Integrating Situation Awareness and Surprise: A Computational Agent Model

To increase realism of virtual agents in serious gaming applications, computational models of human cognitive processes can play an important role. This paper presents a computational agent model that integrates models for situation awareness and surprise. The crucial element in the integration of both models is a mechanism that matches beliefs about the situation from the situation awareness model with expectations about the world from the surprise model. To evaluate the integrated model, it has been implemented within an existing simulator for F16 pilots, and four variants of the model were tested within a domain-specific scenario. The results indicate that the model has promising capabilities to enhance realism of virtual agents.