The tripartite model of representation

Thetripartitemodelofrepresentation

PETERSLEZAK

ABSTRACTRobertCummins[(1996)Representations,targetsandattitudes,Cambridge,MA:Bradford/MIT,p.1]hascharacterizedthevexedproblemofmentalrepresentationas“thetopicinthephilosophyofmindforsometimenow.”Thisremarkissomethingofanunderstatement.ThesametopicwascentraltothefamouscontroversybetweenNicolasMalebrancheandAntoineArnauldinthe17thcenturyandremainedcentraltotheentirephilosophicaltraditionof“ideas”inthewritingsofLocke,Berkeley,Hume,ReidandKant.However,thescholarly,exegeticalliteraturehasalmostnooverlapwiththatofcontemporarycognitivescience.Ishowthattherecurrenceofcertaindeepperplexitiesaboutthemindisasystematicandpervasivepatternarisingnotonlythroughouthistory,butalsoinanumberofindependentdomainstodaysuchasdebatesovervisualimagery,symbolicsystemsandothers.Suchhistoricalandcontemporaryconvergencessuggestthatthefundamentalissuescannotariseessentiallyfromthetheoreticalguisetheytakeinanyparticularcase.

…ifmenhadbeenbornblindphilosophywouldbemoreperfect,becauseitwouldlackmanyfalseassumptionsthathavebeentakenfromthesenseofsight.(GalileoGalilei,1610)

Mentalrepresentation:“thetopicforsometimenow”

RobertCummins(1996,p.1)hasrecentlycharacterizedthevexedproblemofmentalrepresentationas“thetopicinthephilosophyofmindforsometimenow.”However,thisremarkissomethingofanunderstatement.Infact,thesametopicwascentraltothefamouscontroversybetweenAntoineArnauldandNicolasMale-brancheinthe17thcentury,andalsocentraltotheentirephilosophicaltraditionof“ideas”inthewritingsofLocke,Berkeley,Hume,ReidandKant.Thispatternofrecurrenceisastrikingfact.However,thecognitivescienceliteraturehasalmostnooverlapwiththatofthehistoryofearlymodernphilosophy.Thismutualneglectisremarkableinviewoftheintimateconnectionoftheirconcerns.Iamconcernedheretorevealsomethingoftherichandmutuallyilluminatingconnectionsbetweenthesedisjointliteratures.Inprinciple,suchmutualilluminationcanmakeavaluableandperhapsnovelcontributionbothtocontemporarycognitivescienceandalsotothescholarshipofearlymodernphilosophy.Thepossibilityofmutualbene󰁎tiseven

PeterSlezak,PrograminCognitiveScience,SchoolofHistory&PhilosophyofScience,UniversityofNewSouthWales,Sydney,NSW2052,Australia,email:p.slezak@unsw.edu.au

ISSN0951-50/print/ISSN1465-394X/online/02/030239–32Ó2002Taylor&FrancisLtdDOI:10.1080/0951508021000006085

240P.SLEZAK

moreevidentwhenwenoticethattheparallelsextendbeyondmerelyhavingcommonconcerns.Thatis,notonlytheproblemsofthe17thcentury,butthesamesolutionsarebeingrehearsedtodayattheforefrontofresearchincognitivescience.Descartes’de´ja`vu:Edelman’sTraite´del’homme

ApreliminaryindicationofthemodernrelevanceofearlyphilosophymaybeseeninEdelman’s(1998)workonperception.Despiteitsconcernwiththelatesttheoriesofperception,thecentralproblemisstatedintermsidenticalwiththatoftheentiretraditionofwriterson“ideas”sincethe17thcentury.Edelmanwrites:“Advancedperceptualsystemsarefacedwiththeproblemofsecuringaprincipled(ideally,veridical)relationshipbetweentheworldanditsinternalrepresentation.”Edelman’sboldnewsolution“isacallfortherepresentationofsimilarityinsteadofrepresenta-tionbysimilarity.”However,thismighthavebeentakenverbatimfromDescartes’sTraite´del’homme(1662/1972)orDioptrics(1637/1985)wherehesaid“theproblemistoknowsimplyhow[images]canenablethesoultohavesensoryperceptionsofallthevariousqualitiesoftheobjectstowhichtheycorrespond—nottoknowhowtheycanresembletheseobjects”(Descartes,1637/1985,pp.1,165).InthesameveinasEdelman,Meyering(1997)pointsoutthat,despiteitsadvocatestoday(Wright,1993),resemblancecannotbeanalyzedwithoutcircularity.Aswewillsee,thisissuearisesinescapablyaspartofadeeperproblemconcerningthenatureofrepresentation.

Thetripartiteschema

Inarecentarticle,Bechtel(1998,p.299)statestheessentialsofamoderntheoryofrepresentation:“Thereare…threeinterrelatedcomponentsinarepresentationalstory:whatisrepresented,therepresentation,andtheuseroftherepresentation.”

Z:SystemUsingY®Y:Representation®X:ThingRepresented

Bechtel’sschemaarticulatesatripartiteconceptionofideasasrepresentativesinterveningbetweenthemindandtheworld.Aswewillsee,amongtheproblematicassumptions,Bechtel’sdiagram(modi󰁎edhere)anddiscussioncruciallyfailtodistinguishinternalandexternalrepresentations(seeAbell&Currie,1999).Impor-tantly,Bechtel’sconceptioninthisregardisnotidiosyncratic,butaccuratelyre󰁏ectsanalmostuniversalconceptionincognitivescience(Dennett,1978a;Lloyd,forth-coming;Newell,1986,p.33;Rumelhart&Norman,1983).Aswewillnotepresently,thesametripartiteconceptioninthecaseofthepictorialtheoryofimagesinherentlyinvolvesthesameassimilationofinternalandexternalrepresentations,andtherebyencouragestheillegitimatepostulateofauserorexternalobserver—thenotorioushomunculus.IwillsuggestthatthesametacitassimilationofexternalandinternalrepresentationsisattheheartofSearle’s(1980)“refutation”ofsymbolicAIandalsoleadstothedoctrinethatwethink“in”language(Carruthers,1996;Slezak,2002).TheassimilationjustnotedinBechtelwillalsobeseenintheseeminglyunrelatedproblemofconsciousnessandthemind–bodyproblem(Place,1956).The

THETRIPARTITEMODELOFREPRESENTATION

241

tripartiteschemeappearsobviousandinnocuousenough,thoughithasbeenremarkablyfraughtwithdif󰁎culties.Indeed,theinescapabilityandubiquityofthispictureinoneformoranotherisapparentfromthefactthatBechtel’sdiagramisavariantoftheschemewhichweseethroughoutthelonghistoryofthesubject.“ThevisionofallthingsinGod”

Thus,forexample,nothingcouldseemmoreremotefrommoderntheoriesincognitivesciencetodaythanMalebranche’s(1712/1997)17thcenturydoctrineof“thevisionofallthingsinGod”—thetheorythatideasareobjectsofourperceptionthatexistinGod’smind.Onthecontrary,however,despitethetheologicaltrap-pings,itisinstructivetorecognizetheprofoundaf󰁎nityofMalebranche’sviewswiththoseattheveryforefrontoftheorizingtodayinpsychologyandarti󰁎cialintelli-gence:Malebranche’stheoryisjustBechtel’stripartitemodel(Nadler,1992),andthemodernproblemofrepresentationishowtoavoidthenotoriousdif󰁎cultiesclearlyarticulatedbyhiscriticArnauld(1683/1990).

Althoughtheseparallelsneedtobedefendedwithdetailedargumentandexegesis,itissigni󰁎cantthatFodorhasoccasionallymadetheaf󰁎nitiesexplicit.RecentlyhesuggeststhathisownRepresentationalTheoryoftheMind(RTM)maybeunderstoodonthemodeloftheclassicalempiricistconception:

Justforthepurposesofbuildingintuitions,thinkofmentalrepresentationsonthemodelofwhatEmpiricistphilosopherssometimescalled“Ideas”.Thatis,thinkofthemasmentalparticularsendowedwithcausalpowersandsusceptibleofsemanticevaluation.(Fodor,1998,p.7)

Inthislight,itishardlysurprisingthatmodernproblemsmightbesimplythereinventionofoldproblemsinanewguise.Fodorendorsestheclassicalconceptionofideasthoughherejectsaconceptionofrepresentationbymeansofresemblingimages.Hesays“Toa󰁎rstapproximation,…theideathattherearementalrepresentationsistheideathatthereareIdeasminustheideathatIdeasareimages”(1998,p.8).Despitesuchdisclaimers,wewillseethatthereisparticularironyinthefactthattheproblemforimagesmaybe,atadeeperlevel,theproblemforFodor’sRTMaswell.

IndependentlyofFodor’sexplicitallusion,hisconceptionofarepresentationaltheoryofmindhasalwaysbeenevocativeoftraditionalaccounts.Thus,itmaybejustafac¸ondeparler,butFodor’s(1978)analysisofpropositionalattitudeshasconsistentlybeenexpressedintermsof“relationsbetweenorganismsandinternalrepresentations”whichare“sentence-likeentities”(1978,p.198),thatis,“formulaeinanInternalRepresentationalSystem”(1978,p.194)andwhoseintentionalcontentsrefertothingsintheworld.Fodorearlierexplainedthisidiombythesameanalogywithtraditionaltheories:

Thisis,quitegenerally,thewaythatrepresentationaltheoriesofthemindwork.Soinclassicalversions,thinkingofJohn(construedopaquely)isa

242P.SLEZAK

relationtoan“idea”—viz.,toaninternalrepresentationofJohn.(1978,p.200)

Fodorspeaksofinternalrepresentationsasthe“immediateobjects”ofbeliefs,therebyrevealingtheclosesimilarityofhistheorywiththeclassicalLockeantheoryofideasasthe“immediateobjects”ofperception.ThisconceptionofinternalrepresentationsasbeinginarelationtoapersonisanexplicittripartiteschemewhichFodortakestospecify“apriori”conditionsonpropositionalattitudes.Thismaybe,atbest,anawkwardlocutionand,atworst,encouraginganotoriouslyproblematictheory.

ItisimportanttoacknowledgethatFodorandBechtel,likemosttheorists,arefullyawareofthefatalproblemlurkinghereinprinciple.However,awarenessoftheprobleminprincipledoesnotnecessarilyprecludefallingvictiminpractice.Forexample,wewillseethatproponentsofpictorialimageryhavebeenrepeatedlychargedwithcommittingthehomunculuserror.Notwithstandingtheiradvocates’protestationsofinnocence[1]andfullawarenessofthehazards,therearegroundsforseeingpictorialtheoriesasproblematicinthetraditionalmanner.Thechargeisthattherepresentationalformatcannotbemadetoworkwithouttacitlyinvokingtheveryabilitiesitissupposedtoexplain(Pylyshyn,1973,1978,1981,inpress;Slezak,1992,1995,inpress).Theerrorisnotcon󰁎nedtoimageryandismadeunwittinglybyfailingtonoticethattheaccessingmechanismscannotperformtheirfunctionontheirowninviewoftheparticularpropertiesascribedtotherepresentation.AsBechtel(1998,p.299)notes,aprocesswhichusesarepresentationas“standin”mustbecoordinatedwiththeformatoftherepresentation.However,thenatureoftheformatmaybesuchastorequireauserwhichisnotmerelyaprocessinaninnocuoussense.Speci󰁎cally,takinginternalrepresentationstobetoocloselymodeledonourexternalrepresentationalartifactsclearlyrisksrequiringthe“user”toshareourrelevantperceptualandcognitiveabilities,therebybeggingthequestioninthetraditionalmanner.Wewillseethattheassimilationofinternalandexternalrepresentationsinjustthiswayisfrequentlymadeasanexplicitdoctrine.Advertingtothevirtuesofcomputationalmodelswhichultimatelydischargetheirhomunculiandpaybacktheirloanonintelligence(Dennett,1978b)isnotsuf󰁎cientasapleaofinnocencetothesecharges(Kosslynetal.,1979).AsRorty(1979,p.235)hasputit,thereisnoadvanceinreplacingthelittlemanintheheadbyalittlemachineinthehead.Inparticular,Iwillarguelaterthatthecommonappealtoaninternalsymboliclanguageanalogoustoaformalsystemappearstobeguiltyofthesamecharge.Thedispute,then,isaboutwhetherthetheoreticalmodelssucceedinavoidingthewell-knowndif󰁎cultydespitetheirauthors’intentions.

Thus,althoughrejectingthechargeofbeing“ontologicallypromiscuous”(1978,p.179),Fodor’slocutionmaybesymptomaticofthedeepdif󰁎cultieswhichpervadetheproblemofrepresentation.Signi󰁎cantly,Fodorsaysthathisconceptioncorrespondspreciselywiththeviewthatpsychologistshaveindependentlyarrivedat.TotheextentthatFodoriscorrectinthisobservation,notonlyphilosophershavebeenpreytothedeeplycompellingmistakesoftheorizingaboutthemind.

ItisnoaccidentthatGibson’s“ecological”approach,likethecloselyrelated

THETRIPARTITEMODELOFREPRESENTATION

243

“situatedcognition,”aretheoriesofdirectrealismwhichhavebeenproposedasalternativestotherepresentationalismofcomputationaltheories.ThisismerelyoneforminwhichtheMalebranche–Arnaulddebateisbeingrehearsedtoday.ThiscelebrateddebateisdescribedbyNadler(19)asadebatebetweenan“objecttheory”ofideasandan“acttheory,”respectively.Heexplains

…theobjecttheoryofideasinvolvesacommitmenttoarepresentationalistorindirectrealisttheoryofperception,suchasMalebranche(and,onthetraditionalreading,Locke)putforth.Anacttheoryofideas,ontheotherhand,formsthecoreofArnauld’sperceptualdirectrealism.Ifideasarerepresentationalmentalacts[ratherthanentities],thentheycanputthemindindirectcognitivecontactwiththeworld—nointerveningproxy,notertiumquid,getsintheway.(19,p.6)

AsNadler(19,p.6)pointsout,Malebranche’s“visioninGod”isa“theologizationofcognition”accordingtowhichthecontentsofourownthoughtsaredependentupontheirdivinesourceinthemindofGod.However,althoughMalebranche’stheologicalandepistemologicalconcernsarewoventogether,thethreadsmaybeseparatedandhisdoctrineofideasidenti󰁎edasthefamiliar,compellingandwidelyheldtheoryuntilthepresenttime.Althoughthereisroomforscholarlydispute[2],mostcommentatorsshareareadingofMalebrancheaccordingtowhichideasareintermediariesorproxiesrepresentingexternalobjectsandinterveningbetweenthemindandtheworld.Thissame“representativetheoryofperception”hasbeenmorefamiliarasJohnLocke’s“veilofideas”inthetraditionreferredtoasthe“wayofideas.”Onthisview,ideasareinternalmentalobjectsofsomekindtowardwhichthemind’soperationsaredirected.NadlerechoesBechtel,describingMalebranche’stheoryasassumingthattherearethreeelementsinthenormalperceptionorknowledgeoftheworld(Nadler,19,p.81).AsArnauldexplainedinhiscritique,Malebranche“regardsthisrepresentationasbeingactuallydistinctfromourmindaswellasfromtheobject”(1683/1990,p.63).AcrucialandfrequentlyquotedpassagefromMalebranchehimselfexplains:

Thus,itdoesnotseethembythemselves,andourmind’simmediateobjectwhenitseesthesun,forexample,isnotthesun,butsomethingthatisintimatelyjoinedtooursoul,andthisiswhatIcallanidea.Thus,bythewordidea,Imeanherenothingotherthantheimmediateobject,ortheobjectclosesttothemind,whenitperceivessomething,ie.,thatwhichaffectsandmodi󰁎esthemindwiththeperceptionithasofanobject.(1712/1997,p.217)

Situatedcognition:the“canonical”cottagecheesecase

Signi󰁎cantly,JohnYoltonhasexpressedahopethatfromthestudyofearlythinkers“wemaybeabletounderstandhowwecanhaverepresentation(cognitivity)andrealismtoo”(1996,p.x).Thisis,ofcourse,acommentontheperennialproblem

244P.SLEZAK

posedbythetripartitescheme.Yolton’sremarksonearlierthinkersisapttodescribethecentralproblemoftheoriestoday:

Thepivotalconceptfortheaccountsofperceptualacquaintanceintheseventeenthandeighteenthcenturiesisthatofobjectspresenttothemind.Dependingonhowthatconceptwasinterpreted,thoseaccountsmovedbetweenanindirectnessofknowledge(becauseonlyarepresentative,proxyobjectcanbepresenttothemind)andastrongdirectrealismwheretheobjectknownwas,insomeway,itselfpresenttoorinthemind.(1984,p.6)

Recentproponentsof“situatedcognition”havebeencomplainingofexactlythesameindirect,mediatedconceptionincomputationaltheoriesofcognition,recog-nizingthattheseembodyessentiallytheLocke–Malebranche’sschemeofrepresen-tationsinterveningbetweenmindandworld.Forexample,Greeno(19)unknowinglyechoesArnauld:

Iampersuaded…thatinnormalactivityinphysicalandsocialsettings,weareconnecteddirectlywiththeenvironment,ratherthanconnectedin-directlythroughcognitiverepresentations.

…Anindividualinordinarycircumstancesisconsideredasinteractingwiththestructuresofsituationsdirectly,ratherthanconstructingrepresen-tationsandinteractingwiththerepresentations.(19,p.290)

GreenocitestheWeightWatcherwhohadstudiedcalculusbutneverthelessanswersaquestionaboutadailyallotmentofcottagecheesebymeansofasimple,directlyphysical,operationdividingupaportionofcheese,ratherthanbyanysymboliccomputationsuchasamultiplicationonfractions.Ergo,reasoningisnotsymbolicbut“situated.”TheWeightWatchercaseissupposedtoillustratethethesisthattheperson’sactionsaresomehowunmediatedbymentalrepresentations[3].ThecauseforGreeno’sconcernisthemodernversionofLocke’sview:“Itisevidentthatthemindknowsnotthingsimmediately,butonlybytheinterventionoftheideasithasofthem”(Locke,1690,BookIV,ChapterIV).

DeuxCarte´siens:plusc¸achange,plusc’estlameˆmechose

ItisamusingtonoticehowMalebranche’sattempttoarticulatethispictureisechoedtodaybyFodor.Malebranchewrote:

Ithinkeveryoneagreesthatwedonotperceiveobjectsexternaltousbythemselves.Weseethesun,thestarsandanin󰁎nityofobjectsexternaltous;anditisnotlikelythatthesoulshouldleavethebodytostrollabouttheheavens,asitwere,inordertobeholdalltheseobjects.(1712/1997,p.217)

Fodorwritesinthesamevein:

Itis,torepeat,puzzlinghowthoughtcouldmediatebetweenbehaviorand

THETRIPARTITEMODELOFREPRESENTATION

245

theworld…Thetroubleisn’t—anyhow,itisn’tsolely—thinkingthatthoughtsaresomehowimmaterial.It’sratherthatthoughtsneedtobeinmoreplacesthanseemspossibleifthey’retodothejobthatthey’reassignedto.Theyhavetobe,asitwere,“outthere”sothatthingsintheworldcaninteractwiththem,buttheyalsohavetobe,asitwere,“inhere”sothattheycanproximallycausebehavior.…it’shardtoseehowanythingcouldbeboth.(Fodor,1994a,p.83)

MalebrancheandArnauldarenotchosenformentionhereatrandomormerelyinretrospectfortheircurrentinterest.AsGaukroger(1990)hasnoted,Malebranche’sSearchaftertruthwas“themostin󰁏uentialphilosophicaltreatiseofthesecondhalfoftheseventeenthcentury,eclipsedonlyattheendofthatcenturybyLocke’sEssay”(1690/19,p.1).

Inparticular,Malebranche’sdoctrineswereatthecenterofafamouscontro-versywithAntoineArnauldwhosetreatiseOntrueandfalseideas(1683)wasareplytoMalebranche.Indeed,thisdebatewasnotonlyamajore´ve´nementintellectuelofitstime,asMoreau(1999)hasrecentlydescribedit,butonewhoseechoesmaybeheardthroughoutthesubsequentcenturiesofspecu-lationaboutthemind.Moreau’s(1999)recentbook-lengthstudyinFrenchisperhapsthe󰁎rstdevotedtothedisputeassuch,andatteststoitsimportanceasanintellectualcausece´le`breinthe17thcentury[4].Nadlerwritesthat,follow-ingthe󰁎rstroundwithArnauld’scritiqueofMalebranche,

Forthenextdecade,untilArnauld’sdeathin1694,thesetwomenengaged

inapublicdebatethatattractedtheattentionofintellectualcirclesthroughoutEurope.Sidesweretakeninarticles,reviewsandlettersintheforemostjournalsoftheday,andtheissuesweredebatedbyothersashotlyastheywerebytheprimarycombatantsthemselves.…itremains…oneofthemostinterestingepisodesinseventeenth-centuryintellectualhistory.(19,p.2)

Nadleraddsthatthedebateisindispensableforunderstandingthecentralphilosophicalissuesoftheperiod,andthisisespeciallytrueinrelationtotheworkofDescartes.AsthetitleofMoreau’s(1999)bookindicates,MalebrancheandArnauldwere,despitetheirdifferences,󰁎rstandforemostDeuxCarte´siens.ArnauldinsistedthathisconceptionswerefaithfultothoseofDescartesand,asNadlernotes,“Arnauldwouldremaincommittedtolapense´ecarte´siennefortherestofhislife”(19,p.34).Thoughdifferingoverthedoctrineofideasandperceptualacquaintance,bothacceptedthefundamentalprinciplesofDescartes’sphilosophy(Nadler,19,p.59).Arnauld’sviewtakesonspecialinteresttodaysincehiscritiqueofMalebrancheconstitutesawayoutoftheanalogousproblematicconceptionsofmoderncognitivescience.

246P.SLEZAK

Precursors:pointlessexercise?

`proposofhistoricalre󰁏ections,withsomejustice,StephenGaukroger(1996)inhisA

landmarkintellectualbiographyofDescarteshasdescribedasa“pointlessexercise”theeffortstoshowtheextenttowhichDescartes,forexample,wasaprecursorofmoderncognitivescience.However,insomecaseswemaydiscernsomethingmorethanfortuitous,independentreinvention.Thereisamoreinterestingkindofrecurrencewhichdeservesattentionbecauseitisamanifestationofdeeper,andthereforemoreilluminating,causes—achronicmalaisewhoserecurrenceissymp-tomaticofdeeppathology.

Notinganticipationsofcurrenttheoriesislikelytoberevealinginbothdirec-tions:precursorsofcognitivescienceprovideanindependent,extensivesourceofinsightintocontemporaryissuesand,conversely,arethemselveselucidatedinnovelwaysunavailabletotraditionalscholarship.(Forpreliminarystepsinthisdirection,seeYolton,1984,1996,2000;Slezak,1999,2000.)Thus,beyondmerelynotingtheparallels,IwouldliketooffersomepreliminarydiagnosisofthemalaiseanditsetiologyalongthelinesofArnauld’sdefenseofDescartes’viewagainstMale-branche.

Tables&chairs:bumpingintothings

FromYolton’sstatementofearlierconcerns,wecanseetheirrelevancetocontem-poraryissues:

Fromthescholastics’intelligiblespecies,throughtheCartesian’sobjectivereality,toBerkeley’sandHume’stalkofideasastheverythingsthem-selves,weseewritersonperceptionstrivingforsomewaytosaythatweperceivephysicalobjects.…Oneofthewaysinwhichsomeofthewriterstriedtopreservetheaccuracy,ifnotthedirectness,ofperceptualawarenesswasbytalkingofaconformityoragreementbetweenideasandobjects;otherwisetheysaidideasrepresentobjects.(1996,pp.1–2)

Thisis,ofcourse,justthemodernproblemofintentionalityor“psychosemantics”whichCumminsdescribesasjustthatofsaying“insomeilluminatingway,whatitisforsomethinginthemindtorepresentsomething”(1996,p.1).Despitetheseemingsimplicityofthephenomenon,theburgeoningliteratureatteststothefactthatthereisaconsensus,atleast,onFodor’sjudgmentthat“ofthesemanticityofmentalrepresentationswehave,asthingsnowstand,noadequateaccount”(1985b,p.28).Typically,Stalnaker,too,says,“Thereislittleagreementabouthowtodosemantics,orevenaboutthequestionsthatde󰁎nethesubjectofsemantics”(1991,p.229).Likewise,B.C.Smithconfesses,“Itshouldbeadmittedthathowthisallworks—howsymbols‘reachoutandtouchsomeone’—remainsanalmosttotalmystery”(1987,p.215).

Inareportonthestateandprospectsofinterdisciplinarycognitivescience,Fodor(1985a)jokedthatphilosophersarenotoriousforhavingbeenpreytoabsurd,eccentricworriessuchasthe“fearthatthereissomethingfundamentallyunsound

THETRIPARTITEMODELOFREPRESENTATION

247

abouttablesandchairs.”Nevertheless,heoptimisticallycontrastedsuch“mere”philosophicalworrieswiththosethatoccasionallyturnouttobe“real,”aswiththerepresentationalcharacterofcognition.Triumphantly,Fodorpointstothefactthattoday,unlikeotherproprietaryconcerns,thisproblemisnolongerjustaphiloso-phers’preoccupationbecauseitssolutionhasbecomeofgeneralimportanceasapreconditionofprogressinseveraldisciplinesofcognitivescience.However,thereisanacuteunintendedironyinFodor’scontrast,becausetheproblemofrepresen-tationattheforefrontofcognitivesciencetodayis,infact,identicalwiththephilosophicalanxietyabouttablesandchairs.Invariousmoreorlessindependentdomains,cognitivescientistshavesimplyrediscoveredtheverysamesterileconun-drumswhichhavekeptphilosophersbusysinceDescartes.

WeseearevealingcluetothiscommonalityinJackendoff’s(1992,p.161)questionwhichisareductioadabsurdumofcontemporarysymbolic,computationaltheories:Inviewofthe“internalism”and“narrow”syntacticcharacterofcomputa-tionalsymbols,Jackendoffasksfacetiously:“Why,ifourunderstandinghasnodirectaccesstotherealworld,aren’twealwaysbumpingintothings?”Jackendoff’ssatireisevocativeofSamuelJohnson’sfamousresponsetoBerkeley’s“ingenioussophistry”:“Irefuteitthus,”hesaid,thatis,bykickingastone.Inbothcases,appealingtobumpingintothings,theresponsesbringintoreliefthewayinwhichclassicalandmoderntheoriesentailadisconnectionofthemindandtheworld.ThesuggestiveparallelbetweenJackendoffandJohnsonisnoaccident.JackendoffcapturespreciselytheparadoxchargedagainstLockeandalsoMalebranche,whoNadler(1992,p.7)says“isoftenportrayedbyhiscriticsasenclosingthemindina‘palaceofideas,’forevercutofffromanykindofcognitiveorperceptualcontactwiththematerialworld.”Ofcourse,Berkeley’sidealismisjusttheworryabouttherealityoftablesandchairs,andBerkeley’sreactiontoLocke’s“ideas”isanalogoustoFodor’sreactiontoSimon’ssymbols—“methodologicalsolipsism.”Seeminglyisolatingthoughtinarealmofitsown,therepresentationsintervenebetweenmindandworld—twoitemswhosesystematicconnectionswitheachotherbecomemysterious.Thetraditionalproblem,rediscoveredincognitivescience,ishowtomakesenseoftherelationbetweenthesethreeelements—mind,representationandworld—seeminglyessentialtoanymodelofcognition.The“philosophicktopick”ofideas

Inhisrecentbook,Yolton(1996,p.43)mentionstheanonymousauthorofapamphletwrittenin1705titledPhilosophickessayconcerningideaswhosays,“ThereishardlyanyTopickweshallmeetwiththattheLearnedhavedifferedmoreaboutthanthatofIdeas.”Itisaremarkablefactthatlittlehaschangedinthisregardconcerningthe“Topick”indispute,theunderlyingreasonsfortheproblemandthesolutionsadopted[5].Althoughtheterminologyofef󰁏uvia,essences,modesandsubstanceshasbeenreplacedbyinformationprocessingjargon,theessentialissuesareunchanged.Thus,Palmer’s(1978)articleon“Fundamentalaspectsofcognitiverepresentation”says“Anyonewhohasattemptedtoreadtheliteraturerelatedtocognitiverepresentationquicklybecomesconfused—andwithgoodreason.The

248P.SLEZAK

󰁎eldisobtuse,poorlyde󰁎ned,andembarrassinglydisorganized.”AfterenumeratingadozendistinctconceptionsPalmeradds“Thesearenotcharacteristicsofascienti󰁎c󰁎eldwithadeepunderstandingofitsproblem,muchlessitssolution”(1978,p.259).ThesituationdoesnotappeartohaveimprovedinthetwodecadessincePalmerwrote.ItisnoaccidentthatPalmer’slamentandhislitanyechoYolton’sanonymousauthorbecausethetheoreticaldisarray,likethedoctrinesthemselves,arenotunrelated.Suf󰁎cientevidenceofthisisthefactthatthe18thcenturyauthor’sanalysisoftheproblemanditscausesremainsappropriatetoday.

…inconsideringtheMind,somemendonotsuf󰁎cientlyabstracttheirThoughtsfromMatter,butmakeuseofsuchTermsascanproperlyrelatetoMatteronly,andapplythemtotheMindinthesameSenseastheyarespokenofMatter,suchasImagesandSignatures,Marks,andImpressions,CharactersandNotesofThings,andSeedsofThoughtsandKnowledge.(quotedinYolton,1956/1993,p.96)

Translatedintocurrentterminology,thisisaninsightfuldiagnosisofthelatestdisputesconcerningrepresentationincognitivesciencetoday.Itis,infact,are-statementofArnauld’sorthodoxCartesianviewwhichinsiststhatmentalrepre-sentationscannotbeproperlycharacterizedintermstakentoodirectlyfromthoseaptforourexternal,materialrepresentations—theproblemof“original”versus“derived”intentionality.“MalebrancheanTheatre”?

Dennett’s(1991)referencetoa“CartesianTheater”hasgivenwidecurrencytothistermandtherebyservedtodrawattentiontothesupposedprovenanceofaconceptionwhichis,indeed,attheheartofphilosophicalpuzzlesaboutthemindandconsciousness.Indeed,therelatedmistakesofthe“Theater”andthehomun-culusareattheheartofmuchtheorizingincognitivescience.However,fullyacknowledgingthevalueofDennett’sanalysis,itremainsthathisterminology,atleast,perpetuatesanhistoricalsolecism.ConcedingthatDennettwasnotconcernedwithexegetical,scholarlyniceties,itremainsimportanttocorrectaseriouserrorofmisattribution.The“Theater”inquestionismoreappropriatelyascribedtoMale-branchethantoDescartes.Although“MalebrancheanTheater”doesnothavethesamepleasingsonority,thereisgoodphilosophicalreasontocorrecttheusagebesidesmerehistoricalpedantry.Itisimportanttorecognizethatacommitmenttothepictureofaninnerpersonobservingasceneonthestageofconsciousnessisindependentof,anddoesnotfollowdirectlyfrom,dualism.Dennettrecognizesthisinhistalkof“Cartesianmaterialism”(1991,p.107)whichhesaysis“theviewyouarriveatwhenyoudiscardDescartes’dualismbutfailtodiscardtheimageryofacentral(butmaterial)Theaterwhere‘itallcomestogether’.”However,DennettseemstoblameDescartesforholdingthis“Theater”conceptiontogetherwith,ordirectlyasaconsequenceof,hisdualism.Nevertheless,contrarytoDennett’simplication,whileundeniablyaCartesiandualist,DescarteswasemphaticallynotaCartesianmaterialistaswell.Thatis,hewasnotguiltyoftheTheaterfallacyinthis

THETRIPARTITEMODELOFREPRESENTATION

249

sense.Onthecontrary,despitepositingtheanatomicalconvergenceofnerve󰁎lamentsinthepinealgland,Descartesdidnotsubscribetothepictureofanobserverintheproblematic“Theater”becauseheexplicitlyarguedagainstpositingconcomitantrepresentationsofakindwhichwouldrequirethenotorioushomun-culus.Descartes’“ghostinthemachine”isnotthisobserver,butapositbasedonentirelydifferent,independentconsiderations—namely,the“Cogito”argumentandthelimitationsofmachines(DiscourseV).TherationaleforDescartes’immaterialsoulisquitedifferentandindependentofthe“Theater”conceptionwhichheexplicitlyrepudiatesinhisDioptrics(1637)andTreatiseofman(1662).

QuiteapartfromtheevidenceofDescartes’owntexts,amplesupportfortheseascriptionsisfoundinArnauld’swritingswhicharticulateanact-theoryasanalternativetoMalebranche’srepresentative“veilofideas”(seeNadler,19,pp.34,118,126,footnote36).Arnauldsawthis“directperception”viewasfaithfultoDescartes,andDescartessaysincorrespondencethatArnauld“hasenteredfurtherthananyoneelseintothesenseofwhatIhavewritten”(ATIII,331).Thus,Arnauld’sCartesianpositionadoptspreciselyDennett’sstanceagainstthe“MalebrancheanTheater.”

Reinventions:synchronicanddiachronic

IfMalebrancheandArnauldanticipatedcontemporaryconcernsaboutrepresenta-tionincognitivescience,thenitisclearthatthecurrenttheoreticalproblemhasnothingtodowiththetheoreticalframeworkofsymbolic,computationalapproachesasuniversallyassumed.Itisparticularlysigni󰁎cant,then,thattherecurrencesofinterestherearefoundnotonlythroughouthistorybutinseeminglyunrelateddomainsofcognitivesciencetoday.Thisrecurrenceofessentiallythesamedisputeinwidelyvaryingcontextscon󰁎rmsthattheunderlyingproblemdoesnotariseessentiallyfromthespecialfeaturesofanyoneofthem.Givenaseductivemistakeconcerningrepresentationassuch,multiple,seeminglyindependent,reinventionsarejustwhatwewouldexpectto󰁎nd.Iwillpresentlysuggestthatwemaydiscernthesameunderlyingproblemattheheartofnotoriousdisputessuchasthe“ImageryDebate,”Searle’sChineseRoomconundrum,thethinking-in-languagedebateandanumberofotherswhichhavebeenprominentandrecalcitrant.Norepresentations?

The“cognitiverevolution”ofthe1960swascharacterizedbyarenewedrecognitionoftheindispensabilityofinternalrepresentationsfollowingtheirrepudiationbySkinnerianbehaviorism.Thereisconsiderableironyinrecentapproacheswhichappeartorejectinternalrepresentationsonceagain(Brooks,1991;Clark&Toribio,1994;Freeman&Skarda,1990;Greeno,19;vanGelder,1998).NotwithstandingEliasmith’s(1996)claim,theseviewsarenotplausiblyseenasareturntobehav-iorismsince,strictlyspeaking,theydonotrejectinternalrepresentationsatall(seeMarkman&Dietrich,2000).Nevertheless,theseapproachesandtheirrhetoricaresymptomsoftheprofounddif󰁎cultiesposedbythephenomena.Particularlyinview

250P.SLEZAK

oftherevolutionaryhypeassociatedwiththelatestfashions,itissoberingtonoticethatArnauld’s(1683/1990)critiqueofMalebrancheexactlypre󰁎gurestheserecentattacksonrepresentationaltheories.ItisnocoincidencethatArnauld’streatiseOntrueandfalseideasisconcernedtorepudiatewhathedescribesas“imaginaryrepresentations,”saying,“Ican,Ibelieve,showthefalsityofthehypothesisofrepresentations”(1683/1990,p.77)for“onemustnotmakeuseofallegedentitiesofwhichwehavenoclearanddistinctideainordertoexplaintheeffectsofnature,whethercorporealorspiritual”(1683/1990,p.65).

Illusions&misrepresentation:“curiousandmelancholyfact”

Inseekingtounderstandthepersistenceandrecalcitranceoftheproblemsofintentionality,itisinstructivetoexamineonefacetoftheissuewhichrevealstheseductivenessofthemistake.Theproblemofmisrepresentationhasarisenforcausalorco-variationtheoriesofintentionalcontent(Dretske,1986;Fodor,1994a)sincethesetheoriesseemtobeunabletocapturethewayamismatchmightarisebetweenarepresentationandtheworld.Ifamentalesetoken“mouse”mightbecausednotonlybymicebutalsobyshrews,thenthesymbolmustipsofactomean“shrew”andcannotbeinerror.Itseemsnottohavebeennoticedthatthismodernphilosophicalproblemofmisrepresentationisavariantofthewell-knownclassical“ArgumentfromIllusion”(Reynolds,2000)whichwasemployedinsupportofLocke’s“ideas”andA.J.Ayer’s(1940)sense-dataastheimmediateobjectsofperception.Theparallelshouldnotbesurprisingsince,afterall,anillusionintherelevantsense(thatis,anhallucination)ispreciselyamisrepresentation.Theproblemofmisrepresen-tation,then,appearstobeoneoftheloosethreadswhichmaybepulledtounraveltherestofthetangledball(seeSlezak,forthcoming).

RespondingtoAyer(1940),Austin(1962,p.61)remarkedonthe“curious”and“melancholyfact”thatAyer’spositiononsense-dataechoesthatofBerkeley.ItisanevenmoremelancholyfacttodaythatFodor’s“real”problemsofrepresenta-tionalsoechoBerkeley.QuestionsofveridicalityforLocke’sideasandAyer’ssense-dataarosefrompreciselythesameassumptionsasFodor’s—namely,theassumptionofbeingabletocomparerepresentationsandtheworld.

TheearlierFodorianpassagefromMalebrancheisfollowedbyaparagraphthatexplicitlyarticulatesthe“ArgumentfromIllusion”:

Itshouldbecarefullynotedthatforthemindtoperceiveanobject,itisabsolutelynecessaryfortheideaofthatobjecttobeactuallypresenttoit—andaboutthistherecanbenodoubt;butthereneednotbeanyexternalthinglikethatidea.Foritoftenhappensthatweperceivethingsthatdonotexist,andthatevenhaveneverexisted—thusourmindoftenhasrealideasofthingsthathaveneverexisted.When,forexample,amanimaginesagoldenmountain,itisabsolutelynecessarythattheideaofthismountainreallybepresenttohismind.Whenamadmanorsomeoneasleeporinahighfeverseessomeanimalbeforehiseyes,itiscertainthatwhatheseesisnotnothing,andthatthereforetheideaofthisanimalreally

THETRIPARTITEMODELOFREPRESENTATION

251

doesexist,thoughthegoldenmountainandtheanimalhaveneverexisted.(1712,p.217)

Illusionsinthissensearecasesinwhichthecorrespondencebetweenrepresen-tationsandworldfails—misrepresentationsofexactlythesortrelevanttothecontemporarypuzzleforsymbolic,computationalaccountsofcognition.Inthemoderncase,asposedbyFodor(1994a)andDretske(1986),theproblemis,givencausationbetweentheseelements,howtoexplainthepossibilityofillusion;intheclassicalcasetheproblemis,givenillusion,howtoexplaincausation.Themodernproblemofmisrepresentationarisesbecausecausalorcorrelationaltheoriesdon’tappeartopermitadistinctionbetweentrueandfalserepresentations.Ifadogcausesarepresentationof“cat”inmentalese,onthecausalaccountitmustipsofactocountasmeaning“dog”andis,therefore,notamistakenrepresentationofcat.Con-versely,theclassicalArgumentfromIllusion,startsfromtheotherend,asitwere.Beginningwiththedistinctionbetweentrueandfalserepresentations,theArgumentrecognizesthatthesecannotbothbecorrelatedwithanexternalreality,andconcludesthatinbothveridicalandnon-veridicalcasestheremustbesomeotherobjectofdirectperception,the“idea”orsense-datum(seeReynolds,2000).Inviewoftheseanalogies,therefore,IsuggestitisnocoincidencethatFodor’s(1980)“methodologicalsolipsism”isstronglyevocativeofaBerkeleyanidealism.

TheMalebranche–Lockeargumentforrepresentativeideasrecognizesthatillusionscannotbecausedintheusualwaybyexternalobjects—essentiallyFodor’spuzzleexpressedinreverse:Fodorarguesthat,ifideasarecausedbyexternalobjects,wecan’thaveillusions.Theparallelshereappeartobemorethansuper󰁎cialorterminological[6].MyclaimisnotthattheproblemofmisrepresentationandtheArgumentfromIllusionaredirectedtowardsthesameends,butonlythattheyarisefromanidenticalconceptualschemeandaremirror-imagesofoneanother:Theclassicalargumentasserts:ifthereareillusions,thenthereisnodirectconnectionorcorrelationwiththeexternalworld(i.e.theremustbeintermediateobjectsofperception);conversely,Fodor’sargumentasserts:Ifthereisadirectconnection(i.e.causalcorrelation)withtheexternalworld,thentherecanbenoillusions.Theseareequivalentcontrapositives:ifwetake“I”5illusion,“C”5correlation,thentheMalebranche–Lockepropositionis[I®,C]andtheFodor–Dretskepropositionis[C®,I].

Inpassing,wemaynotethatadegreeofconfusionhasbeenintroducedinthesediscussionsbythefailuretodistinguishcruciallydifferentkindsof“illusion.”Anillusioninthesenserelevanttotheargumentconcerningideas,sense-dataorrepresentationsis,strictlyspeaking,hallucination.However,certainotherphenom-enacommonlyreferredtoas“illusions”inthiscontextsuchasmiragesorbentsticksinwaterarenotillusoryatallinanimportantsense.TheseareveridicalperceptionsofthelightpatternsenteringtheeyeunlikecognitiveerrorssuchastheMu¨ller–Lyerillusion.RichardGregory(1997),forexample,hasexplicitlyassimilatedthesephenomena,butnotheoryofcognitiveprocessescouldexplainthe“illusion”inthecaseofmiragesandseeminglybentsticksduetorefractedlight.Gregory’smistakeinthisregardisinterestingandperhapsnomeremistake.Assumingthatour

252P.SLEZAK

knowledgeoftheactualconditionsintheworldmustbeusedincharacterizingmentalrepresentationsispreciselytheseductiveerrorwhichIamconcernedtoexposeinitsvariousguises.Astickappearingbentinwaterisacaseof“theworldgonewrong”injustthesenseofthisfelicitousphraseusedbyFodor,aswewillseepresently,

Truthconditionsasexplanatory?

Inboththecaseofmisrepresentationandthatofillusionthepuzzlearisesfromacommitmenttothetripartiteconceptioninwhichrepresentationsintervenebetweenthemindandtheworldandaresomehowcorrelatedwithit.Inparticular,thequestionsofveridicalityforLocke’sideasarosefromtheimpossibilityofanycomparisonbetweenrepresentationsandtheworld,exceptfromtheperspectiveofanindependentoutsideobserver.AsBerkeleyrecognized,theverydistinctionbetweentrueandfalseideascannotbemadewithoutcomparingrepresentationsandtheworld.Ofcourse,thisperspectiveisunavailabletotheminditself.Correspond-ingly,anexplanatorytheorycannotmaketacitappealtosuchaperspectivewithoutcommittingthehomunculuserror.Thismeansthattheveridicalityorotherwiseofmentalrepresentationdoesnotserveanexplanatoryroleandis,therefore,notalegitimatepartofatheoryofmind.InBerkeley’sidealistresponsetothisproblemwecanseetheprecursortoFodor’sproblemarisingfromacommitmenttotruthconditionsformentalrepresentations.SecuringtheveridicalconnectionbetweenrepresentationsandtheworldthroughcausationsimplybindstheminsuchawayastoprecludeerrorandthuscausationfunctionsforFodorinthewaythatamysteri-ouscorrespondenceworkedforLocke.

Ofcoursetheproblemofexplainingerrorandthatofexplainingtrutharetwosidesofthesamecoin.Accordingly,thepuzzleofmisrepresentationissymptomaticoffundamentalproblemsintheconceptionofmentalrepresentationsassemanti-callyevaluable.FodorisemphaticaboutthecentralityoftruthpreservationforthecomputationalRTM.RegardingthefactthatmentalprocessestendtopreservesemanticpropertiesliketruthFodorsays

Thisis,inmyview,themostimportantfactweknowaboutminds;nodoubtit’swhyGodbotheredtogiveusany.Apsychologythatcan’tmakesenseofsuchfactsasthatmentalprocessesaretypicallytruthpreservingisipsofactodeadinthewater.(Fodor,1994a,p.9)

Fodor’sdilemmaarisesfromthefactthatcontentdoesn’tappeartosuperveneonmentalprocessesand,therefore,“semanticsisn’tpartofpsychology”(Fodor,1994a,p.38).

Mypoint,then,isofcoursenotthatsolipsismistrue;it’sjustthattruthreferenceandtherestofthesemanticnotionsaren’tpsychologicalcate-gories.(1980,p.253)

Itseemsthatwecan’tdopsychologywiththesemanticnotions,butwecan’tdo

THETRIPARTITEMODELOFREPRESENTATION

253

psychologywithoutthemeither.ThisformulationofFodor’sdilemmaisreminiscentofaremarkbyDennettinaquitedifferentcontextinwhichheexplained:

…psychologywithouthomunculiisimpossible.Butpsychologywithho-munculiisdoomedtocircularityorin󰁎niteregress,sopsychologyisimpossible.(Dennett,1978b,p.123)

MysuggestionisthatFodor’sandDennett’sdilemmasappeartobethesamebecauseatrootthepuzzleofsemanticsisaversionofthehomunculusproblem.JustasDennett(1978a,p.122)pointedoutthatnothingisintrinsicallyarepresentationofanythingbutonlyforsomeonewhoistheinterpreter,sonothingisamisrepresen-tationforthesamereason.Thatis,Fodor’scurrentproblemofmisrepresentationmightbeaccountedforbynotingthatitarisesfromthedemandfortacitlyadoptingthestanceofexternalinterpreter:Theveryproblemitselfcannotbecoherentlyformulatedexceptintermsofjudgmentswhicharenotpartoftheexplanatoryenterprise.Theveridicalityofrepresentationsisnotapropertywhichcanplayanyroleinthefunctioningofrepresentationsortheexplanationofthem.Likethepictureonajigsawpuzzle,themeaningofrepresentationsconceivedassemanticallyevaluableinthiswayisforourownbene󰁎tandnotintrinsictothearrangementsofinterlockingcomponents.Thesenseinwhichamentalrepresentationdoesitsworkisnotonewhichrequiresjudgmentofitstruth-valuesincethisisonlypossiblefromthepointofviewofanobserver,thetheorist,forwhomtherepresentationisconstruedasanexternalsymbol.

Theveryconcernwithmisrepresentationarisesfromtacitlyadoptingaques-tionableassumptionendorsedbyDavidson(1975)thathavingabeliefrequiresalsohavingtheconceptofbelief,includingtheconceptoferror.Davidsonsays“someonecannothaveabeliefunlessheunderstandsthepossibilityofbeingmistaken,andthisrequiresgraspingthecontrastbetweentruthanderror—trueandfalsebelief”(1975,p.22).However,itseemsthatanimalsmighthavebeliefseveniftheyareunabletoknowthattheyhavethemandre󰁏ectontheirtruth-value.Acatcansurelybecorrectinthinkingthatamouseisinacertainholewithouthavingtheconceptsofbeliefandtruth.

Thejudgmentoftruthorerrorinabeliefmustbedistinguishedfrommerelyhavingabeliefwhichistrueorfalse.Weastheoristsmayjudgetrueandfalsebeliefs(justaswemayjudgepictorialresemblance)sincethesearemeta-linguisticorsecond-orderbeliefs,buttruthanderrorarenotintrinsicpropertiesofrepresenta-tionsassuch,onlytothejudgmentsmadeaboutthem.Catpsychologymustbepossiblewithoutinvokingcatepistemology.

Itisnotabigleapfrommisrepresentationandillusiontonoticethatimagesareaspeciesofthesamegenus.Imageryinvolvesillusoryornon-veridicalexperiencesofexactlythesortrequiredfortheclassicalargumentforsense-data.TheproverbialPinkElephantofinebriatedapprehensionisavisualimageparexcellence,notrelevantlydifferentfromMalebranche’sgoldenmountainorsubjects’imaginingsinthecelebratedexperimentsofShepardandMetzler(1971)andKosslyn(1994).Aswewillseepresently,ofcourse,ifmyconjecturedparalleliswarranted,itisperhapsnosurprisethattheimagerydebatehasbeenamongthemostpersistentlyintractable

254P.SLEZAK

disputesincognitivesciencealsoarisingfromthetheoristdoingtheworkofthetheory.TwinEarth

Putnam’s(1975)TwinEarthpuzzles,too,seemtobeanunnoticedvariantontheproblemofmisrepresentationwesawearlier.Inthefamiliarscenario,insteadofmyTwinEarthdouble,wemaysubstitutemyselfafterhavingbeenunknowinglytransportedtoTwinEarth.There,likemytwinintheoriginalstory,IwillrefertoXYZas“water.”However,onthisvariationoftheoriginalscenariomyterm“water”nowfailstorefercorrectlyratherthanbeingacorrectreferencewithatermhavingadifferentmeaning.SincemytwinandIareidentical,thetwoscenariosmustalsobeindistinguishable.Thatis,theproblemof“wide”and“narrow”meaningisjusttheproblemofmisrepresentationinanotherguise.

InsteadofthinkingofTwinEarth,then,wemayimaginealternativelythatonthisearth,GodmighthaveswitchedallH2OtoXYZwithoutmyknowledge.InsteadoftakingtheoriginalTwinEarthstoryasshowingthatmytwinmustmeansomethingotherthan“water,”wemayequallyconcludethatmyuseofthetermissimplyinerrorwhentheworldshavebeensurreptitiouslyswitched.TheTwinEarthscenariois,indeed,simplyanotherwayoftellingDretske’s(1986)storyofthemagneticmicro-organismswhicharefooledinto“thinking”thatupisdown.Or,inadifferentcase,asFodor(inMillikan,1991,p.161)hasputit,“it’snotthefrogbuttheworldthathasgonewrongwhenafrogsnapsatabee-bee.”Undoubtedly,iftheworldissuf󰁎cientlyperverse,oritiscontrivedtoalterthingsincertainways,ourconceptsmayaccidentally“fail”toreferintheusualmanner.Itisnotclearwhysuchpossibilitiesshouldbeofinteresttoatheoryofrepresentationfortheirdescriptiondependsonknowledgefroma“God’sEye”perspectiveavailabletothetheorist.WhethertheliquidsubstanceisreallyXYZorH2Oisknownonlytotheexternalomniscienceofthetheoristandhasnoexplanatoryroleinatheoryofrepresentation.Inthissense,thephilosophicalconcernwithmisrepresentationisanalogoustothespuriousassimilationofmiragesandseeminglybentstickstogenuinecognitiveillusions,asnotedearlier.Inbothcases,theactualtruthabouttheworldisinvokedirrelevantlytoexplaincognition.Philosophersasthree-year-olds?

Ironically,themistakeIamindicatingisnotunknownincognitivescience:Inthecasesofinteresthere,philosophersarelikethethree-year-oldsandautisticsinthemuch-discussed“falsebelieftask”ofWimmerandPerner(1983)(seeCarruthers&Smith,1996;Davies&Stone,1995a,b).Likethree-year-olds,philosophersfailtodiscountwhattheyknowtobethetruthabouttheworldintheir“theoryofmind.”ThesurreptitiousswitchingofXYZforH2O,bee-beesfor󰁏iesormagnetic“up”for“down”arewaysofmaking“theworldgowrong”preciselyanalogoustoswitchingthecandywhilethechildislookinginthe“falsebelief”paradigm.Knowinghowtheworldreallyis,philosopherstrulyascribefalsebelief,justasthethree-year-olds

THETRIPARTITEMODELOFREPRESENTATION

255

falselyascribetruebelief.Inbothcases,beliefattributionsareindependentofanyfactsaboutthebeliever,dependinginsteadonirrelevantexternalfactsabouttheworld.Inthesecases,thebeliever’sstateofmindcanremain󰁎xedandyetthebeliefscanbemadetochangefromtruetofalsebymanipulatingtheworld.Thechild,likethephilosophernaivelytakesthispossibilitytoberelevanttoa“theoryofmind”inascribingmentalrepresentations.Justi󰀪edtruebelief?

Incasetheforegoinganalogymaybethoughtfar-fetchedormerelywhimsical,itisperhapsworthnotingenpassantthatyetanothernotoriousphilosophicalpuzzlemaybeseentobemerelyaversionofthesameproblem.Gettier(1963)paradoxesmaybeseenasaspeciesofmisrepresentationinwhichtheworldconspirestomakeapropositiontrueforreasonswhichareentirelyindependentofaperson’sgroundsforbelievingit.Inthesecasestheproblemcanonlybedescribedbecausethetheoristknowsthetruthaboutcircumstanceswhichmakeabeliefaccidentallytrue,eventhoughtheactualcircumstancesareirrelevanttotheagent’sownreasonsforbelievingtheproposition.TheGettiercasesarestructurallyidenticalwiththoseofmisrepresentationandTwinEarthbecausethetruthorfalsityofthementalrepresentation(i.e.thestateoftheworld)isvariedindependentlyoftheagent’sbelief-󰁎xingmechanisms.Suchconsiderationsinallcasesshouldbeirrelevanttotheproblemofunderstandingmentalrepresentation.ThemoraloftheGettiercases,likethatofmisrepresentation,isthattheonlysensible,andperhapstheonlypossible,theoryofknowledgeisonethatinvokesjusti󰁎cationsandnottruthfroma“God’sEye”perspective[7].Anyadequate,orevencomplete,accountofaperson’spsychologywouldhavetoinvokeonlytherelationofbeliefstoavailableevidenceandnottheiractual,ultimatetruth-value.Theworldcanmisleadusinvariousways,givingusgoodreasonsforthingsthatmaybefalse,badreasonsforthingsthatmaybetrueandgoodreasonsforthingsthatmaybetrueforotherreasons.Noneofthisshouldoccasionphilosophicalanxietiesforthoseinterestedinpsychology.Doesthespeedometerofabicyclemisrepresentwhenthebikeisriddenonrollersandnotmoving?Onceagain,itistheworldthathasgonewrong,knowntousasexternalobservers.However,psychologyhasnoobligationtoexplainwhytheworldmaygowrong.Thus,conceivably,onemightcontrivethingssothatCabernetSauvignonreplacedtheusualliquidinsomeone’sveins.However,suchapossibilityisofnomoretheoreticalconcernformedicalsciencethanDretske’s(1986)disorientedmicrobesareofinteresttocognitivescience.Idea-objects

Bechtel’s(1998,p.299)re-statementofthetripartitemodelinMalebranche’stermsmakesexplicitthewidelyheldassumptionswhicharethepotentialsourceofthedif󰁎cultiesinunderstandingrepresentation[8].Inparticular,Bechtel’sassimilationofinternalandexternalrepresentationsisacknowledgedwhereheliststhesortsof“high-level”representationswhichhavebeenpostulatedbycognitivescientists.

256P.SLEZAK

Theseinclude“conceptsthatmightdesignateobjectsintheworldorlinguisticsymbols,󰁎guresanddiagramswhichwecanuseinreasoningandproblemsolving”(1998,p.305).Bechtelsuggeststhatifcognitiondoesrequiresuchhigher-levelrepresentations,“themostplausibleanalysisisthatsuchrepresentationsarebuiltupontheselow-levelrepresentationsandperhapsinherittheircontentfromthem”(Bechtel,1998,p.306).However,thedif󰁎cultyisthatthedistinctionbetween󰁎guresordiagrams,ontheonehand,andrepresentationsoperatinginthefrog’sretina,forexample,isnotsimplyamatterofhigherandlower“levels”inanunproblematicsense.Linguisticsymbols,󰁎guresanddiagramswhichweuseinreasoningandproblemsolving,asBechtelsays,areobviously“used”inasensewhichisdifferentandpreciselyinappropriateforinternalmentalrepresentations.“Higher”conceptsofthiskindcouldnotinherittheircontentfromlow-levelconceptsmentionedbecausethedifferencehereisnotoneoflevel,butofakindwhichpreciselyde󰁎nesthedistinctionbetweenoriginalandderivedintentionality.

InthisregardBechtel’saccountaccuratelyre󰁏ectstheassumptionsbuiltintothefoundationalnotionofsymboliccomputation,asAllanNewellexplains:

Theideaisthatthereisaclassofsystemswhichmanipulatesymbols,andthede󰁎nitionofthesesystemsiswhat’sbehindtheprogramsinAI.Theargumentisverysimple.Weseehumansusingsymbolsallthetime.Theyusesymbolsystemslikebooks,theyuse󰁎shasasymbolforChristianity,sothereisawholerangeofsymbolicactivity,andthatclearlyappearstobeessentialtotheexerciseofmind.(1986,p.33)

ThispassageisstrikingfortheexplicitnesswithwhichNewellassimilatesinternalmentalrepresentationswithourexternalcommunicativesymbols.Theassimilationofrepresentationsofradicallydifferentkindsappears,then,tobeamongthefoundationalassumptionsofcognitivescience.Itwasself-consciouslyarticulatedinDennett’s(1978a)reviewofFodor’s(1975)importantworkThelanguageofthoughtwhichwasthephilosophicalmanifestofortheclassicalsymbolicapproachtocognition:

Whatisneededisnothinglessthanacompletelygeneraltheoryofrepre-sentation,withwhichwecanexplainhowwords,thoughts,thinkers,pictures,computers,animals,sentences,mechanisms,states,functions,nerveimpulses,andformalmodels(interalia)canbesaidtorepresentonethingoranother.(1978,p.91)

Thehoped-foruni󰁎cationistobeachievedbyshowingthattheseseeminglyheterogeneousitemsareall,infact,variantsofacommon,underlyingscheme.Dennettmakesthisexplicit,explaining:

Itwillnotdotodivideandconquerhere—bysayingthatthesevariousthingsdonotrepresentinthesamesense.Ofcoursethatistrue,butwhatisimportantisthatthereissomethingthatbindsthemalltogether,andweneedatheorythatcanunifythevariety.(1978a,p.91).

Thepictorialaccountofimageryisperhapstheclearestexampleoftakingour

THETRIPARTITEMODELOFREPRESENTATION

257

externalartifactsasthemodelforinternalrepresentations.Moregenerally,thedif󰁎cultiesarisefromanequivocationonthenotionof“understanding”whichcanmeaninterpretingameaningfulrepresentationasintelligible,orexplainingitasinscience.Wewillseepresentlythatthiscon󰁏ationisevidentlyattheheartofSearle’s(1980)ChineseRoomscenario,forSearleaskswhetherhe,ashomunculusinthesystem,canunderstandthesymbols.Thiscriterionshouldbeirrelevanttothequestionofwhetherasystemhas“original”intentionality,butSearle’smistakeisnothisalone.Thatis,theChineseRoomscenarioaccuratelycapturestheorthodoxassumptionsoftheSimon–Newell“physicalsymbolsystemhypothesis.”Searle’sargumentis,therefore,bestunderstood,notasarefutationof“strong”AI,butasareductioadabsurdumofthewidelyheldassumptionsonwhichAIandcognitivesciencearebased.Iwillsuggestthatthisstandardformalistorlogicistviewconceivedonthemodelofanuninterpretedlogicalcalculusisproblematicasanaccountofthewayanintelligentsystemisrelatedtotheexternalworldbydependingontheintentionsoftheexternaluserwhosuppliestheinterpretationofthemeaninglessformalsymbols(seeBirnbaum,1991;Rosenschein,1985).Embrac-ingthisconception,Nilsson(1987,1991),likeNewellabove,explicitlyinvokesourexternalsymbolicartifactssuchasbookstodefendhisviewagainstthe“proceduralist”positionthatrepresentationsaretobeusedbythesystemitselfratherthanunderstoodorascribedmeaningbythedesigner.Undoubtedlyrecogniz-ingtheproblemswithinvokinganintelligentuserorunderstander,mosttheoristswouldnotknowinglyembracesuchanaccount,buttheirintentionsmaybeinconsistentwiththeactualpropertiesoftheirmodelbyvirtueofassimilatingexternalandinternalrepresentations[9].Itisnotdif󰁎cultto󰁎ndleadingtheoristsexplicitlyendorsingthisassimilation.Thus,RumelhartandNorman(1983)wrote:

Wede󰁎neasymboltobeanarbitraryentitythatstandsfororrepresentssomethingelse.By“entity”wemeananythingthatcanbemanipulatedandexamined…Humansalsouseexternaldevicesassymbols,suchasthesymbolsofwritingandprinting,electronicdisplaysorspeechwaves.(p.78)

InarecentstatementforanencyclopaediaentryonrepresentationDanLloydexplains:

Humansarerepresentinganimals,andwehavebuiltaworldcrammedwithrepresentationsofmanykinds.Consider,forexample,thenumberandvarietyofpictorialrepresentation:paintings,photographs,movingpictures,linedrawings,caricatures,diagrams,icons,charts,graphs,andMaps.Addthevarietyoflinguisticrepresentationsinsigns,titles,textsofallkinds,andespeciallyspokenwordsandsentences…Humanlife,inshort,islargelyacycleofmakingandinterpretingrepresentations.(Lloyd,forthcoming)

Bycontrast,Block(1986)recognizesthat“Therepresentationonthepagemustbereadorheardtobeunderstood,butnotsofortherepresentationinthebrain”(p.83).However,despitemakingthisdistinction,Block’sdiscussionappearsto

258P.SLEZAK

lapseintothecharacteristicerror.Blockasks“whatitistograsporunderstandmeaning?”(p.82).Ofcourse,wedon’tgrasporunderstandthemeaningofourownmentalrepresentations,wejusthavethem.InDennett’s(1978b)felicitousphrase,therepresentationsmustunderstandthemselves.Arnauld’swordsappeartodirectlyaddresstheoriststoday:

Tosaythatourideasandourperceptions(takingthesetobethesamething)representtousthethingsthatweconceiveandthattheyaretheirimages,istosaysomethingcompletelydifferentfromsayingthatpicturesrepresenttheiroriginalsandaretheimagesofthem,orthatspokenorwrittenwordsaretheimagesofourthoughts.Forinthecaseofideaswemeanthatthethingsweconceiveareobjectivelyinourmindandinourthought.Andthiswayofbeingobjectivelyinthemindissopeculiartothemindandtothought,sinceitiswhatspeci󰁎callygivesthemtheirnature,thatoneseeksinvainanythingsimilaroutsidethemindandthought.AsIhavealreadyremarked,whathasthrownthequestionofideasintoconfusionistheattempttoexplainthewayinwhichobjectsarerepre-sentedbyideasbyanalogywithcorporealthings,buttherecanbenorealcomparisonbetweenbodiesandmindsonthisquestion.(Arnauld,1683/1990,p.66)

Symbols&Searle:themeaningofmeaning

Ihavebeensuggestingthatacrucialequivocationondistinctmeaningsof“meaning”hasledtothepostulationofsymbolshavingmeaninginanobserver-rel-ativesenseinwhicharepresentationisnecessarilyapprehendedandunderstoodbysomeone.Cummins(1996)clearlypointstothismistakeofconstruinginternalrepresentationsasiftheymayfunctionthroughbeingunderstood.Thequestionofmeaningofmentalrepresentationsisregularlyconfusedbetweenwhetherrepresen-tationsareintelligibleandwhethertheyareexplainable.Searle(1980)tradesdirectlyonthisconfusionbyaskingwhetheranintelligentunderstandercaninterpretthesymbolswhicharethesubstrateofthought.Buthoweverwemightexplaininten-tionality,itcannotdependonwhetheranyonecan“understand”thegoings-oninamachineoraheadinthesenseofapprehendingthem.Theonlysenseinwhichthesegoings-onaretobeunderstoodisthequitedifferentsenseofscienti󰁎cexplanation.Itisnoaccidentthatthesameperniciousequivocationhasbedeviledlongstandingdisputesinthesocialsciencesbetweensubjectivistadvocatesofverstehenasamethodand“positivist”advocatesoferkla¨ren(Slezak,1990;Winch,1957).How-ever,understandingquaparticipantisnotthesameasunderstandingquascientist[10].Undeniably,Searle’s(1980)ChineseRoomdemonstratesthatcomputationalsymbolsaremeaninglessintheformersense,butthisisnomoreproblematicthanthemeaninglessness,inthissense,ofactionpotentialsorsynapticactivations.Predictably,theproblemmaybeseenarisingindebatesoverconnectionistsystemswhere,inthiscase,itistheallegedabsenceof“explicit”symbolsindispute.

THETRIPARTITEMODELOFREPRESENTATION

259

However,thecriterionistakentobewhethersymbolsmaybe“directlyreadoff”or“immediatelygrasped”(Ramseyetal.,1991).Theobviousquestionis:bywhom?

Searle’sconundrumisremarkablyevokedbyGlanvill’sresponsein1661toDescartes’ownversionofacodingorinformationprocessingtheoryofperception.“Buthowisit,andbywhatArtdoththesoulreadthatsuchanimageorstrokeinmatter…signi󰁎essuchanobject?DidwelearnsuchanAlphabetinourEmbryo-state?”(quotedinYolton,1984,p.28).EchoingSearle,Glanvillsuggeststhatthe“motionsofthe󰁎lamentsofnerves”learnthequalityofobjectsbyanalogywiththewayinwhichapersonlearnstounderstandalanguage,forotherwise“thesoulwouldbelikeaninfantwhohearssoundsorseeslipsmovebuthasnounderstandingofwhatthesoundsormovementssignify,orlikeanilliteratepersonwhoseeslettersbut“knowsnotwhattheymean”(1984,p.28).Itissigni󰁎cantthat,unknowingly,YoltonalsoevokesSearleintheChineseRoomwhendiscussingLocke’sconceptionintermsofa“perspectivebox”orcameraobscura.Yoltonasks“Wastheresometemptationtothinkofourawarenessbeinglikethefaceattheperspectiveboxscanningtheimagesonthewallofthebox?”(1984,p.127).Arnauld’sact-theoryofdirectrealism

Arnauldproposeda“directperception”accountagainstMalebranche’sindirect,object-mediatedtheory.ForArnauld,combatingthetripartiteviewmeansthatideasarenotdistinctentitiesbutjustthoseveryactivitiesofthemindwhichareessentiallyrepresentativeperse.

SinceGoddesiredourmindshouldknowbodies,andthatbodiesshouldbeknownbyourmind,itwasundoubtedlysimplerforhimtorenderourmindcapableofknowingbodiesimmediately,thatis,withoutrepresenta-tiveentitiesdistinctfromperceptions…andbodiescapableofbeingknownimmediatelybyourmind,ratherthanleavingthesoulpowerlesstoseethemotherwisethanbymeansofcertainrepresentativeentities.(VFI,222–3,quotedinNadler,19,p.97)

Arnauldinsiststhatenclosingthemindina“palaceofideas”asBerkeleywastodoisanabsurdconclusiontodrawfromtherepresentativetheory.Speci󰁎cally,aswehaveseen,Arnaulddiagnosestheabsurdityasduetoamistakeorfalseanalogybetween“beingpresenttothemind”inthesenseofhavingideas,thinkingorperceiving,ontheonehand,andbeingpresenttotheeyesinseeing.Ofcourse,thisistheassimilationofinternalandexternalrepresentationswehaveseen.Thatis,“seeing”withthemindorsoulisconfusedwithseeingwiththeeyesorbody.Arnauldarguesthatphilosophershavetriedtoexplainhowwethinkorperceivewiththemind—mentalvisionorlavuespirituelle—byanalogywithopticalvisionortrueseeing[11]withtheeye—lavuecorporelle.ThisisthesameinsightexpressedintheepigraphfromGalileo.AsNadlerpointsout,Arnauldinsiststhattheproblemwiththisanalogyisthatitrestsonfalseassumptionsand“Onemustnotbaseone’sreasoningaboutthementalactofperceptionorobservationson,orbeliefsabout,thephysiologicalprocesseswhichconstitutebodilyseeing.”

260P.SLEZAK

Nadler(19,p.93)construesArnauld’sdistinctionbetweenmentalseeingandbodilyvisionasarisingfromanadherencetoastrictCartesiandualism.However,NadlerappearstobemakingasimilarmistaketotheonewehavealreadynotedbyDennettinhisdiagnosisofthe“CartesianTheater.”Arnauld,likeDescartes,avoidstheproblemsinherentinrepresent-ativeideasquaintervening,apprehendedentitiesbecausehehasabettertheoryofperceptualandintellectualactivity—namely,intermsofmentalprocesseswhicharethemselvesinherentlyreferential.ThisistoavoidthenecessityofanycentralobservinghomunculusandtheTheaterbyavoidingaconceptionofrepresentativeentitieswhichrequireanintelligentperceivertocontemplatethem.ContrarytoNadler,thisconceptionisentirelyindepen-dentofdualism“whichrulesoutanysuchanalogies”betweenmentalandcorporealvision.Arnauld’sviewismoresubtle.TheanalogyisruledoutnotbecauseofanydualismbutbecauseArnauldconceivesmentalactivityasitselfessentiallyrepresentativeandtherebydispenseswithideasassurrogateobjectstobeobservedbythemind’seye:“…Idonotseeanyneedforthisalleged‘representativeentity’inordertoknowanyobject,beitpresentorabsent”(VFI,221,quotedinNadler,19,p.96).

Notmuchofarevolution?

Bycontrastwiththeusualhype,Chomskyrecentlyexpressedskepticismregard-ingtheradicalnoveltyoftheso-calledCognitiveRevolutionsaying“itwasn’tallthatmuchofarevolutioninmyopinion”(1996,p.1).Chomskysuggeststhatthesameconvergenceofdisciplinaryinterestshadtakenplaceinthe17thcenturyinwhathecalls“‘the󰁎rstcognitiverevolution,’perhapstheonlyrealone”(p.1).Chomsky(1966)beganhisCartesianlinguisticsbyquotingWhitehead,whosaidthattherecenthistoryofintellectuallifemaybeaccuratelydescribedas“livingupontheaccumulatedcapitalofideasprovided…bythegeniusoftheseventeenthcentury.”Chomskywasconcernedtoshowthatareturntoclassicalconcernsandappreciationoftheirparallelswithcontemporarydevelopmentsisvaluableinhelp-ingtoadvancethestudyoflanguage[12].Ihavebeensuggestingthat,withsomeinterestingdifferences,Chomsky’spointholdsforrecentspeculationaboutthemindoutsideoflinguisticsaswell.Indeed,Nadler(1992,p.73)notesthatitisboth“strangeandnotalittleembarrassing”thattheMalebranche–Arnaulddebateshouldremainlargelyignoredormisunderstoodbyphilosophersandhistoriansalike.Thesamegoesforcognitivescientists.Thus,Chomsky’sexamplemaybeusefullyfollowedinrelationtothesedifferentissuessuchasthehotlycontestedrepresent-ationaltheoriesofthemind.However,thedifferencesfromthecaseoflinguisticsarerevealing.Thoughequallyneglectedincognitivescience,notonlythegoodideasofthe17thcenturyarebeingrediscovered:Wearenotonlyreinventingthesametheoriesandrelivingthesamedebates,butalsorehearsingthesamenotoriousmistakes.

THETRIPARTITEMODELOFREPRESENTATION

261

Imagery:thepictorialtheory

The“ImageryDebate”isperhapsthemostremarkablemodernduplicationof17thcenturycontroversies.Inthisre-enactment,amongthedramatispersonaePylyshynplaysArnauldagainstKosslyn’sMalebranche.Kosslyn(1994)claimstohaveresolvedthedebateinfavorofhis“pictorial”theory,butthereremaingroundsforskepticism.Themise-en-sce`neisfaithfuleventotheextentoftheacrimonyofthedisputes.Moreimportantly,thecentralerroridenti󰁎edbyArnauldofascribingcorporealpropertiestomentalonesisexactlytheonechargedbyPylyshyn(1973,1978,1981,inpress)againstKosslyn.

Despiteitscomputationalandneurosciencetrappings,Kosslyn’s(1994)picto-rialaccountofimagerytakesmentalimagestorepresentbyvirtueofarelationofresemblancetotheirobjectsandbyvirtueofactuallyhavingspatialpropertieswhichtheyrepresent.Furthermore,“depictive”representationsina“visualbuffer”aretakentohavethespeci󰁎cfunctionofpermittingare-inspectionofimagesbythehighervisualapparatus.Thereissaidtobean“equivalence”betweenimageryandperceptionaccordingtowhichthe“higher”cognitiveprocessingapparatusforvisualperceptionissimplyappliedtoanalternativeinputotherthantheretina—namely,thevisualbuffer.Thus,onthepictorialaccount,amentalimageisconceivedtobea“surrogatepercept”(Pinker&Finke,1980).Inthisway,animagemaybe“reprocessedasifitwereperceptualinput…therebyaccomplishingthepurposesofimagerythatparallelthoseofperception”(Kosslyn1987,p.155).Ofcoursethisisanimplicitendorsementofthetripartiteconceptionofmentalrepresentationinimagery.

Dennett(1991)takesthispictorialtheoryofimagerytobeaparadigmexampleofthe“Theater”misconceptionand,notsurprisingly,this“quasi-perceptual”modelhasbeenrepeatedlychargedwiththeerrorofimportingan“homunculus.”Thechargeisvigorouslyrejectedonthegroundsthat“thetheoryisrealizedinacomputerprogram”(Kosslynetal.,1979,p.574),butundischargedhomunculicanlurkincomputationalmodelsjustaseasilyasintraditionaldiscursivetheories(seeSlezak,1995,inpress).

Thus,Kosslynetal.(19)offeradiagramofthevisualimagerysystemwhichisaprofusionofinterconnectedboxesandarrows.Theboxlabeled“visualbuffer”containsanotherboxlabeled“attentionwindow”whichisleftunexplained.Thisboxis,infact,theobserverinthe“theater”whichisthesourceofthetraditionalproblem.TheelaboratediagramisreducibletothesametripartiteschemawehaveseeninMalebranche.Signi󰁎cantly,followingDescartes,Arnauldexplicitlypointedtotheseductiveerroroftakingpicturesasanappropriatemodelofmentalrepresen-tation(Arnauld,1683/1990,p.67),andhecitesthecameraobscuraasanerroneousmodelforimagery.

Inarevealingmisunderstanding,Kosslyn(1980,p.30)haschargedalternativepropositionalor“tacitknowledge”theories(Pylyshyn,1973)asbeing“noimagery”accounts,butdenyingpictorialimagesisnottodenyimageryperse.Rather,inanArnauldianspirit,thedenialofpictorialformatforrepresentationsisactuallytodenytheproblematichomunculusanditspseudo-explanation.Ineffect,aswesee,

262P.SLEZAK

thedebatebetweenMalebrancheandArnauldisbeingreplayedthroughoutthehistoryofspeculationofthemindandcontemporarycognitivescience.Thus,oneneednotgobackasfarasthe17thand18thcenturiestodiscoverthesameconcerns.F.C.Bartlett’s(1932)theoryofschematainhisbookRemembering,wasareactiontotheoriesof“󰁎xed,lifelessandfragmentarytraces”orimageswhicharemerely“reduplicative,”capableonlyofbeingre-excited.ReminiscentofArnauld’srejectionof“super󰁏uousentities”andhisview“thatideaandperceptionarethesamething,”Bartlettwishestosubstituteacognitiveprocessforobjectswhicharepictorialor“reduplicativetraces”(1932,p.215)[13].Phenomenologicalfallacy

Kosslynclaimstohaveclinchedthedebateaboutimagerybyappealingtothe󰁎ndingsofneurophysiologyandneuroanatomy[14].Topographicallyorganizedregionsofcortexor“retinotopicmapping”aresaidto“supportdepictiverepresen-tations,”thatis,picturesinsomesense.Thus,forexample,amonkeymaybegivenavisualstimuluslikeadartboardtolookat.Ifthebraintissueistreatedinacertainway,itcanbeshowntohavealikenessofthedartboard“etched”onthecortex.Theresultwasanticipatedandperfectlyunderstoodbyonepsychologist30yearsbefore:

Atsomepointtheorganismmustdomorethancreateduplicates…Theneedforsomethingbeyondandquitedifferentfromcopyingisnotwidelyunderstood.Supposesomeoneweretocoattheoccipitallobesofthebrainwithaspecialphotographicemulsionwhich,whendeveloped,yieldedareasonablecopyofacurrentvisualstimulusInmanyquartersthiswouldberegardedasatriumphinthephysiologyofvision.Yetnothingcouldbemoredisastrous…(Skinner,1963,p.285)

Skinnerwasacutelysensitivetothesourceofhomunculipseudo-explanationsevenifhisbehavioristremedyisnolongerattractive.

Kosslyn’sTVscreenmetaphorrevealsthelinkbetweenseeminglyunrelatedproblemsincognitivescience.Forexample,intheclassicstatementofmaterialism,U.T.Place(1956)arguedthattheimplausibilityandrejectionofmaterialismasasolutiontothemind–bodyproblemisbasedonthequalitativefeaturesofsubjectiveexperience.Althoughthesefeatureshaverecentlybeensupposedtoconstitutethe“hard”problemofconsciousness(Chalmers,1996),Placesuggestedthattheyarethesourceofthe“phenomenologicalfallacy.”AnticipatingDennett(1991),Placewrote,thisis“themistakeofsupposingthatwhenthesubjectdescribeshisexperi-ence,howthingslook,sound,smell,taste,orfeeltohim,heisdescribingtheliteralpropertiesofobjectsandeventsonaparticularsortofinternalcinemaortelevisionscreen.”

Thinkinginlanguage

Onthefaceofit,thepersistentdoctrinethatwethink“in”languageisnotobviouslyconnectedwiththeotherswehaveconsideredsuchaspictorialimagery.Neverthe-

THETRIPARTITEMODELOFREPRESENTATION

263

less,thesetwotheoriesarevariationsonthesametheme.Symptomaticisthefactthatbothdependonadeepintuitive,introspectiveappeal.Justasweseemtobelookingatpictureswhenweimaginevisually,soweappeartotalktoourselveswhenwethink.Indeed,Carruthers(1996),whoseekstorevivewhatheacknowledgestobeanunfashionabledoctrine,explicitlybaseshisargumentagainstFodor’sLan-guageofthoughtonsuchevidenceofintrospection.Thisistheevidencethatwesometimes󰁎ndourselvesinasilentmonologue,talkingtoourselvesinournaturallanguage,sottovoce,asitwere.

However,inaneglectedarticle,Ryle(1968)suggestedthattheveryideathatwemightthink“in”languageisunintelligible,andtheundeniableexperienceoftalkingtoourselvescannotsupportanyclaimaboutthevehiclesofthought.Itissigni󰁎cantthatRylementionsenpassantamongtheequallyproblematicalcases,thatinwhichweclaimtoseethingsinour“mind’seye”—takentoinvolvementalpicturesofsomekind.Ryle’scomparisonandhiswarningisunwittinglycon󰁎rmedbyCarruthers(1996,1998),whoexplicitlyinvokesKosslyn’spictorialaccountofimageryassupportforhisownanalogoustheory.Indoingso,however,Carruthersonlybringsintoreliefthenotoriousdif󰁎cultiesofhisownmodelwhichreliesonarepresentationalformat—sentencesofnaturallanguage—whichis,likepictures,paradigmaticallythekindrequiringanexternalintelligentobserver(seeSlezak,2002).

Stillmoreprocessing:triadicordyadic?

Aswehaveseen,thetraditionaldif󰁎culty,rediscoveredinvariousformstoday,arisesfromtheproblematicthree-partrelationbetweenworld,ideasandconsciousness.Cognitionwithoutthesebasicfeaturesseemsinconceivable,andyettheyleadtoseeminglyintractabledif󰁎culties.Initsessentials,theMalebranche–Lockeaccountmaybecapturedinthefollowingschematicdiagram:

ExternalWorld®Ideas®Consciousness

ThecommonalitybetweenthisschemaandmodernonesisclearlyrevealedinadiagramofUlricNeisser(1976).Initsessentialsthediagrammayberepresentedasfollows:

StorageStorageStorage¯­¯­¯­External®Retinal®Processing®Moreprocessing®Stillmoreprocessing®ConsciousnessWorldimageNeisser’sschemaobviouslyabstractsfromthedetailsofanyspeci󰁎caccount,butdespiteitsmildlywhimsicalcharacter,itpurportstobeaseriousgenericsketchofinformationprocessingtheories.Forexample,thewelterofboxesandarrowsinKosslyn’sdiagramreducestoNeisser’spicturewhichhasthevirtueofnotdisguisingitsessentialcommitmenttotheproblematicthirdelement“consciousness.”

Oneearlyattempttoavoidthedif󰁎cultiesinherentinthisaccountisinthereactionsofLocke’scriticJohnSergeant:

2P.SLEZAK

Heneverfoundasatisfactoryanswertothequestionofthenatureofideas,buthewasconvincedthattheyfunctionedtodenytheminddirectaccesstothingsbyrestrictingittosomekindofthirdentity.Cognitionwasthusmadetoconsistinatriadicrelationinvolvingtheknowingmind,theobjectorreferent,andtheideasbymeansofwhichthemindcametoknowthings.Sergeantwishedtoreducetheprocessofknowingtoadyadicrelationconsistingonlyoftheknowingmindandtheobjectknown.(quotedinYolton,1956/1993,p.103)Noideas?

However,givenaproblematicthree-partrelation,thereareonlyafewwaystogetadyadicrelation.First,eliminatingthemiddleterm,ideasorrepresentations,permitstwoconstruals.Ononeview,whichYoltondescribesas“wildlyimpossible”butwhichhasnonethelessbeenexplicitlyheld,anobjectisitselfsomehowliterallypresentto,orin,themind.ThisistheviewwesawparodiedbyMalebranchesayingthatthesouldoesnotstrollintheheavensamongthestars.

ExternalWorld®Consciousness

Noexternalworld?

Ofcourse,adyadicrelationcanalsobeobtainedbydroppingoneoftheotherrelatainsteadofthemiddleone.Thus,wemighteliminatetheexternalworldtogetBerkeley’sidealism.

Ideas®Consciousness

Berkeley’sstrategyhasnotbeenpopularrecentlyamongcognitivescientists,thoughitispreciselytheparadoxofFodor’s“methodologicalsolipsism.”Noconsciousness?

Ofcourse,wehaveanotherchoicebesidesgettingridoftheworldorgettingridoftherepresentations.Wecangetridofconsciousness!Securingadirectconnectionbetweencognitionandtheworldcanalsobeachievedbydispensingwiththeagent[15].

ExternalWorld®Ideas

Despiteitscounter-intuitiveness,insomerespectsthisoptionispreferabletotheothers.Ofcourse,theagentasanelementinmodelsofperceptionisthelocusofthepotentialhomunculus.Theparadoxicalrejectionofconsciousnessisequivalenttothe󰁎rstofthesethreeoptions,beingthesameasrejectingrepresentationswhentheseareconceivedincertainways.Therefore,expressedlessparadoxically,theGibsonianor“situated”caseagainstrepresentationsmaybebestunderstood,notasanoutrightrejectionofaninternalmentalmediumassuch,butratherasthe

THETRIPARTITEMODELOFREPRESENTATION

265

Arnauldianrejectionofacertainparticularconceptionofrepresentationswhichareexplanatorilyquestion-begging.

Conclusion

ThirtyyearsagoinhisbookPsychologicalexplanation,Fodor(1968,p.vii)remarked:“Ithinkmanyphilosopherssecretlyharbortheviewthatthereissomethingdeeply(i.e.conceptually)wrongwithpsychology,butthataphilosopherwithalittletraininginthetechniquesoflinguisticanalysisandafreeafternooncouldstraightenitout.”Today,thesuspicionofdeepconceptualproblemsattheheartofcognitivescienceisperhapsmoreclearlyseentobejusti󰁎ed,althoughFodor’sjokewasintendedtore󰁏ectasmuchuponphilosophyasuponpsychology.Notoriously,deepconceptualproblemsattheheartofphilosophyhavebeennomoredispelledthanthoseinpsychology.Ihavebeensuggestingthatbyadoptingabroaderperspectivewemayseewhythesorryfortunesofthetwodisciplineshavebeeninextricablylinked.

Acknowledgements

Thispaperhasbene󰁎tedgreatlybythemosthelpfulcommentsandcriticismsoftheeditor,W.Bechtel,andtwoanonymousrefereesofPhilosophicalPsychology.VersionsofthematerialhavebeenpresentedattheDepartmentofPhilosophy,UniversityofSydney,theSixteenthAnnualMeetingoftheJapaneseCognitiveScienceSociety,Tokyo,August1999,theAnnualMeetingoftheAustralasianAssociationforHistory,Philosophy&SocialStudiesofScience,Melbourne,June2001,andtheTwentyThirdAnnualConferenceoftheCognitiveScienceSociety,Edinburgh,August2001.IamparticularlygratefulforthecommentsofStephenGaukroger,RonGiere,PatLangley,RobertNola,ZenonPylyshynandJohnSutton.Notes

[1]Kosslynwriteswithevidentannoyance:“Onceandforall,the‘homunculusproblem’issimply

notaproblem.Wethoughtthiswouldbeobviousgiventhatthetheoryisrealizedinacomputerprogram,butitseemsnecessarytoaddressthiscomplaintagain”(Kosslynetal.,1979,p.574).[2]Thescholarly,exegeticalnicetiesneednotconcernushere.However,Nadler(1992,p.8)points

outthatthestandardreadingpresentedhereismistakenandevenacaricature,thoughithasbeenalmostuniversallyheldamongcommentatorsincludingArnauld,Locke,Leibniz,BerkeleyandReid.

[3]The“situated”critiqueconfusesconsciouscalculationwiththe“sub-personal”useofsymbols

andtheirpurelycausal,functionalroleincognition(seeSlezak,1999).AsPasnausays:“Surely,anymoderndirect-realisttheoryofperceptionwillallowcausalintermediariesbetweenobjectandpercipient:noonewoulddreamofdenyingthetitleofdirectrealismtoatheoryofperceptionmerelybecauseittoleratescausalintermediaries”(Pasnau,1997,p.300).

[4]Aboveall,Moreauexpressesirritationattheviewwhichhasbeenwidelyheld,especiallyamong

“noscontemporainsanglo-saxons,”thatthisaffairisatbasenothingotherthanoneofopposingtemperaments—“Arnauld-la-teigne”against“Malebranche-le-grognon”—thatis,Arnauld-the-nuis-

266P.SLEZAK

[5]

[6][7]

[8]

[9]

[10]

[11]

[12][13]

[14]

[15]

anceagainstMalebranche-the-grouch(1999,p.16).Infact,numerousgreat󰁎guresofthetimebecameembroiledinthedebate,suchasLeibnizwhofolloweditclosely(seeNadler,19,p.5).InhisearlierworkonLockeandthewayofideas,Yolton(1956/1993)givesalongerpassagefromthesameauthorwhichisworthrepeatinghereastestimonytothecuriouspersistenceofthepuzzleofideas,albeitindifferentguises:“…likeMenblunderinginthedark,theyfeelafterthemto󰁎ndthem;somecatchatthemunderoneAppearance,someunderanother;somemakethemtobeMaterial,othersSpiritual;somewillhavethemtobeEf󰁊uvia,fromtheBodiestheyRepresent,othersTotallyDistinctEssences;someholdthemtobeModes,othersSubstances;someassertthemAlltobeInnate;othersNone:SothatonewouldthinktheremustneedsbeaverygreatIntricacyinthatwhichhasgivenRise,notonlytosuchaVarietybutalsosuchaContradictionofOpinions”(quotedinYolton,1956/1993,p.96).

Iamgratefultoananonymousrefereeforcriticismswhichhavehelpedmetotryclarifyingthisargument.

Thismoralseemstoextendinanobviouswaytothebroaderconcernwithknowledgeinphilosophyofscience.Althoughitcannotbepursuedhere,anotherdeepparallelmaybeseeninthemoreorlessdistinctliteraturesofperceptualrealismandscienti󰁎crealism.Heretootheproblemoftruthanderrorarisesforrealismintheformoftheso-called“pessimisticmeta-induc-tion”fromhistory—analogoustotheArgumentfromIllusion.ItisnoaccidentthatMach’sinstrumentalismwasharshlycriticizedinLenin’s(1927)Materialismandempirio-criticismasbetrayingsciencebydesertingtoaBerkeleyanidealism.SeealsoPopper(1963a,b)onBerkeleyasprecursortoMachandEinstein.SeealsoZahar(1981).

Amongthefewexplicitattemptstoarticulatetherelevantdistinctionshere,seeCummins(1996,p.87)onthedistinctionbetween“meaning”and“meaningfor”—thelatterdescribedbyCum-minsas“athree-placerelationbetweenarepresentation,aconcept,andacognitivesystem.”Seediscussiononp.130.

Conversely,itmaybethatresearchers’actualtheoriesorprogramsareexemplaryandfreeofproblems,butonlytheir“meta-theoretical”analysessufferfromthefatal󰁏aws.IamgratefultoPatLangleyandRonGiereforthispoint.

Chomsky(1962)hasdrawnattentiontothewayinwhichtraditionalgrammarsproduceanillusionofexplanatorycompletenesswhile,infact,theyhaveseriouslimitationsfromascienti󰁎c,explanatorypointofview.Theapparentsuccessoftraditionalgrammarsdependsonbeing“pairedwithanintelligentandcomprehendingreader.”Thisisanotherversionofthehomun-culusfallacybecauseitisjustthisabilityoftheintelligent,comprehendingreaderthatthetheoryissupposedtoexplain.

Nadler(19)oddlychoosestoputscarequotesaroundthetermsreferringtothephysiologicalprocessesofcorporealvision,i.e.withtheeyes,asopposedtowhathecalls“trueseeing”bythemind.Thisseemsclearlytoreversetheexpectation,sincesurelyitisseeingwiththeeyeswhichisthenormalcaseandmentalvisionthemetaphoricaloranalogicaluseoftheterms.

ButseeskepticaldiscussionofChomsky’sclaimsandreferencesinBuroker(1996,p.ix).

AnArnauldianaccountisseenmorerecentlyinphilosophyintheguiseofso-called“adverbial”accountsofexperience.SeeTye(1984),whosays“havingavisualexperienceisamatterofsensinginacertainmannerratherthansensingapeculiarimmaterialobject”(1984,p.196).Headds“tothesatisfactionofmostpeople”(1994,p.vii).TheexceptionsKosslynhasinmindarephilosopherswhoarepresumablyimmunetorationalpersuasion.Indismissiveremarksheexplains,“Ifullyexpectphilosopherstocontinuetodebatethematter;afterall,thatistheirbusiness”(1994,p.409).

ThispointhasbeenmadeinexactlythesametermsbyJohnSutton(1998).

References

ABELL,C.&CURRIE,G.(1999).Internalandexternalpictures.PhilosophicalPsychology,12,429–445.

THETRIPARTITEMODELOFREPRESENTATION

267

ARNAULD,A.(1683/1990).Ontrueandfalseideas,withintroductoryessaybyS.GAUKROGER(Trans.).

Manchester:ManchesterUniversityPress.

AUSTIN,J.L.(1962).Senseandsensibilia.Oxford:OxfordUniversityPress.AYER,A.J.(1940).Thefoundationsofempiricalknowledge.London:Macmillan.

BARTLETT,F.C.(1932).Remembering:astudyinexperimentalandsocialpsychology.Cambridge:Cam-bridgeUniversityPress.

BECHTEL,W.(1998).Representationsandcognitiveexplanations:assessingthedynamicist’schallenge

incognitivescience.CognitiveScience,22,295–318.

BIRNBAUM,L.(1991).RigorMortis:aresponsetoNilsson’s“Logicandarti󰁎cialintelligence.”Arti󰁉cial

Intelligence,47,57–77.

BLOCK,N.(1986).Advertisementforasemanticsforpsychology.MidwestStudiesinPhilosophy,X,

615–78.

BROOKS,R.A.(1991).Intelligencewithoutrepresentation.Arti󰁉cialIntelligence,47,139–159.

BUROKER,J.V.(1996).Introduction.InA.ARNAULD&P.NICOLE,Logicortheartofthinking,J.V.

BUROKER(Ed.,Trans.).Cambridge:CambridgeUniversityPress.

CARRUTHERS,P.(1996).Language,thoughtandconsciousness.Cambridge:CambridgeUniversityPress.CARRUTHERS,P.(1998).Thinkinginlanguage?Evolutionandamodularistpossibility.InP.CAR-RUTHERS&J.BOUCHER(Eds)Languageandthought:interdisciplinarythemes,Cambridge:CambridgeUniversityPress.

CARRUTHERS,P.&SMITH,P.K.(1996).Theoriesoftheoriesofmind.Cambridge:CambridgeUniversity

Press.

CHALMERS,D.(1996).Theconsciousmind.Oxford:OxfordUniversityPress.

CHOMSKY,N.(1962).Explanatorymodelsinlinguistics.InE.NAGEL,P.SUPPES&A.TARSKI(Eds)

Logic,methodologyandphilosophyofscience.Stanford:StanfordUniversityPress.CHOMSKY,N.(1966).Cartesianlinguistics.NewYork:Harper&Row.

CHOMSKY,N.(1996).Powerandprospects:re󰁊ectionsonhumannatureandthesocialorder.Sydney:Allen

&Unwin.

CLARK,A.&TORIBIO,J.(1994).Doingwithoutrepresenting?Synthese,101,401–431.

CUMMINS,R.(1996).Representations,targetsandattitudes.Cambridge,MA:Bradford/MIT.

DAVIDSON,D.(1975).Thoughtandtalk.InS.GUTTENPLAN(Ed.)Mindandlanguage.Oxford:

ClarendonPress.

DAVIES,M.&STONE,T.(Eds)(1995a).Folkpsychology.Oxford:Blackwell.DAVIES,M.&STONE,T.(Eds)(1995b).Mentalsimulation.Oxford:Blackwell.

DENNETT,D.C.(1978a).Acureforthecommoncode.InBrainstorms.Montgomery,VT:Bradford

Books.

DENNETT,D.C.(1978b).Arti󰁎cialintelligenceasphilosophyandaspsychology.InBrainstorms.

Montgomery,VT:BradfordBooks.ENNETTD,D.C.(1991).Consciousnessexplained.London:Penguin.

DESCARTES,R.(1662/1972).Treatiseofman,commentarybyT.S.HALL(Trans.).Cambridge,MA:

HarvardUniversityPress.

DESCARTES,R.(1637/1985).Dioptrics.InThephilosophicalwritingsofDescartes,2vols,J.COTTINGHAM,

R.,STOOTHOFF&D.MURDOCH(Trans.).Cambridge:CambridgeUniversityPress.

DRETSKE,F.(1986).Misrepresentation.InR.J.BOGDAN(Ed.)Belief:form,contentandfunction.Oxford:

OxfordUniversityPress.

EDELMAN,S.(1998).Representationisrepresentationofsimilarities.BehavioralandBrainSciences,21,

449–498.

ELIASMITH,C.(1996).Thethirdcontender:acriticalexaminationofthedynamicisttheoryofcognition.

PhilosophicalPsychology,9,441–463.

FODOR,J.A.(1968).Psychologicalexplanation.NewYork:RandomHouse.FODOR,J.A.(1975).Thelanguageofthought.NewYork:Crowell.FODOR,J.A.(1978).Propositionalattitudes.TheMonist,61,4.

FODOR,J.A.(1980).Methodologicalsolipsismconsideredasaresearchstrategyincognitivepsychology.

BehavioralandBrainSciences,3,63–109.

FODOR,J.A.(1985a).ApresentationtotheNationalScienceFoundationWorkshoponInformationand

268P.SLEZAK

Representation.InB.H.PARTEE,S.PETERS&R.THOMASON(Eds)ReportofWorkshoponInfor-mationandRepresentation.Washington,DC:NSFSystemDevelopmentFoundation.FODOR,J.A.(1985b).Fodor’sguidetomentalrepresentation.Mind,Spring,55–97.

FODOR,J.A.(1994a).Theelmandtheexpert:mentaleseanditssemantics.Cambridge,MA:MIT.FODOR,J.A.(1994b).Concepts:apotboiler.Cognition,50,95–113.

FODOR,J.A.(1998).Concepts:wherecognitivesciencewentwrong.Oxford:OxfordUniversityPress.FREEMAN,W.J.&SKARDA,C.A.(1990).Representations:whoneedsthem?InJ.L.MCGAUGH,N.

WEINBERGER&G.LYNCH(Eds)Brainorganizationandmemorycells,systemsandcircuits.Oxford:OxfordUniversityPress.

GALILEI,G.(1610/1983).Thestarrymessenger.InS.DRAKE(Ed.)Telescopes,tides&tactics.Chicago:

UniversityofChicagoPress.

GAUKROGER,S.(1990).Thebackgroundtotheproblemofperceptualcognition.InA.ARNAULD,On

trueandfalseideas,S.GAUKROGER(Trans.).Manchester:ManchesterUniversityPress.GAUKROGER,S.(1996).Descartes:anintellectualbiography.Oxford:OxfordUniversityPress.GETTIER,E.L.(1963).Isjusti󰁎edtruebeliefknowledge?Analysis,23,121–123.

GREENO,J.G.(19).Situations,mentalmodelsandgenerativeknowledge.InD.KLAHR&K.KO-TOVSKY(Eds)Complexinformationprocessing:theimpactofHerbertA.Simon.Hillsdale,NJ:LawrenceErlbaum.

GREGORY,R.L.(1997).Knowledgeinperceptionandillusion.PhilosophicalTransactionsoftheRoyal

SocietyLondonB,352,1121–1128.

JACKENDOFF,R.(1992).Languagesofthemind.Cambridge,MA:Bradford/MIT.KOSSLYN,S.M.(1980).Imageandmind.Cambridge,MA:HarvardUniversityPress.

KOSSLYN,S.M.(1987).Seeingandimagininginthecerebralhemispheres:acomputationalapproach.

PsychologicalReview,94,148–175.

KOSSLYN,S.M.(1994).Imageandbrain:theresolutionoftheimagerydebate.Cambridge,MA:MIT.KOSSLYN,S.M.,PINKER,S.,SMITH,G.E.&SCHWARTZ,S.P.(1979).Onthedemysti󰁎cationofmental

imagery.TheBehavioralandBrainSciences,2,535–581.

KOSSLYN,S.M.,SOKOLOV,M.A.&CHEN,J.C.(19).ThelateralizationofBRIAN:acomputational

theoryandmodelofvisualhemisphericspecialization.InD.KLAHR&K.KOTOVSKY(Eds)Complexinformationprocessing:theimpactofHerbertA.Simon,Hillsdale,NJ:LawrenceErlbaum.ENINL,V.I.(1927).Materialismandempirio-criticism:criticalnotesconcerningareactionaryphilosophy.London:MartinLawrence.

LLOYD.D.(forthcoming).Representation.InMacmillanencyclopediaofcognitivescience.London:

Macmillan.

LOCKE,J.(1690/19).Anessayconcerninghumanunderstanding,A.D.WOOZLEY(Ed.).London:

Collins.

MALEBRANCHE,N.(1712/1997).Thesearchaftertruth,T.M.LENNON&P.J.OLSCAMP(Eds,Trans.).

Cambridge:CambridgeUniversityPress.

MARKMAN,A.B.&DIETRICH,E.(2000).Indefenseofrepresentation.CognitivePsychology,40,138–171.MEYERING,T.C.(1997).Representationandresemblance:areviewessayofRichardA.Watson’s

“RepresentationalideasfromPlatotoPatriciaChurchland.”PhilosophicalPsychology,10,221–230.MILLIKAN,R.G.(1991).SpeakingupforDarwin.InB.LOEWER&G.REY(Eds)Meaninginmind.

Oxford:BasilBlackwell.MOREAU,D.(1999).DeuxCarte´siens:lapole´miqueentreAntoineArnauldetNicolasMalebranche.Paris:

Vrin.

NADLER,S.(19).ArnauldandtheCartesianphilosophyofideas.Manchester:ManchesterUniversity

Press.

NADLER,S.(1992).Malebrancheandideas.Oxford:OxfordUniversityPress.NEISSER,U.(1976).Cognitionandreality.NewYork:Freeman.

NEWELL,A,(1986).Thesymbollevelandtheknowledgelevel.InZ.PYLYSHYN&W.DEMOPOULOS

(Eds)Meaningandcognitivestructure.Norwood,NJ:Ablex.

NILSSON,N.J.(1987).CommentaryonMcDermott.ComputationalIntelligence,3,202–203.NILSSON,N.J.(1991).Logicandarti󰁎cialintelligence.Arti󰁉cialIntelligence,47,31–56.

THETRIPARTITEMODELOFREPRESENTATION

269

PALMER,S.E.(1978).Fundamentalaspectsofcognitiverepresentation.InE.ROSCH&B.LLOYD(Eds)

Cognitionandcategorization.Hillsdale,NJ:LawrenceErlbaum.

PASNAU,R.(1997).TheoriesofcognitioninthelaterMiddleAges.Cambridge:CambridgeUniversityPress.PINKER,S.&FINKE,R.(1980).Emergenttwo-dimensionalpatternsinimagesrotatedindepth.Journal

ofExperimentalPsychology:HumanPerceptionandPerformance,6,244–2.

PLACE,U.T.(1956).Isconsciousnessabrainprocess?BritishJournalofPsychology,47,44–50.

POPPER,K.R.(1963a).AnoteonBerkeleyasprecursorofMachandEinstein.InConjecturesand

refutations.London:Routledge&KeganPaul.

POPPER,K.R.(1963b).Threeviewsconcerninghumanknolwedge.InConjecturesandrefutations.

London:Routledge&KeganPaul.

PUTNAM,H.(1975).Themeaningof“meaning.”InK.GUNDERSON(Ed.)Language,mindand

knowledge:Minnesotastudiesinthephilosophyofscience,Vol.7.Minneapolis:UniversityofMinnesotaPress.

PYLYSHYN,Z.(1973).Whatthemind’seyetellsthemind’sbrain:acritiqueofmentalimagery.

PsychologicalBulletin,80,1–24.

PYLYSHYN,Z.(1978).Imageryandarti󰁎cialintelligence.InC.W.SAVAGE(Ed.)Minnesotastudiesinthe

philosophyofscience,Vol.IX.Minneapolis:UniversityofMinnesotaPress.

PYLYSHYN,Z.(1981).Theimagerydebate.InN.BLOCK(Ed.)Imagery.Cambridge,MA:MIT.PYLYSHYN,Z.(inpress).Mentalimagery:insearchofatheory.BehavioralandBrainSciences.

RAMSEY,W.,SITCH,S.&GARON,J.(1991).Connectionism,eliminativismandthefutureoffolk

psychology.InW.RAMSEY,S.STICH&D.RUMELHART(Eds)Philosophyandconnectionisttheory.Hillsdale,NJ:LawrenceErlbaum.

REYNOLDS,S.L.(2000).TheArgumentfromIllusion.Nouˆs,34,604–621.

RORTY,R.(1979).Philosophyandthemirrorofnature.Princeton:PrincetonUniversityPress.

ROSENSCHEIN,S.J.(1985).FormaltheoriesofknowledgeinAIandrobotics.NewGenerationComputing,

3,345–357.

RUMELHART,D.E.&NORMAN,D.A.(1983).Representationinmemory.CenterforHumanInformation

Processing,TechnicalReportCHIP116,UniversityofCalifornia,SanDiego.

RYLE,G.(1968).Apuzzlingelementinthenotionofthinking.InP.F.STRAWSON(Ed.)Studiesinthe

philosophyofthoughtandaction.Oxford:OxfordUniversityPress.EARLES,J.(1980).Minds,brainsandprograms.BehavioralandBrainSciences,3,417–424.

SHEPARD,R.N.&METZLER,J.(1971).Mentalrotationofthree-dimensionalobjects.Science,171,

701–703.

SKINNER,B.F.(1963).Behaviorismat󰁎fty.Science,140,951–58.

SLEZAK,P.(1990).Mannotasubjectforscience?SocialEpistemology,4,327–342.

SLEZAK,P.(1992).Whencanimagesbereinterpreted:non-chronometrictestsofpictorialism.In

Proceedingsof14thConferenceoftheSocietyforCognitiveScience.Hillsdale,NJ:LawrenceErlbaum.SLEZAK,P.(1994).Situatedcognition:empiricalissue,paradigmshiftorconceptualconfusion.In

Proceedingsof16thConferenceoftheSocietyforCognitiveScience.Hillsdale,NJ:LawrenceErlbaum.SLEZAK,P.(1995).The“philosophical”caseagainstvisualimagery.InP.SLEZAK,T.CAELLI&R.

CLARK(Eds)Perspectivesoncognitivescience.Norwood:Ablex.

SLEZAK,P.(1999).Situatedcognition:empiricalissue,paradigmshiftorconceptualconfusion?InJ.

WILES&T.DARTNALL(Eds)Perspectivesoncognitivescience,Vol.2.Norwood,NJ:Ablex.

SLEZAK,P.(2000).Descartes’startlingdoctrineofthereverse-signrelation.InS.GAUKROGER,J.

SCHUSTER&J.SUTTON(Eds)Descartes’naturalphilosophy.London:Routledge.

SLEZAK,P.(2002).Thinkingaboutthinking:language,thought&introspection.LanguageandCom-munication,22,353–373.

SLEZAK,P.(inpress).Mentalimagery:de´ja`vualloveragain?BehavioralandBrainSciences.

SLEZAK,P.(forthcoming).Images,illusions,mistakes&misrepresentations:theworldgonewrong.In

P.STAINES,H.CLAPIN&P.SLEZAK(Eds)Representationinmind.Westport,CT:Praeger.SMITH,B.C.(1987).Thecorrespondencecontinuum.CSLIReport87–71.

STALNAKER,R.(1991).Howtodosemanticsforthelanguageofthought.InB.LOEWER&G.REY(Eds)

Meaninginmind.Oxford:BasilBlackwell.

270P.SLEZAK

SUTTON,J.(1998).Philosophyandmemorytraces:Descartestoconnectionism.Cambridge:Cambridge

UniversityPress.

TYE,M.(1984).Theadverbialapproachtovisualexperience.ThePhilosophicalReview,93,195–225.VANGELDER,T.(1998).Thedynamicalhypothesisincognitivescience.BehavioralandBrainSciences,

21,615–665.

WIMMER,H.&PERNER,J.(1983).Beliefsaboutbeliefs:representationandconstrainingfunctionof

wrongbeliefsinyoungchildren’sunderstandingofdeception.Cognition,13,103–128.WINCH,P.(1957).Theideaofasocialscience.London:Routledge.WRIGHT,E.(Ed.)(1993).Newrepresentationalisms.London:Averbury.

YOLTON,J.W.(1956/1993).Lockeandthewayofideas.Bristol:ThoemmesPress.

YOLTON,J.W.(1984).PerceptualacquaintancefromDescartestoReid.Minneapolis:UniversityofMin-nesotaPress.

YOLTON,J.W.(1996).Perceptionandreality:ahistoryfromDescartestoKant.Ithaca:CornellUniversity

Press.

YOLTON,J.W.(2000).Repliestomyfellowsymposiasts.InS.GAUKROGER,J.SCHUSTER&J.SUTTON

(Eds)Descartes’naturalphilosophy.London:Routledge.

ZAHAR,E.(1981).SecondthoughtsaboutMachianpositivism:areplytoFeyerabend.BritishJournalfor

thePhilosophyofScience,32,267–276.