Before engaging in a group venture agents may seekcommitments from other members in the group and,based on the level of participation (i.e. the numberof actually committed participants), decide whether itis worth joining the venture. Alternatively, agents candelegate this costly process to a (beneficent or non-costly) third-party, who helps seek commitments fromthe agents. Using methods from Evolutionary GameTheory, this paper shows that, in the context of PublicGoods Game, much higher levels of cooperationcan be achieved through such centralized commitmentmanagement. It provides a more efficient mechanismfor dealing with commitment free-riders, those who arenot willing to bear the cost of arranging commitmentswhilst enjoying the benefits provided by the paying commitment proposers.We show also that the participationlevel plays a crucial role in the decision of whetheran agreement should be formed; namely, it needs tobe more strict in the centralized system for the agreementto be formed; however, once it is done right, it ismuch more beneficial in terms of the level of cooperationas well as the attainable social welfare. In short, ouranalysis provides important insights for the design ofmulti-agent systems that rely on commitments to monitoragents’ cooperative behavior.
|Publication status||Published - 4 Feb 2017|
|Event||31st AAAI Conference on Artificial Intelligence - San Francisco, United States|
Duration: 4 Feb 2017 → 9 Feb 2017
|Conference||31st AAAI Conference on Artificial Intelligence|
|Period||4/02/17 → 9/02/17|
Han, T. A., Pereira, L. M., Martinez-Vaquero, L. A., & Lenaerts, T. (2017). Centralized vs. Personalized Commitments and their influence on Cooperation in Group Interactions. Paper presented at 31st AAAI Conference on Artificial Intelligence, San Francisco, United States.