A Game-Theoretic Social Model for Multiagent Systems

Abstract

A collective of autonomous decision-making agents that is tasked to function cooperatively must take into account the effects of their actions on other agents as well as themselves. Although game theory provides a framework within which to achieve rational behavior, classical approaches are limited in their ability to account for sophisticated social relationships such as cooperation, altruism, and negotiation, since they rely on the assumptions of categorical utilities and individual rationality. Conditional game theory is an extension of classical theory that overcomes these limitations by allowing conditional utilities, which permit players to modulate their preferences by the preferences of others, thereby creating an emergent social model as the conditional preferences propagate through the network. New solution concepts that simultaneously take into consideration both group and individual preferences create a framework within which to accommodate sophisticated social relationships.