Epistemic semantics for counterfactuals (Q1185041)

The main purpose of this paper is to propose a modification of the well- known Alchourrón/Gärdenfors/Makinson (briefly, AGM) ``partial meet'' semantics for belief revision, in such a way as to permit its use in Ramsey's ``epistemic rule for conditionals'', i.e. the rule that \(\alpha>\beta\in K\) iff \(\beta\in K*\alpha\), where \(K*\alpha\) is the result of revising the theory \(K\) by the proposition \(\alpha\). As is notorious, such a rule is incompatible with AGM revision in all but certain trivial situations. The author's rather intricate semantics is inspired by David Lewis' concept of ``imaging'', originally devised in the context of probabilities. Reviewer's comments: Although apparently devised independently, the author's construction is rather close to that of \textit{H. Katsuno} and \textit{A. Mendelzon} ``On the difference between updating a knowledge base and revising it'', Technical Report KRR-TR-90-6, Computer Science Department, University of Toronto, 1990. However, in certain respects it is more complex. Indeed theories are already taken by the author as more complex objects than mere sets of formulae closed under a consequence operation. The author also borrows from Stalnaker a constraint of ``expansiveness'' on his selection functions, whose effect is to ensure that when proposition \(\alpha\) is consistent with \(K\), then \(K*\alpha\) is identical, at least in its classical part, with \(C_ n(K\cup\{\alpha\})\) --- unlike the situation for Katsuno and Mendelzon update but like that for AGM revision.