Belief ascription under bounded resources

Natasha Alechina

Brian Logan

pp. 179-197

in: F. Liu (ed), Knowledge, rationality & action, Synthese 173 (2), 2010.

Abstract

There exists a considerable body of work on epistemic logics for resource-bounded reasoners. In this paper, we concentrate on a less studied aspect of resource-bounded reasoning, namely, on the ascription of beliefs and inference rules by the agents to each other. We present a formal model of a system of bounded reasoners which reason about each other’s beliefs, and investigate the problem of belief ascription in a resource-bounded setting. We show that for agents whose computational resources and memory are bounded, correct ascription of beliefs cannot be guaranteed, even in the limit. We propose a solution to the problem of correct belief ascription for feasible agents which involves ascribing reasoning strategies, or preferences on formulas, to other agents, and show that if a resource-bounded agent knows the reasoning strategy of another agent, then its ascription of beliefs to the other agent is correct in the limit.