Two modellings for theory change (Q1100194): Difference between revisions

In this very clear paper, the author shows the close connection between the logic of theory change, as recently developed by \textit{C. E. Alchourrón}, \textit{P. Gärdenfors} and the reviewer [e.g. ``On the logic of theory change: partial meet contraction and revision functions'', J. Symb. Logic 50, 510-530 (1985; Zbl 0578.03011)] and the earlier logic of conditionals developed by R. Stalnaker and \textit{D. Lewis} [e.g. the latter's ``Counterfactuals'' Cambridge, Mass.: Harvard University Press (1973)]. He does this by showing (theorems 1, 2) that the theory-revision functions satisfying postulates \((+1)-(+8)\) of the former reference can be represented by system-of-spheres models that are similar to, though a little different from, those of the latter reference, and by showing (section 4) that the transitively relational partial meet semantics for contraction in the former reference can also be transformed into such a system-of-spheres semantics. The pivot of this translation is the observation that there is a simple bijection between the maximal subtheories of a given theory T that fail to contain a proposition \(A\in T\), and the maximal consistent theories that contain \(\neg A\). In addition to the above, the author describes a way (theorems 3, 4) of characterizing theory-revision functions by a modelling that uses a relation ordering the propositions of the theory rather than ordering its maximal proper subtheories. Readers will need to be familiar with the references mentioned above; so prepared, they will also appreciate the perspicacity and elegance of the author's work of interconnection. The reviewer would mention two questions that are suggested by the author's results. One: Can a more or less analogous job be done for contraction and revision functions for which only the basic postulates \((\pm 1)-(\pm 6)\) of the former reference are assumed, i.e. for partial meet contraction that is not necessarily transitively or otherwise relational? This question seems to remain open. Two: The characterization of theory revision and contraction function via a relation ordering propositions of the theory, given by the author in theorems 3 and 4, is shown to work in a rather indirect way, via system-of-spheres models. Can it be proven directly? This question has been answered positively by \textit{P. Gärdenfors} and the reviewer [``Revision of knowledge systems using epistemic entrenchment'', in Proceedings of the Second Conference on Theoretical Aspects of Reasoning about Knowledge (M. Y. Vardi (ed.), Los Altos: Morgan Kaufmann, 1988)].