The session on Games consisted of four presentations discussing applications of mathematical game theory in multi agent systems. The first two presentations presented the work of Gleb Polevoy and Mathijs de Weerdt on agents that need to divide their available time over multiple projects, respectively multiple group interactions. An example of the former is the division of time over different research project, and an example of the latter is the active participation in different social media groups. When agents have to divide time over multiple projects, they need to consider the benefits of participating in each of the projects. The benefit of a project depends on whether enough agents investigate time in the project to be successful. When agents participate in multiple group interactions, the benefit of each group interaction depends on the participation of the other agents in the group. The conditions under which (social optimal) Nash equilibria exist, are investigated for agents dividing their available time over multiple projects, respectively multiple group interactions.

The third presentation presented a new gift-giving game developed by Elias Fernández Domingos, Juan Carlos Burguillo and Tom Lenaerts. Gift-giving games can be used to study the emergence of trust, fairness and generosity. Two types of agents were compared, namely reactive and anticipative agents. The anticipative agents showed some of the characteristics of human decision making in the experiments.

The last presentation of the session addressed repeated task allocation problems. This work by Qing Chuan Ye and Yingqian Zhang investigated the influence of past task allocation outcomes on future task allocation outcomes. Two types of agents were investigated, agents that only consider optimality in terms of costs, and agent that consider optimality in terms of primarily fairness and secondarily costs. The experiments demonstrated that the latter agents have an incentive to continue participating, which result in a higher social welfare of the agents.