Resource bounded belief revision (Q1583794)

The AGM paradigm of belief change is formulated without explicit reference to the believing agent -- all attention is focused on what is believed. But in so far as the role of the agent is made explicit, it may be said to have unlimited powers of memory and of deductive reasoning. In this paper (which is based on a section of the author's doctoral thesis, with the same title, University of Amsterdam, January 2000) the author outlines a formal account of belief revision for agents that are limited in both respects. The construction is guided by the informal treatment of \textit{G. Harman} [Change in view: Principles of reasoning (Cambridge MA: MIT, 1986)], which is written from the perspective of a philosopher, and develops semiformal ideas of \textit{C. Cherniak} [Minimal rationality (Cambridge MA: MIT, 1986)]. The formal expression also appears to be influenced by basic ideas of the theory of Turing machines. A distinctive feature of the author's approach that any belief set is understood to contain, as well as a set \(E\) (typically finite) of currently explicit beliefs, another set \(A\) of currently active propositions, i.e. ones that are within the current field of attention of the human or other agent. Half a dozen basic operations are defined, representing the transfer of a proposition in, out and between these two sets, as well as applications of a ``single step'' of a background inference relation. These operations may then be composed to form chains which, the author claims, suffice to lead from any one belief state to any other. The author observes that the AGM construction may be seen as a limiting case of the presented one.