________ . ________ Product Information: Underlying Theory . Cognitive science, psychology, and Artificial Intelligence (AI) have made substantial advances in the last decade in using computer-based models to understand cognitive processes. Computational architectures that embody theories and models of human thought and reasoning have led to software systems that can emulate or mimic the way people think and solve problems in specific tasks. iGEN builds on this body of research, with the goal of enabling practical engineering applications of this technology. iGEN takes the idea that human beings are the best example of intelligence available to us, and applies the cognitive science research as the basis for tools to build cognitive agents. Intelligent human-like software programs that think like humans offer many potential advantages over traditional 'dumb' software or conventional AI: their actions should be more human-like and understandable to the people that need to interact with them; the knowledge the agents need should be more readily obtainable from human experts in the same field of work; and the agent's internal reasoning and thought processes should be easier to analyze and debug. At the heart of iGEN is a computational architecture that provides the cognitive 'engine' for the iGEN applications. This engine embodies a number of features that were designed to increase iGEN's ability to meet real-world application engineering needs while maintaining its sound cognitive science foundation. These engineering needs include: Openness and extensibility. The iGEN cognitive engine uses an architecture based on cognitive research, but the architecture minimizes the number of 'built-in' psychological theories. As a result, it has retained an open architecture which allows different component-level theories (e.g., of vision, audition, grasp/reach, memory decay, and so on) to be built and inserted into specific applications as needed and desired by the end-user.   Scalability and compatibility with complex expertise. Unlike the highly constrained settings in which much basic cognitive research has been carried out, real-world cognitive agents must operate in large complex problem environments. They have to be able to incorporate the sophisticated strategies used by true human experts, such as those identified by research on naturalistic decision making. iGEN was deliberately designed to be compatible with these expert strategies, through its use of pattern-directed attention and highly chunked goal structures. This, in turn, has given iGEN cognitive agents an ability to scale up to very complex and dynamic applications that would be unapproachable by other methods.   Flexibility in representational granularity. Because iGEN was designed to support as large a range of applications as possible, it does not have any 'least common denominator' of granularity for representing knowledge and/or cognitive processes. Unlike other architectures which are tied to fixed cognitive cycle time (typically a small fraction of a second), the iGEN cognitive engine allows the cognitive agent builder to select a level of detail that is appropriate for the application at hand. This allows knowledge about the application domain to be programmed at the level most appropriate for the needs of the specific application. Separation of competence and performance. There are many limitations of human cognition that an agent-builder won't want to replicate in a cognitive agent, such as limited processing speed, a propensity for errors, a memory which forgets information, and so on. iGEN differentiates between these limiting factors which model human performance and the overall unconstrained abilities which give rise to human competence. In applications such as decision support or intelligent tutoring, the cognitive agent developer typically wants the advisory or tutorial reasoning within the agent to be as competent as possible. On the other hand, when the cognitive agent is a simulation of a human (such as an equipment operator), then producing realistic human performance is paramount. The iGEN cognitive engine captures expertise in a way that allows unconstrained execution to model human competence, but also allows specific performance-constraining factors to be incorporated to create execution as a human performance model.   Support for multi-tasking and time-critical applications. Cognitive agent applications typically need to help (or simulate) people who need to make time-critical decisions and who are working on many tasks simultaneously (i.e., are multi-tasking). iGEN was designed to deal with competing demands for attention and a constantly-changing set of circumstances. The iGEN cognitive engine uses a memory-centric, situation-based attention process that allows the cognitive agent to be highly responsive to changing situations, to be able to interrupt itself when necessary, and to prioritize among many competing demands on the basis of the current context. The iGEN cognitive engine, called BATON, is an implementation of a broader framework for modeling human information processing. That framework is described in the research literature under the name COGNET. The various publications and white-papers below give more details on COGNET, BATON, and their links to specific portions of the research literature. Download research publications on the iGEN cognitive architecture Related research and background literature Broadbent, D. (1958). Perception and Communications. New York: Pergammon Press. Card, S., Moran, T., & Newell, A. (1983). The Psychology of Human Computer Interaction. Hillsdale, NJ: Lawrence Erlbaum Associates. Chi, M., Glaser, R., & Farr, M. (1988). The nature of expertise (Hillsdale, NJ: Lawrence Erlbaum Associates). Cohen, M. S., Freeman, J. T., and Wolf, S. (1996). Meta-recognition in time stressed decision making: Recognizing, critiquing, and correcting. Human Factors, 38, 206-219. Ericsson, K., & Kinsch, W. (1995). Long-term working memory. Psychological Review, 102(2), 211-245. Ericcson, K., & Smith, J. (Ed.) (1991). Toward a general theory of expertise: Prospects and limits (Cambridge: Cambridge University Press). Flavell, J. (1976). Metacognitive aspects of problem solving. In L. Resnick (Ed), The Nature of Intelligence. Hillsdale, NJ: Lawrence Erlbaum Associates. Glenn, F. (1989). The Case for Micro-Models. In Proceedings of Human Factors Society 33rd Annual Meeting, Santa Monica, CA: Human Factors and. Ergonomics Society. Hayes-Roth, B. (1985). A blackboard architecture for control. Artificial Intelligence, 26, 251-321. Hayes-Roth, B., & Hayes-Roth, F. (1979). A Cognitive Model of Planning. Cognitive Science, 3, 275-310. Hoffman R. (1992). The Psychology of Expertise: Cognitive Research and Empirical AI (New York: Springer-Verlag). Klein, G. (1989), Recognition-primed decisions. In W.B. Rouse (Ed.), Advances in Man-Machine Systems Research, Vol. 5 (Greenwich, CT: JAI Press), 47-92. McGaugh, J. (2000). Memory—a century of consolidation. Science, 287, 248-251. Newell, A. (1990). Unified theories of cognition. Cambridge, MA: Harvard Univ. Press. Nii, P. Blackboard Systems: The Blackboard Model of Problem Solving and the Evolution of Blackboard Architectures. Part One, AI Magazine, 7, 2, 38-53, Part Two, AI Magazine, 7, 3, 82-106, 1986. Pew, W. and Mavor, A. (1998). Modeling Human and Organizational Behavior: Applications to Military Simulations. (National Research Council) (Chapter 3, pp. 51-111). Washington, D.C.: National Academy Press. Also available at http://www.nap.edu. Ritter, F. E., Shadbolt, N. R., Elliman, D., Young, R., Gobet, F. & Baxter, G. D. (1999). Techniques for modeling human performance in synthetic environments: A supplementary review. Technical Report No. 62, ESRC Centre for Research in Development, Instruction and Training, Dept. of Psychology, University of Nottingham, UK. Selfridge, O. (1959). Pandemonium: A paradigm for learning, in Proceedings of the Symposium on the Mechanization of Thought Processes, pp. 511-529. VanLehn, K. (1996). Cognitive skill acquisition. Annual Review of Psychology, 47, 513-539