A previously proposed model for memory based on neurophysiological considerations is reviewed. It is extended by introducing positive feedback of a set of neurons onto itself and allowing the neurons to saturate. The resulting model is analyzed by distinctive features, categorical perception, and as it applies to probability learning. (Editor/RK)

Describes alternative theoretical positions regarding (a) the component information processes used in analogical reasoning and (b) strategies for combining these processes. Also presents results from three experiments on analogical reasoning. (Author/RK)