Towards an Expectation-based Theory of Defense

Multi-Agent Systems should not be conceived as only cooperative. As open systems situations of concurrence, competition and conflict often arise. Starting from this perspective it is relevant not only pro-social interaction modeling, but also a theory of trust and monitoring, giving special relevance to issues of security and defense: how can an agent prevent that dangerous actions of other agents and dangerous events will frustrate his own goals? In this paper a general model of defense for intentional agents is presented as well as a general ontology of defensive goals and acts. Defensive goals are activated at different point of the temporal trajectory which goes from antecedents of the expected vehicle of attack to the expected damage and are directed to prevent that different components of the expected damaging sequence of events will happen.