Belief formation and belief updating under ambiguity: Evidence from experiments

This paper investigates beliefs in an ambiguous environment. In contrast to many previous studies, the beliefs regarding possible scenarios are measured independently from attitudes. We use laboratory experiments to estimate the entire distribution of subjective beliefs and examine how beliefs are updated, incorporating new information. We find that beliefs and updating rules are quite heterogeneous. For most subjects, we can reject the objective equality hypothesis that beliefs are uniformly distributed. The unbiased belief hypothesis cannot be rejected overall; Most subjects display no bias towards pessimism/optimism in beliefs. The Bayesian updating hypothesis can be rejected; Most subjects under-adjust beliefs in response to new information. Finally, we find that subjects adjust their beliefs symmetrically to good news and to bad news.