The cognitive demands on cooperation in social dilemmas: An fMRI study

Abstract

This study uses fMRI to investigate the cognitive demands of decision-making in two types of cooperation games: a prisoner's dilemma (PD) eliciting a temptation to free-ride, leading to a dominant, self-interested response, and a stag hunt (SH) that has no dominant response but offers pay-off incentives that make mutual cooperation collectively beneficial but risky. Consequently, the PD poses greater conflict between self- and collective interest, greater demands for computational reasoning to derive the optimal solution, and greater demands for mentalizing to infer the intentions of others. Consistent with these differences between the two games, the results indicate that the PD is associated with increased activity in the anterior cingulate gyrus, prefrontal cortex, parietal lobe, and temporoparietal junction. With less conflict, the demands for computation and mentalizing are reduced in the SH, and cooperation levels increase dramatically. The differences in brain activation elicited by the different incentive structures of the PD and the SH appear to be independent of individual differences in revealed social preferences.

Keywords:

• Cooperation
• fMRI
• Prisoner's dilemma
• Cognition
• Game theory

Acknowledgments

We thank Armin Heinecke and Professor Wim Fias for their assistance regarding data analysis. This study was funded by ID BOF 1931 from the University of Antwerp.

Notes

¹The economic games in the study by Kuo et al. (2009) did not involve cooperation and defection, and therefore the dominance-solvable games they used lacked the conflict between self-interest and collective interest that is so typical of the PD.

²Candidates who indicated that they were ambivalent about their handedness were asked to fill in a handedness questionnaire (Oldfield, 1971).

³No participants raised any doubts or questioned the procedures before, during, or after the experiment.

⁴To avoid boredom or habituation during the course of the experiment, the pay-off matrices for the games were varied. All but one of the matrices for PD and SH in Appendix 1 were retrieved from previously published studies on cooperation and were matched as best as possible to have similar pay-offs to defection.

⁵The paper we refer to is based on the same data as the current paper but discusses the analysis of neural activity with respect to individual differences in social value orientation.