Towards a ``sophisticated'' model of belief dynamics. I: The general framework (Q935023)

The author proposes a way of representing belief states that is richer than those that are standard in decision theory, game theory, and the theory of belief change, so as to support less idealized and more realistic versions of such theories. More specifically, the paper seeks to take into account three phenomena: (1) The fact that certain statements may be `out of mind' for a given agent at a given time, i.e. not currently in the agent's language; (2) The set of statements currently `in mind' may change with time; (3) The agent may not be omniscient regarding logical relations between propositions. Part I of the paper, here reviewed, outlines and discusses the representation. Part II of the paper (announced) will apply it to the logic of belief change. Reviewer's comment: The constructions appear to have only a limited effect on the third requirement. For statements in the language of classical propositional logic, logical omniscience is avoided only when the elementary statements from which they are formed are not all simultaneously `in mind'. For example, when two elementary statements both happen to be in the current language, the agent is assumed to be aware of all classical equivalences between truth-functional ones formed from them, no matter how complex.