Toward a Unif ied Theory of Reasoning

This article describes a theory that uses mental models to integrate deductive, inductive, and probabilistic reasoning. It spells out the main principles of the theory and illustrates them with examples from various domains. It shows how models underlie inductions, explanations, estimates of probabilities, and informal algorithms. In all these cases, a central principle is that the mind represents each sort of possibility in a separate mental model and infers whatever holds in the resulting set of models. Finally, the article reviews what has been accomplished in implementing the theory in a single large-scale computer program, mReasoner.