Adaptive Risk Choice Behavior Using MAS

This paper analyzed risk choice game by using multi agent simulation and theoretical analysis. The game defined here is that decision-makers can choose some lotteries sequentially. The sequential choice means that they can select the next lottery by using outcomes of all the previous lotteries they chose. The number of types of lotteries are two, with the same expected payoff but different degree of risk. We modeled it mathematically strictly and analyzed some features with two approaches: multi agent model and theoretical analysis. According to the analyses, we observed that relative high-risk behaviors, which is not the highest, easily survive. Further, more risky strategies are most adaptive when mutants always can invade. Such insights not only correspond to human behavior described by the prospect theory, but also make it possible to explain a rational mechanism which cannot be explained by this theory.