Modeling Plausible Behaviour of Cognitive Agents

This paper describes a formal framework for the representation of agents that do not necessarily share common beliefs and goals. In this framework, that we consider to be a plausible model of cognitive agent behaviour in a multiagent system, the agents are able to reason based on assumptions while facing inconsistencies, manifest preferences over their beliefs, select and dynamically modify their attitude towards the beliefs of other agents. We use the same framework to support agent communication and rational reactions to communicative acts, and to develop plausible models of agents, without access to the internal structure of those agents. The main idea of this work is that default reasoning enhanced with preferences, as formerly proposed by the first author for the monoagent case, may uniformly support plausible solutions for the design and modeling of many agent features in open multiagent systems 1.