Hypothesis Testing and Ambiguity Aversion∗

We study a model of non-Bayesian updating, based on the Hypothesis Testing model of Ortoleva (2012), for ambiguity averse agents. Agents ranks acts following the MaxMin Expected Utility model of Gilboa and Schmeidler (1989) and when they receive new information they update their set of priors as follows: If the information is such that all priors in the original set of priors assign to it a probability above a threshold, then the agent updates every prior in the set using Bayes’ rule. Otherwise: she looks at a prior over sets of priors; she updates it using a rule similar to Bayes’ rule for second order beliefs over sets; finally, she chooses the set of priors to which the updated prior over sets priors assigns the highest likelihood. JEL: C61, D80, D81