Reconciling Group and Individual Preferences of Multi-Agent Systems

Many multi-agent systems are mixed-motive, such that the interests of the individuals are not entirely compatible with each other or with the group as a whole. In such situations it is imperative to reconcile these differences and arrive at a joint decision that yields acceptable performance for the group and for each of its members. Social utilities provide a natural mathematical model to account for social attributes such as compromise, coordination, negotiation, and altruism. Satisficing game theory provides an alternative solution concept that is compatible with social utilities and which admits principled negotiation protocols that are designed to reach solutions that provide adequate performance for the group and are acceptable to each of the individuals.