ABSTRACT
Introduction
A belief revision system must be able to decide, in the face of contradictory information, which (if any) beliefs should be abandoned. What general principle should be followed to make this decision? In Martins and Shapiro (1988), several specific tasks were described which must be solved in order for a system to successfully revise beliefs in the face of contradictory evidence: inference, dependency recording, nonmonotonicity, and disbelief propagation. A cognitive theory of belief revision should be able to say which beliefs should be retained or abandoned while allowing solutions to the subproblems to be covered by the same theory.
