Skip to content Skip to navigation

Bayes, Bounds, and Rational Analysis

Thomas Icard
Philosophy of Science
2017

Abstract: While Bayesian models have been applied to an impressive range of high-level cognitive phenomena in recent years, methodological challenges have been leveled concerning their particular use in cognitive science, and specifically their role in the program of rational analysis. The focus of the present article is on one strand of these criticisms, namely, computational impediments to probabilistic inference, and related puzzles about empirical confirmation of these models. The proposal is to rethink the role of Bayesian methods in rational analysis, and specifically to adopt an independently motivated notion of rationality appropriate for computationally bounded agents, and then to explore broad conditions under which (approximately) Bayesian agents would in fact be rational. The proposal is illustrated with a characterization of computational costs in an abstract manner inspired by ideas in thermodynamics and information theory.