This article presents findings from a research project that uses an interactive simulator of an imaginary fishing dispute. Subjects operating the simulator play the role of a state leader, while the computer program controls the behavior of a contending state as well as provides all the environmental data associated with the conflict. The simulator is designed to operate in real time according to a predefined set of rules and feedbacks. Therefore, it is not a game, where each party takes turns, but a microworld, where both the operator and the computer opponent can initiate actions independently of one another. Because the subjects are able to learn from experience and improve their policies, it might be expected that they should eventually discover an appropriate policy that maximizes their payoffs. However, experiments with university students reveal that the majority of the subjects fail to reach optimal strategies. It appears that there is a preferred bias toward choosing particular types of strategies that satisfy normative standards, even though these strategies produce lower payoffs. (Contains 6 tables, 2 figures and 6 notes.)