Beliefs, rational choice and Bayesian learning

Abstract

People act in the light of their beliefs, yet sociological theories of action have been weak in explaining why people hold the beliefs they do. I propose a means of integrating beliefs into rational choice theory using a Bayesian learning model in which people act in accordance with the beliefs they hold about the world. By observing the outcomes of their actions they modify their beliefs. This approach, in contrast with many others, recognizes that beliefs evolve in the light of experience. The approach is applied to show how the evolution of beliefs about the returns to effort in education can give rise to observed patterns of class differences in educational expectations and so to class differences in mobility chances, at least as these are mediated through educational attainment. I discuss the general issue of learning in the Bayesian approach and outline some limitations and areas of further research.