ACT-R Workshop John R. Anderson Daniel Bothell Christian Lebiere Niels A. Taatgen Schedule of Events: 9:00-10:30: ACT-R from CMU’s Perspective 11:00-12:30: Architecture 1:30-3:30: Extensions 4:00-5:30: Future of ACT-R from a non-CMU Perspective And lots of Interaction! I am John Anderson. On behalf of my co-organizers, Dan Bothell, Christian Lebiere, and Niels Taatgen I would like to welcome you to the ACT-R workshop. It is very gratifying to us to see so many of you here.

ACT-R Workshop Schedule Opening: ACT-R from CMU’s Perspective 9:00 - 9:45 Overview of ACT-R -- John R. Anderson 9:45 – 10:30 Details of ACT-R 6.0 -- Dan Bothell Break: 10:30 – 11:00 Presentations 1: Architecture 11:00 – 11:30 Functional constraints on architectural mechanisms -- Christian Lebiere 11:30 – 12:00 Retrieval by Accumulating Evidence in ACT-R -- Leendert van Maanen 12:00 – 12:30 A mechanism for decisions in the absence of prior reward -- Vladislav D. Veksler Lunch: 12:30 – 1:30 Presentations 2: Extensions 1:30 – 2:00 ACT-R forays into the semantic web -- Lael J. Schooler 2:00 – 2:30 Making Models Tired: A Module for Fatigue -- Glenn F. Gunzelmann 2:30 – 3:00 Acting outside the box: Truly embodied ACT-R -- Anthony Harrison 3:00 - 3:30 Interfacing ACT-R with different types of environments and with different techniques: Issues and Suggestions.-- Michael J. Schoelles Break: 3:30 – 4:00 Panel: 4:00 – 5:30: Future of ACT-R from a non-CMU Perspective Danilo Fum, Kevin A. Gluck, Wayne D. Gray, Niels A. Taatgen, J. Gregory Trafton, Richard M. Young This is the schedule for the workshop. Dan and I are going to give two different views of ACT-R I am going to provide some general background and general facts about ACT-R. Many in the audience will already know this but they still might find it interesting to reflect on that. Then Dan is going to provide an overview of the newest developments in ACT-R 6.0. Others in the audience may find this more detail than they think they want but they may find it interesting to hear about what the nitty gritty is like The rest of the day is between these two levels -- we will hear a set of talks about specific ideas about the ACT-R architecture and how to extend it into new domains. Finally there will be a symposium where we will get some perspectives on the future of ACT-R.

Overview of ACT-R John R. Anderson Carnegie Mellon University Outline: 9:10: Big picture of what ACT-R is about 9:20: Evolution of the Procedural Module 9:30: Evolution of the Declarative Module 9:35: How ACT-R spreads Truth of the matter is that I don’t have very much to say. So you should feel free to interrupt with questions. Dan has a lot to say. To make sure I don’t intrude on Dan’s time I have put time markers on these parts of my presentation.

ACT-R is Not Monolithic It is a community brought together by common theoretical assumptions and a commitment to the “No Magic” Principle -- cognitive theory has to run and it has to predict data. While ACT-R may be sustained from CMU it no longer resides at CMU. The community motto is “Let a thousand flowers grow” It is a set of software for purposes of simulation. This software consists of a core LISP implementation, but there are many theoretically-motivated extensions and alternative practicality-motivated alternative implementations. In some cases the software provides the best definitions of what the theoretical claims are. It is a theory that attempts to formalize and operationalize certain aspects of our understanding of the human mind. This includes assumptions that are more core and those that are more peripheral. It changes as our knowledge grows and has different interpretations in different hands. First to state what will be obvious to some ACT-R is not monolothic. In fact there are three distinct senses of ACT-R

ACT-R: The Oldest Core Principles 1. The Procedural-Declarative Distinction a. The declarative component originated in Anderson & Bower (1973) HAM network representation of memory. b. The procedural component originated in Newell’s (1973) production system theory of cognitive control. c. Both the procedural and declarative components have evolved far from these origins. The Symbolic-Subsymbolic Distinction In addition to the symbolic level that represented knowledge there is a subsymbolic level that controls access to that knowledge. The subsymbolic level was initially designed to reflect the 1970s & 1980s ideas about neural processing. Guided by rational analysis the subsymbolic level was updated in 1993 to reflected the likelihood that the information was useful. This was the birth of ACT-R.

Evolution from ACT-R 2.0 (1993) to ACT-R 6.0 (2007) 1. There were 3 driving forces: a. The emergence of a user community around the publicly available ACT-R 2.0. b. The realization that the “No Magic” principle required that we be able to model the processing all the way from input to output. c. The insistence on not making assumptions that could not be cashed out into neurally plausible computations. This converged in the modular architecture of ACT-R 6.0: The allowed community members to try variations on existing ideas and extensions but keep what they wanted. We borrowed the modular organization of EPIC for the perceptual-motor modules. There was growing evidence that, while the brain was a complex parallel machine, different regions had their specializations.

Modules in ACT-R 6.0 Modules are high capacity, parallel, and asynchronous Manual Vocal Visual Aural Imaginal Declarative Goal Procedural Buffers provide narrow paths of communication -- only hold a chunk in ACT-R terms. Production system that contains rules that recognize patterns and react

ACT-R Module-Region Mappings

9:10: Big picture of what ACT-R is about Outline: 9:10: Big picture of what ACT-R is about 9:20: Evolution of the Procedural Module 9:30: Evolution of the Declarative Module 9:35: How ACT-R spreads What I want to do next is say a little about how the two most core modules, the procedural and the declarative, have evolved from their origins.

The Procedural Component in ACT-R has Evolved from Computer Science Notation to Description of the Brain’s Action Selection 600517 23523 4

The First Real ACT-R Production Rule If the goal is to process a column and the top digit is not smaller than the bottom digit, Then write the difference between the digits as the answer Responds to a Particular Pattern that Appears in the Buffers of a Set of Modules Selects an Action Which consists of requests to other Modules Imaginal> Top: 7 Relation: >= Bottom: 3 Goal> Task: Process-Column Declarative> Type: subtraction Minuend: 7 Subtrahend: 3 Task: Subtracting Request Difference 23

The Second Real ACT-R Production Rule If the goal is to process a column and the top digit is not smaller than the bottom digit, Then write the difference between the digits as the answer Responds to a Particular Pattern that Appears in the the Buffers of a Set of Modules Selects an Action Which consists of requests to other Modules Goal> Task: Subtracting Goal> Task: Next-column Harvest Difference Declarative> Type: subtraction Difference: 4 Manual> Action: write Digit: 4

Attributes of Production Rules Production rules are stimulus-response bonds that have “gone over to the cognitive side” because among the stimuli they respond to are past memories, mental images, and control states. Respond to conjunctions of elements in the various buffers. These buffers can represent relational structures -- e.g. A above B. Note how innocuous the use of variables is -- it basically copying information from one brain region to another. Stewart, T.C. and Eliasmith, C. (2008). Building production systems with realistic spiking neurons. 30th Annual Meeting of the Cognitive Science Society. Stocco, A., Lebiere, C., & Anderson, J. R. (in revision). Conditional routing of information to the cortex: A model of the role of basal ganglia in high-level cognition. Psychological Review

Learning of New Production Rules New Problem Situations Declarative Representations Requires Deliberation Deduction From 1st Principles Following instructions (e.g. Multicolumn Subtraction) Analogy to Prior Experiences (e.g. Past Tense Model) Interpreted Production Compilation Traces Feed Into New Production Rules Eventually Produces

Origin of One of the Subtraction Rules Production compilation compresses general-purpose processing of knowledge into special case rules -- replacing deliberation by action.

Reinforcement of Competing Productions Retrieve-Instruction (Reinforcement 10) If the goal is to process a column Then retrieve an operator for that kind of column Request-Difference-Subtract (Reinforcement 14) If the goal is to process a column and the top digit is not smaller than the bottom digit, Then subtract the bottom from the top Request-Difference-Borrow (Reinforcement 14) If the goal is to process a column and the top digit smaller than the bottom digit, Then add 10 to the top digit and set as a subgoal to borrow from the column to the left. 28 Request-Difference-Wrong (Reinforcement 14 or 0) If the goal is to process a column Then subtract the smaller from the larger

Utility Learning for Competing Productions Considerable simplification of ACT-R utility learning based of reinforcement-like learning results from the basal ganglia Every time a rule created it is rewarded with the utility of its parent Standard ACT-R soft-max rule for choosing among productions according to their noisy utilities So that is a quick effort to review how the procedural component works in ACT-R -- although as I will discuss later there are variations in this. Now let me try to give you an equally brief snapshot of the current state of the declarative component in ACT-R.

9:10: Big picture of what ACT-R is about Outline: 9:10: Big picture of what ACT-R is about 9:20: Evolution of the Procedural Module 9:30: Evolution of the Declarative Module 9:35: How ACT-R spreads What I want to do next is say a little about how the two most core modules, the procedural and the declarative, have evolved from their origins.

What has Happened to the Declarative Component in ACT-R? It has bifurcated into two completely separate things: An increasingly watered-down set of principles for the representation of knowledge, which comes to be the contents of module buffers. This is clearly a place where important new thinking is required. An increasingly empirically well-founded set of principles (with a foundation in rational analysis) for how the brain performs controlled retrieval of information from declarative memory.

Buffers and Declarative Memory Buffers associated with modules provide narrow paths of communication. The contents of the buffers are called chunks. Records of these chunks are placed in declarative memory. These can be later retrieved and placed in the declarative buffer. Manual Vocal Visual Aural Imaginal Declarative Goal Procedural 28

Chunk Activation Reflects Probability of Use Environmental Equation: Posterior odds that memory i will be needed in context C Prior odds that i is needed: recency and frequency Likelihood ratio of element j in context given i is needed Momentary Activation of memory i Weighting of Source j In developing the ACT-R theory I took the analyses I had developed with Lael and read them right into the ACT-R theory of memory. Activation Equation: Base-level Activation of memory i Association Strength from j to i

Fan Experiment: Pirolli & Anderson (1985)

Growth of Activation Activation Level

Recognition Latencies intercept latency scale Recognition Time (ms.) 18 r = .986 is a parameter-free measure of the match between theory and data.

9:10: Big picture of what ACT-R is about Outline: 9:10: Big picture of what ACT-R is about 9:20: Evolution of the Procedural Module 9:30: Evolution of the Declarative Module 9:35: How ACT-R spreads What I want to do next is say a little about how the two most core modules, the procedural and the declarative, have evolved from their origins.

Temporal Module: An Example of How One Can Extend ACT-R Pacemaker Gate Start Signal Accumulator Memory Comparison Matching Selection Execution Productions Declarative Module Visual Module Manual Module External World Retrieval Buffer Manual Buffer Visual Buffer Goal Buffer Problem Buffer Pacemaker Gate Start Signal Accumulator

Other Module Extensions for ACT-R Salvucci’s Emma Module for Eye Movements. My new Metacognitive Module. Spatial Modules (Gunzelmann, Harrison & Trafton). Fatigue Module (Gunzelmann) ???? Reasoning Module LarKC (Schooler)????

Module Modifications SNIF-ACT (Fu & Pirolli): Procedural and Declarative. Threaded Cognition (Salvucci & Taatgen): Goal Spacing Effect (Pavlik): Declarative. Blending (Lebiere): Declarative. Race/A (van Maanen & Van Rijn): Declarative Visual Saliency (Byrne): Visual. Gray, Veksler, & and others of the RPI Co: Procedural. Bothell & Leabra: Visual.

You Don’t Need to Change ACT-R to Have an Interesting Model Fum & Stocco: Sugar Factory Lebiere, Wallach, & Taatgen: Sugar Factory Altmann & Trafton: Tower of Hanoi Lewis & Vasishith: Parsing Taatgen: Acquistion Past Tense Model Anderson (2007) & Everybody (recently): Everything in fMRI And indeed most of the published ACT-R models.

Getting ACT-R out of the Narrow Confines of Laboratory Experiments Best & Lebiere: MOUT St. Amant & Ritter: Segman Bothell, Douglass, Lee: Unreal Tournament Harrison & Trafton: Robotics Destefano: Space Fortress Schoelles: Lots of Interfaces

ACT-R is Not Monolithic It is a community brought together by common theoretical assumptions and a commitment to the “No Magic” Principle -- cognitive theory has to run and it has to predict data. While ACT-R may be sustained from CMU it no longer resides at CMU. The community motto is “Let a thousand flowers grow” It is a set of software for purposes of simulation. This software consists of a core LISP implementation, but there are many theoretically-motivated extensions and alternative practicality-motivated alternative implementations. In some cases the software provides the best definitions of what the theoretical claims are. It is a theory that attempts to formalize and operationalize certain aspects of our understanding of the human mind. This includes assumptions that are more core and those that are more peripheral. It changes as our knowledge grows and has different interpretations in different hands. That is basically all I wanted to say. I hope I have given you some sense for the truth in this slide. Some people in response to this express a certain degree of discomfort in that ACT-R is not a theory you can nail down and prove wrong. I think you can understand what is going on here by appreciating the distinctions beween frameworks, theories, and models. Searching around on the web for a discussion of this I found this quote from 1983 which I will leave you with.

Be Fruitful and Multiply! (p. 12 Architecture of Cognition, 1983) 2007 That is basically basically Be Fruitful and Multiply!