下载
加入VIP
  • 专属下载特权
  • 现金文档折扣购买
  • VIP免费专区
  • 千万文档免费下载

上传资料

关闭

关闭

关闭

封号提示

内容

首页 *新书上架*【约翰·罗伯特·安德森】认知结构

*新书上架*【约翰·罗伯特·安德森】认知结构.pdf

*新书上架*【约翰·罗伯特·安德森】认知结构

九月虺
2009-09-17 0人阅读 举报 0 0 0 暂无简介

简介:本文档为《*新书上架*【约翰·罗伯特·安德森】认知结构pdf》,可适用于人文社科领域

ril,':t:*:iiir':i:,','',n':liContents{:ProductionSystemsandACTKnowledgeRePresentationSpreadofActivationControlofCognitionr,MemoryforFactsLi"ProceduralLearningl',:,LanguageAcquisition,NotesReferencesIndexofAuthorsGeneralIndexfrF$,"t*i,i:lilf'i't,dEHFEEEEI'TEaaSFa'o{Tot:rqsn*{'fF€HSgE:{arE€r,EvsFelifTEiFqag€a:rHoE"*sEgtEircrEgEsn=HIaAgFIi$'awwBtiEi:EsEEgFrtEF=dgEEPFgHTgFuj€HHFE:ruFgs,Si,Fg,FsfixB'*i*gg'ngEa*f!rtsEnN=E!'{r!lEitBsPFHAad"ltig=tEFgFrriOF'IE'gc:tl'it)do:lEgrl*Jil'*FoiJiilIJFbO'ftogCYDV'bFlHAo^o)ttHqlbaolcgF'fDFg'!iHFfDr:tPrefaceInthemids,whenIwasanundergraduate,threeoftheactiveareasofresearchinpsychologywerelearningtheory,psycholinguistics,andcognitivepsychologyAtthattimetherewasnocoherentconnectionamongthethree,butthequestionofhowlanguagecouldbeacquiredandintegratedwiththerestofcognitionseemedinterestingHowever,therewasnoobviouswaytotacklethequestionbecausethefieldjustdidnothavetherelevantconceptsThereweretheoptionsofpursuingagraduatecareerinlearningtheory,cognitivepsychology,orpsycholinguisticsIchosecognitivepsychologyandIbelieveIchosewisely(orluckily)WhenIwenttoStanfordin,GordonBowerassignedmeahotissueofthetime:tounderstandthecategoricalstructureinfreerecallAsweanalyzedfreerecallitbecameclearthatweneededacompletemodelofthestructureofhumanmemory,withaparticularfocusonmeaningfulstructureBowerandIworkedonthisforanumberofyears,firstdevelopingthemodelFRAN(Anderson,),thenHAM(AndersonandBower,)ThroughthisworkwecunetoappreciatetheessentialroleofcomputersimulationindevelopingcomplexmodelsofcognitionHAM,themajorproductofthateffort,wasacompletemodelofthestructuresandprocessesofhumanmemory,havingasitscentralconstructaproPositionalnetworkrepresentationItsuccessfullyaddressedagreatdealofthememoryliterature,includingafairnumberofexperimentsonsentencememorythatweperformedexplicitlytotestitCriticaltopredictingaparticularexperimentalphenomenonwasdecidinghowtorepresenttheexperimentalmaterialinoursystemUsuallyourintuitionsaboutrepresentationdidleadtocorrectpredictions,butthatwasnotaverysatisfactorybasisforvllvlllPrefaceprerJictionThememorysystemhadtobeunderstoodasacomponentofamoregunur"tcognitivesystemThereasonforchoosirrgaparticulairepresentationlayinthepropertiesofthegeneralryri"rr,Forinstance,choiceofarepresentationforaJentencedependedontheoperationofanaturallanguagipurrur,whic'h,unlikethefactsrePresentedinHAM,isaskillth"ruisafundamentaldistinctionbetweendeclarativeknowledge,whichreferstofactsweknow,andproceduralknowledfe,whichreferstoskillsweknowhowtoperformHAM,athe"oryofthedeclarativesystem,wasincompletebecausethe,pruruntationchoseninthedeclarativesystemdepended'operationsofcomponentsoftheproceduralsystemsuchastheParserThefirststeptodeveloPinsamorecompletetheorywasidentifyinganappropriatefJrmalisrntomodelproceduralknowledglMyfititeffortwastousetheaugmentedtransitionnetworks(ATNs),whichhadseemedsoaPPropriatefornaturallanguageprocessingAtheoryoflanguageacquisitionwasdevetopeJittthatframeworkHowever,ATNsprovedtobeatoncetoorestrictiveascomPutationalformalismsandtoopowerfulasmodelsofhumancognitionJlresedifficultiesledmetoconsidertheproductionsystemmodelsAllenNewellhadbeenpromoting,whichhadsomesimilaritiestoATNsIhavetoadmitthataIfirst,whenIfocusedonsurfacefeatures'ProductionsystemsseemedunattractiveButasIdugdeeper,Ibecamemoreandmoreconvincedthattheycontainedkeyinsightsintothenatureofhumanproceduralknowledgern"ecrsystem,theproductofmythregyearsatMichigan,g,anddescribedinAnderson(),wasasynthesisoftheHAMmemorysystemandproductionsystuTarchitectureInthissystemaptodttionsystemwasusedforthefirsttimeasaninterpret"tofaPropositionalnetworkAcTalsoborrowedthespreidingactivationconceptfromotherresearchers,suchasCollinsandbuillianItsproductionsystemoperatedonaworkingmemorydefinedbytheactiveportionofitslongtermmemorynetfuorkInacomputersimulationversionofACT,calledACTE,itwaspossibleto,,program,,productionsetsthatmodeledvarioustasksMuchofmybookdescribessuchproductionsetsthatmodeledvarioussmalltasks'However,ultho,ghACTEansweredthequestionofwherepropositional,"pr"J"r,tationscamefromtheycamefromtheutiorofproductionsnowthenaturalquestionwas,wheredoprodulior,,comefromThisledtotheissueofalearningtheoryforproductionsystemsPaulKline,charlesBeasley,andPrefaceixIthendevelopedACTF,thefirstproductionsystemtocontainanextensivetheoryofproductionacquisitionAviableversionofACTFwasrunninginAftermovingtoCamegieMellonin,Ispentthenextfouryearstryingto"tune"theACTsystemandmakeitaddresslanguageacquisitionMyattemptsproducedsomemajorreorganizationsinthetheory,includingchangesinthespreadingactivationmechanisms,atheoryofproductionpattemmatching,augmentationstothearchitecturetohandlegoals,andadditionstotheproductionlearningmechanisms,allofwhicharedescribedinthisbookPaulKlineandDavidNevescollaboratedintheearlypartofthiseffortInthelateryearsIwasguidedbytheACTresearchgrouP,whichhasincludedFrankBoyle,GaryBradshaw,ReneeElio,RobFarrell,Billfones,MattLewis,PeterPirolli,RonSauers,MiriamSchustack,andfeffShragerIwasabletofinallyaPPlyACTtolanguageacquisitionThenewversionofACTiscalledACT*(tobereadACTstar)Thetheoryhasevolvedfromtheoriginalconcernwithcategoricalfreerecalltoaddress,llleverwideningsetofquestionsEachtheoryalongthewayraisedquestionsthatthenextoneansweredFinally,thewideningcircleshaveexpandedtoencompassmyoriginalinterestinleamingtheory,psycholinguistics,andcognitivepsychologyWhileitisundoubtedlynotthefinaltheory,itachievesagoalsetfifteenyearsearlierACT'Iisatheoryofcognitiuearchitecturethatis,atheoryofthebasicprinciplesofoperationbuiltintothecognitivesystemACTstandsforAdaptiveControlofThoughtItisworthreviewingwhatthistitlemeansandwhyitisaptFirst,thistheoryconcernshigherlevelcognitionorthoughtAmajorpresuppositioninthisbookisthathigherlevelcognitionconstitutesaunitaryhumansystemAcentralissueinhigherlevelcognitioniscontrolwhatgivesthoughtitsdirection,andwhatcontrolsthetransitionfromthoughttothoughtAswillbecomeapparent,productionsystemsaredirectedatthiscentralissueAmajorconcemformehasbeentounderstandtheprinciplesbehindthecontrolofthoughtinawaythatexposestheadaptivefunctionoftheseprinciplesUltimately,understandingadaptivefunctionbringsustoissuesofhumanevolution,whicharelargelyexcludedfromthisbook(butseeAnderson,c)ItneedstobeemphasizedthatproductionsystemsaddresstheissueofcontrolofcognitioninaprecisewaythatisrelativelyunusualincognitivepsychologyOthertyPesoftheoreticalanalysesmayproduceprecisemodelsofspecifictasks,butxPrefacehowthesystemsetsitselftodoaparticulartaskinaparticularwayislefttointuitionlnaproductionsystemthechoiceofwhattodonextismadeinthechoiceofwhatproductiontoexecutenextCentraltothischoicearetheconflictresolutionstratgi",(seeChap*lThusproductionsystemshavefinallysucceededinbanishingthehomunculusfrompsychology'InwritingthisbookIhavebeenhelpedbyagleat*TIpeoplewhoprovidedcommentsondrafts'GaryBradshawlqtltChase,ReneeElio,IraFischler,SusanFiske'|imGreeno'KeithHolyoak,Billlones,PaulKline,SteveKosslyn'MattLewis'BrianMacWhin"eyMichaelMcCloskey,AllenNewell,anePerlmutter,RolfPfeifer,stevenPinker,PeterPirolli,zenonPylyshyn,RogerRatcliff,LanceRips,PaulRosenbloom'Miriamschustack,andfeffshragerhave*^*"ttedononeormoreofthesechaptersRoberti'rederking,IaYMcClelland,EdSmith,andtwoanonymousreviewershavereadthemanuscriptinitsentiretyEriclvannerhasbeenanexcePtionaleditorreadingandcommentingontheentirebookLynneRederhasgonethroughthisboikwithme,wordbywordandideabyidea'u*poJigweaknessesandprovidinganumberoftheideasder"iop"aiereTheACTresearchg,p,aswellasmy"ThinkirrJiclasses,havegoneoI"thesJideasmanytimeswithme'tlurrypeoplehaworkedhardtogettheprogramsafexperirnents,urrningMyerafuatestudents'GaryBfd:h*'it"rr""Elio,BiillJner,MittLewis,PeterPirolli,andMiriamSchustack,havetu"r,valuablecollaboratorsTakashiIwasawaandGordonPammwrotecountlessexperimentrunningProfr*,anddataanalysesBarbaraRiehleandRaneWinslowgreatlyassisteJintestingsubiectsandTakashilwasawaandfir*kgoyledidmuchworkonthegeometryproiect'Twoout,t"iai"g'rrrdergraduates,RobFarrellandRonSauers'haveL"reJpor,siUfifortheGRAPESsimulationoftheacquisitionofLISPprogru*mingskills'Amonghermanyotherduties'RanewinslowhadfrimaryrespotiiuitityforprepalilsTd"oorai"tingthis*i^t"'ipt,andIamdeeplygrateful^tohutforallofherresourcefulnessandhardworkMonicawallace*u""rponsibleforthelastmonthsofshepherdingthebookintopublicationFinallv,IwanttothankthoseSovernmentagencieswhohave*J"itosupportmyresearchResearchgrantsfromNIEandNIMH**rucriticalinthedevelopmentofACTEandACTF,andacurrentgrantfromtheInformationsciencesLu"nofNSF(tsrgor)hassupportedthegeometryiPrefacexproiectMysimulationworkhastakenplaceonthesUMEXfalifityatSianford,supportedbygran!PLf'RfromNIHoncarnegieMellon'scomputerfacilities,supporte{uycontractFiKfromAnfeandonourPsycholgEYvAx,purchasedthroughgrantBNSfromNSFMylanguageacquisitio,,"iuuichissupportedbIacontractfromONRiwouldliketoexpressspecialappreciationtothePersonnelandTrainingResearchProgramofoNR(MarshaltTl|HenryHalff)andtotheMemoryindCognitionProgramatNSFgoeioungffresefi^ogroupshavebeenlongstan{ilgttqiortersoflheACTrer"urihatdha*teprovidedthestablefundingthataproiectofthisscaleneedsthecurrentsourcesofsuppo,r"rioodrgtCfromONRandBNSfromitSfMarshallFarr,HenryHalff,andfoeYounghavealsoprovidedvaluableinputtotheresearchP"donaliio(nFloCNs)pnle!iDqrlooqollrEAF'arFgAtaalHo)(')Kv,FoIOgEHfiHtIHfr'EbFn<t'xIrFgcssEtilEH$tieiiaIeiHrEilti',rEiiillggilleegiligggalgiiEFillii*i*r'EiilBirIBlfiliialisHirf*gt*lEi,*rnrFuaTheArchitectureofCognitionlearningtheory,orofsomeyettobeconceivedgeneralpurPoseleamingstrategythatarecommontoalldomains(Chomsky,,p)ThisfacultyapproachholdsthatdistinctcognitiveprinciplesunderlietheoPerationofdistinctcognitivefunctions'tneunitaryapproachholdsthatallhigherlevelcognitivefunctionscanbeexplainedbyonesetofprinciplesInsomewaysthefacultyapproachseemsiustplaincommonsenseManycognitivetheories,extendingbackatleasttothephrenologyofGall(seeBoring,,foradiscussionofthisandthemore"reputable"facultytheories)haveheldthisviewItstruthmightalmostseematautology:clearlyweperformdifferentintellectualfunctions,soitmightseemthatwemusthavedifferentfacultiesforthesefunctionsThefacultyproposalsthathavebeenadvancedhavealwaysgottenintodifficultiesintheirspecifics,butitneverhasbeenclearwhetherthereisanythingfundamentallywrongwiththefaculffaPproachTheearlyproposalsforunitarysystems(forexample,stimulusresponse,orSR,theories)werealsoshowntobebasicallyinadequateHowever,theunitarytheoryfoundanimportantmetaphorinthemoderngeneralpurPosecomPuterand,perhapsmoresignificantly,insymbolicProgramminglanguages,whichshowedhowasinglesetofprinciplescouldspanabroadrangeofcomputationaltasksItalsobecameclearthatthesetofcomputationalfunctionswasunlimited,meaningthatgeneralprocessingprincipleswereessentialtospanbroadrangesoftasksItmadenosensetocreateaspecialsystemforeachconceivablefunctionAnumberofcandidatesforageneralsystemhavebeenoffered,includinggeneralproblemsolvers(FikesandNilsson,NewellandSimon,Sacerdoti,),generalinferencesystems(GreenandRaphael,LMcDermottandDoyle,Robinson,),andgeneralschemasystems(BobrowandWinograd,Minsky,LRumelhartandOrtony,SchankandAbelson,)MyresearchhasbeenpredicatedonthehypothesisthatproductionsystemsprovidetherightkindofgeneralcomputationalarchitectureforachievingaunitarymentalsystemTheparticularlineofproductionsystemtheoriesIhavedevelopedallgounderthenameACTThisbookwilldescribeaspecialACTinstantiationcalledACf(tobereadACTstar)Aswillbecomeclear,ACT*isnotjustarandommemberoftheACTseriesItisProductionSystenrsandACTtheproductIhavebeenworkingtowardforthepastsevenyearsInACT*thesamecoresystemifgivenonesetofexperiencesdevelopsalinguisticfacility,ifgivenanothersetofexperiencesdevelopsageometryfacility,andifgivenanothersetofexperiencesdevelopsaprogrammingfacilityThereforeACfisverymuchaunitarytheoryofmindAncuurNTsFoRaUNrrenyTHroRyoFMrrvpqOnethingthatdistinguishesusfromothercreaturesisourabilitytoacquirecomplexskillsAlldistinctivelyhumanactivi:tiessuchasmathematics,language,chess,computerproIgramming,sculptureareacquiredskillsTheremaybeasignificantinnatecomponenttotheirsuccessfulacquisition,butwiththepossibleexceptionoflanguageitistotallyimplausibletosuggestthatwehaveevolvedspecialfacultiesor"organs"formathematics,chess,computerprogramming,orsculpturePeoplebecomeexpertatactivitiesforwhichtherewasnopossibilityofanticipationinourevolutionaryhistov,andtheessenceofthehumangeniusisjustthisplasticityItisunnecessarytoproposespecialorgansforspeciilabilitieswhenwecanfashionarticulateabilitieswherethereisnopossibilityofapriororganIfalltheseabilitiesarefashionedfromthesameinitialsystem(whichhardlyneedbeatabularasa),theninanimportantsensetheadulthumanmindisaunitaryconstructionILanguageisanimportantspecialcasethatmightbetheexceptiontotheruleItisnottotallyimplausibletoproposethatithashadalongevolutionaryhistoryinwhichvariouslanguagespecificadaptationshaveoccurredHowever,itseemsmoreplausiblethatthelanguagespecificadaptrtionsarefewandminor,thatthelanguagefacultyisreallythewholecognitivesystemInourevolutionwemayhavedevelopedorenhancedcertainfeaturestofacilitatelanguage,butoncedeveloped,thesefeatureswerenotconfinedtolanguageandarenowusedinnonlinguisticactivitiesThusthemindisageneralpoolof,'rr,,basicstructuresandprocesses,whichhasbeenaddedtounderevolutionarypressuretofacilitatelanguageTheadditionshaveIbeenusedinskills,forexample,computerprogramming,thatwerenotanticipatedintheoriginalevolutionarydevelopmentsPartoftheevidenceforthisviewaretheremarkablecommunalitiesbetweenlanguageandotherskills,whichwillbediscussedlaterinthisbookThereisatendencytoregardtheexistenceof"languageareas"andotherlocalizationsoffunctioninthebrainasstrongTheArchitectureofCognitionevidenceforfacultiesHowever,thereisnothingnecessaryaboutthisinference,asshownbyacomputeranalogy:twopro,gramscanoccupydifferentareasofcomputer*ettory,muchastwodifferentcognitiveabilitiesmightlieintwoseparateregionsofthebrainHowever,thetwoprogramsmayhaveideniicalprinciplesForinstance,IcanhaveoneACTsimulationdoinglanguageandanotherdoinggeometryThus,thereneedbenoconnectionbetweendistinctphysicallocationanddistinctcognitiveprinciplesTherealissueconcernstheuniquenessofihestructureandprocessesunderlyingcognitivefunctions,nottheirphysicallocationrAnothermajorreasonfornotbelievinginanorganforlanguageorforothercognitiveactivitiesisthattheboundariesbeitweentheseorgansJannotbedrawnaprioriItisprettyclear,'wheretheactivityofthelungleavesoffandthatofthecirculatorysystemtakesover,butthiscannotreallyusaidforcognitiveficulgiesThelungandtheheaitareboihinvolvedinanactivitysuchqg$gG,butitispossibletoidentifytheirittinctivecontribuiioffIthasbeenproposedthatthereisalanguagefaculty,anumberfaculty,adeductionfaculty,andaptoblttsolvingfaculty,butiftherearesuchfaculties,theiractivitiesareterriblyintertwinedinatasklikecomputerprogrammingWhenwelookatanexpertProgrammercreatingaprogram,w€cannotseparatethecontributionsofthevariousi"rlti"tIndeed,ifweappliedanyreasonablecriterionforindividuatingfaculties,wewouldhavetoconcludethatcomputerprogramitgwasaseparatefacultyThisisbecausesomeofiheoreprinciplesforthisskillorganization,suchasstrategiesforcreatingrecursiveprograms,aPPlyacnosstheentirerangeofprogram*ingbehaviorsandareseldomifeverevokedelsewhereSincJitisnonsensetosuggestaprogrammingfaculty,weshouldbemoreskepticalofotherproposedfacultiesAnexpert'sexecutionofaskillisspecialinthatastrongtaskllspecificcognitiveorganizationhasdevelopedthroughexten{rin""*p"rienceThisisnotthecasewiththenovice,butanalysisofnovicebehaviorgivesnomorecomforttothefacultyapproachTheremarkablefeatureofnovicesisthattheyareableioputtogethersomanydifferentfacetsofknowledgetosolveataskAnoviceprogrammerbringstogetherrecentfactslearnedaboutprogrammingandtheprogramminglanguage,factsfrommathematics,realworldexperiencesasanalogies,generalproblemsolvingskills,deductivestrategies,linguisticanalyseballtosolvetheproblemThenovice'sattemptsatsynthesizingthisknowledgecanbeterriblyofftarget,butthisisonlyforIIIItProductionSystemsandACTlackoftherightknowledge,notbecauseofafundamentalinco

用户评价(0)

关闭

新课改视野下建构高中语文教学实验成果报告(32KB)

抱歉,积分不足下载失败,请稍后再试!

提示

试读已结束,如需要继续阅读或者下载,敬请购买!

评分:

/35

VIP

在线
客服

免费
邮箱

爱问共享资料服务号

扫描关注领取更多福利