Models of emotion, particularly those based on the Ortony, Clore, and Collins (OCC) account of emotions, have been used as part of agents' decision making processes to explore their effects on cooperation within social dilemmas [7, 19, 22]. We analyse two different interpretations of OCC agents. Firstly, Emotional agents that decide their action using only a model of emotions. To analyse the possibility of evolutionary stability of these agents we use the Prisoner's Dilemma game. We contrast the results with the second interpretation of an OCC agent, the Moody agent [7], which additionally uses a psychology-grounded model of mood. Our analysis highlights the different strategies that are needed to achieve success as a society in terms of both stability and cooperation, in the iterated Prisoner's Dilemma. The Emotional agents are better suited playing against a mixed group of agents with differing strategies than the Moody agents are. The Moody agents are more successful than the Emotional agents when only one strategy exists in the society.

, , , ,
18th International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2019

Collenette, J., Bloembergen, D., Atkinson, K., & Tuyls, K. (2019). Stability of human-inspired agent societies. In Proceedings of the International Joint Conference on Autonomous Agents and Multiagent Systems, AAMAS (pp. 1889–1891).