Realistic Belief Revision.
Greg Restall, John Slaney · 1995
Abstract In this paper we consider the implications for belief revision of weakening the logic under which belief sets are taken to be closed. A widely held view is that the usual belief revision functions are highly classical, especially in being driven by consistency. We show that, on the contrary, the standard representation theorems still hold for paraconsistent belief revision. Then we give conditions under which consistency is preserved by revisions, and we show that this modelling allows for the gradual revision of inconsistency. 1 Realistic Logics Belief Revision is a rich and diverse eld. The unit of study is most often a belief set | a set K of sentences (propositions, whatever) closed under a consequence relation. In Gardenfors ' canonical text [6], and in nearly all other studies of belief revision, the notion of consequence is taken to be superclassical. Consequence at least includes classical propositional consequence. This is a theoretical simpli cation. No-one believes that belief is closed under that sort of consequence. If it were, we would believe all