Comparative Table of Cognitive Architectures (started on October 27, 2009; last update: June 18, 2012)

  Features↓ vs. Architectures→
(download this table)		 4D/RCS ACT-R ART BECCA biSoar CERA-CRANIUM Chrest Clarion CogPrime CoJACK Disciple Epic FORR GLAIR GMU-BICA HTM Leabra LIDA NARS Nexting Pogamut Polyscheme Recommendation Architecture REM Soar Ymir other architectures to consider including:
  Contributors to this table:   →
Curator: Alexei Samsonovich
Last update: 2010-09-20
James Albus
Andrea Stocco and Christian Lebiere Stephen Grossberg BECCA block diagram
Brandon Rohrer, updated 8/3/2011
B. Chandrasekaran and U. Kurup Raul Arrabales
Fernand Gobet and Peter Lane
Ron Sun Ben Goertzel
Frank Ritter & Rick Evertsz
George Tecuci
Shane Mueller, Andrea Stocco Susan L. Esptein Stuart C. Shapiro Alexei Samsonovich
Jeff Hawkins
David C. Noelle Stan Franklin Pei Wang Akshay Vashist, Shoshana Loeb
Cyril Brom
Nick Cassimatis
L. Andrew Coward
Ashok K. Goel, J. William Murdock, and Spencer Rugaber
Andrea Stocco / John Laird Kristinn R. Thórisson  
  Iconic link to diagrams                      
meta.pdf
 
 
 
birdviewsmall.gif
     
 
 
            Icarus
SAL
Tosca
  Basic Overview                                                      
  Knowledge and experiences are represented using images, maps, objects, events, state, attributes, relationships, situations, episodes, frames Chunks and productions, Visual 3D boundary and surface representations; auditory streams; spatial, object, and verbal working memories; list chunks; drive representations for reinforcement learning; orienting system; expectation filter; spectral timing networks Perceptual experiences are represented features and combinations of features. Knowledge about the world is represented as transitions between experiences. The representational framework of Soar plus diagrams – the diagrammatic part can also be combined with any symbolic general architecture, such as ACT-R Single percepts, complex percepts, and mission percepts. Single and complex behaviors. chunks and productions chunks, rules, and NNs CogPrime is a multi-representational system.  The core representation consists of a hypergraphs with uncertain logical relationships and associative relations operating together.  Procedures are stored as functional programs; episodes are stored in part as “movies” in a simulation engine; and there are other specialized methods too. Beliefs-Desires-Intentions (BDI) architecture with events, plans/intentions (procedural), beliefsets (declarative) and activation levels Disciple shell includes general modules for user-agent interaction, ontology representation, problem solving, learning, and tutoring, as well as domain-independent knowledge (e.g., knowledge for evidence-based reasoning). The problem solving engine of a Disciple cognitive assistant (see the top part of Figure 6) employs a general divide-and-conquer strategy, where complex problems are successively reduced to simpler problems, and the solutions of the simpler problems are successively combined into the solutions of the corresponding complex problems. To exhibit this type of behavior, the knowledge base of the agent contains a hierarchy of ontologies, as well as problem reduction rules and solution synthesis rules which are expressed with the concepts from the ontologies. Production Rules, Working memory entries  Descriptives are  shared knowledge resources  computed on demand and refreshed only when necessary  Advisors are domain-dependent decision rationales  for actions. Measurements are  synopses of problem solving experiences.   SNePS, simultaneously a logic-based, assertional frame-based, and propositional graph based representation schemas, mental states Sparse distributed representations. Memory and representations are distributed across a hierarchy of nodes. Within each node representations are large sparse binary vectors. patterns of neural firing rates and patterns of synaptic strengths. Sensory events drive patterns of neural activation, and such activation-based representations may drive further processing and the production of actions.  Knowledge that is retained for long periods is encoded in patterns of synaptic connections, with synaptic strengths determining the activation patterns that arise when knowledge or previous experiences are to be employed. Perceptual knowledge - nodes and links in a Slipnet-like net with sensory data of various types attached to nodes;                          Episodic knowledge - boolean vectors (Sparce Distributed Memory;            Procedural knowledge - schemes a la Schema Mechanism beliefs, tasks, and concepts Facts, Rules, frames (learned/ declared),  symbolic representation of raw sensory inputs, expectation generation and matching rules, operators Relational constraints, constraint graphs, first-order literals, taxonomies, weight matrixes. A large set of heuristically defined similarity circumstances, each of which is a group of information conditions that are similar and have tended to occur at the same time in past experience. One similarity circumstance does not correlate unambiguously with any one cognitive category, but each similarity circumstance is associated with a range of recommendation weights in favour of different behaviours (such as identifying categories in current experience). The predominant weight across all currently detected similarity circumstances is the accepted behaviour Tasks, methods, assertions, traces Procedural knowledge: Rules. Semantic knowledge: relational graph structure, Episodic memory: episodes of relational graph structures Distributed modules with traditional Frames 4CAPS
AIS
Apex
Atlantis
CogNet
Copycat
DUAL
Emotion Machine
  Main components Behavior Generation, World Modeling, Value Judgment, Sensory Processing, Knowledge Database.  These are organized in a hierarchical real-time control system (RCS) architecture Chunk-based declarative memory; buffers; procedural knowledge encoded as productions Many model brain regions, notably laminar cortical and thalamic circuits BECCA is a solution to the general reinforcement learning problem. It consists of two parts, an unsupervised feature creator and a model-based incremental learner. Both are incremental and on-line, designed for a physically embodied agent operating in an unstructured environment. DRS, for diagram representation; perceptual and action routines to get information from and create/modify diagrams. Physical layer global workspace, Mission-specific layer global workspace, core layer contextualization, attention, sensory prediction, status assessment, goal management. Attention, sensory memory, short-term memory, long-term memory procedural knowledge and declarative knowledge (each in both implicit and explicit forms --- rules and NNs) The primary knowledge store is the AtomSpace, a neural-symbolic “weighted labeled hypergraph” with multiple cognitive processes acting on it (in a manner carefully designed to manifest cross-process “cognitive synergy”), and other specialized knowledge stores indexed by it.  The cognitive processes are numerous but include: an uncertain inference engine (PLN, Probabilistic Logic Networks), a probabilistic evolutionary program learning engine (MOSES, developed initially by  Moshe Looks), an attention allocation algorithm (ECAN, Economic Attention Networks, which is somewhat neural net like), concept formation and blending heuristics, etc.   Work is under way to incorporate a variant of Itamar Arel’s DeSTIN system as a perception and action layer.  Motivation and emotion are handled via a variant of Joscha Bach’s MicroPsi framework called CogPsi. Beliefsets for long term memory, plans, intentions, events, goals
 Cognitive processor (including production rule interpreter and working memory), long term memory, production memory, detailed perceptual-motor interfaces (auditory processor, visual processor, ocular motor processor, vocal motor processor, manual motor processor, tactile processor) Advisors are organized into three tiers. Tier-1 Advisors are fast and correct, recommend individual actions, and are consulted in a pre-specified order. Tier-2 Advisors trigger in the presence of a recognized situation, recommend (possible partially ordered) sets of actions, and are consulted in a pre-specified order. Tier-3 Advisors are heuristics, recommend individual actions, and are consulted together. Tier-3 Advisors' opinions express preference strengths that are combines with weights during voting to select an action. A FORR-based system learns those weights from traces of its problem-solving behavior. 1) Knowledge Layer containing Semantic Memory, Episodic Memory, Quantified & conditional beliefs, Plans for non-primitive acts, Plans to achieve goals, Beliefs about preconditions & effects of acts, Policies (Conditions for performing acts), Self-knowledge, Meta-knowledge; 2) Perceptuo-Motor Layer containing implementations of primitive actions, perceptual structures that ground KL symbols, deictic and modality registers; 3) Sensori-Actuator Layer containing sensor and effector controllers. memory systems: working, semantic, episodic, procedural, iconic (I/O); plus: cognitive map, reward system, the engine HTM is a biologically constrained model of neocortex and thalamus. HTM models  cortex related to sensory perception,  learning to infer and predict from high dimensional sensory data. The model starts with a hierarchy of memory nodes. Each node learns to pool spatial pattern using temporal contiguity (uising variable order sequences if appropriate) as the teacher. HTMs are inherently modality independent. Biologically the model maps to cortical regions, layers of cells, columns of cell across the layers, inhibitory cells, and non-linear dendrite properties. All representations are large sparse distributions of cell activities. At the level of gross functional anatomy, most Leabra models employ a tripartite view of brain organization.  The brain is coarsely divided into prefrontal cortex, the hippocampus and associated medial-temporal areas, and the rest of cortex -- "posterior" areas.  Prefrontal cortex provides mechanisms for the flexible retention and manipulation of activation-based representations, playing an important role in working memory and cognitive control.  The hippocampus supports the rapid weight-based learning of sparse conjunctive representations, providing central mechanisms for episodic memory.  The posterior cortex mostly utilizes slow statistical learning to shape more automatized cognitive processes, including sensory-motor coordination, semantic memory, and the bulk of language processing. At a finer level of detail, other "components" regularly appear in Leabra-based models.  Activation-based processing depends on attractor dynamics utilizing bidirectional excitation between brain regions. Fast pooled lateral inhibition plays a critical role in shaping neural representations.  Learning arises from an associational "Hebbian" component, a biologically plausible error-driven learning component, and a reinforcement learning mechanism dependent on the brain's dopamine system. Cognitive cycle (action-perception cycle) acting as a cognitive atom. Higher level processes implemented as behavior streams. Cognitive cycle includes sensory memory, perceptual associative memory, workspace, transient episodic memory, declarative memory, global workspace, procedural memory, action selection, sensory motor memory an inference engine and an integrated memory and
control mechanism		 Learning, Reasoning, Imagining, Attention Focus, Time awareness, Expectation generation and matching


 procedural knowledge encoded as reactive rules;
episodic and spatial memory encoded within a set of graph-based structures		 Modules using data structres and algorithms specialized for specific concepts.  A focus of attention for exchaning information among these modules.  A focus manager for guiding the flow of attention and thus inference. Specialized modules for representing and making inferences about specific concepts. Condition definition and detection (cortex); Selection of similarity circumstances to be changed in each experience (hippocampus); Selection of sensory and other information to be used for current similarity circumstance detection (thalamus); Assignment and comparison of recommendation weights to determine current behaviour (basal ganglia); Reward management to change recommendation weights (nucleus accumbens etc.); Management of relative priority of different types of behaviour (amygdala and hypothalamus); Recording and implementation of frequently required behaviour sequences (cerebellum) Task-Method-Knowledge (TMK) models provide functional models of what agents know and how they operate.  They describe components of a reasoning process in terms of intended effects, incidental effects, and decomposition into lower-level components.  Tasks include the requirements and intended effects of some computation.  Methods implement a task and include a state-transition machine in which transitions are accomplished by subtasks. Working memory encoded as a graph structure; Knowledge represented as rules organized as operators; semantic memory; episodic memory; menal imagery; reinforcement learning Distributed heterogeneous module-based network, interacting via blackboards, realtime performance ERE
Gat
Guardian
H-Cogaff
Homer
Imprint
MAX
Omar
PRODIGY
PRS
Psi-Theory
R-CAST
RALPH-MEA
Society of Mind
Subsumption architecture
Teton
Theo
  Primary original reference(s)   Anderson, J.R., & Lebiere, C. (1998). The Atomic Components of Thought. Mahwah: Lawrence Erlbaum Associates. Grossberg, S. (1987). Competitive learning: From interactive activation to adaptive resonance. Cognitive Science 11: 23-63. B Rohrer, "An implemented architecture for feature creation and general reinforcement learning," Workshop on Self-Programming in AGI Systems, Fourth International Conference on Artificial General Intelligence, Mountain View, CA, Aug 3-6, 2011. (1MB pdf) http://www.sandia.gov/rohrer/doc/Rohrer11ImplementedArchitectureFeature.pdf Unmesh Kurup & B. Chandrasekaran, "Modeling Memories of Large-scale Space Using a Bimodal Cognitive Architecture." Proceedings of the International Conference on Cognitive Modeling July 27-29, 2007, Ann Arbor, MI. (CD-ROM, 6 pages.) Arrabales, R. Ledezma, A. and Sanchis, A. "A Cognitive Approach to Multimodal Attention". Journal of Physical Agents. Volume 3. Issue 1. Pages 53-64. January 2009.
 Feigenbaum, E. A., & Simon, H. A. (1962). A theory of the serial position effect. British Journal of Psychology, 53, 307-320.

Gobet, F. & Lane, P. (2005). The CHREST architecture of cognition: Listening to empirical data. In D. Davis (Ed.), Visions of mind: Architectures for cognition and affect. (pp. 204-224). Hershey, PA: Information Science Publishing.		 Sun, R. (2004). The CLARION cognitive architecture: Extending cognitive modeling to social simulation. In: Ron Sun (Ed.), Cognition and Multi-Agent Interaction. Cambridge University Press: New York.     The most representative early paper on Disciple is [74]. Other key references include [75-78]. Most recent representative publications include [79,80].

[74] Tecuci G., Disciple: A Theory, Methodology and System for Learning Expert Knowledge, Thèse de Docteur en Science, University of Paris-South, 1988.
[75] Tecuci G., Building Intelligent Agents: An Apprenticeship Multistrategy Learning Theory, Methodology, Tool and Case Studies, San Diego: Academic Press, 1998.
[76] Tecuci G., Boicu M., Bowman M., Marcu D., with a commentary by Burke M., An Innovative Application from the DARPA Knowledge Bases Programs: Rapid Development of a Course of Action Critiquer, AI Magazine, 22, 2, pp. 43-61, 2001.
[77] Tecuci G., Boicu M., Boicu C., Marcu D., Stanescu B., Barbulescu M., The Disciple-RKF Learning and Reasoning Agent, Computational Intelligence, 21, 4, pp. 462-479, 2005.
[78] Tecuci G., Boicu M., and Comello J., Agent-Assisted Center of Gravity Analysis, CD with Disciple-COG and Lecture Notes used in courses at the US Army War College and Air War College, GMU Press,, 2008.
[79] Tecuci G., Boicu M., Marcu D., Boicu C., Barbulescu M., Disciple-LTA: Learning, Tutoring and Analytic Assistance, Journal of Intelligence Community Research and Development, 2008.
[80] Tecuci G., Schum D.A., Boicu M., Marcu D., Hamilton B., Intelligence Analysis as Agent-Assisted Discovery of Evidence, Hypotheses and Arguments. In: Phillips-Wren, G., Jain, L.C., Nakamatsu, K., Howlett, R.J. (eds.) Advances in Intelligent Decision Technologies, SIST 4, pp. 1-10. Springer-Verlag, Berlin Heidelberg, 2010.
 Kieras, D. & Meyer, D.E. (1997). An overview of the EPIC architecture for cognition and performance with application to human-computer interaction. Human-Computer Interaction., 12, 391-438. ;    Meyer, D.E., & Kieras, D.E. (1997). A computational theory of executive cognitive processes and multiple task performance: Part I. Basic mechanisms. Psychological Review 63: 81-97. Susan L. Epstein. 1994. For the Right Reasons: The FORR Architecture for Learning in a Skill Domain. Cognitive Science, 18(3): 479-511.   Samsonovich, A. V. and De Jong, K. A. (2005). Designing a self-aware neuromorphic hybrid. In K.R. Thorisson, H. Vilhjalmsson, and S. Marsela (Eds.). AAAI-05 Workshop on Modular Construction of Human-Like Intelligence: AAAI Technical Report, volume WS-05-0  George D, Hawkins J (2009) Towards a Mathematical Theory of Cortical Micro-circuits. PLoS Comput Biol 5(10). "On Intelligence" Jeff Hawkins with Sandra Blakeslee, 2005 O’Reilly, R.C. (1996), ‘Biologically Plausible Error-Driven Learning Using Local Activation Differences: The Generalized Recirculation Algorithm’, Neural Computation, 8, 895–938. Franklin, S. and F. G. Patterson, Jr. (2006). The Lida Architecture:  Adding New Modes of Learning to an Intelligent, Autonomous, Software Agent. Integrated Design and Process Technology, IDPT-2006, San Diego, CA, Society for Design and Process Science.       Akshay Vashist, Shoshana Loeb (2010), Attention Focusing Model for Nexting Based on Learning and Reasoning. BICA 2010. (Submitted).   Cassimatis, N.L., Trafton, J.G., Bugajska, M.D., & Schultz, A.C. (2004). Integrating cognition, perception and action through mental simulation in robots. Journal of Robotics and Autonomous Systems 49 (1-2): 13-23. Coward, L. A. (1990). Pattern Thinking. New York: Praeger. Murdock, J.W. and Goel, A.K. (2001). Meta-Case-Based Reasoning: Using Functional Models to Adapt Case-Based Agents. Proceedings of the 4th International Conference on Case-Based Reasoning (ICCBR'01). Vancouver, Canada, July 30 - August 2, 2001. Laird, J.E., Rosenbloom, P.S., & Newell, A. (1986). Universal Subgoaling and Chunking: The Automatic Generation and Learning of Goal Hierarchies. Boston: Kluwer.
Laird, J.E., Newell, A., & Rosenbloom, P.S., (1987). SOAR: An architecture for general intelligence. Artificial Intelligence 33: 1-64.		 Thórisson, K. R. (1996). Communicative Humanoids: A Computational Model of Psycho-Social Dialogue Skills. Ph.D. Thesis, Media Laboratory, Massachusetts Institute of Technology.

Thórisson, K. R. (1999).  A Mind Model for Multimodal Communicative Creatures and Humanoids.  International Journal of Applied Artificial Intelligence, 13(4-5): 449-486.		  
  Other key references       www.sandia.gov/rohrer B. Chandrasekaran. Multimodal Cognitive Architecture:
Making Perception More Central to Intelligent Behavior. Proceedings of the AAAI National Conference on Artificial Intelligence, 2006, pp. 1508-1512.		   Gobet, F. & Lane, P. (2005). The CHREST architecture of cognition: Listening to empirical data. In D. Davis (Ed.), Visions of mind: Architectures for cognition and affect. (pp. 204-224). Hershey, PA: Information Science Publishing.        Kieras, D. EPIC Architecture Principles of Operation. ftp://www.eecs.umich.edu/people/kieras/EPICtutorial/EPICPrinOp.pdf   Epstein, S. L. and S. Petrovic. 2010 Learning Expertise with Bounded Rationality and Self-awareness. In Autonomous Search. Y. Hamadi, E. M., F. Saubion, editors. Springer.

     www.numenta.com   Franklin, S., & Ferkin, M. H. (2008). Using Broad Cognitive Models and Cognitive Robotics to Apply Computational Intelligence to Animal Cognition. In T. G. Smolinski, M. M. Milanova & A.-E. Hassanien (Eds.), Applications of Computational Intelligence in Biology: Current Trends and Open Problems (pp. 363-394): Springer-Verlag.         Coward LA (2001) The Recommendation Architecture: lessons from the design of large scale electronic systems for cognitive science. Journal of Cognitive Systems Research 2:111-156; Coward LA (2005) A System Architecture Approach to the Brain: from Neurons to Conscious-ness. Nova Science Publishers, New York. SEE http://cs.anu.edu.au/~Andrew.Coward/References.html Murdock, J.W. and Goel, A.K. (2003). Localizing Planning with Functional Process Models. Proceedings of the Thirteenth International Conference on Automated Planning & Scheduling (ICAPS'03). Trento, Italy

Ulam, P., Goel, A.K., Jones, J., and Murdock, J.W. (2005). Using Model-Based Reflection to Guide Reinforcement Learning. Proceedings of the IJCAI 2005 Workshop on Reasoning, Representation and Learning in Computer Games. Edinburgh, UK		   Thórisson, K. R. (1999).  A Mind Model for Multimodal Communicative Creatures and Humanoids. International Journal of Applied Artificial Intelligence, 13(4-5):449-486. 
Thórisson, K. R. (2002). Machine Perception of Multimodal Natural Dialogue. P. McKevitt (Ed.), Language, vision & music . Amsterdam: John Benjamins.  
Thórisson, K. R. (1998). Real-Time Decision Making in Face to Face Communication. Second ACM International Conference on Autonomous Agents , Minneapolis, Minnesota, May 11-13, 16-23.  
Thórisson, K. R. (1997). Layered Modular Action Control for Communicative Humanoids. Computer Animation '97, Geneva, Switzerland, June 5-6, 134-143. 
Thórisson, K. R. (2002). Natural Turn-Taking Needs No Manual: A Computational Theory and Model, from Perception to Action. In B. Granström (Ed.), Multimodality in Language and Speech Systems. Heidelberg: Springer-Verlag.		  
  Most recent representative reference(s) Albus, J. S., and Barbera, A. J.  (2005) "RCS: A cognitive architecture for intelligent multi-agent systems," Annual Reviews in Control,  Vol. 29, Issue 1 87-99
 Anderson, 2007 http://cns.bu.edu/~steve www.sandia.gov/rohrer
B Rohrer, M Bernard, JD Morrow, F Rothganger, P Xavier, "Model-free learning and control in a mobile robot," 5th Intl Conf on Natural Computation, Tianjin, China, Aug 14-16, 2009.           B Rohrer, JD Morrow, F Rothganger, PG Xavier, "Concepts from data," Brain-Inspired Cognitive Architectures Symposium, AAAI Fall Symposium Series, Washington DC, Nov 5-7, 2009. 		   Arrabales, R. Ledezma, A. and Sanchis, A. "CERA-CRANIUM: A Test Bed for Machine Consciousness Research". International Workshop on Machine Consciousness. Hong Kong. June 2009. Gobet, F., Lane, P.C.R., Croker, S., Cheng, P. C-H., Jones, G., Oliver, I., and Pine, J. M. Chunking mechanisms in human learning. Trends in Cognitive Science, 5:236-243, 2001.

Gobet, F., & Lane, P. C. R. (2010). The CHREST architecture of cognition: The role of perception in general intelligence. The Third Conference on Artificial General Intelligence. Lugano, Switzerland.		 http://www.cogsci.rpi.edu/~rsun/clarion.html Goertzel, Ben (2009).  OpenCogPrime: A Cognitive Synergy Based Architecture for Embodied General Intelligence, Proceedings of ICCI-2009.   See also http://opencog.org
Goertzel, Ben et al. (2010).  OpenCogBot: Achieving Generally Intelligent Virtual Agent Control and Humanoid Robotics via Cognitive Synergy, Proceedings of ICAI-10, Beijing		 Evertsz, R., Pedrotti, M., Busetta, P., Acar, H., & Ritter, F. E. (2009).  Populating VBS2 with realistic virtual actors. In Proceedings of the 18th Conference on Behavior Representation in Modeling and Simulation.  1-8.  09-BRIMS-04.   Epstein, S. L. and S. Petrovic. 2010 Learning a Mixture of Search Heuristics. Metareasoning in Thinking about thinking, Cox, M. T. and A. Raja, editors. MIT Press.
 Shapiro, S.C., & Bona, J.P. (in press). The GLAIR cognitive architecture. AAAI Technical Report FS-09-01. Menlo Park, CA: AAAI Press Samsonovich, A. V., Ascoli, G. A., De Jong, K. A., and Coletti, M. A. (2006). AAAI Technical Report WS-06-03, pp. 129–134. Menlo Park, CA: AAAI Press.  George D, Hawkins J (2009) Towards a Mathematical Theory of Cortical Micro-circuits. PLoS Comput Biol 5(10) O'Reilly & Munakata (2000) Franklin, Stan. (2007). A foundational architecture for artificial general intelligence. In Advances in artificial general intelligence: Concepts, architectures and algorithms, Proceedings of the AGI workshop 2006, ed. Ben Goertzel and Pei Wang:36-54. Amsterdam: IOS Press. Rigid Flexibility: The Logic of Intelligence, Springer, 2006
and see http://sites.google.com/site/narswang/home
   Brom, C., Pešková, K., Lukavský, J.: What does your actor remember?
Towards characters with a full episodic memory. Proceedings of 4th
International Conference on Virtual Storytelling, LNCS, Springer		 Cassimatis, N. L., Bignoli, P., Bugajska, M., Dugas, S., Kurup, U.,
Murugesan, A., & Bello, P. (in press). An Architecture for Adaptive
Algorithmic Hybrids. IEEE Transactions on Systems, Man, and Cybernetics.
Part B.		 Coward LA, Gedeon TO (2009) Implications of Resource Limitations for a Conscious Machine. Neurocomputing 72:767-788;  Coward, L. A. (2010). The Hippocampal System as the Cortical Resource Manager: a model connecting psychology, anatomy and physiology. Advances in Experimental Medicine and Biology 657, 315 - 364. Murdock, J.W. and Goel, A.K.  (2008). Meta-Case-Based Reasoning: Self-Improvement through Self-Understanding. Journal of Experimental & Theoretical Artificial Intelligence, 20(1):1-36 Laird, 2008 (AGI) Ng-Thow-Hing, V., K. R. Thórisson, R. K. Sarvadevabhatla, J. Wormer and Thor List (2009). Cognitive Map Architecture: Facilitation of Human-Robot Interaction in Humanoid Robots. IEEE Robotics & Automation Magazine, March, 16(1):55-66.   Jonsdottir, G. R.   
  Implementation http://www.isd.mel.nist.gov/projects/rcslib/
C++, Windows Real-time, VXworks, Neutral Messaging Language (NML),
Mobility Open Architecture Simulation and Tools (MOAST) http://sourceforge.net/projects/moast/
Urban Search and Rescue Simulation (USARSim) http://sourceforge.net/projects/usarsim/		 Lisp, TCL/Tk Nonlinear neural networks (feedback, multiple spatial and temporal scales) MATLAB, Python Implemented in the Soar framework .Net Framework, code written in C#. Runtime based on CCR (Concurrency and Coordination Runtime) and DSS (Decentralized Software Services), part of Robotics Developer Studio 2008 R2. http://www.conscious-robots.com/en/robotics-studio/2.html  Lisp, Java Java The OpenCogPrime system implements CogPrime within the open-source OpenCog AI framework, see http://opencog.org.   The implementation is mostly C++ for Linux, some components in Java; also a Scheme shell is used for interacting with the system. overlay to JACK® Disciple was initially implemented in Lisp and is currently implemented in Java Original version: Common Lisp; EPIC-X: C++ Common Lisp, Java Common Lisp Matlab, Python NuPIC development environment available for PC and Mac. It is available under a free research license and a paid commercial license. "Emergent" open source software written largely in C++ (runs on many
platforms)		 Extensible framework in Java. Open source, in Java and Prolog In Progress (C++, Java, Perl, Prolog) Java, Python; features ACT-R binding and Emergent binding Java Smalltalk Common Lisp, using Loom as the underlying knowledge-engine; Java version in development C with interfaces to almost any language; Java CommonLisp, C, C++, 8 networked computers, sensing hardware  
  Funding program, project and environment in which the architecture was applied
(added by Jim Albus)		 Please see the long list:
http://members.cox.net/bica2009/cogarch/4DRCS.pdf		     Sandia National Labs, internal R&D funding               Office of Naval Research National Science Foundation   DARPA IPTO BICA, virtual indoor/outdoor environments Numenta licesnses its software to various commercial partners.               NSF Science of Design program; Self Adaptive Agents project; Turn-based strategy games      
  Support for Common Components                                                       
  Working memory? yes Not explicitly defined Yes. Recurrent shunting on-center off-surround network that obeys LTM Invariance Principle and Inhibition of Return rehearsal law Yes. Current percepts are folded together with a decayed version of recent percepts. This creates a short history of salient percepts that functions as working memory. Diagrammatic working memory added yes Yes, although we call it short-term memory. Auditory short-term memory and visuo-spatial short-term memory are implemented. separate structure yes The active beliefs and intentions The framework of Disciple supports features and components that are common for many cognitive architectures, including working memory (reasoning trees), semantic memory (ontologies), episodic memory (reasoning examples), and procedural memory (rules). Communication is based on natural language patterns learned from the user. Yes Yes   yes: includes mental states of the Self   Many Leabra working memory models have been published, mostly focusing on the role of the prefrontal cortex in working memory. Explicitly included as a workspace with significant internal structure including a current situational model with both real and virtual windows, ans a conscious contents queue,  the active part of the memory Yes declarative Each specialist implement its own. Yes. Frequency modulation placed on neuron spike outputs, with different modulation phase for different objects in working memory. No commitment to a specific theory of working memory Relational graph structure Functional Sketchboard, Content Blackboard, Motor Feedback Blackboard, Frames  
  Semantic memory? yes Encoded as chuncks Yes, limited: Associations between chunks Yes.  Semantic information for a feature is obtained from the history of the agent's experience.     no (to be implemented) Yes. Implemented by the network of chunks in long-term memory. yes. in both implicit and explicit forms (chunks/rules and NNs) yes encocoded as (weighted) beliefs in beliefsets, uses ACT-R's declarative memory equations Not explicitly In many descriptives Yes, in SNePS  yes: includes schemas semantic meaning can be encoded in sparse distributed representations Many Leabra models of the learning and use of semantic knowledge, abstracted from the statistical regularities over many experiences, have been published.  These include some language models.
 Implemented automatically as part of declarative memory via sparse distributed memory the whole memory is semantic frames none Each specialist implement its own. Yes. Similarity circumstances (= cortical column receptive fields) that are often detected at the same time acquire ability to indirectly activate each other. Uses Powerloom as underlying knowledge engine; OWL based ontological representation Relational graph structures Frames  
  Episodic memory? yes Not explicitly defined Yes, limited: Builds on hippocampal spatial and temporal representations Yes. Individual transitions from experience to experience are an episode.   no (to be implemented) No, although “episodic” links are used in some simulations.

 yes. in both implicit and explicit forms (chunks and NNs) yes Not explicetly defined, but would be encocoded as beliefs in beliefsets. No Yes, in task history and summary measurements Yes, temporally-related beliefs in SNePS  yes: includes frozen mental state assemblies   Many Leabra models of episodic memory have been published, mostly focusing on the role of the hippocampus in episodic memory. Both declarative memory and transient episodic memory encoded via sparse distributed memory the part of memory that contains temporal information Yes (Implicit) declarative or a spreading activation network Each specialist implement its own. Yes. Similarity circumstances (= cortical column receptive fields) that change at the same time acquire ability to indirectly activate each other. An episodic memory is indirect activation of a group of columns that all changed at the same time at some point in the past. Because the hippocampal  system manages the selection of cortical columns that will change in response to each sensory experience, information used for this management is applicable to construction of episodic memories  Yes: Defined as traces through the procedural memory, below Encoded as graph structures (snapshots of working memory) Functional Sketchboard, Content Blackboard, Motor Feedback Blackboard  
  Procedural memory? yes Explicitly defined Yes. Multiple explicitly defined neural systems for learning, planning ,and control of action Yes. Common sequences of transitions are reinforced and are more likely to be executed in the future. Extends Soar's procedural memory to diagrammatic components yes. Some processors generate single or complex behaviors. Yes. Implemented by productions.  yes. in both implicit and explicit forms (chunks/rules and NNs) yes Plans and intentions with activation levels   Yes: production rules Yes, Advisors can be weighted by problem progress, and repeated sequences of actions can be learned and stored Yes, in PML implemented in Lisp, could be compiled from KL  yes: includes primitives   A fair number of Leabra models of automatized sequential action have been produced, with a smaller number specifically addressing issues of motor control.  Most of these models explore the shaping of distributed patterns of synaptic strengths in posterior brain areas in order to produce appropriate action sequences in novel situations.  Some work on motor skill automaticity has been done.  A few models, integrating prefrontal and posterior areas, have focused on the application of explicitly provided rules. Schemas a la Drescher the part of memory that is directly related to
executable operations		 Yes (Implicit) rules Each specialist implement its own. Yes. Recommendation weights associated in the basal ganglia with cortical column receptive field detections instantiate procedural memory Yes: Defined as tasks, which are functional elements, and methods, which are behavioral elements Rules Frames, Action Modules, (limited implementation)  
  cognitive map? yes   Yes. Networks that learn entorhinal  grid cell and hippocampal place field representations on line No explicit. The set of all transitions form a world model. This serves as an implicit cognitive map. Cognitive map emerges yes. Mission-specific processors build 2D maps. No.   yes           yes   Leabra contains the mechanisms necessary to self-organize topographic representations.  These have been used to model map-like encodings in the visual system.  At this time, it is not clear that these mechanisms have been successfully applied to spatial representation schemes in the hippocampus.
   as part of the memory   graph based and Bayesian There is a spatial secialist that has this functionality. No. Requirement to conserve resources by using any one similarity circumstance (= receptive field) to support multiple behaviours precludes the existence of unambigous cognitive maps No commitment to a specific theory of cognitive maps   Limited body-centric spatial layout of selected objects  
  reward system? Yes, Value Judgment processes compute cost, benefit, risk   Yes. Model how amygdala, hypothalamus, and basal ganglia interact with sensory and prefrontal cortex to learn to direct attention and actions towards valued goals. Used to help explain data about classical and instrumental conditioning, mental disorders (autism, schizophrenia), and decision making under risk. Yes. Reward is specified in the definition of the task. It can be an arbitrary function of observed and unobserved states.   yes. Status assessment mechanism in core layer.   yes. in the form of a motivational subsystem and a meta-cognitive subsystem (MCS determines rewards based on the MS) Yes, Value Judgment processes compute cost, benefit, risk Uses ACT-R memory equations, so memories and plans get strengthened.   No Yes; Advisor weights are acquired during self-supervised learning.   yes   Leabra embraces a few alternative models of the reward-based learning systems dependent on the mesolimbic dopamine systems, including a neural implementation of temporal difference (TD) learning, and, more recently, the PVLV algorithm.  Models have been published involving these mechanisms, as well as interactions between dopamine, the amygdala, and both lateral and orbital areas of prefrontal cortex. Feeling & emotion nodes in perceptual associative memory experience-based and context-sensitive evaluation       Yes. Some receptive field detections are associated with recommendations to increase or decrease recently used behavioural recommendation weights Functional descriptions of tasks allow agents to determine success or failure of those tasks by observing the state of the world.  Success or failure can then be used to reinforce decisions made during execution. appraisal-based reward as well as user-defined internal/external reward No  
  Iconic memory (including Interface and Imagery)? Yes, image and map representations Propositional (Based on chunks) Emerges from role of top-down attentive interactions in laminar models of how the visual cortex sees No explicit. An implicit iconic memory emerges from the transision history between vision-based features.    no (to be implemented) Yes.   Intended to emerge implicitly via combination of sensory, declarative and simulative memory.  Not implemented/tested yet. not particularly, would be application and IO system specific.   Part of visual perceptual processor Yes, in some descriptives   yes (imagery was not implemented)   In Leabra, iconic memory can result from activation-based attractor dynamics or from small, sometimes transient, changes in synaptic strength, including mechanisms of synaptic depression. Imagery naturally arises from patterns of bidirectional excitation, allowing for top-down influences on sensory areas.  Little work has been done, however, in evaluating Leabra models of these phenomena against biological data. Defined using various processed pixel matrices   Not directly implemented. Extension possible. not sure There is a spatial secialist that has some of this functionality. Yes. Maintained on the basis of indirect activation of cortical columns recently active at the same time Spatial relationships are encoded using the underlying knowledge engine. Explicitly defined No  
  Perceptual memory (if understood separately from iconic and working memory)? Yes, pixels are segmented and grouped into entities and events   Yes. Model development of laminar visual cortex and explain how both fast perceptual learning with attention and awareness, and slow perceptual learning without attention or awareness, can occur. Yes. All different sensory modalities and combinations of modalities are treated the same within BECCA. Yes. Diagrammatic elements of perceptual memnory Yes, but only as preprocessing buffers in CERA sensor services. Yes.   Yes,but currently only for vision.  Further modalities will be implemented later. gets input from the world as events.  these events are processed by plans.   Yes     yes: input-output buffer (in psychology, very short-term perceptual memory and iconic memory are synonyms)   Different aspects of perceptual memory can be supported by activation-based learning, small changes in synaptic strengths, frontally-mediated working memory processes, and rapid sparse coding in the hippocampus. Semantic net with activation passing. Nodes may have sensory data of various sorts attached.       There is a spatial secialist that has some  of this functionality. Yes. Based on indirect activation of cortical columns on the basis of recent simultaneous activity     Perceptual representations at multiple levels (complexity) and timescales (see “visual input” and “auditory input” below).   
  Attention and consciousness Yes,  can focus attention on regions of interest.  Is aware of self in relation to the environment and other agents.   Yes. Clarifies how boundary, surface, and prototype attention differ and work together to coordinate object and scene learning. Adaptive Resonance Theory predicts a link between processes of Consciousness, Learning, Expectation, Attention, Resonance, and Synchrony (CLEARS) and that All Conscious States Are Resonant States. Yes. The most salient feature at each time step is attended. Uses Soar's attention framework Yes, attention is implemented as a bias signal induced from the core layer to the lower levels global workspaces. Other aspects of consciousness are also considered. No explicit representation of the self implemented so far. Attention plays an important role in the architecture, as it for example determines the next eye fixation and what will be learnt.   Yes,  can focus attention on regions and topics of interest.  Is aware of self in relation to the environment and other agents. Represented by transient events, goal structure, and intentions/beliefs whose activation level is above the threshold   Emergent Phenomenon of working memory Yes, some Advisors attend to specific problem features   Distribution of attention is described by a special attribute in instances of schemas.

Consciousness can be identified with the content of the mental state "I-Now".		 We have modeled covert attentional mechanisms within HTMs although this is not in currently released product. In Leabra, attention largely follows a "biased competition" approach, with top-down activity modulating a process that involves lateral inhibition.  Lateral inhibition is a core mechanism in Leabra, as is the bidirectional excitation needed for top-down modulation.  Models of spatial attention have been published, including models that use both covert shifts in attention and eye movements in order to improve object recognition and localization.  Published models of the role of prefrontal cortex in cognitive control generally involve an attention-like mechanism that allows frontally maintained rules to modulate posterior processing.  Virtually no work has been done on "consciousness" in the Leabra framework, though there is some work currently being done on porting the Mathis and Mozer account of visual awareness into Leabra. Implemented a la Global workspace theory with global broadcasts recruiting possible actions in reponse to the current contents and, also, modulating the various forms of learning.   Atttntion via expectation generation and matching.   Attention. Yes. Cortical columns have receptive fields defined by groups of similar conditions that often occurred at the same time in past sensory experiences, and are activated if the receptive field occurs in current sensory inputs. Attention is selection of a subset of currently detected columns to be allowed to communicate their detections to other cortical areas. The selection is on the basis of recommendation strengths of active cortical columns, interpreted through the thalamus, and is implemented by placing a frequency modulation on the action potential sequences generated by the selected columns. Consciousness is a range of phenomena involving pseudosensory experiences. A column can also be indirectly activated if it was recently active, or often active in the past, or if it expanded its receptive field at the same time as a number of currently active columns. Indirect activations lead to conscious experiences.  Implicit in the fact that methods constrain how and when subtasks may be executed (via a state-machine); each method can be in only one state at a time, corresponding to the attended-to portion of a reasoning task (and directly linked to the attended-to knowledge via the tasks requirements and effects).   Within-utterance attention span. Situated spatial model of embodied self (but no semantic representation of self that could be reasoned over).   
  Visual input? Yes, color, stereo Propositional (based on chunks) Natural static and dynamic scenes, psychophysical displays. Used to develop emerging architecture of visual system from retina to prefrontal cortex, including how 3D boundaries and surface representations form, and how view-dependent and view-invariant object categories are learned under coordinated guidance of spatial and object attention. Yes Yes. Diagrams are the visual input. Yes, both real cam and synthetic images from the simulator. Visual input is currently coded as list structures or arrays.   Currently handled via interfacing with external vision processing tools.  Tighter interlinkage with a hierarchical NN vision system is a topic of current research. currently depends on the particular model   Interaction of visual motor and visual perceptual processors Possible but not implemented perceptual structures in PML symbolic HTMs are inherently modality independent although we have applied them to vision tasks. We offer a Vision Framework for programmers and a Vision Toolkit requiring no programming skills. An advanced Leabra model of visual object recognition has been produced which receives photographic images as input. possible but not implemented   Yes. Both raw video and symbolic representations of raw visual inputs. both symbolic and subsymbolic (tailored for the purposes
ov virtual worlds)		   Currently implemented by emulation of action potential outputs of populations of simulated sensory neurons Not implemented, but would be handled by the underlying knowledge engine Propositional or relational Yes. Temporally and spatially accurate vector model of upper human body, including hands, fingers, one eye. Via body-tracking suit, gloves and eyetracker.  
  Auditory input? No, not yet Propositional (based on chunks) Natural sound streams. Used to develop emerging architecture of visual system for auditory streaming and speaker-invariant speech recognition Yes No no (to be implemented) Auditory input is currently coded as text input (segmented either as words, phonemes, or syllables).   Currently the only auditory input we handle is speech, via an external speech-to-text engine.  The architecture supports it in principle.     Auditory and speech objects, spatialized auditory information Yes.  FORRSooth is an extended version that conducts  human-computer dialogues in real time Has been done using off-the-shelf speech recognition textual commands We have worked with customers applying HTMs to auditory tasks. While a few exploratory Leabra models have taken low-level acoustic features as input, this modality has not yet been extensively explored. possible but not implemented   No. none   Currently implemented by emulation of action potential outputs of populations of simulated sensory neurons Not implemented, but would be handled by the underlying knowledge engine Support for text-based communication Yes. Speech recognition (BBN Hark), custom real-time prosody tracker with H* and L* detection.  
  Special modalities? Yes, LADAR, GPS, odometry, inertial   Yes. SAR, LADAR, multispectral IR, night vision, etc. BECCA is modality agnostic. It can handle inputs originating from any sensory modality.   SONAR, Laser Range Finder.     Reads text from the Internet ;-)       Accepts data from external databases agents have used speech and navigation   HTMs have been applied to vision, audition, network sensors, power systems, and other tasks. All senses are special, are they not?   no built-in modalities, but allow plug-in sensors
and actuators		 Possible to extend.         No  Multimodal integration and realtime multimodal communicative act interpretation  
                                                         
  Support for Common Learning Algorithms                                                      
  Reinforcemenent learning Yes.  Learns parameters for actuator backlash, braking, steering, and acceleration. Yes, for productions (linear discount version) Yes: CogEM and TELOS models of how amygdala and basal ganglia interact with orbitofrontal cortex etc BECCA is an RL algorithm and represents a novel approach to RL. Extends Soar's chunking for diagrammatic components no (could be implemented in specific processors) No. yes Yes.   Strengthening/weakening of plans An expert interacts directly with a Disciple cognitive assistant, to teach it to solve problems in a way that is similar to how the expert would teach a less experienced collaborator. This process is based on mixed-initiative problem solving (where the expert solves the more creative parts of a problem and the agent solves the more routine ones), integrated learning and teaching (where the expert helps the agent to learn by providing examples, hints and explanations, and the agent helps the expert to teach it by asking relevant questions), and multistrategy learning (where the agent integrates complementary learning strategies, such as learning from examples, learning from explanations, and learning by analogy, to learn general concepts and rules). No Yes   yes (a version of it)   Leabra embraces a few alternative models of the reward-based learning systems dependent on the mesolimbic dopamine systems, including a neural implementation of temporal difference (TD) learning, and, more recently, the PVLV algorithm.  Models have been published involving these mechanisms, as well as interactions between dopamine, the amygdala, and both lateral and orbital areas of prefrontal cortex. Yes, for perception, episodic and procedural memories via base-level activation   Not decided.     Yes. Increases in recommendation weights associated in the basal ganglia with cortical column receptive field detections, on the basis of rewards. Yes.  Learns criteria for selecting alternative methods for accomplising tasks, and also alternative transitions within a methods state-transition machine. Yes, for operators (SARSA/Q-learning)  Yes, in a recent implementation (RadioShowHost)  
  Bayesian Update Not implemented, but could be. Yes, for memory retrieval No, sort of: Includes some Bayes effects as emergent properties No   no (could be implemented in specific processors) No.   Yes Retrieval of beliefs and access to current intentions No Yes   no HTM hierarchies can be understood in a belief propagation/Bayesian framework. While Leabra does not include a mechanism for updating knowledge in a Bayes-optimal fashion based on singular experiences, it's error-driven learning mechanism does approximate maximum a posteriori outputs given sufficient iterated learning. No   Yes.     No Not implemented, but could be. No No  
  Hebbian Learning Not implemented, but could be No Yes, sort of: Hebbian learning law is insufficient. Both Hebbian and anti-Hebbian properties are needed. Not formally, although the principle of associating co-occurring signals is used extensively.   no (could be implemented in specific processors) No. yes Yes   No Yes, with respect to groupings of Tier-3 Advisors.   yes Yes An associational learning rule, similar to traditional Hebbian learning, is one of the core learning mechanisms in Leabra. No   No. the episodic memory features hebbian learning   Yes, but with an overlay management that determines whether Hebbian learning will occur at any point in time Not implemented, but could be No No  
  Gradient Descent Methods (e.g., Backpropagation) Not implemented, but could be No Yes, but not Backpropagation No. BECCA does not use artificial neural networks.   no (could be implemented in specific processors) No. yes We don't use this sort of algorithm explicitly, no   No No   no   A biologically plausible error-correction learning mechanism, similar in performance to the generalized delta rule but dependent upon bidirectional excitation to communicate error information, is one of the core learning mechanisms in Leabra. No         No Not implemented, but could be No No  
  Learning of new representations Yes.  Learns maps and trajectories in new environments Production compilation (forms new productions) Yes. Multiple kinds of self-organization Yes   no (to be implemented) Yes. Chunks and templates (schemata) are automatically and autonomously created as a function of the interaction of the input and the previous state of knowledge. new chunks, new rules, new NN representations Yes, although this has been tested only in simple cases.   No Yes, can learn new Advisors    yes Yes All "active" representations in Leabra are, at their core, patterns of neural firing rates.  These vectors of activity may be interpreted -- encoded and decoded -- in different ways, however.  By analogy, all digital computer representations are strings of bits, but they may be interpreted as structures and pointers and the like.  The way in which vectors of activity at "hidden" layers of neural units are interpreted is almost always a matter of learning in Leabra models.  In this way, internal representations are always learned from experience. New representations (instrutionalist lerning)  for perception, episodic and procedural memories   Capable of representing newly learned knowledge.     Yes. A new representation is a new subset of receptive field detections, with some slight changes to some receptive fields Yes.  Learns refinements of methods for existing tasks and can respond to a specification of some new task by adapting the methods for some similar task. Chunking (forms new rules); also mechanisms to create new episodes, new semantic memories No  
                                                         
  Common General Paradigms Modeled                                                      
  Main general paradigms     Visual and auditory information processing Goal-directed behavior   Global Workspace Theory. Multimodal sensory binding.  Learning (e.g. implicit learning, verbal learning); acquisition of first language (syntax, vocabulary); expertise; memory; some problem solving; concept formation   Control of virtual-world agents.  Natural language processing.  We are now starting to work with humanoid robots but that's early-stage. BDI Disciple agents have been developed for a wide variety of domains, including manufacturing [74], education [75], course of action critiquing [76], center of gravity determination [77,78], and intelligence analysis [79]. The most recent Disciple agents incorporate a significant amount of generic knowledge from the Science of Evidence, allowing them to teach and help their users in discovering and evaluating evidence and hypotheses, through the development of Wigmorean probabilistic inference networks that link evidence to hypotheses in argumentation structures that establish the relevance, believability and inferential force of evidence [80]. Disciple agents are used in courses at various institutions, including US Army War College, Joint Forces Staff College, Air War College, and George Mason University [numbers refer to the above citations]. Human performance, multi-tasking, PRP Procedure, air traffic control Constraint solving; game playing; robot navigation; spoken dialogue reasoning, belief change voluntary perception, cognition and action   Learning, learning, learning. Take biological constraints seriously. Build more controlled processes on a foundation of more automatic processes. Global Workspace Theory reasoning with insufficient knowledge and resources Expectation generation and matching via learing and reasoning on stored knowledge and sensory inputs. 3D virtual worlds Reasoning All cognitive processes are implemented through sequences of receptive field activations, including both direct detections and indirect activations. At each point in the sequence the behaviour with the predominant recommendation weight across the currently activated receptive field population is performed. This behaviour may be to focus attention on a particular subset of current sensory inputs or to implement a particular type of indirect activation (prolong current activity, or indirectly activate on the basis of recent simultaneous activity, past frequent simultaneous activity, or past simultaneous receptive field change). Recommendation weights are acquired through rewards that result in effective sequences for cognitive processing. Frequently used sequences are recorded in the cerebellum for rapid and accurate implementation. Reflection.  Adaptation in response to new functional requirements.   Integrated behavior-based and classical AI; blackboards; distributed implementation  
  other general paradigms       Perception, abstraction               Verbal working memory, visual working memory   reasoning without a goal self-regulated learning (at the stage of design)   Tripartite brain organization.         Modle finding   Planning, Reinforcement learning   Modular architecture ("schema-style") allows easy expansion of common and custom features and principles. Solves to some extent the scaling problem.   
  Problem Solving Uses rules and/or search methods for solving problems Yes Yes, depending on what is meant Yes Yes, problem solving is the main application. yes, implicit. Yes. At moment, CHREST solves problems mostly by pattern recognition. yes Yes Yes   No Yes   yes   In the traditional AI meaning of "problem solving", involving the generation of a sequential plan to meet a novel goal, little work has been done in Leabra.  Some Leabra models of sequential action can generalize when performing in a novel situation, but none of these models have addressed the traditional AI planning problem. Yes       Yes A simple example is fitting together two objects. First step is activating receptive fields often active in the past shortly after fields directly activated by one object were active. Because objects have often been seen in the past in several different orientations, this indirect activation is effectively a "mental rotation". Receptive fields combining information from the indirect activation derived from one object and the direct activation from the other object recommend movements to fit the objects together. A bias is placed upon acceptance of such behaviours by taking on the task.  Yes Yes No  
  Decision Making Makes decision based on Value Judgment calculations Yes  Yes Yes   yes Yes. yes Yes Yes   No Yes   yes   Much work has been done on Leabra modeling of human decision making in cases of varying reward and probabilistic effects of actions, focusing on the roles of the dopamine system, the norepinepherine system, the amygdala, and orbito-frontal cortex. Yes   Uses Inference (both statistical and logical).     There can be extensive indirect activation steps, with (slight) changes to receptive fields at each step. Eventually, one behaviour has a predominant recommendation strength in the basal ganglia, and this behaviour is the decision. Yes, specifically in selecting methods to perform a task and selecting transitions within a method Yes Yes, using hierarchical decision modules as well as traditional planning methods  
  Analogy Not implemented
   Yes in rule discovery applications Yes   not implemented No.   Yes
     No Yes, via pattern matching yes     Some preliminary work has been done on using dense distributed representations in Leabra to perform analogical mapping. No   Somewhat       Yes, using the functional specification (requirements and effects) of tasks. Limited No  
  Language Processing Not implemented
 Yes  Yes     not implemented Only acquisition of language.   Comprehension and generation fully implemented.  Dialogue is something we're actively working on.
     Limited Yes Yes     Many Leabra language models have been produced, focusing on both word level and sentence level effects. Beginning stage   Yes.   Yes Yes Not implemented Yes Yes.  
  Working Memory Tasks ? Yes Yes       No. yes Unclear exactly what this means.  The system does many tasks involving working memory. Yes   Visual and Verbal WM Tasks Unclear exactly what this means       Leabra models of the prefrontal cortex have explored a variety of working memory phenomena. Yes         Yes   Yes Yes.  
  perceptual illusions             No.         No     yes: modeled perceived flipping of the Necker cube     No         Yes     No.  
  implicit memory tasks             Yes. yes       No           In principle         Yes. Depend on indirect receptive field activations on the basis of recent simultaneous activity        
  metacognitive tasks             No. yes       No Yes   yes: modeled perceived flipping of the Necker cube     NO           Yes   No.  
  social psychology tasks             No. yes       No           No                  
  personality psychology tasks             No. yes       No Personality and emotion can be modeled through Advisors   yes: modeled perceived flipping of the Necker cube     No               No.  
  motivational dynamics             No. yes       No           No                  
  Common Specific Paradigms Modeled                                                      
  Stroop ? Yes (multiple models)  Yes     no No.   We aren't doing modeling of human cognition and so we haven't tried the system on standard “cognitive modeling test problems.  So no.     no         Yes. Not yet           No ? No.  
  Task Switching Yes Yes (multiple models) Yes Yes   not yet No. yes See above Yes   yes         Yes. Not yet   Yes (necessary feature for nexting)     Yes Not implemented ? Yes.  
  Tower of Hanoi/London Yes Yes No       No. yes See above Yes   no         I think there might have been some preliminary work on Tower of London, but I'm not sure. Not yet           Yes Yes No.  
  PRP ? Yes         No.   See above     yes         Not that I know of. Not yet           No   ?  
  Dual Task Yes Yes         No. yes See above     yes         Not that I know of. Not yet   No.     Yes Not implemented Yes No.  
  N-Back ? Yes Yes       No.   See above     yes         Yes, but there is still work to be done, here. Not yet           No   ?  
                                      Not yet                  
  visual perception with comprehension Yes   Yes     not yet Yes.   We have done this in the virtual world.  Currently research aims at doing it for humanoid robots also     yes     yes Yes A powerful object recognition model has been constructed. Not yet   Yes.     Yes Not implemented   Yes (see “special modalities” above)  
  spatial exploration, learning and navigation Yes   Yes Yes Yes yes To some extent. yes See above     no Yes   yes   Preliminary work only. Not yet         Yes Yes implemented but not compared to human behavior No.  
  object/feature search in an environment Yes.  Search for targets in regions of interest.
   Yes Yes   not imeplemented Yes.   See above     no     yes   Object localization naturally arises in the object recognition model. Not yet   yes.     Yes. 
    Yes.  
  learning from instructions Yes.  Learn from subject matter experts.   No, unless you mean supervised learning     Not imeplemented No. yes Yes, in the virtual world context     no Yes Yes quasi-implemented   Some preliminary work on instruction following, particularly in the domain of classification instructions, has been done in Leabra. Not yet   yes.     Yes.   Yes, from new task specifications implemented but not compared to human behavior No.  
  pretend-play No   Yes: Outline of architecture for teacher-child imitation     no  No.   See above     no     quali-implemented   Not that I know of. Not yet       some       No.  
                                                         
  Meta-Theoretical Questions (added by Stephen Grossberg)                                                      
  Uses only local computations? No.  Uses information from battlefield information network, apriori maps, etc.   Yes. All operations defined by local operations in neural networks. No     Yes.   No.  Mix of local and global       Yes   yes Yes, HTM is a biological model.   Yes, throughout the architecture           Yes, in the existing implementation   ?  
  Unsupervised learning? Yes.  Updates control system parameters in real-time   Yes. Can categorize objects and events or alter spatial maps and sensory-motor gains without supervision. Yes   yes Yes.   Yes.       no Yes   yes Yes, uses time as primary learning method.   Perceptual, episodic and procedural, each in both instructionalist and selectionist modes   yes       Yes.  Adapts in response to failures through situated action + reinforcement learning, or through generative planning + abstraction.   No.  
  Supervised learning? Yes.  Learns from subject matter experts.   Yes. Can learn from predictive mismatches with environmental constraints, or explicit teaching signals, when they are available.  No   not implemented Yes.   Yes.      no Yes   yes Yes, networks can be supervised at top node.   No   yes       Yes, from new task specifications   No.  
  Arbitrary mixtures of unsupervised and supervised learning? Yes   Yes. E.g., the ARTMAP family of models. No   not implemented Yes.   Yes     no Yes   yes     No   yes       Yes.  Can develop a new method from a task specification and then later adapt that method based on experience.   No.   
  Can it learn in real time? Yes   Yes. Both ART (match learning) and Vector Associative Map (VAM; mismatch learning) models use real-time local learning laws. Yes   yes Yes.   Yes     no Yes   yes HTMs can learn on-line, meaning learning while inferring. Real time depends on size of network and nature of problem.   Yes   Sometimes.       Yes   No.   
  Can it do fast stable learning; i.e., adaptive weights converge on each trial without forcing catastrophic forgetting? Yes.  Uses CMAC algorithm that learns fast from error correction.  When no errors, learning stops.   Yes, theorems prove that ART can categorize events in a single learning trial without experiencing catastrophic forgetting in dense non-stationary environments. Mismatch learning cannot do this, but this is adaptive in learning spatial and motor data about changing bodies. Yes     Yes.   Yes      no Yes   in short, yes. Learning in GMU-BICA consists in storage of mental states and in creation of new schemas, without forgetting. Learning "weights" applies to the neuromorphic cognitive map. In most cases, learning occurs in one shot. Yes   Yes               No.   
  Can it function autonomously? Yes.  Operates machines and drives vehicles autonomously.   Yes. ART models can continue to learn stably about non-stationary environments while performing in them. Yes   yes Yes.   Yes.   Yes. Key motivation for selecting BDI   yes Yes   yes. Yes   Yes           Yes   Yes; however, no actual implementations have yet pushed the architecture on this issue.   
  Is it general-purpose in its modality; i.e., is it brittle? It is general purpose and robust in real world environments.   ART can classify complex non-stationary data streams. The FACADE 3D vision model clarifies how multiple types of visual data (e.g., edge, texture, shading, stereo) are processed by laminar cortical circuits. BECCA is designed to be general purpose and robust to large and small changes in tasks.   yes It is general-purpose and not brittle.   It is general purpose and robust in real world environments.     no General purpose   general-purpose. Yes   General purpose           General purpose   The architecture to some extent addresses the brittleness problem, but not to full autonomy.   
  Can it learn from artbitrarily large databases; i.e., not toy problems? Yes.  All applications are real-world and real-time.   Yes. Theorems about ART algorithms show that they can do fast learning and self-stabilizing memory in arbitrarily large non-stationary data bases. ART is therefore used in many large-scale applications http://techlab.bu.edu/ Planned   not tested Yes. Simulations on the acquisition of language have used corpora larger than 350k utterances. Simulations with chess have used databases with more than 10k positions
Discrimination networks with up to 300k chunks have been created.		   Yes.       no Yes   in principle, yes. Yes   ???           Yes, up to the limitations of the underlying knowledge engine   No.   
  Can it learn about non-stationary databases; i.e., environmental rules change upredictably? Yes.  Battlefield environments change unpredictably.   Yes. See above. Yes     Yes.   Yes.       no Untested but we believe so   in principle, yes. It can be implemented for on-line learning.   ???           Yes   No.   
  Can it pay attention to valued goals?      Yes. ART derives its memory stability from matching bottom-up data with learned top-down expectations that pay attention to expected data. ART-CogEM models use cognitive-emotional resonances to focus attention on valued goals. Yes   yes Yes, through attention mechanisms.   Yes     yes Yes   yes.     Yes           Yes.  Goals are encoded as the intended effects of tasks.   Yes.   
  Can it flexibly switch attention between unexpected challenges and valued goals? Yes.  Makes decisions about what is most important based on rules of engagement and situational awareness.   Yes. Top-down attentive mismatches drive attention reset, shifts, and memory search. Cognitive-emotional and attentional shroud mechanisms modulate attention shifts. Yes   yes Yes.   Yes.   Yes.   yes Yes   yes: by switching roles of mental states.     Yes           Not implemented   Yes.   
  Can reinforcement learning and motivation modulate perceptual and cognitive decision-making?     Yes. Cognitive-emotional and perceptual-cognitive resonances interact together for this purpose. Yes   yes No.   Yes Yes, cognitive   no Yes   yes.     Yes           Yes   Yes.   
  Can it adaptively fuse information from multiple types of sensors and modalities?     ART categorization discovers multi-modal feature and hierarchical rule combinations that lead to predictive success. Yes   yes Yes.   Yes, in principle, but this has not been tested except on vision and language and speech input...     no Yes   yes: by using / creating appropriate schemas. Yes   In principle, but not implemented           In principle, but not implemented   Yes, although it depends on the particular use of the architectural features.   
  Etc.                                                      
  Statement by the contributor                                                      
  Most statements included here were written as summaries of the BICA-2009 CogArch panel presentations. Their short versions will be included in the AI Magazine symposium report.     Biologically-relevant cognitive architectures should clarify how individuals adapt autonomously in real time to a changing world filled with unexpected events; should explain and predict how several different types of learning (recognition, reinforcement, adaptive timing, spatial, motor) interact to this end; should use a small number of equations in a larger number of modules, or microassemblies, to form modal architectures (vision, audition, cognition, ...) that control the different modalities of intelligence; should reflect the global organization of the brain into parallel processing streams that compute computationally complementary properties within and between these modal architectures; and should exploit the fact that all parts of the neocortex, which supports the highest levels of intelligence in all modalities, are variations of a shared laminar circuit design and thus can communicate with one another in a computationally self-consistent way.
 BECCA was designed to solve the problem of natural-world interaction.   The research goal is to place BECCA into a system with unknown inputs and outputs and have it learn to successfully achieve its goals in an arbitrary environment.  The current state-of-the-art in solving this problem is the human brain.  As a result, BECCA's design and development is based heavily on insights drawn from neuroscience and experimental psychology.  The development strategy emphasizes physical embodiment in robots. Chandrasekaran called attention to the lack of support in the current family of cognitive architectures for perceptual imagination, and cited his group's DRS system that has been used to help Soar and Act-R  engage in diagrammatic imagination for problem solving.     at the panel, I pointed out that the list of tasks needs to be greatly expanded, to include, for example, implicit learning tasks, meta-cognitive tasks, social psychology tasks, personality psychology tasks, motivational dynamics,  and so on, all of which have been simulated using the CLARION cognitive architecture. The architecture described in this column was pioneered in the proprietary Novamente Cognition Engine, and is now being pursued in the open-source OpenCogPrime system, build with in the OpenCog framework...  In my presentation in the BICA-2009 CogArch panel, I discussed the AGI Roadmap Initiative (see http://agi-roadmap.org) and also the need for a glossary of AGI terms to help us compare cognitive architectures.       FORR (FOr the Right Reasons) is highly modular. It includes a declarative memory for facts and a procedural memory represented as a hierarchy of decision-making rationales that propose and rate alternative actions. FORR matches perceptions and facts to heuristics, and processes action preferences through its hierarchical structure, along with its heuristics’ real-valued weights. Execution affects the environment or changes declarative memory. Learning in FORR creates new facts and new heuristics, adjusts the weights, and restructures the hierarchy based on facts and on metaheuristics for accuracy, utility, risk, and speed.      Hierarchical Temporal Memory (HTM) is a model of neocortex and thalamus. It is highly constrained and guided by anatomy and physiology at the levels of cortical regions, cellular layers, cellular connectivity, local inhibitory neurons, and non-linear integration of synapses along dendrites. HTMs build hierarchical models of the spatial and temporal statistics in sensory data. The models can be used for inference and prediction. HTMs have been commercially applied to numerous commercial problems. Numenta aims to be a catalys t for commercial applications of neocortical models of compution and inference. It publishes its algorithms and code for research and commercial deployment.     Though NARS can be considered as a "cognitive architecture" in a broad sense, it is very different from the other systems. Theoretically, NARS is a normative theory and model of intelligence and cognition as "adaptation with insufficient knowledge and resources", rather than a direct simulation of human cognitive behaviors, capabilities, or functions; technically, NARS uses a unified reasoning mechanism on a unified memory for learning, problem-solving, etc., rather than integrates different techniques in an architecture. Therefore, accurately speaking it is not after the same goal as many other cognitive architectures, though still related to them in various aspects.     Nick Cassimatis argued that repositories should focus on standardizing task environments and problems within them rather than details about cognitive architectures themselves. Theoretical arguments indicate that any system which must learn to perfom a large number of different behaviors will be constrained into this recommendation architecture form by a combination of practical requirements including the need to limit information handling resources, the need to learn without interference with past learning, the need to recover from component failures and damage, and the need to construct the system efficiently.      The actual systems built in the Ymir architecture, notably Gandalf, the Cognitive Map for Asimo, and the SuperRadioHost systems, have shown the Ymir framework to be quite flexible and extensible. Ymir was not initially built to solve general-purpose intelligence, and it is become clear that none of the Constructionist methodologies (i.e. most efforts to date relying on “first order” manually-constructed software) will be able to handle the multitude of topics related to general intelligence, such as global attention, global learning, flexible task learning and integration, etc. For this we are working on new methodologies relying on Constructivist approaches, which emply self-organization and automatic architectural growth (see my keynote paper From Constructionist to Constructivist A.I. (2009), in AAAI Fall Symposium Series - Biologically Inspired Cognitive Architectures, Washington D.C., Nov. 5-7,175-183. AAAI Tech Report FS-09-01, AAAI press, Menlo Park, CA.).