The copycat project: A model of mental fluidity and analogy-making
Douglas R. Hofstadter, Melanie S. Mitchell · 1995
Copycat is a computer program designed to be able to discover insightful analogies, and to do so in a psychologically realistic way. Copycat's architecture is neither symbolic nor connectionist, nor a hybrid of the two; rather, the program has a novel type of architecture situated somewhere in between these extremes. It is an emergent architecture, in the sense that its top-level behavior emerges as a statistical consequence of myriad small computational actions, and the concepts that it uses in creating analogies can be considered to be a realization of "statistically emergent active symbols " (Hofstadter, 1985). The use of parallel, stochastic processing mechanisms and the implementation of concepts as distributed and probabilistic entities in a network make Copycat somewhat similar in spirit to connectionist systems. However, as will be seen, there are important differences, and we claim that the middle ground in cognitive modeling occupied by Copycat is at present the most useful level at which to attempt to understand the fluidity of concepts and perception that is so clearly apparent in human analogy-making. * We thank past and present members of the Fluid Analogies Research Group, and in particular,