Bounded rationality in games

Abstract

How do people reason about others in strategic situations and how does that affect their behavior? These questions have been at the forefront of game theory since its inception in the first half of the twentieth century. Traditionally, the focus has been on the question how "rational" players behave. As already observed by VonNeumann and Morgenstern (1944), however, the question how rational players should behave cannot be separated from the question how non-rational players behave. Even if one is concerned only with rational behavior, the interactive nature of the problem makes that one has to deal with all possible types of players: What is optimal for a rational player depends on what he expects his opponents to do, and these opponents may be boundedly rational. It is therefore desirable to have a theory of behavior in strategic settings that encompasses both perfect rationality and forms of bounded rationality. In this talk, I describe a general theoretical framework that takes into account that individuals may have limited capacities to reason about others, and sometimes only have access to a very coarse description of the game. I discuss the strategic implications of such a framework.