CeDEx Brown Bag Seminar - Tong Fang

Title: The Difficulty of Cooperating in Long Finitely Repeated Prisoners’ Dilemmas

Abstract: A finitely repeated Prisoners’ Dilemma (PD) game models the social dilemma in a frequent interaction. Backward induction indicates no cooperation irrespective of game parameters and the horizon, but recent studies found evidence against this. In this study, we used stage game incentives to predict cooperation in a finitely repeated PD game: the percentage gain of unilateral defection and the percentage gain of mutual cooperation in terms of mutual defection. We conducted an orthogonal 2*2 between-subject experiment, where a fixed pair played the same PD game for 85 periods. We found the incentive of mutual cooperation positively correlates with average cooperation rate, while the incentive of unilateral defection is not a robust predictor.