The role of knowledge in logic-based rational interactions

Research on the theory of rational interaction has concentrated on designing the logical database axioms that need to be present for an agent to make a reasoned decision in an encounter. It is critical for an agent's designer to understand exactly what that agent needs to know in order to complete its analysis of an encounter and decide on an action. Previous research has assumed that common knowledge of outcome preferences was necessary, but it is shown here that this if often not the case. An explanation is given of the levels of knowledge that each agent requires of other agents' preferences and rationality axioms to resolve common interactions.<<ETX>>