Cognitive Architecture

You might also like

Download as ppt, pdf, or txt
Download as ppt, pdf, or txt
You are on page 1of 54

CognitiveArchitecture

B.KiranMaruthi09005047
M.Sonika09005054
D.V.Ramana09005059

OUTLINE
WhatisCognitiveArchitecture?
PlausibilityofCognitiveArchitectures
TypeIdentityTheory
Functionalism
HistoryofCognitiveArchitecture
GeneralCharacteristics
Consciousness
UnifiedTheoryofCognition
SOARcasestudy

IntelligentAgents
Entitieswhichobservethroughsensorsandactuponthe
environmentusingactuatorsanddirecttheiractivitytowards
achievinggoals.

WhatisCognitiveArchitecture?
Blueprintforintelligentagents.

Itproposes(artificial)computationalprocessesthatact
likecognitivesystems(human)
Anapproachthatattemptstomodelbehavioralaswellas
structuralpropertiesofthemodeledsystem.

Aim:tomodelsystemsthataccountsforthewholeof
cognition,i.e.,systemswithArtificialConsciousness
whichcannotonlyrespondbutalsothink,perceiveand
believelikeahuman!

ArtificialConsciousness
ArtificialConsciousnessisbroadlyclassified
asaccessandphenomenalconsciousness.
Brainprocessesneuralimpulsesfrom
theeyesanddeterminesthatthisimageis
physicallyunstablepatternrecognizability.
Whataboutpain,anger,motivation,attention,feelingof
relevance,modelingotherpeople'sintentions,anticipating
consequencesofalternativeactions,orinventing?

PlausibilityofArtificial
Consciousness
AviewskepticalofACisheldbytypeidentitytheorists
consciousnesscanonlyberealizedinparticular
physicalsystemsbecauseconsciousnesshaspropertiesthat
necessarilydependonphysicalconstitution

However,forfunctionalists,
anysystemthatcaninstantiatethesamepatternof
causalroles,regardlessofphysicalconstitution,will
instantiatethesamementalstates,including
consciousness
Alongtheselines,sometheoristshaveproposedthat
consciousnesscanberealizedinproperlydesignedand
programmedcomputers.

TypeIdentityTheory
Thementaleventscanbegroupedintotypesand
associatedwithtypesofphysicaleventsinthebrain.
Forexample,mentaleventpainresultsinphysicaleventin
thebrain(likeCfiberfirings)
Wehavetwototallydifferentversionsoftypeidentity
theorybasedonthedefinitionofwhatkindofidentityis
associatedwithmentalandphysicalevents.
UllinPlace(1956)CompositionalIdentity
Feigl(1957)andSmart(1959)ReferentialIdentity

CompositionalTypeIdentityTheory
U.T.Place'snotionofidentityisdescribedasarelationof
composition.
Everymentalprocessiscomposedofasetofphysical
sensationstowhichitreacts.Butcanweassociatethem
basedpurelyoncomposition?
"lightningisanelectricaldischarge"istrue.

ReferentialTypeIdentityTheory
ForFeiglandSmart,theidentitywastobeinterpretedas
theidentitybetweenthereferentsoftwodescriptionswhich
referredtothesamething.
themorningstarandtheeveningstarareidenticalinthe
sensethatbothofthemrefertotheVenus.
Sensationsandbrainprocessesdoindeedmeandifferent
thingsbuttheyrefertothesamephysicalphenomenon.This
iscalledasTheFregeandistinction
Conclusion:Alloftheversionssharethecentralideathat
themindisidenticaltosomethingphysical.

IMPOSSIBRU!!!

MultipleRealizability
Objectionstothetype
identitytheory
HilaryPutnampopularizedit
inlate1960s.
Itstatesthatthesame
mentalproperty,state,or
eventcanbeimplementedby
differentphysicalproperties,
statesorevents.

Putnam'sFormulation
Doallorganismshavethesamebrainstructures?Clearly
not!

Paincorrespondstocompletelydifferentphysicalstatesand
yettheyallexperiencethesamementalstateof"beingin
pain."
Shouldrobotsbeconsideredaprioriincableofexpereincing
painjustbecausetheydidnotpossesthesame
neurochemistryashumans?
Putnamconcludedthattypeidentityismakingan
implausibleconjecture.

Functionalism
Coreideaisthatmentalstatesareconstitutedsolelybytheir
functionalrole
Theyarecausalrelationstoothermentalstates,sensory
inputs,andbehavioraloutputs.
Brainsarephysicaldeviceswithneuralsubstratethat
performcomputationsoninputswhichproducebehaviours.
Accordingtothistheoryitispossibletobuildsiliconbased
deviceswhicharefunctionallyisomorphictothehumans
aslongassystemperformsappropriatefunctions.

VariationsofFunctionalism
MachineStatefunctionalismHilaryPutnam
MentalstateislikeautomatonstateofaTuringMachine.
Eachstatecanbedefinedexclusivelyintermsofits
relationstotheotherstatesaswellasinputsandoutputs.
Beinginpainisthestatewhichdisposesonetocry"ouch"!

VariationsofFunctionalismCont...
PsychofunctionalismJerryFordor
Theroleofmentalstates,suchasbeliefanddesire,is
determinedbythefunctionalorcausalrolethatisdesignated
forthemwithinourbestscientificpsychologicaltheory.
Ifsomenewmentalstatefromfolkpsychologycomes,itis
considerednonexistentasithasnofundamentalrolein
cognitivepsychologicalexplaination.
Sometheoreticalcognitivepsychologicalstateswhichare
necessaryforexplainationofhumanbehaviourbutarenot
foreseenbynormalfolkpsychology,alsoexistinthesystem.

QuickQuestion

WhatdifferencedoesthecolourREDmake?

Qualia
FromtheLatin,meaning"whatkind".
referstothesubjectivequalitiesofsensoryperceptionand
thefeelingtheygenerate.
Qualiaisnotonlytherednessofred,butthewaythat
rednessmakesusfeel.
Qualiaare,inessence,ourownuniqueandpersonal
perceptionsofourenvironment.

Mary'sthoughtexperiment
FrankJacksonofferstheknowledgeargumentforqualia.
Mary,thecolourscientistknowsallthephysicalfactsabout
colourandtheexperienceofcolourwithotherpeople.

Confinedfrombirthtoaroomthatisblackandwhite.
Whensheisallowedtoleavetheroom,itmustbeadmitted
thatshelearnssomethingaboutthecolourredthefirsttime
sheseesitspecifically,shelearnswhatitisliketosee
thatcolour.
Thisattackstheknowledgecompletenessoffunctionalism.

AbilityHypothesis
Nemirowclaimsthat"knowingwhatanexperienceislikeis
thesameasknowinghowtoimaginehavingtheexperience".
HearguesthatMaryonlyobtainedtheabilitytodo
something,nottheknowledgeofsomethingnew.
Marygainedanabilityto"remember,imagineandrecognize."
Knowingwhatit'sliketoseeredismerelyasortofpractical
knowledge,aknowinghow(toimagine,remember,orre
identify,acertaintypeofexperience)ratherthana
knowledgeofpropositionsorfacts.

FunctionalIsomorphism
Putnamdefinedtheconceptoffunctionalisomorphismas:
Twosystemsarefunctionallyisomorphicifthereisa
correspondencebetweenthestatesofoneandthestatesof
theotherthatpreservesfunctionalrelations.

Presently...
Functionalismiswidelyacceptedandresearchtodevelop
cognitiverobotsison!

CognitiveArchitecture
UsingPutnam'sMultipleRealizabilityformulationand
functionalism,DavidChalmersinlate1960ssuggestedthe
possibiltyofmechanismsandstructuresthatunderlie
Cognition:
processorsthatmanipulatedata
memoriesthatholdknowledgeand
interfacesthatinteractwithanenvironment.

HistoryofCognitiveArchitecture
19692000(timeline)

GPS (Ernst & Newell, 1969) Means-ends analysis, recursive subgoals

1970

ACT (Anderson, 1976) Human semantic memory

CAPS (Thibadeau, Just, Carpenter) Production system for modeling reading

1975

Soar (Laird, & Newell, 1983) Multi-method problem solving, production systems, and problem spaces

Theo (Mitchell et al., 1985) Frames, backward chaining, and EBL

1980

PRS (Georgeff & Lansky, 1986) Procedural reasoning & problem solving

BB1/AIS (Hayes-Roth & Hewitt 1988) Blackboard architecture, meta-level control

1985

Prodigy (Minton et al., 1989) Means-ends analysis, planning and EBL

MAX (Kuokka, 1991) Meta-level reasoning for planning and learning

1990

Icarus (Langley, McKusick, & Allen,1991) Concept learning, planning, and learning

3T (Gat, 1991) Integrated reactivity, deliberation, and planning

1995

CIRCA (Musliner, Durfee, & Shin, 1993) Real-time performance integrated with planning

AIS (Hayes-Roth 1995) Blackboard architecture, dynamic environment

2000

EPIC (Kieras & Meyer, 1997) Models of human perception, action, and reasoning

APEX (Freed et al., 1998) Model humans to support human computer designs

Characteristics
Holism,e.g.Unifiedtheoryofcognition
Thearchitectureoftentriestoreproducethebehaviorofthe
modelledsystem(human),inawaythattimelybehavior
(reactiontimes)ofbotharecomparable
Othercognitivelimitationsareoftenmodeledaswell
Robustbehavior
Parameterfree
ArtificiallyConscious

ArtificialConsciousness
ThefunctionsofconsciousnesssuggestedbyBernardBaars:
DefinitionandContextSetting
AdaptationandLearning
AnticipationFunction
PrioritizingandAccessControl
DecisionmakingorExecutiveFunction
AnalogyformingFunction
MetacognitiveandSelfmonitoringFunction
AutoprogrammingandSelfmaintenanceFunction
DefinitionalandContextsettingFunction.

Learning

Reactiontimeforconsecutivereadings?
HumanimprovementviaPractise

Anticipation
Machineneedsflexible,realtimecomponentsthatpredict
worlds.
Aconsciousmachineshouldmakecoherentpredictionsand
plans,forenvironmentsthatmaychange.
Executedonlywhenappropriatetosimulateandcontrolthe
realworld.
Significantresearchonroleofconsciousnessincognitive
models.Examples:CLARION,OpenCog

UnifiedTheoryofCognition
BookwrittenbyAllenNewell
Newell'sgoal:
Todefinethearchitectureofhumancognition,whichisthe
waythathumansprocessinformation.Thisarchitecture
mustexplainhowwereacttostimuli,exhibitgoaldirected
behavior,acquirerationalgoals,representknowledge,and
learn.

Newell'sCognitiveModel
NewellintroducesSoar,anarchitectureforgeneral
cognition.
Soaristhefirstproblemsolvertocreateitsownsubgoals
andlearncontinuouslyfromitsownexperience.
Soarhastheabilitytooperatewithintherealtime
constraintsofintelligentbehavior,suchasimmediate
responseanditemrecognitiontasks.

Soar
WhatisSoar?
HistoryofSoar
ArchitectureofSoar
EvolutionofSoarandpresentversion

WhatisSoar?
Soarisasymboliccognitivearchitecture.
AnAIprogramminglanguage.
Itprovidesa(cognitive)architecturalframework,within
whichyoucanconstructcognitivemodels.

Itcanbeconsideredasanintegratedarchitecturefor
knowledgebasedproblemsolving,learning,andinteraction
withexternalenvironments.

History
CreatedbyJohnLaird,AllenNewell,andPaulRosenbloom
atCarnegieMellonUniversityin1983.

JohnLairdAllenNewellPaulRosenbloom

It'sSoarnotSOAR!
Historically,SoarstoodforState,OperatorAndResult,
becauseallproblemsolvinginSoarisregardedasasearch
throughaproblemspaceinwhichyouapplyanoperatorto
astatetogetaresult.
Overtime,thecommunitynolongerregardedSoarasan
acronym:thisiswhyitisnolongerwritteninuppercase

ScreenshotSoarDebugger

ProblemSpaces
Soarrepresentsalltasksascollectionsofproblemspaces.
Problem spaces are made up of a set of states and operators
that manipulate the states.
Soar begins work on a task by choosing a problem space,
then an initial state in the space. Soar represents the goal of
the task as some final state in the problem space.

StructureofSoar
Soarcanbedividedinto3levels:
MemoryLevel
DecisionLevel
GoalLevel

MemoryLevel
Ageneralintelligencerequiresamemorywithalarge
capacityforthestorageofknowledge.
Avarietyoftypesofknowledgemustbestored,including:
declarativeknowledge
proceduralknowledge
episodicknowledge

LongtermProductionMemory
AllofSoar'slongtermknowledgeisstoredinasingle
productionmemory.
Eachproductionisaconditionactionstructurethat
performsitsactionswhenitsconditionsaremet.
Memoryaccessconsistsoftheexecutionofthese
productions.
Duringtheexecutionofaproduction,variablesinits
actionsareinstantiatedwithvalue.

WorkingMemory
Theresultofmemoryaccessistheretrievalofinformationinto
aglobalworkingmemory.
ItisthetemporarymemorythatcontainsallofSoar'sshort
termprocessingcontext.Ithas3components:
Thecontextstackspecifiesthehierarchyofactivegoals,
problemspaces,statesandoperators
objects,suchasgoalsandstates(andtheirsubobjects)
preferencesthatencodetheproceduralsearchcontrol
knowledge

Soar
Architecture

Preferences
Thereisonespecialtypeofworkingmemorystructure
thepreference
Preferencesencodecontrolknowledgeaboutthe
acceptabilityanddesirabilityofactions.
Acceptabilitypreferencesdeterminewhichactionsshould
beconsideredascandidates.
Desirabilitypreferencesdefineapartialorderingonthe
candidateactions.

DecisionLevel
Thedecisionlevelisbasedonthememorylevelplusan
architecturallyprovided,fixed,decisionprocedure.
Thedecisionlevelproceedsinatwophaseelaboratedecide
cycle.
Duringelaboration,thememoryisaccessedrepeatedly,in
parallel,untilquiescenceisreached;thatis,untilnomore
productionscanexecute.
Thisresultsintheretrievalintoworkingmemoryofallof
theaccessibleknowledgethatisrelevanttothecurrent
decision.
Afterquiescencehasoccurred,thedecisionprocedure
selectsoneoftheretrievedactionsbasedonthepreferences
thatwereretrievedintoworkingmemory.

GoalLevel
Ageneralintelligencemustbeabletosetandwork
towardsgoals.Thislevelisbasedonthedecisionlevel.
Goalsaresetwheneveradecisioncannotbemade;thatis,
whenthedecisionprocedurereachesanimpasse.
Impassesoccurwhentherearenoalternativesthatcanbe
selected(nochangeandrejectionimpasses)orwhenthere
aremultiplealternativesthatcanbeselected,but
insufficientdiscriminatingpreferencesexisttoallowa
choicetobemadeamongthem(tieandconflictimpasses).

ImpasseResolution
Wheneveranimpasseoccurs,thearchitecturegeneratesthe
goalofresolvingtheimpassewhichbecomesthesubgoal.
Alongwiththisgoal,anewperformancecontextiscreated.

Thecreationofanewcontextallowsdecisionstocontinueto
bemadeintheserviceofachievingthegoalofresolvingthe
impasse.
Astackofimpassesispossible.
Theoriginalgoalisresumedafteralltheimpassestackis
cleared.

LearningthroughChunking
Inadditiontoallabovelevels,ageneralintelligence
requirestheabilitytolearn.

Alllearningoccursbytheacquisitionofchunks
productionsthatsummarizetheproblemsolvingthat
occursinsubgoals,amechanismcalledChunking
Theactionsofachunkrepresenttheknowledgegenerated
duringthesubgoal;thatis,theresultsofthesubgoal.

EvolutionofSoar
YEAR

VERSION

IMPLEMENTED IN

1982

Soar 1

Lisp

1983

Soar 2

Lisp/OPS5

1984

Soar 3

1986

Soar 4

1989

Soar 5

1992

Soar 6

1996

Soar 7

Tcl/tk

Soar9:InterestingDevelopement
UnifyingCognitiveFunctionsandEmotionalAppraisal
Thefunctionalandcomputationalroleofemotionisopento
debate.
Appraisaltheoryistheideathatemotionsareextracted
fromourevaluations(appraisals)ofeventsthatcause
specificreactionsindifferentpeople.
Themaincontroversysurroundingthesetheoriesargues
thatemotionscannothappenwithoutphysiologicalarousal.

Appraisal'sDetector
Thistheoryproposesthatanagentcontinuallyevaluatesa
situationandthatevaluationleadstoemotion.
Theevaluationishypothesizedtotakeplacealongmultiple
dimensions,suchas
goalrelevance
goalconduciveness
causalityandcontrol
Thesedimensionsareexactlywhatanintelligentagent
needstocomputeasitpursuesitsgoalswhileinteracting
withanenvironment.

Conclusion
Thiseldstillhasfartotravelbeforeweunderstandfully
thespaceofcognitivearchitecturesandtheprinciplesthat
underlietheirsuccessfuldesignandutilization.
However,wenowhaveovertwodecadesexperiencewith
constructingandusingavarietysucharchitecturesfora
widerangeofproblems,alongwithanumberofchallenges
thathaveariseninthispursuit.
Ifthesceneryrevealedbytheseinitialstepsareany
indication,thejourneyaheadpromisesevenmore
interestingandintriguingsitesandattractions.

Soar9:AppraisalDetector

References
1)SOAR:AnArchitectureforGeneralIntelligence,JohnE.
Laird,AllenNewell,PaulS.Rosenbloom,1986.
2)ApreliminaryanalysisoftheSoararchitectureasabasis
forgeneralintelligence,JohnE.Laird,AllenNewell,Paul
S.Rosenbloom,1989.
3)http://en.wikipedia.org/wiki/Cognitive_architecture
4)http://cs.gmu.edu/~eclab/research.html
5)http://en.wikipedia.org/wiki/Unified_theory_of_cognition
6)http://cll.stanford.edu/research/ongoing/icarus/
7)http://en.wikipedia.org/wiki/Artificial_consciousness
8)http://plato.stanford.edu/entries/functionalism/

References
9)ASurveyofCognitiveArchitectures,DavidE.Kieras,
UniversityofMichigan.
10)ConnectionismandCognitiveArchitecture:ACritical
Analysis,JerryA.FodorandZenonW.Pylyshyn,Rutgers
CenterforCognitiveScience,RutgersUniversity,New
Brunswick,NJ.
11)HumanCognitiveArchitecture,JohnSweller,University
ofNewSouthWales,Sydney,Australia.
12)http://cogarch.org/index.php/Soar/Architecture
13)http://code.google.com/p/soar/wiki/Documentation

References
14)AGentleIntroductiontoSoar:AnArchitecturefor
HumanCognition:2006Update,JillFainLehman,John
Laird,PaulRosenbloom.
15)http://sitemaker.umich.edu/soar/home

You might also like