Proceedings of the Annual Meeting of the Cognitive Science Society

Understanding the nature of commonsense reasoning is one ofthe deepest questions of cognitive science. Prior work hasproposed analogy as a mechanism for commonsense reasoning,with prior simulations focusing on reasoning about continuousbehavior of physical systems. This paper examines howanalogy might be used in commonsense more broadly. The twocontributions are (1) the idea of common sense units,intermediate-sized collections of facts extracted fromexperience (including cultural experience) which improvesanalogical retrieval and simplifies inferencing, and (2)analogical chaining, where multiple rounds of analogicalretrieval and mapping are used to rapidly constructexplanations and predictions. We illustrate these ideas via animplemented computational model, tested on examples froman independently-developed test of commonsense reasoning.