The goal of this research is to determine whether a computer based training game (HEART-SENSE) can improve recognition of heart attack symptoms and shift behavioral issues so as to reduce pre-hospitalization delay in seeking treatment. Since treatment delay correlates with adverse outcomes, this research could reduce myocardial infarction mortality and morbidity. In Phase I we created and evaluated a prototype virtual village in which users encounter and help convince synthetic personas to deal appropriately with a variety of heart attack scenarios and delay issues. Innovations made here are: (1) a design for a generic simulator package for promoting health behavior shifts, and (2) algorithms for animated pedagogical agents to reason about how their emotional state ties to patient condition and user progress. Initial results show that users of the game exhibit a significant shift in intention to call 9-1-1 and avoid delay, that multi-media versions of the game foster vividness and memory retention as well as a better understanding of both symptoms and of the need to manage time during a heart attack event. Also, results provide insight into areas where emotive pedagogical agents help and hinder user performance. Finally, we conclude with next steps that will help improve the game and the field of pedagogical agents and tools for simulated worlds for healthcare education and promotion.