Companions: A cognitive architecture based on analogical processing

Ken Forbus

Northwestern University

Understanding the nature of common sense reasoning is one of the central problems of cognitive science. Dedre Gentner and I hypothesize that the robustness of human reasoning arises from analogical processing. One way we are testing this hypothesis is by creating a new cognitive architecture, Companion Cognitive Systems, which uses our cognitive simulations of analogical matching, retrieval, and generalization centrally. Our goal is to create flexible, broadly capable reasoning and learning systems that can operate as partners with people over extended periods of time. This talk will outline the key analogical processing ideas and some results of learning experiments involving sketch understanding with a baseline model. The features of the next generation of the architecture, which is being brought on-line now, will be discussed, along with work underway to learn how to play strategy games and to solve physics problems.

Bio: Kenneth D. Forbus is the Walter P. Murphy Professor of Computer
Science and Education at Northwestern University.  His research
interests include qualitative reasoning, analogy and similarity, sketch
understanding, spatial reasoning, cognitive simulation, reasoning system
design, articulate educational software, and the use of AI in computer
gaming.  He received his degrees from MIT (Ph.D. in 1984).  He is a
Fellow of the American Association for Artificial Intelligence and a
Fellow of the Cognitive Science Society.  He is an Associate Editor of
Cognitive Science and serves on the editorial boards of Artificial
Intelligence and AAAI Press, and the Advisory Board of the Journal of
Game Development.  His email address is forbus AT northwestern DOT edu.

Deepak Ramachandran

Last modified: Mon Mar 27 11:36:13 CST 2006