Establishing cooperation in a mixed-motive social dilemma. An fMRI study investigating the role of social value orientation and dispositional trust

Abstract

When people are confronted with social dilemmas, their decision-making strategies tend to be associated with individual social preferences; prosocials have an intrinsic willingness to cooperate, while proselfs need extrinsic motivators signaling personal gain. In this study, the biological roots for the proselfs/prosocials concept are explored by investigating the neural correlates of cooperative versus defect decisions when participants engage in a series of one-shot, anonymous prisoner’s dilemma situations. Our data are in line with previous studies showing that prosocials activate several social cognition regions of the brain more than proselfs (here: medial prefrontal cortex, temporo–parietal junction, and precuneus BA 7 (Brodmann area 7), and that dispositional trust positively affects prosocials’ decisions to cooperate. At the neural level, however, dispositional trust appears to exert a greater marginal effect on brain activity of proselfs in three social cognition regions, which does not translate into an increase in cooperation. An event-related analysis shows that cooperating prosocials show significantly more activation in the precuneus (BA 7) than proselfs. Based on previous research, we interpret this result to be consistent with prosocials’ enhanced tendency to infer the intentions of others in social dilemma games, and the importance of establishing norm congruence when they decide to cooperate.

Keywords:

• Cooperation
• Trust
• Social value orientation
• Decision-making
• Prisoner’s dilemma.

Notes

¹ See Figure 1 for a pay-off matrix of a PD used in this experiment. In the PD game, subjects are presented with a choice between two options shown in a matrix: cooperate or defect (respectively options “L” and “K” in Figure 1). Mutual cooperation (both players choose “L”) is collectively the best choice. However, a greedy person can always obtain a better outcome by defecting (choosing “K”), hoping to benefit from the other player’s cooperation. From an economic viewpoint, defecting is the dominant strategy and mutual defection the only Nash equilibrium: you gain more if the other player cooperates, and you don’t lose anything if the other player defects. This option, however, is collectively deficient, as the best mutual outcome can only be reached if both players cooperate.

² The one-shot PD was contrasted with a coordination dilemma where no conflict was present.

³ However, while in our previous study (Emonds et al., 2011), we attributed differences in TPJ activity between prosocials and proselfs to differences in mentalizing activities (an interpretation on which we base our current hypotheses), we point out that recent results of lesion studies indicate that the contribution of these region to social cognition may be more general and not specific to reasoning about other people’s believes (Apperly, Samson, Chiavarino, Bickerton, & Humphreysn, 2004; Apperly, Samson, Chiavarino, & Humphreys, 2007). Patients with Alzheimer’s disease show atrophy in the posterior cingulate and temporoparietal cortex which accompany deficits in social cognition (Herholz, 2003), but they maintain relatively intact social awareness (Bozeat, Gregory, Ralph, & Hodges, 2000). Furthermore, in an experiment which set out to test the domain specificity of false belief reasoning, patients with TPJ lesions were indeed impaired in belief reasoning, but they also showed deficits during performance on a photograph task which imposed similar executive demands (Apperly et al., 2007). This particular study is consistent with findings of Saxe and Wexler (2005) that corroborates that TPJ regions are activated both during false belief reasoning and when plainly reading about the thoughts and feelings of story characters. These findings do not necessarily contradict a role of the TPJ in belief reasoning (Samson, Apperly, Chiavarino, & Humphreys, 2004), but its role may be more related to mentalizing semantics rather than theory of mind reasoning (Apperly et al., 2007).

⁴ Most research in this domain uses the bivariate SVO variable rather than a continuous measure (see Bogaert et al. (2008) for a review). Although information loss is sometimes a drawback, the advantage of the bivariate measure is that it reduces noise by eliminating those individuals with an undefined SVO. Previous research has compared analyses using either a bivariate or a continuous SVO measure, and found no statistically significant differences (Declerck & Bogaert, 2008; Sheldon, 1999).

⁵ Age, gender, and cooperative decision in previous round were added as control variables.