No bio entered.
Game theory, artificial intelligence, agent-based models, genetic algorithms.
The application of game theory features in computational environments has been extensively per- formed in distinct fields, as the desire to reproduce strategic behaviour applied to solving complex optimization problems is part of a growing niche in artificial intelligence. This paper introduces an experimental and exploratory approach, combining game theory and Genetic Algorithms to create a model that simulates social interaction situations, economic learning and biological evolution. The introduced construct aimed at allowing for the evaluation of how players change their strategies over time, towards an optimal outcome. Specific 2 × 2 strategic form games as objects of analysis and three distinct strategy selection rules (the Nash equilibrium, Hurwicz Criterion and a Random method) were used in this study. Such games act as strategic scenarios simulations and are treated as individuals of a genetic population, from which they are evaluated, selected and reproduced. The outcome indicated an optimized player environment, demonstrating the maximization of the overall individual payoffs, while transforming the games in such a way that agents can coordinate their strategies and achieve an Evolutionary Equilibrium with optimal payoffs. Additionally, the mutation of the populations into a robust set of fewer isomorphic games featuring the strong characteristics from well-performing individuals is observed, as a consequent to the evolutionary learning process.