A preliminary analysis of interactive effects between network structure and decision criteria on the global behavior

Although, it is said that the society consists of individually-motivated agents, a decision making of each agent´s is affected more or less by that of other agents´. In this paper, we focus on the possible factors that influence to the agents´ interactions and manage the behavior of whole system. We take up three factors, i.e.; a payoff structure, an agent´s decision criterion and a network structure. These three factors seem largely affect the behavior of multiagent system, and they seem to be crucial in terms of finding the way to make whole behavior a desirable condition. We show the empirical results of agents´ behaviors in the some typical types of network structures, such as small-world network and random network where each agent has two alternative actions, decides rationally based on its own decision criterion, and is partially affected by the actions of its neighbors. The results show that these three factors seem important to manage system behavior, especially a network structure affects agent´s behavior more than the individual decision criteria does, in Stag-Hunt game.