Oxytocin promotes altruistic punishment

Abstract

The role of neuromodulators in the enforcement of cooperation is still not well understood. Here, we provide evidence that intranasal applied oxytocin, an important hormone for modulating social behavior, enhances the inclination to sanction free-riders in a social dilemma situation. Contrary to the notion of oxytocin being a pro-social hormone, we found that participants treated with oxytocin exhibited an amplification of self-reported negative social emotions such as anger towards free-riders, ultimately resulting in higher magnitude and frequency of punishment of free-riders compared to placebo. Furthermore, we found initial evidence that oxytocin contributes to the positive effects of a punishment institution by rendering cooperation preferable in the oxytocin condition for even the most selfish players when punishment was available. Together, these findings imply that the neural circuits underlying altruistic punishment are partly targeted by the oxytonergic system and highlight the importance of neuromodulators in group cohesion and norm enforcement within social groups.

Keywords: altruistic punishment; neuroendocrinology; norm enforcement; oxytocin; social dilemma.

Fig. 1.

Relative frequency of assigned punishment points imposed on defectors (A) and on cooperators (B) as a function of oxytocin. Assigned punishment points are depicted for the OT group in gray bars and for the placebo group in white bars (n = 144). (A) Punishment points imposed on defectors (M = 4.32, SD = 3.79) are significantly higher in the OT group than on defectors (M = 2.20, SD = 3.70) in the placebo group (Mann–Whitney U-test; z = −2.334, P = 0.0196, two-sided). (B) Punishment of cooperators is almost nonexistent; subjects in the OT group (M = 0.72, SD = 2.48) and the placebo group (M = 0.10, SD = 0.45) show non-distinguishable inclinations to punish cooperators (Mann–Whitney U-test; z = −0.578, P = 0.5630, two-sided).

Fig. 2.

Punishment rate by punished types. The bar C→C represents the proportion of all cooperators punished by other cooperators; analogously, C→D is the proportion of all defectors punished by cooperators; and so on. (A) In the placebo group about 22% (15 out of 68) of all defectors were punished by cooperators, whereas only 6% (4 out 68) were punished by other defectors. Punishment of cooperators (2 out of 76) played only a minor role. (B) In the OT group 43% (19 out of 44) of all defectors were punished by cooperators. A few defectors punished cooperators (4 out of 100).

Fig. 3.

(A) Relative frequency of cooperation levels with and without a punishment option in the OT (grey bars) and placebo (white bars) groups (n = 144). (B) The choice to cooperate is the dependent variable in a Probit regression with clustered standard errors on the participant level. Coefficients represent marginal effects on the probability to cooperate. Note that ***, **, * denote significance at the 1%, 5% and 10% level.