Specified meet contraction (Q946198)

The paper draws its intuitive motivation from two perceived shortcomings in the AGM account of belief contraction: (1) it works through `large' items, namely maximal non-implying subsets of the belief set, rather than remain on the level of sentences in the belief set; (2) it may take us from a finitely based belief set to one that cannot be finitely based. As is well known, AGM partial meet contractions can be presented in a manner that avoids feature (1), e.g., as `safe contractions' (Alchourrón and Makinson 1985) or as contractions via epistemic entrenchment (Gärdenfors and Makinson 1988); but both these presentations leave feature (2) intact. To avoid it as well, the author's presentation uses as auxiliary apparatus a function \(f\) from sentences to sentences, and treats the contraction of sentence \(x\) from a closed belief set \(K = Cn(K)\) as an AGM full meet contraction from \(K\), but of \(f(x)\) rather than of \(x\). This automatically avoids feature (2) since, as shown by Alchourrón and Makinson in 1982, the full meet contraction of a sentence \(x\) from a closed belief set \(K\) is just the intersection of \(K\) with the set of classical consequences of the negation of \(x\), so that if sentence \(k\) axiomatizes \(K\) then the disjunction of \(k\) with the negation of \(x\) will axiomatize the result of the contraction. The author's construction turns out to be surprisingly flexible, and he shows that with suitable conditions on the `sentential selector' \(f\) it yields all and only those AGM partial meet contractions lacking feature (2). It is also shown how variant conditions on the sentential selector function provide links with other kinds of contraction, including the class of finite-based contractions and the `severe withdrawal' contractions of \textit{H. Rott} and \textit{M. Pagnucco} [``Severe withdrawal (and recovery)'', J. Philos. Log. 28, No.~5, 501--547 (1999; Zbl 0934.03024)]. Reviewer's comments: As the author has remarked in correspondence with the reviewer, we may combine results of Rott and Pagnucco with others in the paper under review, to characterize severe withdrawal in terms sentential selector functions with suitable constraints. Likewise for the class finitely-based contractions, this time combining with results of the author's textbook [A textbook of belief dynamics. Theory change and database updating. Dordrecht: Kluwer (1999; Zbl 0947.03023), Theorem 4 on p. 313]. It is unfortunate that these characterizations were not formulated explicitly in the paper under review.