Analyzing the Evolution of Social Exchange Strategies in Social Preference-Based MAS through an Evolutionary Spatial Approach of the Ultimatum Game

摘要

This paper presents a multiagent-based approach of an evolutionary and spatial version of the Ultimatum Game interpreted as Game of Social Exchange Processes, where the agents organized in a complex network evolve their exchange strategies considering their possibly different social preferences. We analyze the possibility of the emergence of the equilibrium/fairness behavior when the agents, trying to maximize their social preference-based utility functions, increase the number of successful interactions. We consider an incomplete information game, since the agents do not have information about the other agents' exchange strategies. For the strategy learning process, a genetic algorithm is used, where the agents aiming at the self-regulation of the exchanges allowed by the game, balance individual and collective goals expressed by their social preferences. We also analyze a second type of scenario, considering an influence politics, when the average of the offer and reserve values of all agents adopting the same social preference form becomes public in a single simulation step, and the agents of the same network, have been influenced by that, imitate those values. At the same time, the network topology is modified, representing some kind of mobility, in order to analyze if the results are dependent on the neighborhood. The model was implemented in Net Logo.