Modeling Unimportant Decisions

Abstract

We propose an incomplete information game, in which rational players depend on finite belief hierarchies in determining their optimal actions. This is done by introducing costs when players climb up their belief hierarchies. With bounded payos, players climb up to a finite order of belief, and just depend on these beliefs in making their decisions. The model is consistent with an experimental literature which shows that people play games with finite belief hierarchies, while it can avoid the problem of misspecification of high order beliefs, which may limit predictive power of some theories.