1、Introduction to ACT-R 5.0Tutorial 24th Annual Conference Cognitive Science SocietyACT-R Home Page: http:/act.psy.cmu.edu,Christian Lebiere Human Computer Interaction Institute Carnegie Mellon University Pittsburgh, PA 15213 clcmu.edu,Tutorial Overview,1. Introduction2. Symbolic ACT-RDeclarative Repr
2、esentation: ChunksProcedural Representation: ProductionsACT-R 5.0 Buffers: A Complete Model for Sentence Memory3. Chunk Activation in ACT-RActivation CalculationsSpreading Activation: The Fan EffectPartial Matching: Cognitive ArithmeticNoise: Paper Rocks ScissorsBase-Level Learning: Paired Associate
3、4. Production Utility in ACT-RPrinciples and Building Sticks Example5. Production CompilationPrinciples and Successes6. Predicting fMRI BOLD responsePrinciples and Algebra example,Motivations for a Cognitive Architecture1. Philosophy: Provide a unified understanding of the mind.2. Psychology: Accoun
4、t for experimental data.3. Education: Provide cognitive models for intelligent tutoring systems and other learning environments.4. Human Computer Interaction: Evaluate artifacts and help in their design.5. Computer Generated Forces: Provide cognitive agents to inhabit training environments and games
5、.6. Neuroscience: Provide a framework for interpreting data from brain imaging.,Approach: Integrated Cognitive Models,Cognitive model = computational process that thinks/acts like a person Integrated cognitive models,Model Predictions,Human Data,Study 1: Dialing Times,Total time to complete dialing,
6、Model Predictions,Human Data,Study 1: Lateral Deviation,Deviation from lane center (RMSE),These Goals for Cognitive Architectures Require1. Integration, not just of different aspects of higher level cognition but of cognition, perception, and action.2. Systems that run in real time.3. Robust behavio
7、r in the face of error, the unexpected, and the unknown.4. Parameter-free predictions of behavior.5. Learning.,History of the ACT-framework,Predecessor HAM (Anderson & Bower 1973)Theory versions ACT-E (Anderson, 1976)ACT* (Anderson, 1978)ACT-R (Anderson, 1993)ACT-R 4.0 (Anderson & Lebiere, 1998)ACT-
8、R 5.0 (Anderson & Lebiere, 2001)Implementations GRAPES (Sauers & Farrell, 1982)PUPS (Anderson & Thompson, 1989)ACT-R 2.0 (Lebiere & Kushmerick, 1993)ACT-R 3.0 (Lebiere, 1995)ACT-R 4.0 (Lebiere, 1998)ACT-R/PM (Byrne, 1998)ACT-R 5.0 (Lebiere, 2001)Windows Environment (Bothell, 2001)Macintosh Environme
9、nt (Fincham, 2001),I. Perception & Attention1. Psychophysical Judgements2. Visual Search3. Eye Movements4. Psychological Refractory Period5. Task Switching6. Subitizing7. Stroop8. Driving Behavior9. Situational Awareness10. Graphical User InterfacesII. Learning & Memory1. List Memory2. Fan Effect3.
10、Implicit Learning4. Skill Acquisition 5. Cognitive Arithmetic6. Category Learning7. Learning by Exploration and Demonstration8. Updating Memory &Prospective Memory9. Causal Learning, 100 Published Models in ACT-R 1997-2002,III. Problem Solving & Decision Making1. Tower of Hanoi2. Choice & Strategy S
11、election3. Mathematical Problem Solving4. Spatial Reasoning5. Dynamic Systems6. Use and Design of Artifacts7. Game Playing8. Insight and Scientific DiscoveryIV. Language Processing1. Parsing2. Analogy & Metaphor3. Learning4. Sentence MemoryV. Other1. Cognitive Development2. Individual Differences3.
12、Emotion4. Cognitive Workload5. Computer Generated Forces6. fMRI7. Communication, Negotiation, Group Decision Making,Visit http:/act.psy.cmu.edu/papers/ACT-R_Models.htm link.,ACT-R 5.0,Environment,Productions (Basal Ganglia),Retrieval Buffer (VLPFC),Matching (Striatum),Selection (Pallidum),Execution
13、(Thalamus),Goal Buffer (DLPFC),Visual Buffer (Parietal),Manual Buffer (Motor),Manual Module (Motor/Cerebellum),Visual Module (Occipital/etc),Intentional Module (not identified),Declarative Module (Temporal/Hippocampus),ACT-R: Knowledge Representation, goal buffer visual buffer retrieval buffer,ACT-R
14、: Assumption Space,ADDITION-FACT,ADDEND1,THREE,ADDEND2,FOUR,SUM,FACT3+4,(,SEVEN,),isa,Chunks: Example,CHUNK-TYPE,NAME,SLOT1,SLOT2,SLOTN,(,),Chunks: Example,(CLEAR-ALL) (CHUNK-TYPE addition-fact addend1 addend2 sum) (CHUNK-TYPE integer value) (ADD-DM (fact3+4isa addition-factaddend1 threeaddend2 four
15、sum seven)(threeisa integervalue 3)(fourisa integervalue 4)(sevenisa integervalue 7),ADDITION-FACT,FACT3+4,ADDEND1,SUM,ADDEND2,THREE,FOUR,SEVEN,isa,isa,INTEGER,isa,VALUE,VALUE,3,7,isa,Chunks: Example,VALUE,4,Chunks: Exercise I,Fact:,The cat sits on the mat.,proposition,action,cat007,sits_on,mat,isa,
16、fact007,agent,object,(Add-DM(fact007 isa propositionagent cat007action sits_onobject mat),Chunks: Exercise II,Fact,The black cat with 5 legs sits on the mat.,proposition,action,cat007,sits_on,mat,isa,fact007,agent,object,cat,isa,color,5,black,legs,Chunks: Exercise III,Fact,Chunk,The rich young profe
17、ssor buys a beautiful and expensive city house.,(Chunk-Type proposition agent action object) (Chunk-Type prof money-status age) (Chunk-Type house kind price status)(Add-DM(fact008 isa propositionagent prof08action buysobject house1001)(prof08 isa profmoney-status richage young)(obj1001 isa housekind
18、 city-houseprice expensivestatus beautiful),proposition,action,buys,isa,fact008,agent,object,prof,isa,prof08,age,young,rich,house,kind,city-house,obj1001,price,expensive,isa,status,beautiful,money- status,A Production is1. The greatest idea in cognitive science.2. The least appreciated construct in
19、cognitive science.3. A 50 millisecond step of cognition.4. The source of the serial bottleneck in otherwise parallel system.5. A condition-action data structure with “variables”.6. A formal specification of the flow of information from cortex to basal ganglia and back again.,Key Properties, modulari
20、ty abstraction goal/buffer factoring conditional asymmetry,Productions,(,p,=,),Specification of Buffer Transformations,condition part,delimiter,action part,name,Specification of Buffer Tests,Structure of productions,ACT-R 5.0 Buffers1. Goal Buffer (=goal, +goal)-represents where one is in the task-p
21、reserves information across production cycles,2. Retrieval Buffer (=retrieval, +retrieval)-holds information retrieval from declarative memory-seat of activation computations 3. Visual Buffers-location (=visual-location, +visual-location)-visual objects (=visual, +visual)-attention switch correspond
22、s to buffer transformation 4. Auditory Buffers (=aural, +aural)-analogous to visual 5. Manual Buffers (=manual, +manual)-elaborate theory of manual movement include feature preparation, Fitts law, and device properties 6. Vocal Buffers (=vocal, +vocal)-analogous to manual buffers but less well devel
23、oped,Model for Anderson (1974),Participants read a story consisting of Active and Passive sentences.Subjects are asked to verify either active or passive sentences.All Foils are Subject-Object Reversals.Predictions of ACT-R model are “almost” parameter-free.,DATA: Studied-form/Test-formActive-active
24、 Active-passive Passive-active Passive-passive Targets: 2.25 2.80 2.30 2.75 Foils: 2.55 2.95 2.55 2.95Predictions:Active-active Active-passive Passive-active Passive-passive Targets: 2.36 2.86 2.36 2.86 Foils: 2.51 3.01 2.51 3.01CORRELATION: 0.978 MEAN DEVIATION: 0.072,250m msec in the life of ACT-R
25、: Reading the Word “The”,Identifying Left-most Location Time 63.900: Find-Next-Word SelectedTime 63.950: Find-Next-Word FiredTime 63.950: Module :VISION running command FIND-LOCATIONAttending to Word Time 63.950: Attend-Next-Word SelectedTime 64.000: Attend-Next-Word FiredTime 64.000: Module :VISION
26、 running command MOVE-ATTENTIONTime 64.050: Module :VISION running command FOCUS-ONEncoding Word Time 64.050: Read-Word SelectedTime 64.100: Read-Word FiredTime 64.100: Failure RetrievedSkipping The Time 64.100: Skip-The SelectedTime 64.150: Skip-The Fired,Attending to a Word in Two Productions,(P f
27、ind-next-word=goalISA comprehend-sentenceword nil =+visual-locationISA visual-locationscreen-x lowestattended nil=goalword looking )(P attend-next-word=goalISA comprehend-sentenceword looking=visual-locationISA visual-location =goalword attending+visualISA visual-objectscreen-pos =visual-location ),
28、 no word currently being processed. find left-most unattended location update state looking for a word visual location has been identified update state attend to object in that location,Processing “The” in Two Productions,(P read-word=goalISA comprehend-sentenceword attending=visualISA textvalue =wo
29、rdstatus nil =goalword =word+retrievalISA meaningword =word )(P skip-the=goalISA comprehend-sentenceword “the“ =goalword nil ), attending to a word word has been identified hold word in goal buffer retrieve words meaningthe word is “the” set to process next word,Processing “missionary” in 450 msec.I
30、dentifying left-most unattended Location Time 64.150: Find-Next-Word SelectedTime 64.200: Find-Next-Word FiredTime 64.200: Module :VISION running command FIND-LOCATION Attending to Word Time 64.200: Attend-Next-Word SelectedTime 64.250: Attend-Next-Word FiredTime 64.250: Module :VISION running comma
31、nd MOVE-ATTENTIONTime 64.300: Module :VISION running command FOCUS-ON Encoding Word Time 64.300: Read-Word SelectedTime 64.350: Read-Word FiredTime 64.550: Missionary RetrievedProcessing the First Noun Time 64.550: Process-First-Noun SelectedTime 64.600: Process-First-Noun Fired,Processing the Word
32、“missionary”,Missionary 0.000isa MEANINGword “missionary“(P process-first-noun=goalISA comprehend-sentenceagent nilaction nilword =y=retrievalISA meaningword =y =goalagent =retrievalword nil ), neither agent or action has been assigned word meaning has been retrieved assign meaning to agent and set
33、to process next word,Three More Words in the life of ACT-R: 950 msec.,Processing “was”Time 64.600: Find-Next-Word SelectedTime 64.650: Find-Next-Word FiredTime 64.650: Module :VISION running command FIND-LOCATIONTime 64.650: Attend-Next-Word SelectedTime 64.700: Attend-Next-Word FiredTime 64.700: Mo
34、dule :VISION running command MOVE-ATTENTIONTime 64.750: Module :VISION running command FOCUS-ONTime 64.750: Read-Word SelectedTime 64.800: Read-Word FiredTime 64.800: Failure RetrievedTime 64.800: Skip-Was SelectedTime 64.850: Skip-Was FiredProcessing “feared”Time 64.850: Find-Next-Word SelectedTime
35、 64.900: Find-Next-Word FiredTime 64.900: Module :VISION running command FIND-LOCATIONTime 64.900: Attend-Next-Word SelectedTime 64.950: Attend-Next-Word FiredTime 64.950: Module :VISION running command MOVE-ATTENTIONTime 65.000: Module :VISION running command FOCUS-ONTime 65.000: Read-Word Selected
36、Time 65.050: Read-Word FiredTime 65.250: Fear RetrievedTime 65.250: Process-Verb SelectedTime 65.300: Process-Verb Fired,Processing “by”Time 65.300: Find-Next-Word SelectedTime 65.350: Find-Next-Word FiredTime 65.350: Module :VISION running command FIND-LOCATIONTime 65.350: Attend-Next-Word Selected
37、Time 65.400: Attend-Next-Word FiredTime 65.400: Module :VISION running command MOVE-ATTENTIONTime 65.450: Module :VISION running command FOCUS-ONTime 65.450: Read-Word SelectedTime 65.500: Read-Word FiredTime 65.500: Failure RetrievedTime 65.500: Skip-By SelectedTime 65.550: Skip-By Fired,(P skip-by
38、=goalISA comprehend-sentenceword “by“agent =per =goalword nilobject =peragent nil ),Reinterpreting the Passive,Two More Words in the life of ACT-R: 700 msec.Processing “the”Time 65.550: Find-Next-Word SelectedTime 65.600: Find-Next-Word FiredTime 65.600: Module :VISION running command FIND-LOCATIONT
39、ime 65.600: Attend-Next-Word SelectedTime 65.650: Attend-Next-Word FiredTime 65.650: Module :VISION running command MOVE-ATTENTIONTime 65.700: Module :VISION running command FOCUS-ONTime 65.700: Read-Word SelectedTime 65.750: Read-Word FiredTime 65.750: Failure RetrievedTime 65.750: Skip-The Selecte
40、dTime 65.800: Skip-The Fired Processing “cannibal”Time 65.800: Find-Next-Word SelectedTime 65.850: Find-Next-Word FiredTime 65.850: Module :VISION running command FIND-LOCATIONTime 65.850: Attend-Next-Word SelectedTime 65.900: Attend-Next-Word FiredTime 65.900: Module :VISION running command MOVE-AT
41、TENTIONTime 65.950: Module :VISION running command FOCUS-ONTime 65.950: Read-Word SelectedTime 66.000: Read-Word FiredTime 66.200: Cannibal RetrievedTime 66.200: Process-Last-Word-Agent SelectedTime 66.250: Process-Last-Word-Agent Fired,Retrieving a Memory: 250 msec,(P retrieve-answer=goalISA compre
42、hend-sentenceagent =agentaction =verbobject =objectpurpose test =goalpurpose retrieve-test+retrievalISA comprehend-sentenceaction =verbpurpose study ), sentence processing complete update state retrieve sentence involving verb,Time 66.250: Retrieve-Answer Selected Time 66.300: Retrieve-Answer Fired
43、Time 66.500: Goal123032 Retrieved,Generating a Response: 410 ms.,(P answer-no=goalISA comprehend-sentenceagent =agentaction =verbobject =objectpurpose retrieve-test=retrievalISA comprehend-sentence- agent =agentaction =verb- object =objectpurpose study =goalpurpose done+manualISA press-keykey “d“ ),
44、 ready to test retrieve sentence does not match agent or object update state indicate no,Time 66.500: Answer-No SelectedTime 66.700: Answer-No FiredTime 66.700: Module :MOTOR running command PRESS-KEYTime 66.850: Module :MOTOR running command PREPARATION-COMPLETETime 66.910: Device running command O
45、UTPUT-KEY,Subsymbolic Level,1. Production Utilities are responsible for determining which productions get selected when there is a conflict.2. Production Utilities have been considerably simplified in ACT-R 5.0 over ACT-R 4.0.3. Chunk Activations are responsible for determining which (if any chunks)
46、 get retrieved and how long it takes to retrieve them.4. Chunk Activations have been simplified in ACT-R 5.0 and a major step has been taken towards the goal of parameter-free predictions by fixing a number of the parameters.As with the symbolic level, the subsymbolic level is not a static level, bu
47、t is changing in the light of experience. Subsymbolic learning allows the system to adapt to the statistical structure of the environment.,The subsymbolic level reflects an analytic characterization of connectionist computations. These computations have been implemented in ACT-RN (Lebiere & Anderson
48、, 1993) but this is not a practical modeling system.,Chunk i,Seven,Three,Four,Addend1,Addend2,Sum,=Goal,isa,write,relation sum,arg1 Three,arg2 Four,+,Conditions,+Retrieval,isa,addition-fact,addend1 Three,addend2 Four,+,Actions,S,ji,Sim,kl,B,i,Activation,Chunk Activation,base activation,activation,=,
49、+,Activation makes chunks available to the degree that past experiences indicate that they will be useful at the particular moment:Base-level: general past usefulnessAssociative Activation: relevance to the general contextMatching Penalty: relevance to the specific match requiredNoise: stochastic is useful to avoid getting stuck in local minima,
