Beliefs About Others: A Striking Example of Information Neglect

Abstract

In many game-theoretic  models of incomplete information, Nature draws players' types from one of several possible states of the world, and players have to update their beliefs about other players' types. We study such inferences in a simple environment with just two ex-ante equally-likely states, in which rational belief-updating does not involve difficult computations.  However, we find that between 50% and 70% of the experimental participants, depending on the scenario, completely ignore the information contained in their own type. Using interventions, we show that this striking information neglect is very robust.

Keywords: Information neglect, Bayesian inference, biased belief-updating