Towards a ``sophisticated'' model of belief dynamics. II: Belief revision (Q1005934)

In Part I of this paper [Stud. Log. 89, No. 1, 81--109 (2008; Zbl 1148.03007)], the author proposed a way of representing belief states which is richer than those customarily employed. A salient feature of the representation is that it allows change over time in the propositions available in the language. The purpose of Part II is to illustrate the fruitfulness of that representation by using it in a reconstruction of the AGM theory of belief change. Familiarity with Part I is presupposed. The author shows that his reconstructed system satisfies the AGM postulates in certain cases, though not in others. The delineation of those cases provides him with a vantage point from which to discuss some candidate counterexamples devised by \textit{H. Rott} [Synthese 139, No. 2, 225--240 (2004; Zbl 1106.03310)]. From the author's perspective, these counterexamples should not be seen as refutations of the postulates, but rather as symptoms of their limited range of applicability. One lesson that he draws from them is that before carrying out revision, we may need to effect some pre-processing of the input. Finally, the author observes that his enriched representation of belief states provides a framework for iterated revision, into which fit as special cases several systems from the literature.