Norm learning in multi-agent societies

Abstract

In Normative Multi-Agent Systems (NorMAS), researchers have investigated several mechanisms for agents to learn norms. In the context of agents learning norms, the objectives of the paper are three-fold. First, this paper aims at providing an overview of different mechanisms employed by researchers for norm learning. Second, it discusses the contributions of different mechanisms to the three aspects of active learning namely learning by doing, observing and com- municating. Third, it compares two normative architectures which have an emphasis on the learning of norms. It also discusses the features that should be considered in future norm learning architectures.