Luis G. Nardin and Jaime S. Sichman

In multiagent systems, agents may form coalitions in order to cooperatively achieve their goals. Assuming that agents are selfish, or potentially unreliable, they should implement some mechanism to deal with the uncertainty arising from the cooperation. Trust is usually chosen as the mechanism for modeling and reasoning about agents’ reliability. Hence, this work presents an agent-based simulation model composed of agents who play the spatial prisoner’s dilemma game and take into account the notion of trust to form coalitions. Moreover, some experiments are performed and their results suggest that in some cases the notion of trust significantly influences the coalition dynamics.

Download PDF