Edited by Richard Arena, Agnès Festré and Nathalie Lazaric
1 Jacques Durieu and Philippe Solal 11.1 INTRODUCTION It is relatively well known that agents do not always make rational choice. Several experimental studies show that many observed behaviors can be both well described ex post and robustly predicted ex ante by a simple family of learning theories. In the last 20 years, the variety of such learning models that have been used in economics has increased tremendously. Most of these models explain the need for agents to learn because, initially, they lack information. Moreover, agents may have limited ability to make optimal decisions under various constraints. In a framework of strategic interaction among agents, game theory has traditionally assumed that players are perfectly rational. However, this approach is complemented by models with boundedly rational players. In game theory, the assumption of boundedly rational behaviors can be justified by a lack of information on the game structure. This means that players have limited knowledge of the strategic environment, the payoff functions or the rationality of other players. Then, agents try to simplify their decision task. A common assumption in the models discussed in this chapter is that players consider that their environment is stationary. One of the main purposes of these models is to investigate whether agents, who face the same game recurrently over time, learn to play an equilibrium of this game. In such a case, the models provide a foundation of equilibrium based on information collected by players along their interactions. By contrast, it is worth noting that...
You are not authenticated to view the full text of this chapter or article.