Do Social Norms Emerge? The Evolution of Agents' Decisions with the Awareness of Social Values under Iterated Prisoner's Dilemma

Abstract

Recently, the study of social norms has attracted much attention in multi-agent system research. We here focus on the case that agents are aware of the social value of each action under prisoner's dilemma games. During decision-making, agents weigh different actions based on the social values and their individual payoffs. We model such agents' behaviours with the notion of social value orientation. An experimental study is conducted on different kinds of agent societies. It is shown that a stable state can be generally reached in these societies, however, global social norms do not always emerge. In particular, we identify dynamic equilibrium phenomena as the ultimate state in some agent societies. Moreover, we show that in dynamic societies, agents eventually develop individualistic or altruistic orientations.