussed(seeCARNAP):necessarytruthsarethosethataretrueatallpossibleworldsineverymodel.Byincludinginadditionan“accessibilityrelation”(meanttoselecttheworldsthatarepossiblerelativetoanygivenpossibleworld),KripkewasableflexiblyandsystematicallytocharacterizemanyothermodallogicsthatareweakerthanthatsuggestedbyCarnap’sdiscussion.Indeed,muchofthelaterprogressofmodallogichasdependedontheideaofKripkemodels,aswellasonthenotionof“Kripkeframes,”whicharejustlikeKripkemodels(specifyingasetofpossibleworldsandanaccessibilityrelation)butwithouttheevaluationofatomicsentences.MeaningAfterthisimportantworkinmodallogic,Kripketurnedhisattentiontothephilosophyoflanguage,revolutionizingthatfieldwithaseriesofpublicationsintheperiodbetween1971and1982.In“IdentityandNecessity”andtheearly,articleversionof“NamingandNecessity,”KripkebeginstodeveloptheexcitingideasandargumentsthatgettheirfullesttreatmentinthebookversionofNamingandNecessityin1980.Theseworkschallengelong-heldassumptionsaboutmeaningwhilerehabilitatingothers,offeranewparadigm(or“picture,”touseKripke’sterm)ofreferenceandmeaning,andpropose,onthebasisofthedevelopingtheoryofmeaning,provocativethesesinmetaphysics,epistemology,andphilosophyofmind.Fundamentally,Kripkearguesthatatraditionalviewofmeaningismistaken.InthetraditionKripkeseesasbeginningwithFregeandRussell(seeFREGEandRUSSELL),names,forexample,refertowhattheydoinvirtueofbeingassociatedwithsomedescriptivecontent.Thereferentofthenameiswhatsatisfiesthedescriptivecontentassociatedwithit.Withanamesuchas,say,“Aristotle,”onemightthinkthedescrip-tivecontentwouldincludetaughtAlexanderorwasthestudentofPlato,andsoon.Kripkepresents,incompellingform,abatteryofargumentsagainstanysuchview.Theseargumentscanprofitablybeseenascominginthreevarieties:(1)modal,(2)semantic,and(3)epistemic.ThemodalargumentbeginswithanobservationforwhichKripkeisnowcelebrated:namesarerigiddesignators.Arigiddesignatorisawordthatdesignatesthesameobjectwithrespecttoanypossiblesituation.So,forexample,wemaysaythatifhehadbeenchosentoleadtheAcademy,AristotlewouldneverhavegoneontoteachAlexander.Whenwemakethatstatement,it’saclaimaboutasituation(orwhatcanalsobecalled467\nDAVIDSOSAa“possibleworld”)that’sdifferentfromourown;inourworld,AristotlewasnotchosentoleadtheAcademyafterPlato.Buteventhoughwe’retalkingaboutadifferentsitua-tion,we’retalkingaboutAristotleinthatsituation.Sothename“Aristotle”maintainsitsreferencetoAristotle,evenwithrespecttopossiblesituationsinwhichAristotlewaschosentoheaduptheacademy,didnotgoontoteachAlexander,orinwhichhislifevariedinanyofthewaysitmighthave.Butnoticethatsincenamesarerigiddesignators,wecanmaketrueclaimsaboutwhatmighthavehappenedthatwouldappeartoberuledoutbythedescriptiontheoryKripkeopposes.Consideranydescriptionwemightthinkispartofthedescriptivecontentofthename“Aristotle”:say,wasborninStagira.Aristotle,ofcourse,mighthavebeenbornelsewhere,ifhisparentshadmovedbeforehewasborn,forexample.Ontheotherhand,noonecanbothbebornandnotbeborninStagira.Sowhilethesentence,“AristotlewasnotborninStagira,”seemstoexpresssomethingthat’spossible,anysentencelike“The...whowasborninStagirawasnotborninStagira,”seemstoexpresssomethingimpossible.Butifpartofwhat“Aristotle”meansiswasborninStagira,thenit’shardtoseewhythesetwosentencesshoulddifferinthisway.Whyiswhat’sexpressedbyonesentencepossibleandwhat’sexpressedbytheotherimpos-sible,whentheyhave,relevantly,thesamemeaning?ThisisanexampleofKripke’smodalargument.Thereareseveralotherwaysofputtingthepointofthemodalargument.Buttheycanbeseenasreducingtoageneralpattern:namesarerigiddesignators,descriptionsarenot;thereforedescriptionscannotgivethemeaningofnames(inthewayproposedbythetraditionalviewofFregeandRussell).Nameshaveadifferentmodalprofilefromdescriptions.EvenifKripkehadgivennoneother,manywouldfindthemodalargumentsuffi-cientlydevastatingtorefutethetraditionalviewofnamesatwhichit’sdirected.ButanimportantpartofthesignificanceofKripke’sworkonmeaningisthathepresents,asnotedabove,abatteryofarguments,eachofwhichisafurther,independentpointagainstthetraditionalviewhechallenges.Kripke’sepistemicargumenthasastructuresimilartothatofhismodalargument.IfwasastudentofPlato’sisliterallypartofthemeaningofthename“Aristotle,”thenweshouldexpectthesentence“AristotlewasastudentofPlato’s”toexpressatrivialaprioritruththatcouldbeknownwithoutanyhistoricalorempiricalinvestigations.Butyoumightbeacompetentuserofthename“Aristotle”withoutknowingthatAristotlewasastudentofPlato’s.PerhapsallyouknowisthatAristotlewassomegreatphiloso-pher.Thedescriptiontheorypredictsthatcertainsentencesshouldbeaprioriwheninrealitytheyarenot.AndKripke’ssemanticargumentssuggestthatthereferentofanameisnotwhat-eversatisfiesthedescriptionsthatmightbeassociatedwithit.Heisaidedherebycom-pellingexamples.Inone,KripkeasksustoimagineacircumstanceinwhichKurtGödeldidnotdiscovertheincompletenessofarithmetic(as,infact,hedid),butratherstolethatresultfromsomeonenamed“Schmidt.”Now,it’splausiblethatsomethinglike“dis-coveredtheincompletenessofarithmetic”isassociatedwiththename“Gödel.”Butnoticethatinthiscase,thatwouldyieldSchmidtasthereferentofthename“Gödel.”Kripkeusesthisasanargumentagainstthedescriptiontheory.SurelyevenwithrespecttoasituationinwhichSchmidtisthediscovereroftheincompletenessofarithmetic,468\nSAULKRIPKE“Gödel”referstoGödelandnottoSchmidt.Butthatmeansthename“Gödel”isnottiedtoitsreferentbymeansofthesatisfactionofthedescriptiondiscoveredtheincom-pletenessofarithmetic.Ifthemeaningof“Gödel”werethedescriptivecontentassoci-atedwithit,thenthenamewouldrefertothewrongperson–itwouldhavethewrongsemantics.Anotherexample:afamousphysicistpicksoutGell-MannasmuchasitdoesFeynman.Still,evenifthat’stheonlydescriptivecontentassociatedwiththename“Feynman,”thenamereferstoFeynmanandnottoGell-Mann.AcknowledgingadebttoJ.S.Mill,Kripkeholdsthatnamesaredenotativebutnon-connotative.Themeaningofanameisexhaustedbyitsreferent.Ratherthanhavinganydescriptivecontentasitsmeaning,adescriptivecontentthatwouldthendetermineareferent,Kripkesuggeststhatthemeaningofthenamejustisthereferentitself.Thisclaimisnowconsideredconstitutiveofapositionknownas“Millianism”inphilosophyoflanguage.Thisleavesopenthequestionofwhyanamehasthereferentitdoes.InplaceofthedescriptiontheoryheassociateswithFregeandRussell,Kripkeoffersanalternative“picture”ofthenamingrelation.Inthecausalaccounthesuggests(sometimescalledthe“historicalchain”account),anamehasthemeaningitdoes–thatis,itrefersasitdoes–invirtueofachainofcausalrelationsbetweenusesofthenameandthereferent.Kripkeexplicitlyadmitsnothavinganythinglikea“theory”:butheproposescausationasthefundamentalmechanismbywhichreferenceisfixed(thoughthesecausalrelationsdonotthemselvesconstitutethemeaning;themeaning,recall,justisthereferent).It’saninterestingfactthatalthoughheattacksadescriptivetheoryofnamingassociatedwithRussell,inotherworkKripkeingeniouslydefendsRussell’stheoryofdescriptionsthemselves.AccordingtoKripke,Russellwaswrongtoviewnamesonthemodelofdescriptions;buthisaccountofdescriptionsthemselveswasunobjectionable.Russell’stheoryofdescriptions(in“OnDenoting”)concernedthemeaningofexpres-sionssuchas“thePresident”or“Theevenprimenumber,”oreven“Plato’smostfamousstudent.”In1966,thephilosopherKeithDonnellanissuedachallengewithanexampleinwhichasentencecontainingadefinitedescriptionseemedtohaveameaningthatwasinconsistentwithwhatwouldbepredictedbyRussell’stheory.Drawingonadistinctionbetweenlanguageuseandlanguagemeaning,anddistin-guishingbetweenspeakerreferenceandsemanticreference,KripkeanswersDonnellan’schallengeanddefendsRussell’stheoryofdescriptions.OneseriousproblemforthesortoftheoryKripke’sargumentssupport(though,again,Kripkehimselfneverexplicitlyadoptsanyparticular“theory”)concernsbeliefandbeliefascription.Ifnamesaremerelydenotativeandarenon-connotative,then,sincethemeaningofanameisexhaustedbyitsreferent,anytwonameswiththesamereferenthavethesamemeaning.Butgivenjustafewotherplausibleassumptions,thisentailsthatthereshouldbenodifferenceinmeaning(andthusnodifferenceintruth-value)betweensentenceslike“LoisLanebelievesClarkKentcanfly”and“LoisLanebelievesSupermancanfly.”But(amongotherproblems)itseemsthatwhatLoisreallybelievesisthatClarkKentcannotfly.Inhis“APuzzleAboutBelief”(1979),Kripkearguesthatthisunwelcomeresultisnotduetoanyfeaturesspecifictothepositioninquestion:ourpracticesofbeliefascrip-tionthemselves,independentofanyspecificassumptionsaboutthemeaningofnames,469\nDAVIDSOSAwillyieldthesameunwelcomeresults.Heusesthenow-infamous(inphilosophyofmindandlanguage!)exampleofPierre,anormalmonolingualFrenchman,whohearsofthatfamousdistantcity,London(whichPierreofcoursecalls“Londres”).OnthebasisofwhathehasheardofLondon,heisinclinedtosay,inFrench,“Londresestjolie.”Takinghimathisword,andtranslating,wecanconcludethathebelievesthatLondonispretty.Later,PierreleavesFranceandmovestoanunattractivepartofLondon.HelearnsEnglishbythe“directmethod,”withoutusinganytranslationbetweenEnglishandFrench.PierreisunimpressedwithhissurroundingsandisinclinedtoassenttotheEnglishsentence“Londonisnotpretty.”Again,takinghimathisword,wecanconcludethathebelievesthatLondonisnotpretty.ButnowheseemstobeinmuchthesamepositionasLoisabove.What’simportant,forKripke’spurposes,isthatweseemtohaveputPierreintothatpositionwithoutexplicitlyappealingtoa“Millian”(namesaremerelydenotative)posi-tion.ThatsuggestsMillianismisnotadistinctivelyproblematicposition.Thesortofpuzzlethat’sputforwardagainstMillianismisreallyaproblemforeveryone,Kripkeargues.ThushedefendsMillianismfromitsmainchallenge.Necessity,apriority,themind–bodyproblem,andessentialismKripke’srevolutioninphilosophyoflanguagewouldhavebeenmorethanenoughtosecurehisimportance.ButKripkewentontotransformhisthesesaboutmeaningintointerestingpositionsinmetaphysics,epistemology,andphilosophyofmind.Perhapsthemostsignificantelementofhismeaningtheory,forthesepurposes,ishisdistinctionbetweenwhat“fixesthereference”ofaterm(whichfornames,hesuggests,istypicallyfundamentallyacausalrelation)andtheactualmeaningofthatterm(which,inthecaseofnames,consistsofthereferentitself).SinceImmanuelKantinthelate1700s,philosophershadtraditionallyseentwosortsofphenomenaasintimatelyrelated.Apropositionwastakentobenecessaryifitcannotpossiblyfailtobetrue,andcountedasaprioriif,roughly,itcanbeknownwithoutthebenefitofempiricalinvestigation.Itwasnaturaltothinkthatallnecessarypropositionsareaprioriandthat,withafewspecialexceptions,thosethatarenotnec-essarycanbeknownonlyaposteriori.Ifapropositionisnecessary,thenoneneedn’tseehowtheworldisasamatteroffactinordertoknowthatproposition.Itstruthdoesnotdependonthestateoftheworld;empiricalinvestigationthusseemsbesidethepoint.And,conversely,ifapropositioniscontingent,thenhowcoulditbeknownapriori?Sinceit’snottrueineverypossibleworld,wewouldhavetoinvestigatetheworldaroundustoseewhetherit’strueinours.(OneexceptionisDescartes’sCogito–Ithink,thereforeIam–whosepremise,andconclusion,eachseemcontingentandyet,inonesense,apriori).Shockingly,Kripkerejectedbothdirectionsofthisallegedintimaterelation.AccordingtoKripke,necessityandapriorityarenotnearlyasintimatelyrelatedashadbeenthought.Therearenecessarytruthsthatcanbeknownonlyaposterioriandaprioritruthsthatarecontingent.Andthesearen’tjustexceptional,unusualcases,butsystematic,standardoccurrences.ConsideranexampleKripkeuses,pickinguponacommentofWittgenstein’s,tosupporthisclaimthatwhat’saprioricanbecontin-470\nSAULKRIPKEgent.Weintroducedtheword“meter”andfixeditsreferencewithrespecttoacertainstandard:thestandardmeterbarinParis.(Thereferencehassincebeenre-fixed,butsetthataside.)Nowtaketheclaimthatthestandardmeterisonemeterlong.Howcanweknowthis?Theideaofmeasuringthestandardmeterisludicrous:ourknowledgethatthestandardmeterbarisonemeterlongisnotthesortofthingthatistobecheckedempirically.Thestandardmeterispreciselywhatfixesthereferenceoftheterm“meter.”Butisitanecessarytruththatthestandardmeterisonemeterlong?Kripkeremindsusthatthestandardmeterbarmighthavebeenlongerthanitinfactis.Indeed,ifjustbeforewefixedthemeaningofourword“meter”withreferencetothatbar,ithadundergonesomesignificanttemperaturechange(thatitdidnot,asamatteroffact,undergo),thenthebarwouldhavebeenlonger(orshorter)thanameter.Ofcourse,inthatcircumstance,we’dusetheword“meter”forthatnewlength.Butit’sstilltruethatthemeterbarinthatcircumstancewouldn’tbeameterlong:we’djustbeusingtheword“meter”foradifferentlength.Weknowapriorithatthestandardmeter’sameterlong;butitmightnothavebeen.Therearepossiblecircumstancesinwhichthestandardmeterbarhasadifferentlength.Conversely,necessitydoesnotentailapriority.Goldhasatomicnumber79andwaterisH2O.AccordingtoKripke,thesearenotthingswecouldhaveknownapriori.Thechemicalcompositionofwaterandtheatomicnumberofgoldwereempiricalsci-entificdiscoveries.Weusedsomesuperficialidentifyingmarkstofixthereferencesofourterms“water”and“gold.”Now,thosemarksdon’tdefinethewords,theydon’tgivetheirmeanings.Theyservedtopickoutkindswhichwetheninvestigatedempirically.Butitisthroughempiricalinvestigationthatonediscoversgold’satomicweightandwater’schemicalcomposition.Nevertheless,Kripkethinksthestatements“goldhasatomicnumber79”and“waterisH2O”arenecessary.There’snopossiblecircumstanceinwhichgoldhasanyatomicnumberotherthan79;andwatercouldn’tbeanythingbutH2O.Theremaybecircumstancesinwhichwhatwecall–inthosecircumstances–“gold”hasadifferentweight,orinwhichwhatwecall“water”hasadifferentchemi-stry,butthosearejustworldsinwhichweusethetermsforotherstuff.(Ofcourse,that’snottosaywe’dbemakingamistakeincallingthatotherstuff“gold”or“water”:inthoseothercircumstances,thewordswouldn’thavethesamemeaningtheyactu-allyhave.)AccordingtoKripke,it’samatterofnecessitythatwaterbeH2Oandthatgoldhaveatomicnumber79.Havingthosechemicalnaturesiswhatmakeswaterandgoldwhattheyare.Sciencecandiscoveressences.ButKripkewasn’tfinishedyet.Beforeclosinghisworkonthesematters,hetakesontwoothershibboleths:(1)atthetimehewroteNamingandNecessity,apopularresponsetothemind–bodyproblem–thetraditionalphilosophicalproblemofthenatureofmindanditsrelationtothephysicalbody–wasakindof“identitytheory.”Theideawastoviewtheproblemassolvedbycontemporaryscienceinmuchthesamewaythatcontemporarysciencehaddiscoveredthenatureof,forexample,heat.Wecansupposethatheatwasoriginallyidentifiedaswhatproducesacertaindistinc-tivesensation.Throughempiricalinvestigation,wefindthatitisthekineticmotionofmoleculesthatproducesthosesensations.So,roughly,heatisthemotionofmolecules.Thethen-popularidentitytheorywantedtoviewtherelationofmindtobodyasakintothatbetweentemperatureandmeanmolecularkineticenergy.Asweinvestigatethebrainfurther,anddiscoverwhichstatesarecorrelatedwithwhich471\nDAVIDSOSAmentalphenomena,welearnwhatthesementalstatesare,justaswelearnedwhattemperatureis.Takepain.Thementalstateofpainappearstobecorrelatedwiththestimulationofwhatarecalled“C-fibers.”Isthatjustwhatpainis?Havewesolvedthemind–bodyproblem?KripkepointsoutthatifweweretotakethementalstateofpaintojustbethestimulationofC-fibers,thenthatwouldconstitutetheempiricaldiscoveryofanecessity,onthemodelofthediscoveryofthechemicalconstitutionofwater(remem-ber:“waterisH2O”isnecessary)orthenatureoftemperature.Butthere’saproblem.Inthesecasesoftheoreticalidentification,ofthescientificdiscoveryofnecessity,thereisanexplanatorynotetobepaidoff:whatexplainstheillusionofcontingency?ForitcertainlyseemsthatwatermighthaveturnedoutnottobeH2O.Aswewereperform-ingthechemicalinvestigations,atleast,itseemedtobeacontingentmatter,possiblyturningoutoneway,possiblyanother.Thereisastandardwaytomakegoodonthisexplanatorydebt:theidentifyingmarksbywhichwefixedthereferenceoftherelevanttermsare,indeed,onlycontingentlyrelatedtotheessenceofthekinds.Sobeingthecolorless,odorlessliquidthatfallsfromtheskyasrain,etc.–thatsetofpropertiesbywhichweidentifywater–isonlycontin-gentlyrelatedtobeingwater.Watermighthaveexistedwithouthavingthoseidentify-ingmarks.SoalthoughwatermustbeH2O,itcanseemasthoughitneednothavebeen,becauseH2Oneednotbeacolorless,odorlessliquidthatfallsfromtheskyasrain,etc.Similarlywithheat.Heatisnecessarilymeanmolecularkineticenergy;butit’snotanecessarytruththatmeanmolecularkineticenergyproducesthesensationofheat.Thatsensationisjustamarkthatweusedtoidentifythephenomenontobeinvestigated.NowcomesKripke’sinsight:inthecaseofpain,there’snoanalogousmove!Themarksbywhichweidentifypainareessentialtoit;paincouldnotexistwithoutbeingfeltaspain.SoifthestimulationofC-fiberscouldoccurwithoutbeingfeltaspain,thiswouldrefutethemind–brainidentitytheory.PainappearstobeonlycontingentlyrelatedtothestimulationofC-fibers.Theidentitytheorymust,accordingtoKripke,denythatappearanceasmistaken.Butitcannotexplainitsplausibilityasitdoesintheanalogouscases.Forthemarkbywhichweidentifypain,itspainfulfeeling,isessentialtopain.ThatleadsustotheothershibbolethKripkeattacked:anti-essentialism.(2)inthe1960sandintothe1970s,influencedbyQuineamongothers,manyphilosopherswereopposedtoessentialism–beliefinmodalitydere–whileacceptingmodalitydedicto(seeQUINE;cf.MARCUS).Inotherwords,itwaswidelyacceptedthatstatementscouldbenec-essarilyorpossiblytrueorfalse(modalitydedicto)butwidelydeniedthatitmadesensetospeakofaparticularindividual’snecessarilyoronlycontingentlyhavingagivenproperty(modalitydere).Kripkearguesthatamaterialobject’smaterialorigin(thestufffromwhichitwasmade)isessentialtoit:itcouldnothavebeenmadefromany-thingelse.Andhearguesthatonehasone’sparentsessentially,sothatonecouldnothavehaddifferentparents.Thesearederenecessities;propertiesthatindividualshavenecessarily.It’struethatthemethodandforceofhisargumentationhere,aselsewhere,islargelyintuitive;butKripkeholdsthatalthough“somephilosophersthinkthatsome-thing’shavingintuitivecontentisveryinconclusiveevidenceinfavorofanything”hehimselfdoesn’t“know,inaway,whatmoreconclusiveevidenceonecanhaveaboutanything,ultimatelyspeaking”(1980:42).472\nSAULKRIPKETruthInhisgroundbreaking“OutlineofaTheoryofTruth,”Kripkemakesanumberofimportantadvancesinourtheoreticalunderstandingoftruth.Thepaperquicklybecameafocusofallsubsequentdiscussions.Amainproblemforourunderstandingoftruthispresentedbytheso-called“liarparadox.”Considersentence(1):(1)Sentence(1)isfalse.Issentence(1)trueorisitfalse?Well,exploitingtheattractiveideathatasentenceistruejustincasewhatitsaysholds,wemightsupposethatsentence(1)–thatis,“Sentence(1)isfalse”–istruejustincasesentence(1)isfalse.Butnowwehaveaproblem:forwearesayingthatsentence(1)istrueifandonlyifitisfalse.Indeed,whethersentence(1)istrueorfalse,itfollowsthatit’sbothtrueandfalse!Tarskiconfrontedthisparadox,orineffectametalinguisticversionofit,andcon-cludedthatlanguagesforwhichtheparadoxarisesare“inconsistent”:theyarelan-guagesinwhichasentenceanditsnegationarejointlytrue.Hesuggestedthatsuchlanguageswereinadequateforatheoryoftruthandproposedreplacingthemwithmoreregimentedlanguages,whoserulespreventedtheparadox.Tarskiproposedahierarchyoflanguages,noneofwhichcontainsa“truthpredicate”thatappliestosentencesofthatverylanguage(atthatsamelevelofthehierarchy)(seeTARSKI,CHURCH,GÖDEL).Atruthpredicate(foralanguageL)isanypredicateTwhichmakesthefollowingschematrueforallinstances(whereoneobtainsaninstanceoftheschemabyreplacing“S”withasentenceofL):ÈÈ˘˘È˘TSistrueifandonlyifSistrue.Byprohibitingtheapplicationofatruthpredicatetosentencesinthesamelanguage,Tarskipreventstheconstructionoftheliarparadox;buthegivesuptheideathattherecanbeasatisfactorytheoryoftruthforEnglish(whichapparentlydoeshaveatruthpredicate–namely,“istrue”–thatappliestosentencesinthesame,English,language).Kripkeshowshow,ifweallow“truth-valuegaps”(i.e.ifweallowsentencesthatareneithertruenorfalse)wecanmakeprogress.Butwedonotsimplyeliminatetheparadoxbyallegingthat(1)isneithertruenorfalse,becausewecanreadilyseethattheparadoxwillrearise,instrengthenedform,withthesentence:(1¢)Sentence(1¢)isnottrue.Perhaps(1¢)isneithertruenorfalse;butthenit’snottrue.Inwhichcase,sincewhatitsaysisthatit’snottrue,itmustbetrue.Sotheparadoxrearisesevenifweallowtruth-valuegaps.Tomakeprogress,Kripkeintroducesthenotionofa“grounded”sentencewithref-erencetothenotionofa“fixedpoint”(which,veryroughly,isaninterpretedlanguagewhoseinterpretationassignstothetruthpredicateallandonlythetruesentencesofthatlanguage).Thisnotionofgroundednessisusefulbecause,accordingtothecon-structiveprocedurebywhichitisunderstood,noteverysentenceofalanguagewillbegrounded.Somesentences,like(1)or(1¢),maybeungrounded,notpartoftheexten-sionofthecandidatetruthpredicate,butalsonotpartoftheanti-extensioneither(wheretheanti-extensionincludesjustthosesentencestowhich,accordingtothe473\nDAVIDSOSAinterpretation,thetruthpredicatedoesnotapply).Thepointisthattheinterpreta-tionmaybepartial,somesentencescharacterizedasonestowhichthecandidatetruthpredicateapplies,otherscharacterizedassentencestowhichthepredicatedoesnotapply,andotherssimplyleftuncharacterized.Kripkesuggeststhatanungroundedsen-tencefailstoexpressapropositionandthisrelievessomeofthephilosophicaldiseaseassociatedwiththeliarparadox.Sentence(1¢)doesnotsaythatsentence(1¢)isnottrue:itdoesn’tsayanything.Ittries,andfails,toexpressaproposition.ItisimpossibleheretoexpoundallofthetechnicaldetailsofKripke’stheory.AmorethoroughpresentationwouldemphasizerelationsbetweenKripke’sviewandKleene’sthree-valuedlogic,woulddiscussKripke’s“fixedpointtheorem”accordingtowhich,givencertainconstraints,therewillbea“minimalfixedpoint,”andwoulddetailthewayinwhichKripke’stheoryexplains(grounds)thetruth-valuesofthosesentencesthathavethem.MuchofthesignificanceofKripke’sworkliespreciselyinthosedetails.Butacoupleofpointsshouldbenoted:toeverysentencetowhichitassignsavalue,Kripke’sconstructionassignstheintuitivelyattractivevalue.Anditsfailuretoassignanyvaluetocertainproblematicsentenceshasanimportantphilosophicalpayoff.Still,thereareproblems.Forexample,thesortofconstructionKripkeproposesfailstoassigntruth-valuestosentencesweintuitivelyexpecttohaveone.Generalizationssuchas“everytruesentenceisatruesentence”areungroundedandleftwithoutatruth-value.AttemptstoextendKripke’stheorytoprovideintuitivelyattractivetruth-valuesforsuchsentencesthreatentounderminethebasicintuitionofgroundednessthatgivesKripke’stheorymuchofitsforce.Andthoughtheideathatungroundedsentencesdonotexpresspropositionscouldinprinciplebeeliminated,withoutthatclaimthetheory’sresponsetotheparadoxlosesmuchofitsphilosophicalattraction.SubstitutionalquantificationInhisclassic,“IsThereaProblemaboutSubstitutionalQuantification?”(1976),Kripkeestablishesanumberofimportantresultsaboutsubstitutionalquantification.Quantification(or“generalization”),whichcanbeexistentialoruniversal,involvessomeschema’sbeingtrueinatleastonecase(existential)orineverycase(universal).Butwhatisa“case”?Thisquestioncanintroducethedifferencebetweensubstitutionalandtheperhapsmorefamiliarobjectual(orsometimes“standard”or“referential”)quan-tification.Atrueobjectualexistentialquantificationrequiresthattherebesomeentityofwhichtheschemaistrue.Atruesubstitutionalexistentialquantification,bycon-trast,requiresthatsomeexpressioncanbesubstitutedforthevariableintheschematoproduceatruesentence.Importantdifferencesbetweenobjectualquantificationandsubstitutionalquantificationarisemostclearlywheneither(1)somenamesare“empty”(thereisnothinginthedomainofdiscourseofwhichtheyarethenames),or(2)noteveryentityinthedomainofdiscoursehasaname.Becausethetruthofgeneralizations,whentheyarereadsubstitutionally,canseemnottorequiretheexistenceofentitiesintherelevantdomainofdiscourse,substitu-tionalquantificationpromisedtosomephilosophersanattractive“ontologicalneu-trality.”Wecouldsayforexamplethat“everyevennumberisdivisiblebytwo”withoutexplicitlycommittingourselvestotheexistenceofevennumbers,ifweweresodisin-clined.Butintheearly1970s,papersbyJ.WallaceandL.Tharpchallengedsomeof474\nSAULKRIPKEtheallegeddistinctivevalueofsubstitutionalquantification.Kripkerefutedanysuspi-cion,whichsomedrewfromtheargumentsofWallaceandTharp,thatsubstitutionalquantificationisunintelligibleorthatintelligiblyinterpreteditreducestoobjectualquantification.Remindingusthatsubstitutionalquantificationpresupposesthenotionofasubsti-tutionclass(theclassofitemsthatcanbesubstitutedforthevariableboundbythesub-stitutionalquantifier),Kripkeemphasizesthattheitemsintheclassmustnotincludetheverysubstitutionalquantifieritself.Manyofthealleged“paradoxes”surroundingsubstitutionalquantificationresultfromignoringthisrequirement.Kripkethenshowsthatitispossible,andinsomecasestrivial,togive(finitelyaxiomatized)theoriesoftruth(inDavidson’ssense)forlanguagescontainingsubstitutionalquantifiers.Theoriesoftruthbasedonsubstitutionalquantificationcan,Kripkeshows,satisfyTarski’sConventionT.Moreover,Kripkeshowsthatinsomecases,asubstitutionalinterpretationofthequantifierswillbeequivalenttoareferentialinterpretation.Inthesecases(whichwillincludeallfirst-orderlanguageswithoutidentity),whetherthequantifiersareinterpretedsubstitutionallyorobjectuallywillmakenodifferencetowhichformulaearesatisfied.Butthisnomoreeliminatesthedifferencebetweensub-stitutionalandreferentialquantificationthandoesthelogicalequivalenceof“PandP”and“PorP”eliminatethedifferencebetweendisjunctionandconjunction.AlthoughKripkeshowsthatthereisnoproblemaboutsubstitutionalquantification,heisskepticalaboutitsroleforinterpretingnaturallanguage.Forexample,Kripkedoesnotthinktheviabilityofsubstitutionalquantificationhasanybearingonwhethertheordinaryexpressions“thereis”or“thereexists”typicallycarryontologicalcommit-ment(indeed,heisconcernedabouttheveryintelligibilityofthe“issue”ofontologi-calcommitment).Moreover,ordinaryexistentialassertionsappeartomakenocommitmenttonameability,aswouldberequiredforsuchquantificationtobeinter-pretedsubstitutionally.Attheendofhispaper,Kripkedrawsaseriesofvaluablemetaphilosophicalmorals.WittgensteinonfollowingaruleInhisinfluentialWittgensteinonRulesandPrivateLanguage(1982),KripkeattemptsanexpositionofWittgenstein’sso-calledprivate-languageargument.KripkelocatesthatargumentearlierinWittgenstein’sPhilosophicalInvestigationsthanwascommonatthetime,earlier,thatis,thaninthesectionsthatbeginwithandfollow§243.In§201,Wittgensteinsays,“thiswasourparadox:nocourseofactioncouldbedeterminedbyarule,becauseeverycourseofactioncanbemadeouttoaccordwiththerule.”Bystartingwiththispassage,Kripkewillemphasizethecentralityfortheprivate-languageargumentofWittgenstein’sconsiderationsonrulefollowing(seeWITTGENSTEIN).Considertheword“plus”orthesymbol“+.”Weusethesetoexpressthemathe-maticalfunctionofaddition.Ofcoursethereareinfinitelymanypossiblesums:nofinitebeingcouldeverperformthemall.Considernowsomesumthatwehaveneverper-formed:Kripkeconsiders68+57.Ofcourse,thatsumis125.Butwecanimagineaskepticchallengingus.Howdoweknowwe’refollowingthesameruleforaddingaswe’veusedinthepast?Whyarewesoconfidentthatwehavealwaysused“+”withtheimplicitintentionthat68+57shouldturnouttostandfor125?AccordingtoKripke,475\nDAVIDSOSAtheskepticintroducesthepossibilitythatwithallofthose(finitelymany)pastuseswereallyexpressedadifferentfunction,the“quus”function(or“quaddition”),whichisdefinedtoequalx+ysolongasx,y<57,andtoequal5otherwise.Sotheskepticchal-lengesusforsomereasontobelievethat,inordertoaccordwithourpastusesof“plus,”weshouldnowsay“68+57equals125”ratherthan“68+57equals5.”Ifwereallydidalwaysuse“plus”forquadditionratherthanforaddition,theninordertodoto68and57whatwehaveinthepastdoneto,say,3and5toget8,weshouldnowget5asourresult.Kripkeadmitsthattheskeptic’shypothesis(thatwehavealwaysmeantquusby“plus”)is“ridiculous,”“fantastic,”“bizarre,”and“wild.”Ifheproposesitsincerely,theskepticissurelycrazy.Butthehypothesisisnotlogicallyimpossible.Ifitisfalse,weshouldbeabletocitesomefactaboutourpastusagewhichestablishesthatby“plus”Imeantplusratherthanquus.Theproblemisthatallcandidatefactscanseemtofail.Ourproblemisphilosophical:thequestionisnot“dowemeanplusby‘plus’?”but“invirtueofwhatdowemeanplusby‘plus’?”Ifwehavenoanswertothatquestionthenwemusttakeseriouslythepossibilitythatmeaningisamyth.Ofcourse,inposingtheparadoxweassumethatlanguageismeaningful.Butwemusteventuallykicktheladderaway:ifnofactaboutuscouldsufficeforourhavingmeantplusratherthanquusinthepast(andtheparadoxisasgeneralasitappearstobe),thentherecanbenofactastowhatwemeanbyanythingatanytime.Meaningisanillusion.MuchofKripke’spurposeinthebookistodevelopandsharpentheproblem(thoughhefindsmaterialinWittgensteintosketcha“skeptical”solution).Hedeftlydeflectsseveralimmediateresponses.Andhedevotesasubstantialsectiontodiscussinga“dispositional”responseaccordingtowhichwemeanplusratherthanquusinvirtueofhavingadispositiontoperformvariouscalculationsinspecificways:wearedisposedtogive125,not5,asthesumof68and57.Thereareimmediateproblemssuchas(1)wemightbedisposedtoperformvariouscalculationserroneouslywithoutthereforenotmeaningplusby“plus”and(2)wemighthavesimplynodispo-sitionwithrespecttocertainadditions(ifthenumbersaretoobig,forexample).Butthebasicthreattoanysuchresponse,asKripkemakesclear,isthatjustbecauseIaminfactdisposedtoperformvariouscalculationsinspecificwaysdoesnotmakeitthecasethatIshouldperformtheminthatway.IfIamperformingaddition,Ishouldderive125from68and57,whatevermydispositionsmightactuallybe.InKripke’sterminology,thedispositionalaccountofmeaningplusby“plus”leavesoutthenormativityofmeaning.Kripke’sdiscussionhashelpedmakeWittgenstein’srule-followingconsiderationsacentralissuenotonlyinthephilosophyofmindandlan-guage,butalsointhephilosophyoflaw,wheretheideaofarule’shavingcontentandnormativeforce,withrespecttopreviouslyuncontemplatedcircumstances,ispre-dictablyimportant.BibliographyofworksbyKripke1959:“ACompletenessTheoreminModalLogic,”JournalofSymbolicLogic24,pp.1–14.1963a:“SemanticalAnalysisofModalLogicI,”ZeitschriftfürMathematischeLogikundGrundlagenderMathematik9,pp.67–96.1963b:“SemanticalConsiderationsonModalLogic,”ActaPhilosophicaFennica16,pp.83–94.476\nSAULKRIPKE1965:“SemanticalAnalysisofModalLogicII,”inTheNewTheoryofModels,ed.J.Addison,L.Henkin,andA.Tarski,Amsterdam:NorthHolland,pp.206–20.1971:“IdentityandNecessity,”inIdentityandIndividuation,ed.M.Munitz,NewYork:NewYorkUniversityPress,pp.135–64.(AlsoinNaming,Necessity,andNaturalKinds,ed.S.Schwartz,Ithaca,NY:CornellUniversityPress,pp.66–101.)1972:“NamingandNecessity,”inSemanticsofNaturalLanguage,ed.D.DavidsonandG.Harman,Dordrecht:Reidel,pp.253–355and763–69.1975:“OutlineofaTheoryofTruth,”JournalofPhilosophy72,pp.690–716.(AlsoinRecentEssaysonTruthandtheLiarParadox,ed.R.L.Marin,Oxford:OxfordUniversityPress,1975,pp.53–81.)1976:“IsThereaProblemAboutSubstitutionalQuantification?,”inTruthandMeaning,ed.G.EvansandJ.McDowell,Oxford:ClarendonPress,pp.325–419.1977:“SpeakerReferenceandSemanticReference,”inContemporaryPerspectivesinthePhilosophyofLanguage,ed.P.French,T.Uehling,andH.Wettstein,Minneapolis:UniversityofMinnesotaPress,pp.6–27.1979:“APuzzleAboutBelief,”inMeaningandUse,ed.A.Margalit,Dordrecht:Reidel,pp.239–83.1980:NamingandNecessity,Cambridge,MA:HarvardUniversityPress,andOxford:BlackwellPublishers.1982:WittgensteinonRulesandPrivateLanguage,Cambridge,MA:HarvardUniversityPress.1986:“AProblemintheTheoryofReference:TheLinguisticDivisionofLaborandtheSocialCharacterofNaming,”inPhilosophyandCulture:ProceedingsoftheXVIIthWorldCongressofPhilosophy,Montréal:ÉditionsduBeffroi,pp.241–7.477\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd200139DavidLewis(1941–)ROBERTSTALNAKERIntroductionDavidLewisisaphilosopherwhohaswrittenaboutawiderangeofproblemsinmeta-physicsandthephilosophyofmindandlanguage,includingthemetaphysicsofpos-sibleworlds,theanalysisofcounterfactualconditionals,causationandprobability,theproblemsofuniversals,ofintentionality,ofpersonalidentity,thefoundationsofdeci-siontheory,ofsettheory,ofsemantics.Adistinctiveandcomprehensivemetaphysicaltheoryhasemergedfromhisdiscussionsofphilosophicalproblems:atheorythatcom-binesrealismaboutpossibleworldswithakindofnominalism,amaterialistaccountofmind,andHumeanskepticismaboutunanalyzednaturalnecessity.ButLewis’sdis-cussionshavealsoyieldedconceptualtoolsthathaveapplicationsbothwithinandoutsideofphilosophythatareindependentofthegrandmetaphysicalscheme,forexample,ananalysisofcommonknowledgethathasbeeninfluentialingametheoryandtheoreticalcomputerscience,andworkongeneralizedquantifiersinnaturallan-guageandontheroleofextra-linguisticcontextintheinterpretationofspeechthathasinfluencedthedevelopmentoflinguisticsemantics.LewisstudiedatHarvardwithW.V.QuineandNelsonGoodman,andtheinfluenceofthosetwophilosophersisevidentinhisownphilosophicalmethod,intheproblemshehasfocusedon,andinthesubstanceoftheviewshedefends.ButLewisdevelopedQuineanandGoodmanianthemeswithadistinctivetwistthattakestheminunantici-pateddirectionsandthathasresultedinatheorythatcombinesfeatureshisteacherswouldapplaudwithfeaturestheywouldabhor.ForQuineandGoodman,therejectionoftheanalytic/syntheticdistinctionmotivatedaholisticphilosophicalmethodology,amethodof“reflectiveequilibrium”thathelpedtomakemetaphysicsrespectablefortheheirsofthepositivisttradition(seeGOODMANandQUINE).(Oneconsequenceofaban-doningthetwodogmasofempiricism,Quinewrote,was“ablurringofthelinebetweenspeculativemetaphysicsandnaturalscience,”Quine1953:20.)Lewisadoptedtheholisticmethod,andacceptedtheinvitationtodometaphysicswithaclearconscience,buthedefendedtheanalytic/syntheticdistinction,andtheintelligibilityoftruthbyconvention.HefollowedGoodmaninseekingareductiveanalysisofcounterfactualconditionals,butrejectedGoodman’sdemandforareductionofthepossibletotheactual.HeadoptedQuine’sstandardsforontologicalcommitment,andforphilo-478\nDAVIDLEWISsophicalclarification,butusedthemtoreachverydifferentconclusionsaboutwhatthereis,arguingthatQuine’s“creaturesofdarkness”–intensions,propositions,pos-sibleworlds–canfindaplaceinaworld-viewthatmeetstherigorousstandardsofade-quacythatQuinesetdown.Theactualworld,accordingtothemetaphysicaltheoryLewisdefends,ismuchasQuineandGoodmanthought.Theironlymistakewastothinkthattheactualworldistheonlyworldthereis.TheemphasisinthisexpositionwillbeonthegeneralmetaphysicalframeworkthatprovidesthecontextforLewis’smanyconstructivephilosophicalanalyses.Iwillbeginwithsomegeneralremarksaboutphilosophicalmethodandmetaphysicsinthenextsection,andafterthatdiscussLewis’smodalrealismandfinallyhisHumeanaccountofcounterfactuals,laws,andcausation.MethodandmetaphysicsDuringthefirsthalfofthetwentiethcentury,theword“metaphysics”hadmostlyapejorativeusewithintheanalyticphilosophicaltradition.Thelogicalempiriciststaughtthatmetaphysicswastheresultofequivocationbetweenquestionsaboutmeaning,whichcalledforadecisionaboutwhatlinguisticframeworktouseandquestionsthatarisewithinthecontextofanacceptedframework.ButQuinenotedthatthemethodsusedwithinthescientificframeworkfordecidingwhichtheoreticalclaimsweretruewerenotverydifferentfromthemethodsusedtomakethepracticaldecisionsaboutwhatlanguageformstoadopt.Inbothcases,onechosethetheoryorframeworkthatdidthebestjobofmakingsenseofone’sexperience.Hearguedthatthelinebetweeninternalandexternalquestions,andbetweendecisionsthatconstitutedlinguisticstipu-lationanddecisionsthatconstitutedempiricaljudgmentswasarbitrary.Ifdecisionsaboutwhatgeneralframeworktotheorizeinarenotseparablefromjudgmentsaboutwhatistrue,thenthereisroomformetaphysicsafterall.“Thequestofasimplestclear-estoverallpatternofcanonicalnotation,”Quinewrote,“isnottobedistinguishedfromaquestofultimatecategories,alimningofthemostgeneraltraitsofreality”(Quine1960:161).Lewis’saccountofhisphilosophicalmethodfollowsthatofQuineandGoodmanclosely.Webeginwithacollectionofopinions.“Somearecommonsensical,somearesophisticated;someareparticular,somegeneral....Areasonablegoalforaphiloso-pheristobringthemintoequilibrium,”Lewis1983b:x)AndlikeQuine,Lewisempha-sizesthatthemethodofreflectiveequilibriumshouldnotbetakentohaverelativistoranti-realistconsequences.Philosophymaybeamatterofopinion,butsomeopinions,evensomethatareinsomephilosopher’sreflectiveequilibrium,mayneverthelessbefalse.ButunlikeQuineandGoodman,Lewisdidnottiehisepistemologicalholismtotherejectionoftheanalytic/syntheticdistinction.HisfirstmajorphilosophicalprojectrespondedtoQuine’scritiqueofthisdistinction,andoftruthbyconvention.LewisacceptedthetermsofQuine’sdemandforananalysis:onemustbreakoutofthetightcircleofconcepts(synonymy,semanticrule,meaning,etc.),andexplainwhatitistobeananalytictruthintermsofthedispositionsandbehavioroflanguageusers.Buthearguedthatthiscouldbedonewiththehelpofageneralanalysisofthenotionofaconvention,andadistinctionbetweentwodifferentnotionsoflanguage:languageasdefinedbyasetofsyntacticandsemanticrulesandlanguageasdefinedbyapopula-479\nROBERTSTALNAKERtionofspeakers.Thedefinitionofanabstractlanguagesimplystipulatesthatthelan-guageisconstitutedbycertainsemanticrulesthatdetermineaclassofanalytictruths.Theworkisdoneinexplaining,intermsofananalysisofconvention,whatitisaboutthebehavior,expectations,dispositionsofagivenpopulationofspeakersforalanguagedefinedinthiswaytobethelanguagespokenbythatpopulation.TherearetwowaysinwhichLewis’saccountofanalyticity,eveniffullyadequateonitsownterms,willfailtosatisfyanunreconstructedQuinean.First,conventionsareexplainedintermsofintentions,beliefs,andknowledge,andsoanexplanationofsemanticnotionssuchasmeaningandanalyticityintermsofconventionwouldnotbeanexplanationthatsolvedtheproblemofintentionality.Quinethoughtthattotheextentthatmentalisticintentionalnotionssuchasbeliefandintentioncouldbeexplainedatall,theywouldbeexplainedintermsoftheintentionalityoflanguage–believing,forexample,intermsofholdingtrue–andsohewouldnotbesatisfiedwithanexplanationofsemanticnotionsthattookbeliefandintentionforgranted.Inthisregard,LewisislikeH.P.Grice,separatingproblemsaboutlinguisticmeaningfromthemoregeneralproblemofintentionality,andtakingtheintentionalityofthoughtasmorebasic.ButLewis,Quine,andGricewouldallagreethatwhichevercomesfirst,anadequateaccountoflinguisticandmentalintentionalitymustultimatelyexplaintheminmaterialisticallyacceptableterms.AsecondwayinwhichtheaccountwilldisappointaQuineanisperhapsthemoresignificantone.AsLewisemphasized,hisaccountofanalyticitymakesreferencetopos-sibleworlds,andsodoesnotprovideaninformativeanalysisofoneofthenotions–necessity–thatQuinewouldhaveputinhistightcircleofproblematicconcepts.ButLewisarguedthatthemetaphysicalnotionsofnecessityandpossibilitydonotbelonginthiscircle,sincetheyarenotsemanticnotions.Analytictruthsarenecessarybecausetheyexpresspropositionsthatarenecessary.Theaccountoftheconventionsoflan-guageexplainwhythesentencesusedbythemembersofsomepopulationexpressthepropositionstheyexpress,butthenecessityorcontingencyofthepropositionsthem-selveshasnothingtodowithconvention,orwithlanguage.Thoughhedefendsanalyticity,Lewisdoesnotassumethatspeakersareauthorita-tiveabouttheconventionsoftheirownlanguage,soabouttheanalytictruths.Evenifthereisasharplinebetweentruthsofmeaningandtruthsoffact,thereisnosharplinebetweenlinguisticintuitionandbeliefsaboutsubstantivetheory.“Our‘intuitions’aresimplyopinions,andourphilosophicaltheoriesarethesame”(Lewis1983b:x).Wecandrawthelinebetweenanalyticandsynthetic,butthedecisionaboutwherewedrawit,likeourotherdecisionsaboutwhattobelieve,isapartofajudgmentabouttheglobaltheorythat,allthingsconsidered,bestmakessenseofourexperience.ModalrealismPossibleworldshaveplayedaprominentroleinLewis’sphilosophicalanalysesfromthebeginning.BeingagoodQuinean,Lewisrecognizedanobligationeithertoadmitthemintohisontology,ortoreducethemtosomethingelse.Andiftheyaretobeaccepted,itshouldbeclearwhatkindofthingtheyare.“Weoughttobelieveinotherpos-sibleworldsandindividuals,”heargues,“becausesystematicphilosophygoesmoresmoothlyinmanywaysifwedo”(Lewis1986b:354).Lewismakesnoattempttomini-480\nDAVIDLEWISmizethecounterintuitivecharacteroftheontologicalcommitmentheispreparedtomake;possibleworlds,asheusestheterm,areconcreteparticulars:otherthingsofthesamekindastheuniverseofwhichweareapart.Merelypossiblehighwaysinlandsthatwillneverbeactualaremadeofconcretethatisjustasrealasthatusedtomaketheactualhighwaysonwhichwedrive.Justaspeoplewholiveatothertimesandplacesintheactualworldareasrealasweare,so,accordingtoLewis’smodalrealism,arethenon-actualpeoplewhoinhabitotherpossibleworlds.Theirnonactualityconsistsinthefactthattheyarespatiotemporallydisconnectedfromus.Lewisgrants–infactempha-sizes–thatthebeliefinapluralityofparalleluniversesconflictssharplywithcommonopinion,andsincehetakescommonopinionseriously,heacknowledgesthatthisisaseriouscosttobebalancedagainstthebenefitsofthismetaphysicaltheory.Butwhilehetakesthe“incredulousstare”thatisacommonresponsetothistheorytoreflectaformidableobjection,hearguesthatmoretheoreticalargumentsagainstmodalrealismfail,asdoattemptstoanalyzepossibleworldsaway,ortogiveamoreinnocentexplanationofwhattheyare.Intheend,hejudgesthatthecostofoffendingcommonopinionisoutweighedbythemanybenefitsthatmodalrealismbrings.Sothestrategyfordefendingmodalrealismcombinesanexpositionofthemanybenefitsofthepossibleworldsframework,responsestotheoreticalargumentsagainstmodalrealism,andargumentsagainstattemptstogetthosebenefitswithoutthecounterintuitivecommitment.Itisusefultodividethedoctrineofmodalrealismintoasemanticandametaphys-icalcomponent.First,thereisthemetaphysicalthesisthatthereisalargepluralityofparalleluniverses,whereasingleuniverseconsistsofeverythingthatisspatiotempo-ralrelatedtoanythinginit.Second,therearethesemanticanalysesthatrelatethispluralityofworldstothemanymodal,epistemic,andintentionalconceptswhoseclari-ficationprovidethebenefitsofmodalrealism.Asanexampleofathesisthatbelongstothesecondcomponent,considertheanalysisofpossibilityastruthinsomepossibleworld.Lewisemphasizesthatthetheorymustbeevaluatedasapackage,andhewouldagreethateachcomponentwouldlooseallplausibilitywithouttheother.Ontheassumptionthatthemetaphysicalthesisisfalse–thatcommonsenseisrightthatouruniverseistheonlyone–thesemanticanalysisofpossibilityhasnoplausibility,sinceonthatassumptionthepossiblecollapsesintotheactual.Ontheotherhand,ifwelookatthemetaphysicalclaiminisolationfromthesemanticanalyses,itlookslikeanextravagantandgratuitousempiricalhypothesis.Whyshouldonebelieveinalltheseotheruniverses?Lewis’sanswer–thatsystematicphilosophygoesmoresmoothlyifwedo–hasforceonlywhenthemetaphysicalhypothesisiscombinedwiththesemanticanalysesthatconnectthehypothesiswiththephenomenathatsystematicphilosophyseekstoexplain.Thepossibleworldsframeworkpromisestoclarifynotonlydedictomodalclaims,suchasthatitisnecessarythatallbachelorsareunmarried,butalsoderemodalclaimssuchasthatnobachelorisessentiallyunmarried.Modalrealismusescounterparttheorytoanalyzeclaimsaboutthemodalpropertiesofthings.Aswiththegeneralmodalrealistthesis,wecandistinguishametaphysicalandasemanticcomponentofLewis’scounterparttheory.Thereisthemetaphysicalclaimthatindividualsexistinonlyonepossibleworld,andthesemanticclaimthatderemodalpropertiesshouldbeanalyzedinsomethinglikethefollowingway:anindividualhasthepropertyofbeingpossibly481\nROBERTSTALNAKERFifandonlyifithasacounterpartthathasthepropertyofbeingF,wherethecoun-terpartrelationisacontextuallydeterminedrelationofsimilarityinrelevantrespects.Aswiththegeneralthesis,Lewiswouldemphasizethatthesemanticandmetaphysi-calpartsofthepackagemustbeevaluatedtogether.Themetaphysicaldoctrinehasbeencriticized(forexample,byAlvinPlantingaandNathanSalmon)onthegroundthatithastheimplausibleconsequencethatallprop-ertiesareessentialproperties;butthiscriticismsimplyassumesthatLewis’ssemanticanalysisofwhatitistohaveapropertyessentiallyismistaken.Thesemanticthesishasbeencriticized(forexamplebySaulKripke)onthegroundthatithastheconsequencethatwhenwesaythatHumphreymighthavewontheelection,wearenotreallytalkingaboutHumphrey.ButLewisrightlyinsiststhat,onthecounterpartanalysis,itisHumphreyhimselfwhohas,intheactualworld,thepropertyofbeingapossiblewinner.Itisjustthathehasthispropertyinvirtueofhisresemblancetosomeoneelsewho(inanotherpossibleworld)hasthepropertyofbeingawinner.Thecounterpartsemanticsmaybemorecomplexandlessstraightforwardthanthestandardanalysis,butgiventhemetaphysicalthesis,itgivesabetteraccountofourmodalbeliefs.Andifoneacceptsthegeneraldoctrinethatotherpossibleworldsareparalleluniverses,itseemsmostreasonabletothinkthatnoonecaninhabitmorethanoneofthem.Whateverone’sverdictabouttheplausibilityofmodalrealismasawhole,itseemsclearthatcounterparttheorybelongsinthepackage.Bothmodalrealism’scentralmetaphysicalthesisanditssemanticanalysesofneces-sity,possibility,andothermodalnotionsconflictwithunreflectivecommonopinion.Itisnotonlythatitstrainscredibilitytohypothesizethatthereisavastpluralityofpar-alleluniverses,italsoseemscounterintuitivetomanypeopletoclaimthatouropinionsaboutwhatmightorwouldhavehappenedareopinionsabouttheexistenceofsuchparalleluniverses.Asnotedabove,Lewisgrantsthatmodalrealismconflictswithunre-flectivecommonopinion,andthatthisconflictisastrikeagainstthetheory,buthearguesthatthecostisoutweighedbythebenefits.Sinceheagreesthatifsomealter-nativeaccountcouldprovidethebenefitswithoutthecost,modalrealismwouldnotbedefensible,itisanimportantpartofitsdefensetocriticizeattemptstoreconciletheexplanationsthatthepossibleworldsframeworkprovideswithamoremodestaccountofwhatpossibleworldsare.“Ersatzmodalrealism”isLewis’slabelfortheattempttogetthebenefitsofmodalrealismwithoutthecostsbyexplainingpossibleworldsassomethingotherthanpar-alleluniverses.Mostofhiscriticaldiscussionofthisprojectisdevotedtoattemptstoreducepossibleworldstosomekindoflinguisticobject:statedescriptions,maximalconsistentsetsofsentences,completenovels.Thisisacommonandseductivestrategyforexplainingwhatpossibleworldsare,butthereisalotwrongwithit,asLewis’scriti-cismsbringout.Themostseriousproblemisthatthiskindofexplanationseemstofore-closeoneofthemostimportantusesofpossibleworlds:torepresentthecontentsofspeechactsandpropositionalattitudes.Ifsentences,orsetsofthem,aretorepresentpossibleworldsadequately,theymustbeinterpretedsentences–sentenceswiththeirtruthconditions.Wewillhaveaseriouscircularityifwetrytocombinethiskindofexplanationofpossibleworldswithanexplanationofthetruthconditionsofasen-tenceintermsofthepossiblecircumstances,orpossibleworlds,inwhichthesentencewouldbetrue.Thereare,however,philosophicalaccountsofpossibleworldsthatagree482\nDAVIDLEWISwithLewisthatpossibleworldsarenon-linguisticthings,suitableforrepresentingtruth-conditionalcontent,whiledisagreeingwiththethesisthatpossibleworldsaresomethinglikeotheruniversesparalleltoourown.Accordingtothesimplestandmoststraightforwardattempttoexplainwhatpos-sibleworldsareinawaythatiscompatiblewithactualism(athesisthatcommonopinionmightregardastriviallytrue:thatwhatactuallyexistsisallthereis),possibleworlds(orlessmisleadingly,possiblestatesoftheworld)areakindofproperty:waystheworldmightbe,ormighthavebeen.Thisisobviouslynotareductionofpossibleworldstosomethingelse:itisintendedsimplyasacharacterizationofthekindofthingthatapossibleworldis.Tocallpossibleworldspropertiesistosaytwothingsaboutthem:first,theyarethingsthatare,ormaybe,instantiated.Second,theyarethekindofthingthatis(atleastprimafacie)independentoflanguageandthought.Apropertyofsomething(suchasthepropertyofbeingthefirstchildborninthetwenty-firstcentury)isdifferentbothfromthething(ifany)thathasthepropertyandfromathoughtorapredicatethatexpressestheproperty.Thesignificanceofthischaracteri-zationisthatitprovidesuswithawaytoreconcileacommitmenttotheexistenceofpossibleworlds,construedasnon-mentalentities,withtheapparentlycontradictorythesisthattheactualworldistheonlyworldthereis.Anyonewhotakesliterallytheclaimthattherearepossibleworldshastorespondtothisprima-facieparadox:unrealizedpossibilities–counterfactualsituations–aresituationsthatturnedoutnottoexist.Howcantherebesituations,orworlds,thatdon’texist?Anyresponsetothisproblemwillmakeadistinctionbetweenasenseinwhichnon-actualpossibleworldsexist,andasenseinwhichtheydonot.Lewis’sstrategyistodistinguishtwodifferentscopesforthequantifier.Quantifiersareoftenrestrictedtosomecontextuallydeterminedsubdomainofallthereis,andoneverygeneralrestric-tion,accordingtoLewis,istothedomainofthingsthatinhabittheactualworld.Whenwesaythattherearenotalkingdonkeys,wenormallymeanthattherearenoactualtalkingdonkeys.Butthereisalsoanunrestrictedquantifier,whichrangesoverabsolutelyeverythingthereis.Commonopinionmaynotdistinguishwhatexistsfromwhatactuallyexists,butLewiswouldsaythatthedistinctionisimplicitintheirmodaldiscourse.Theactualistresponsetothispuzzlemakesthedistinction,notintermsofadifferenceofdomain,butintermsofanambiguityintheterms“possibleworld”and“actualworld.”Justaswecandistinguishthepropertyofbeingthefirstchildborninthetwenty-firstcenturyfromthatchild,sowecandistinguishthepropertyofbeingauniverseofacertainkindfromauniversethatisofthatkind.Accordingtotheactualist,thereare(andactuallyare)manywaystheworldmighthavebeen,butthereisonlyoneworldthatisoneofthoseways.Thisconstrualofpossibleworldsaspropertiesallowsusmanyofthebenefitsofthepossibleworldsframework(forexample,theformalsemanticanalysisofmodalandepistemicnotions,theclarificationofcounterfactualsandcausalandtemporalstruc-tures,therepresentationofprobabilityasameasureonstatespaces,therepresenta-tionofmentalandlinguisticcontent,andofspeechcontexts)withouteitherdenyingtheontologicalcommitmenttopossiblestatesoftheworld,orchallengingpretheoreti-calcommonopinion.Itdoesnotproduceincredulousstarestosaythattherearemanywaystheworldmighthavebeen.ToseewhyLewisresiststhisactualistinterpretationofpossibleworldsweneedtoconsideranotheroneofhismetaphysicalprioritiesthat483\nROBERTSTALNAKERhasitsrootintheQuine–Goodmanlegacy:apenchantfornominalism(seeGOODMANandQUINE).Oneofthebenefitsofmodalrealism,accordingtoLewis,isthatitprovidesuswithananalysisofproperties:ofwhatpropertiesare,andwhatitistohaveaproperty.Properties,accordingtoLewis’smodalrealism,arejustsets,andtohaveapropertyisjusttobeoneofitsmembers.Thedomainofallpossibiliaprovidesananswertothestandardobjectiontotheidentificationofpropertieswiththeirextensions,ananswerthatthosewithonlytheimpoverisheddomainoftheactualthingscannotavailthem-selvesof.Distinctpropertiescanhavethesameextensionintheactualworld,buttheyaredistinguishedbythedifferenceintheirextensionsinotherpossibleworlds.Evenifintheactualworld,allandonlycreatureswithakidneyarecreatureswithaheart,thetwopropertiesaredistinguishedbythefactthattherearepossiblecreatureswithoneproperty,butnottheother.Somodalrealismoffersthevirtuesofasimpleextension-alistaccountofpropertieswithoutthedefectsofactualistversionsofthataccount.Lewisrecognizesthatanadequatetheoryofpropertiesneedstodistinguishbetweendifferentkindsofproperties.Someproperties(sets)andrelations(setsofn-tuples)arenaturalorfundamental.Amongthefundamentalrelations,somearespatiotemporalrela-tions.TheseareprimitivedistinctionsofLewis’stheory,buthearguesthattheyaredis-tinctionsthatanyplausiblemetaphysicaltheorymustmake.Withjusttheseprimitiveconceptsforclassifyingpropertiesandrelations,Lewissuggests,wecangiveafullchar-acterizationofthelogicalspaceofpossibleworldswhilecontinuingtomaintainthatpropertiesarenothingbutsets.World-mates(inhabitantsofthesamepossibleworld)areindividualsthatstandinspatiotemporalrelationswitheachother.Apossibleworldisfullycharacterizedbyspecifyingasetofworld-mates,andbysayingwhichfundamentalpropertiestheyhave,andhowtheyarerelatedbythefundamentalrelations.Allthepropertiesandrelationsofthethingsinanyworldsuperveneonthefundamentalpropertiesandrelationsofthosethings:possibleworldsthatareindiscerniblefromeachotherwithrespecttofundamentalpropertiesandrelationsareidentical.Thisapriorisuperve-nienceclaimisnotsubstantive,sincethefundamentalpropertiesarejustthosethatarenecessarytogiveacompletecharacterizationofapossibleworld.Substantive(andcontingent)metaphysicalhypothesescanbestatedasthesesaboutwhatthefunda-mentalpropertiesoftheactualworldare.So,forexample,materialismisexplainedasthethesisthatonlyphysicalpropertiesandrelationsarefundamental.(Thatis,mate-rialismistrueofpossibleworldwifthefundamentalpropertiesandrelationsofthingsinwareallphysical.)ThethesisofHumeansupervenience,whichwewilldiscussinthenextsection,isthethesisthatonlyintrinsicpropertiesandspatiotemporalrelationsarefundamental.ThetheoryofpropertiesassetsisacrucialpartofLewis’smodalrealism,andunlikemanyofthefruitsofthepossibleworldsframework,thisanalysiscannotbereconciledwiththeactualistinterpretationofpossibleworlds.Lewis’stheorycan,ofcourse,makethedistinctiontowhichtheactualistappealsbetweenpropertiesandtheirinstances–itisjustthedistinctionbetweensetsandtheirmembers–butitwillbenohelpinavoid-ingacommitment,notjusttowaystheworldmightbe,buttoworldsthatarethoseways.Forifpropertiesaresets,andifpossiblestatesoftheworldareidentifiedwithmaximalpropertiesthattheworldmighthave,thenapossiblestateoftheworldisa484\nDAVIDLEWISunitsetwithaworldthatisinthatstateasitsmember.Accordingtotheactualistmeta-physics,thereisonlyonethingtobethememberofsuchaset,andsoifpossiblestatesoftheworldareproperties,andLewisisrightaboutwhatpropertiesare,thereisonlyonepossiblestateoftheworld.Sowhileactualistscanavailthemselvesofmanyofthebenefitsoftheframeworkofpossibleworlds,theywillhavetoforegoLewis’selegantreductiveaccountofproperties.CounterfactualsandcausationLewis’ssecondbookundertooktogiveareductiveanalysisofcounterfactualcondi-tionals,aprojectmotivatedbythesameHumeanskepticismaboutnaturalnecessitythatmotivatedNelsonGoodmantotrytogivesuchananalysismorethantwentyyearsearlier.ForGoodman,thecruxoftheproblemwasthemodalcharacterofcounterfac-tuals:theyseemedtobeaboutunrealizedpossibilities.Thetaskwastoexplainthepossibleintermsoftheactual.Lewis,ofcourse,hadnoproblemwithnon-actualpos-sibilities,andtookcounterfactualsatfacevalueasstatementsaboutcounterfactualpossibleworlds.Butcounterfactuals(andstatementsaboutcauseandeffect,disposi-tionsandpropensities,dependencyandchance)areforthemostpartcontingentstate-ments.Onehastoexplainhowastatementaboutcounterfactualpossibilitiescanbecontingentlytrueorfalseintheactualworld.Forsuchstatementstobecontingent,thecounterfactualworldsthatarerelevanttotheevaluationofaconditionalmustbedeterminedbytheirrelationtotheactualworld.Lewis’sformalsemanticsgivestruthconditionsforconditionalsintermsofathree-placecomparativesimilarityrelationonpossibleworlds(worldxismoresimilartoworldwthanyistow).Theroughideaoftheanalysisisthataconditional,“ifA,thenC”istrue(inapossibleworldw)ifandonlyifCistrueinthosepossibleworldsinwhichAistruethataremostsimilartow.Thisfirstapproximationisnotquiteright,sinceifthereisaninfinitesequenceofevermoresimilarworldsinwhichAistrue,therewillbenoclosestsuchpossibleworlds.Toallowforthiscase,Lewis’sfavoredanalysisisasfollows:“IfA,thenC”istrueinwifandonlyifsomeworldinwhichA&CistrueisclosertowthananyworldinwhichA&~Cistrue.Thisanalysisprovidesanabstractformalsemanticsforcounterfactualconditionals,butwedon’thaveareductiveanaly-sisuntilwehaveexplainedtherelevantrespectsofsimilarity.Thesemanticanalysisisjustthefirststepofalargerproject,adefenseofthedoctrinethatLewislabeled“Humeansupervenience.”TheprojectismotivatedbyaHumeanskepticismaboutrealrelationsbetween“distinctexistences.”FortheHumean,spatiotemporalrelations(suchascontiguity)areacceptable,asarelogicalrelations,orrelationsofideas.Relationsofresemblancebetweenthingsareacceptable,solongastherespectsofresemblancearespelledout,sincetheyareexplic-ableintermofthesharingofspecifiedproperties.Butcausalrelations,andothersinthesamefamily,mustbeanalyzedintermsofglobalregularities,withthehelpofrelationsoftheunproblematickind.Iftherespectsofsimilaritybetweenpossibleworldsthatarerelevanttotheinterpretationofcounterfactualscanbespecified,Lewis’sanalysiswillyieldanaccountofcounterfactualsthatshouldsatisfyaHumean,andsoanaccountthatpermitstheHumeantousecounterfactualstoanalyzerelationsofcausationandcausaldependenceandindependence.485\nROBERTSTALNAKERLewis’sHumeanprojecthasthefollowingseparablecomponents:(1)anabstractsemanticanalysisofconditionalsintermsofcomparativesimilarity;(2)anexplana-tionoftherespectsofsimilaritythatareappropriateforinterpretingofthekindofcon-ditionalsthatarerelevanttotheanalysisofcausalrelations.SinceLewis’sresponsetothisproblemappealstolawsofnature,heneeds(3)anaccountoflawsofnatureintermsofglobalpatternsofparticularfact,andfinally,(4)ananalysisofcausationintermsofcounterfactuals.Thisisanambitiousagenda.Somepartshavebeencarriedoutindetailandwithprecision;inothercases,thereareonlysketchysuggestionsaboutthekindofaccountthatshouldbegiven.Andsomepartsoftheprojectareongoing.Aconditionalistrueiftheconsequentistrueinthepossibleworldinwhichtheantecedentistruethatismostsimilar,inrelevantrespects,totheactualworld.Butwhataretherelevantrespects?Onemightbetemptedtoappealtoanintuitivenotionofoverallsimilarity.Lewisnotesthatwedomakeandunderstandjudgmentsofoverallsimilaritybetweencomplexobjectsuchascities,andwedohaveintuitionsaboutwhichpossibleworldsaremoreandlessalike.Ageneralimpressionisticnotionofsimilaritywouldbebothvagueandcontext-dependent,butasLewisnotes,counterfactualsarebothvagueandcontext-dependent.Therewould,however,beatleasttwoproblemswithrelyingonsuchanotionofsimilarity.First,animpressionisticnotionofsimilar-itywouldbesuitablefortheprojectofHumeanreduction,sincejudgmentsofsimilar-itybetweenworldsmightbebasedinpartoncomparisonofunanalyzedfactsaboutcausalrelations.Butsecond,inanycasetherearecounterexamplesthatshowthatoverallsimilarityisnottherightrelation.Itseemsintuitivelyclearthatsmalleventscanhavelargeconsequences.IfOswaldhadmissedKennedyin1963,thecourseofAmericanpoliticsbetweenthenandnowprobablywouldhavebeenquitedifferent.Butisn’tapossibleworldinwhichOswaldmisses,butsomeoneelsesucceeds,andthecourseofAmericanpoliticsproceedsmuchasitactuallydidmuchmoresimilar,overall,totheactualworld?Ifcertainconspiracytheoristsareright,andtherewerebackupassassinsreadytoactifOswaldfailed,thenitmightbetruethatifOswaldhadn’tkilledKennedy,someoneelsewouldhave,butwedon’twantananalysisofcounterfactualstoensurethatsuchconspiracytheoriesaretrue.Onemightbetemptedtobuildatemporalasymmetryintotheaccountofcom-parativesimilaritythatisrelevanttotheinterpretationofcounterfactuals:per-hapssimilarityofearlierpartsofhistoryshouldhavemuchgreaterweightthansimi-larityoflatertimes.Buttodothiswouldbetoexplainthetemporalasymmetryofcausalandcounterfactualdependenceasaconsequenceofconventionandnotasafactabouttheworld.Lewis’saimwastodefineatemporallyneutralnotionofcom-parativesimilaritybetweenpossibleworlds,anduseittoexplainhowtemporalasym-metriesinthepatternoffactsintheactualworldresultsinadefactoasymmetryofcounterfactualdependence.Lewis’saccountoftherelevantrespectsofcomparativesimilaritybetweenworldsgiveshighestprioritytoavoidinglargeandwidespreadviolationsoflawsofnature.Thesecondpriorityistomaximizeexactagreementofparticularfact.Smallandlocalvio-lationsoflawsofnaturearepermissibletoachievethesecondpriority,andinadeter-ministicworld,such“smallmiracles”willalwaysberequired.Approximateagreementoffactcountsforverylittle:deviationsfromthelaws,evensmallones,inordertoincreaseapproximatesimilarityoffactarenotpermitted,andpossibleworldsthatagree486\nDAVIDLEWISexactlyforaperiodoftimearemoresimilarthanworldsthatagreeonlyapproximately,butoveramuchlongerperiodoftime.Thereisnoattempttomakethisaccountofcom-parativesimilarityprecise,butLewisarguesthatithelpstoexplainsometemporalasymmetries,andthatitpointstothekindofexplanationthatcanvindicateareduc-tiveaccountofcounterfactuals.AHumeancannot,ofcourse,restwithanappealtoanunanalyzednotionoflawofnature.HereLewisendorsesanideaofFrankRamsey’s:thatthelawsofnature(inagivenpossibleworld)arethefactualregularitiesthatareconsequencesofthesim-plestandstrongestsystematizationofthetruthsofthatworld.Thecriteriaforevalu-atingsystemsoftruthsremaintobeexplained,butLewisarguesthatsolongastherelevantcriteriaofstrengthandsimplicityarenon-contingent,thiswillbeanaccountoflawofnaturethatmeetsthestandardsofHumeansupervenience.Counterfactuals,onceexplainedintermsofresemblanceofthefactsandregulari-tiesofthepossibleworlds,arethenavailabletotheHumeanforananalysisofcausa-tion.Thefirststepistodefinecounterfactualdependence:onetruthBcounterfactuallydependsonanotherAifBwouldnothavebeentrueifAhadnotbeentrue.Ifcandearedistincteventsthatoccur,itseemsagoodfirstapproximationtosaythatcisacauseofeifandonlyifbothoccur,andewouldnothaveoccurredifchadnot.Thisproposalwouldaccountformanyofthefeaturesofcausationthatcreateproblemsforasimpleregularityanalysis.Causecanbedistinguishedfromeffectwithoutmakingexplicitappealtotemporalorder,andeventsthatareregularlyconnectedbecauseonecausestheothercanbedistinguishedfromeventsthatareconnectedbecausetheyareeacheffectsofacommoncause.Butcasesofpreemptivecausesshowthatonecannot,ingeneral,identifycausationwithcounterfactualdependence.Supposethehitmanwassuccessful,butifhehadmissed,anotherwaswaitinginthewingstodothejob.Thevictim’sdeathwascausedbythehitman’saction,but,becauseofthebackuppotentialcause,wasnotcounter-factuallydependentonit.Lewis’sfirststrategyforaccommodatingpreemptivecausa-tionwastodefinethecausalrelationasthetransitiveclosureoftherelationofcounterfactualdependence(betweendistinctevents).Thedeathdoesnotdependcoun-terfactuallyontheshooting,buttherewillbeintermediateeventswhicharedependentontheshooting,andonwhichtheeffectisdependent.Thismoveaccountsforsomecasesofpreemption,butnotforall.Asecondstrategyfordealingwithpreemptioncasesarguesthatwhilethemanwouldstillhavediedifthebackupassassinhaddonethejob,hewouldhavediedadifferentdeath,andsodespitethepreemption,theeventthatwastheeffectwasstillcounterfactuallydependentontheactualassassin’sact.Butitisdifficulttofindandmotivateanaccountofthemodalpropertiesofeventsthatwillexplainallcasesofpreemptioninthiswaywithoutintuitivelyimplausibleconse-quences.Eventakingtheresourcesofcounterfactualconditionalsforgranted,theanalysisofcausationhasprovedtobeasurprisinglyrecalcitrantproblem.Thisisnowalivelyareaofongoingresearch.ConclusionLewis’smetaphysicalframework,andhisphilosophicalmethod,providearichcontextfortheclarificationofphilosophicalproblems,thearticulationanddefenseofphilo-487\nROBERTSTALNAKERsophicaltheses,andtheformulationofconstructiveconceptualanalyses.Hehasfor-mulatedanddefendedamaterialisttheoryofmind,withaccountsbothofintentionalstatessuchasbelief,andsensorystatessuchaspain.InthecontextofthedefenseofHumeansupervenience,heexploredtherelationbetweenobjectiveandsubjectiveprob-ability–chanceanddegreeofbelief–andbetweensubjectiveprobabilityandcondi-tionalpropositions.Andheprovidesafoundationforcausaldecisiontheory.Buildingonhisearlyworkonconvention,hedevelopedafoundationforsemanticsandprag-maticsthatclarifiestherelationsbetweenspeechandthought,andthatalsomakessub-stantivecontributionstocompositionalsemantictheoriesfornaturallanguages.Modalrealism’suseofsettheorymotivatedanexplorationofthefoundationsofsettheoryitselfthatclarifiestherelationbetweenmereology(thetheoryofpartsandwholes)andsettheory.EventhoughLewis’sgeneralmetaphysicaltheoryhasacoherenceandunitythattiethedifferentpartstogether,manyofhisconstructiveanalysesareseparablefromthesystemthatprovidesthecontextfortheirdevelopment.Thisisappropriate,givenLewis’spragmaticcost–benefitmethodology:herecognizesthatotherswithdifferentprioritiesmaynotbepreparedtoswallowhissystemwhole.ThosewhorejectmodalrealismorHumeansuperveniencewillstillfindmuchtolearnandtoadoptfromhisphilosophicalwork.Andeventhosewhoareskepticalaboutthismetaphysicaltheorycanappreciatethepowerofasystemthathasgeneratedsomanyclarifyingphilo-sophicalanalyses.BibliographyWorksbyLewis1969:Convention,Cambridge,MA:HarvardUniversityPress.1976:Counterfactuals,Oxford:BlackwellPublishers.1983a:“NewWorkforaTheoryofUniversals,”AustralasianJournalofPhilosophy61,pp.343–77.1983b:PhilosophicalPapers,vol.I,Oxford:OxfordUniversityPress.1986a:OnthePluralityofWorlds,Oxford:BlackwellPublishers.1986b:PhilosophicalPapers,vol.II,Oxford,OxfordUniversityPress.WorksbyotherauthorsQuine,W.V.(1953)FromaLogicalPointofView,Cambridge,MA:HarvardUniversityPress.——(1960)WordandObject,Cambridge,MA:MITPress.488\nBlackwellCompanionstoPhilosophy:ACompaniontoAnalyticPhilosophyEditedbyA.P.Martinich,DavidSosaCopyright©BlackwellPublishersLtd2001IndexNote:Pagereferencesinboldtypeindicatethemaintreatmentofthephilosopher.actions87–8,251–2,277–8,288,297–8,Ayer,A.J.2,3,98,176,179,205–17,304,320–1,446–7;underadescription223–4,227,232,327,335,389297–8,320–1axiomofexistence436adverbsandadverbialmodification304–5axiomofidentification436aestheticsandthearts165–6axiomofinfinity18,97agentcausation288axiomofreducibility97analysis,conceptionsof1–5,76–7,84,91–2,97,118–19,210,214–15,380,Background,the442,444–5428–30;paradoxof293BarcanFormula,the357analyticityandsynonymy2,74–5,82–4,Bauch,Bruno94,9588–9,98,102,105,153,161–2,183–5,Bayesianism106,152205–6,207–9,210,345,456,470–1,Bedeutung13–16;seealsoreference478,480–1;seealsopropositionsandbehabitives221,234meaningbehaviorism184,247,357,413,451,452,Anscombe,Elizabeth229,315–25454anti-realism340–1,380–90,402–8;seealsobelief141–4,293–4,308–10,359,413,realism415,451,469–70,480,481;beliefdereapriori98,183,190–1,207–8,429,anddedicto199;directandself-attribution470–1;seealsoanalyticityandof293synonymyBerkeley,George407Aristotle317–18,323,341,372,373,416,bivalence387–9466Blackburn,Simon179Armstrong,David413–18Blanshard,Brand177assertibility,warranted387–8Bloomsburygroup45–6assertives435–6bodies,material341–2asymmetricdependencytheoryofmeaningBoltzmann,Ludwig68458–9Boolos,George17AugustineofHippo82,396Bouwsma,O.K.3,231Austin,J.L.3,118,218–30,255,288,Boyle’sLaw157,298,299,373335,430,434–5Bradley,F.H.1,23,27Austin,John169braininavat404–6,446authority,seepoliticaltheoryBraithwaite,Richard376autopsychologicalrealm95–7breakdowntheory,the350–1489\nINDEXBrentano,Franz287,452constructivism95–7Broad,C.D.55,57–67,274,348content,wideandnarrow459–60Burge,Tyler424contextprinciple,the12convention321–2,479–80Cantor,Georg18,24,25conventionalimplication256–73Carnap,Rudolf9,39,72,90,94–109,146,conventionalism99,100–1148,150,152,163–4,181–3,185–7,conversation,theoryof256–73232,239,248,373,375,466–7conversationalimplicature256–73Cartesianism,seeDescartesandconversationalmaxims258–73Cartesianismconverseconsequencecondition151categories335CooperativePrinciple258–73categorymistake120–1Copernicus,Nicholas372–3,374causaltheoryofmind413–14correspondencetheoryoftruth,seetruthcausaltheoryofperception446corroboration111causaltheoryofreference,seereferencecounterparttheory481–2causation144–5,157,297–300,318–19,coveringlawmodelofexplanation156320–1,345–6,416–17,431,447,485–7;Craig,William153–4intentional439–41criteria322centralstatematerialism413–14CRTT,seecomputational/representationalcertainty51–2,282theoryofthoughtC-fibers397–8,472;seealsomind–brainidentityDarwinandDarwinianism188–9charity,principleof303Davidson,Donald340,296–314,424,431,ChineseRoom,the439–41452Chisholm,Roderick4,61,281–95decisiontheory142–3Chomsky,Noam419–27,451,452,453,declarations435–6,448–9461,463Dedekind,Richard17,24Church,Alonzo9,102,128–33,207dedictoandderebelief,seebeliefChurch’stheorem129deductivenomologicalmodel155–8Church’sthesis128–9D131–2class14,18demonstratives293coherenceinknowledge284–5Dennett,Daniel60,452commissives221,435–6denotinganddenotingconcepts27–9;seecommonsense47alsoreferencecommunication384;seealsoconversation,Derrida,Jacques432–3theoryofDescartes,René,andCartesianism1,51–2,compositionality83–4,270,300–7,312,55–6,82,119,235–6,242–3,310,312,456341,343,402–3,428–9,470computational/representationaltheoryofdescriptions,theoryof28–33,80,197,thought451,454–9,461264–5,336–7,338–9,467–9computationalstates398descriptivetheoryofreference437,467–9concepts14–15descriptivism327conceptualschemes309,335determinism,seefreedomandfreewillconfirmation99,103–4,150–1,186–7,Devlin,Lord172–3191,193,298;seealsoverificationDewey,John431connectionism452,455DifferencePrinciple,the362,364–5,367consciousness319–20,345–6,438–45;seediophantineproblems135–6alsofirstpersonandfirstpersonpronounsdirectives434–5constativeutterances219–20disciplinarymatrix373constructionalism163–4discourse,indirect198–9490\nINDEXdistributivenormalform275family369Donnellan,Keith338,437,469fanatics331dreamsanddreaming235Feigl,Herbert239Dreske,Fred457feminism369,376dualism,property439Feyerabend,Paul375Duhem,Pierre-Maurice-Marie187firstpersonandfirstpersonpronouns87,Dummett,Michael338,340,378–92,96,211–12,233–4,242–3,292–3,429310–12,343,438,441Dworkin,Ronald172fixedpointlemma101,106Fodor,Jerry451–65egalitarianism370folkpsychology451Einstein,Albert62,95,112–13,137Foot,Philippa350–6E-language419–22force,linguisticorillocutionary12,379,Elementarerlebnisse96–7,106n.4,163434–5elex,seeElementarerlebnisseFoucault,Michel433emotivism175–9,327freedomandfreewill87–8,225,278,288,empiricism95–9,182–94,240–1,312–13;318,348–9seealsologicalempiricismFrege,Gottlob1,6–20,68,70–2,75,83,entrenchmentproperty162–394,95,99,102,130,152,182,378–9,entropy104382–4,390–1,429,433,437,459,epistemology,seeknowledge460–1,467,469;Begriffsschrift7,9–11;ethics45–6,175–9,212–13,276–7,286,GrundlagenderArithmetik(Foundationsof323–4,326,350–5,359Arithmetic)7–8,11–12,18,382;Evans,Gareth242,349,379GrundgesetzederArithmetik(BasicLawsofevents297–8,304–5Arithmetic)7–8,16,17–20,26evidence104,150,186,189–94,199,311;Freud,Sigmund111,112,452seealsoconfirmationFriedman,Michael106n.6evident,the282–4,287;seealsocertaintyfunctionalism247,396–402,452–3excluderwords224,228functions9–10,34–5,131,134excuses173exemplar373–4Galileo372exemplification345Geach,Peter8,9,28exercitives221geometry95experience,metaphysicsof345–6,446–7Gettier,Edmund286explanation155–8;deductivenomologicalghostinthemachine121157Gibbard,Allan179expositives221Gibson,J.J.452,453expressives435–6given,the,andthemythofthegiven190,extensionality197–8240,249–50externalism396,401,403–7,409–10,424,Glymour,Clark151459God,proofsof136–7Gödel,Kurt37,99–100,101,128,133–7,facts191–2,294,306;brute322;185institutional449;seealsostatesofgödelnumbers133–4affairsGoldbach’sconjecture380–1fact/valuedistinction350–1Goodman,Nelson42,151,155–6,160–8,fairness362–3419,425,463,478–9,485fallibilism99,111,188–9Grice,H.P.233,254–73,340,447,480falsification111–13,149,193;seealsogroundedness474inductionandverificationgrue162–3491\nINDEXhaecceity292–3intentionandintentionality73–4,247–51,Hahn,Hans99287–8,289,320–2,325n.2,409–10,Hale,Bob382,387,390,391437–47,448,480;collective448;Hanson,N.R.374,375intrinsicandderived440,444Hare,R.M.179,326–33internalism240–1,401–2;seealsoHarman,Gilbert5externalismHart,H.L.A.169–74,227interpretation,radical302–4,309–11Hegelianism22–3introspectionism442Heidegger,Martin423–3,444irrealism166–8;seealsorealismHelmholtz,H.L.F.von95Hempel,CarlG.2,146,148–59Jaeger,Robert235Hertz,HeinrichRudolf68Jaeger,Werner279Hesse,Mary375James,William41,431Hilbert,David7judgment26,40–1;seealsothoughthistoricalchaintheoryofreference469justice353,355,361–70historicism114justification240,281,347,431–2;seealsoHobbes,Thomas176confirmationHolder,Otto19holism186–7,308,444,456,463,464Kaila,Eino274,279humanism279–80Kant,Immanuel,andKantianism2,21–2,Hume,David110–11,144,176,205,210,23,24,42,48,94,95,137,173,207,297,318–19,335,348,350–1,353,240,328,335,341,345,361,362,416–17,442,454,485–6,487408–9,428–30,470Husserl,Edmund432–3Kaplan,David103,459,460hypotheses104,461Kemeny,John104hypothetical-deductivemethod89,129,Kepler’sLaws157150–2Keynes,JohnMaynard104Kleene,S.C.101idealism22–4,45,216n.6,345,431knowledge33,39,47–57,95–6,98,142,ideasandmeaning383–4,394;seealso144,188–96,208,228,232–3,240–1,meaning,theoryof255,281–7,319–21,341,415,428–30,identification,seereference470–1;byacquaintance33,38–9,341;identity30–1;criteriaof291,416self-knowledge242–3;seealsofirstpersonI-languages419–22,424,425andfirstpersonpronounsillocutionaryacts,seespeechactsKripke,Saul4,106,338,358,359,375,illusion223–4,417466–77,482;Kripkemodels467images,mental319,454Kuhn,Thomas105,106,371–7,463imperatives326–7implicature,seeconversation,theoryoflambdaconversion133incompletenesstheorems133–6Langford,C.H.129,466–7indeterminacyofreference202language,natureof2–3,17,76–9,82–5,indeterminacyoftranslation188100–1,124–7,201,316–17,322,induction43,101,103–5,106,110–13,419–26,434–8,475–6,479–80;private210,241–3,274–5,347;newriddleof475–6induction151,162–3languagegames84,317,322inductivestatisticalexplanation156–7languageofthought454–5,461inferentialrolesemantics456,463–4law,natureof169–73infinity,axiomof18lawandmorality170,172informationtheoreticsemantics457lawofexcludedmiddle388–9inscrutabilityofreference202lawsofnature298,487492\nINDEXlegalpositivism169,172341–2,345–7,380–1,384,402–8,Leibniz,GottfriedWilhelm24,136–7415–16,425,429,448–9,479–81;Lewis,C.I.102,166,216n.3,232,descriptiveandspeculative335–6,341466–7metersandthemeterbar471Lewis,David145,293,478–88Mill,JohnStuart17–18,114,155,208,liarparadox34,473469,470linguisticcompetenceandperformancemind–bodyproblem,seemindsandthe422mentalandmind–brainidentitylinguisticturn428–31mind–brainidentity397–8,471–2Locke,John383–4,410,429,454mindsandthemental64–6,85–7,233–4,locutionaryacts,seespeechacts278,298–9,319,343–4,397–402,logic99–102,275,454;deontic276–7;413–14,429–30,438–9,471;othernatureof186–7,197–8,291,339–40,minds12,215378;seealsomodalityandmodallogicmodalityandmodallogic102,106,logicalatomism37–8199–200,275–6,290,357–9,466–7,logicalempiricism94,95,98,105,106,472,480–5148,152,206–8,239,288models373logicalpositivism94,97–8,148,205,208,modularityofmindandlanguage422,389426n.3,451,463–4logicalsyntax101monism,anomalous298–300logicaltruth77–8mood,linguistic12logicism24,35–6,97–8,99,140–1moodofconsciousness441Moore,G.E.23–4,39,45–56,176–7,227,McDowell,John349,379,387231,233Mackie,J.L.177morality,seeethicsMcTaggart,J.M.E.22,57,62–3Morris,Charles373makingasiftosay257–9Malcolm,Norman231–8Nagel,Ernest154Marcus,RuthBarcan357–60names,causaltheoryof358–9,467–70MarxandMarxism111,112names,proper197–8,358–9,437,467–70;masterargument,the329–30seealsomeaning,theoryofmaterialism,seenaturalisminmetaphysicsnativism461mathematics,philosophyof6–9,12,17–20,naturalkinds39624–8,80–1,133–7naturalselection462–3meaning,theoryof13–15,73,82–5,98–9,naturalisminethics327,350101–3,124–8,148–50,161–2,183–4,naturalisminmetaphysics188–90,193,199,244–8,254–73,300–9,312,322,244,290,344,413–14,428,484337,340–1,378–90,394–6,400–1,necessity71,88–9,102,199–200,404–5,447,455–7,467–70,476;and290–1,480–2;seealsomodalityandideas383–4,394;andunderstandingmodallogic379–89,394;seealsonames,properNetwork,the444–5mechanism234Neurath,Otto98–9Meinong,Alexius30,289Newton,Isaac371,372,373memory236–7,284,319nominalism11,72,163–5,248,382,416,Menn,Stephen200484mentalese,seelanguageofthoughtnoncognitivism390Merleau-Ponty,Maurice226normalscience371metaphor165–6norms276–7,299,476;seealsologic,metaphysics3–4,77,88–9,97–8,101,103,deonticandethics196–7,200–3,216,247–8,287–94,305,Nozick,Robert53493\nINDEXobligation322–3,324,435postmodern,bourgeoisliberalism432observationandobservationsentences190,practices322191–3,201–2,206–7pragmatismandpragmatics99,101,103,omegarule101141–3,430–1ontologicalcommitments200predicates,seepredicationontologicalobjectivityandsubjectivitypredication344–5,436–8448preferences329–31ontologicalrelativity202–3prescriptivism326–7,330ontology,seemetaphysicspresupposition256,337operationalism149Price,H.H.59,218,223ordinarylanguagephilosophy3–4,117–22,PrincipleofInformationalEquilibrium460219–23,255–6PrincipleofTolerance100–1,106,185O’Shaughnessy,Brian416principlesandparameters419,423–4“ought”324,327,332privacy86overlappingconsensus367–8privatelanguage346Oxfordphilosophy,seeordinarylanguageprobability103–5,142–3,274–5;seealsophilosophyinductionpromising322–3pain86–7,398,414,438,472propernames,seenames,properandparadigm371–2,373referenceandmeaning,theoryofparadoxofanalysis46properties291,459,484–5;self-presentingparadoxoftheravens150–1283paradoxes140propositionalattitudes40–1,199,292–4,particularsanduniversals344–5,416–18,308–9,451,470437–8propositionalfunctions35Peano,Giuseppe7,19,24,28,140propositions40,83,105,199,291,485–6;Pearson,Karl155conditional145,340,485–6;contingentPeirce,C.S.431,432290–1;counterfactual298,483–7;perception240,248–50,284,319,415,general27–8;aboutthepast12,316,445–6327;probable284;religiousandperformatives3,219–20theological213;sentencesandperlocutionaryacts,seespeechactsstatements,synthetic207–8,345,470–1;persons252,335,343–4synthetic88–9,470–1;universal99;seephaticacts,seespeechactsalsoanalyticityandsynonymyandphenomenalism98,101,211–12necessityPhillips,D.Z.3proprioception441philosophy,natureof82–4,88–91,97–8,protocolsentences98101–2,187–8,194–5,210–11,222–4,pseudo-problems98227–8,243,335–6,397,479–80;futurepsychoanalysis111of213psychology,empirical193–4,299–300,physicalism64–6,98,99,101,190;seealso423,451–2naturalisminmetaphysicspsycho-physicalparallelism278Piaget,Jean376,417,422Putnam,Hilary106,106n.6,154,Plantinga,Alvin482393–412,414–15,424,459–60Plato1,416,461Pylyshyn,Zenon463Poincaré,JulesHenri97,187politicaltheory323–4,362–8,432quantifiersandquantification9–10,28–9,Popper,Karl99,105,110–1635,70–1,74–5,100,356–9,474–5,possibleworlds357,359,467–8,483480–3quasi-analysis97494\nINDEXQuine,W.V.2,3–4,102,105,106,153,Russell,Bertrand1,2,21–44,51,60,62–4,164–5,181–204,207,288–90,336,68,70–1,75,95,97,99,106n.6,117,340,344,358,375,419,422–3,425,144,152,181,182,197–8,205,222,430,431,452,456,459,464,472,232,236,264–5,288,306,336–7,339,478–9,480341,359,467,469;HumanKnowledge:Itsquotationandextensionality198ScopeandLimits42–3;“OnDenoting”28–33,336–7,339,341;Principiaradicalinterpretation,seeinterpretation,Mathematica29,35–7,96–7;PrinciplesofradicalMathematics24–7,29Ramsey,F.P.34,42,69,104,139–47,400,Russell’sparadox25–6415,487Ryle,Gilbert3,117–23,255,397rationality329Rawls,John4,361–70Salmon,Nathan482realism98,103,105,194–5,202–3,Sartre,Jean-Paul433243–4,291,335,378,380–90;satisfaction127–8intentional452;metaphysical402–9,Schlick,Moritz2,77,95,97,98,99,205,448;modal481–4;naive214;in207,211perception415Schopenhauer,Arthur68reasonsandcauses88,277–8,296–300,science94–5,97,102,103,111–12,194,320335,371–6reducibility,axiomof35–6,140Searle,JohnR.3,178,221,222–3,225–6,reductionism96,214,251,299,438–9233,267–8,339,423–4,434–50reference13–16,73,201–2,336–8,341–2,self-reference290,293,319344–5,378–9,383,394–5,424–5,self-knowledge,seefirstpersonandfirst436–7;causaltheoryof379personpronounsreferring,seereferenceself-referentialexperiencesandcausationreflectiveequilibrium362,366–7,478,445–7479Sellars,Wilfrid4,239–53,293,408,430,Reichenbach,Hans94,98,104,152,154,431,463375semantics,seemeaning,theoryofReid,Thomas47,59sensation248–50,319,415;seealsorelativism,moral354perceptionrepresentation445–6,455–60sense(Sinn),seemeaning,theoryofresentment348–9sense-data38,58–62,222,223–5retributivism331sets34–5,201,289–90,484rhetoricacts,seespeechactsSextusEmpiricus429Richard’sparadox140similarity96,486rigiddesignators467–8Sinn,seemeaningRorty,Richard4,428–33situatedness441Ross,Alf170skepticism48–54,111,194,214,309–ruleofadjudication17111,336,340,342,344,403–6,429,ruleofchange171446ruleofrecognition171Skinner,B.F.454rules,constitutiveandregulative448–9Smart,J.J.C.413,414rules,linguistic420–1Smullyan,Arthur358rulesandrepresentations,seeprinciplesandsocialcontract362–3parameterssocialreality447–9rulesandrulefollowing84–6,88–9,socialrules170–1475–6Socrates227rulesofactionandcriticism246solipsism212;methodological459–60495\nINDEXspace95,97;seealsotimetruth-conditions379,384–8speakerandsentencemeaning255–6;seetruth-valuegaps337alsomeaning,theoryofT-sentences125–7,300–4,307speechacts219–21,434–8,443–4;Turing,Alan129,454,455,462,464locutionary,illocutionary,andTwinEarth395,405,410–11,459–60perlocutionary220–1types,theoryof34–5,97,140Sraffa,Piero69statements,seepropositionsunconscious,the442statesofaffairs38,73–4,290–2,294,underdeterminationoftheory191416–18;seealsofactsunderstandingandmeaning381–2;seealsoStevenson,C.L.175–80,327meaning,theoryofstimulation,sensory193unityofscience98Stout,G.F.57universalizability326,328Strawson,P.F.3–4,32,225,257,263,264,universals,seeparticularsanduniversals334–49,437utilitarianism45–6,173,329–33,361,subjectivism177,350–2,354–5365,366,367subjectivity441utterer’sandutterancemeaning254–6,subjects,seereference266–72;seealsomeaning,theoryofsubstitutionalquantification,seequantifiersandquantificationvalidation210supervenience417,438,484,485,487valuejudgments326Suppe,Frederick376valuerange18–20Suppes,Patrick376vanFraassen,Bas376swampman410,424variables28–9symbols,natureof165–6veilofignorance361,363synonymy,seeanalyticityandsynonymyVendler,Zeno435syntax440verdictives221verificationandverificationprinciple2,78,Tarski,Alfred101,102,124–8,47398–9,122,149–50,205–7,232–3,456;tautologies,seeanalyticityandsynonymyweakandstrong78,122,206;seealsoteleologicalsemantics457–8confirmationTharp,Leslie474–5verification-transcendenttruth-conditionstheories,natureof153–4384–90thinking87,235–6viciouscircleprinciple34,35,140thought(Gedanke)16–17,73–4,382;seeViennaCircle2,77–8,94,97–9,103,148,alsojudgment181–2,205thought,languageof73–4,415volition87–8,251;seealsofreedomandfreetime62,415–16,486–7willtone,linguistic13,379vonNeumann,John37Toulmin,Stephen375vonWright,G.H.274–80transcendentalarguments347transferprinciple329,331–2Waismann,Friedrich99translation,radical,seeinterpretation,Wallace,John474–5radicalWarnock,Geoffrey223,338,339trouserword224White,Stephen460truth101,103,209–10,294,300–1,Whitehead,A.N.34,35–6,42,71,95,97,305–7,338–9,340,416,430–1,473–4;198,288correspondencetheory225,306–7,338,will,seefreedomandfreewill416,430–1;redundancytheory141,Wien,Max95209–10,307,308Winch,Peter231496\nINDEXWittgenstein,Ludwig2–3,9,33,36,40–1,45,70;PhilosophicalInvestigations81–8;45,47–8,68–93,98,122,181,182,TractatusLogico-Philosophicus2,70–82,216n.4,225–6,227,231–2,274,275,288278–9,288,297,306,315,316–17,336,world-mates484346,348,350–1,352,378,382,384,Wright,Crispin381,385–9,390–1386–7,394–5,396–7,403,418,423,425,430,445,452,475–6;OnCertaintyZermelo,Ernst37497