Minority game with local interactions due to the presence of herding behavior

Abstract

In this paper, we introduce a framework to study local interactions due to the presence of herding behavior in a minority game. The idea behind this approach is to consider that some of the agents who play the game believe that some of their neighbors are more informed than themselves. Thus, in this way, these agents imitate their most informed neighbors. The notion of neighborhood here is given by a regular network, a random network or a small world network. We show that under herding behavior the cooperation between the agents is less efficient than that one which arises in the standard minority game. On the other hand, depending on the topology of the network, we show that that the well known curve volatility versus memory, which caracterizes the minority game, is a monotone decreasing curve.