Relational Reasoning in a Neurally Plausible Cognitive Architecture: An Overview

ABSTRACT—Human mental representations are both flexi-ble and structured—properties that, together, present challenging design requirements for a model of human thinking. The Learning and Inference with Schemas and Analogies (LISA) model of analogical reasoning aims to achieve these properties within a neural network. The model represents both relations and objects as patterns of activation distributed over semantic units, integrating these representations into propositional structures using synchrony of firing. The resulting propositional structures serve as a natural basis for memory retrieval, analogical mapping, analogical inference, and schema induction.The model also provides an a priori account of the limitations of human working memory and can simulate the effects of various kinds of brain damage on thinking