Neural signals relating future cooperation from prisoner’s dilemma feedback are outcome-contingent
E. Martínez-Montes,Valentina Paz,Eliana Nicolaisen-Sobesky,Santiago Garat,D. Kessel,V. Gradin,Álvaro Cabana,Francisco Cervantes Constantino
DOI: https://doi.org/10.1101/784033
2019-09-26
bioRxiv
Abstract:Cooperation upholds life in organized societies, but its neurobiological mechanisms remain unresolved. Recent theoretical analyses have contrasted cooperation by its fast versus slower modes of decision making. This raises the question of the neural timescales involved in the integration of decision-related information, and of the participating neural circuits. Using time-resolved electroencephalography (EEG) methods, we characterized relevant neural signatures of feedback processing at the iterated prisoner’s dilemma (iPD), an economic task that addresses cooperation-based exchange between social agents. We then probed the ability of relevant EEG signals to indicate subsequent decision-making, and inspected game and behavioral conditions under which they do so - including timing. Participants played against an assumed co-player, and the neural activation at the game feedback stage was analyzed via event-related potential (ERP) and spectrotemporal analysis methods. As expected from value-based decision-making studies, ERPs were increased by cooperation, including the feedback-related negativity (FRN) and the P3. In addition, we report a late potential (LP) lateralized to left frontal regions, which was also modulated by cooperation. Slow-wave, delta-band activity associated to the LP component (termed ‘LP-delta’) was also found to be reduced for mutual versus one-sided cooperation feedback. We then addressed individually which of these feedback signals related to subsequent cooperation and when. Crucially, the FRN, P3 and LP-delta components were modulated by choice at the next round, although involvement of each component was contingent upon the current outcome type. Immediately after one-sided cooperation (‘sucker’s payoff’), differential signaling indicating future decision-making occurred at the relatively early FRN, decreasing in amplitude for player choices not to cooperate at the next round. For outcomes where the player did not cooperate at the co-player expense, P3 increases were associated with subsequent cooperation by the player. Following mutual cooperation rounds, and at a relatively late stage of feedback processing, modulations by forthcoming decision were found in the form of LP-delta desynchronizations that were prospective of further cooperation. We note that the decision to cooperate in the game can be based on predicting different aspects of the prospective interaction given the past realized outcome (e.g. whether ‘to sustain’ mutual cooperation or ‘to retaliate’ if it is one-sided). The results support information integration for cooperation decision-making at the iPD, beginning at different timescales and circuits, by such updating context. The specific case of verifying a partner’s cooperation entailed a shorter reciprocation response which was predictable by feedback signal changes at the immediately preceding trial. A neurally informed account of fast-mode dual-process mechanisms, and involvement of predictive coding principles in guiding cooperation with a partner at the iPD, are discussed. Highlights *FRN, P3 and late delta-band asymmetric frontal activity encode cooperation feedback *Decision to cooperate in iterated prisoner’s dilemma indicated by prior round signals *Speed of reciprocating partner’s cooperation correlates with single-trial activity
Biology,Psychology