Explaining and Predicting the Behavior of BDI-Based Agents in Role-Playing Games

Virtual characters in games operate in a social context involving other characters and human players. If such socially situated virtual characters are to be considered believable, they should be able to adjust their behavior based on beliefs about the mental states of other characters and human players. Autonomous BDI-based agents are suitable for modeling characters that exhibit such intentional behavior. In this paper, it is illustrated how agent-based characters can infer the mental state of other players by observing those other players’ actions in the context of a declarative game specification. The game specification can be utilized in explanation and prediction of agents’ behavior, and as such can form the basis for developing socially aware characters.