A Computational Account of Everyday Abductive Inference


In this paper, we review the main qualitative characteristics of everyday inference in human cognition and present a computational account that is consistent with them. This includes both a representational framework and associated processes that produce abductive explanations in a flexible, incremental, and efficient manner. We clarify our approach with examples and report some initial empirical results. In closing, we examine related work and suggest directions for future research.

Proceedings of the Thirty-Third Annual Conference of the Cognitive Science Society