Psilocybin and MDMA reduce costly punishment in the Ultimatum Game

Human social decision-making can be disrupted across psychiatric disorders and is often studied with paradigms that probe interpersonal mechanisms such as trust, norm enforcement and cooperation. The Ultimatum Game (UG) is a widely used social decision-making task in which a responder may accept or reject monetary splits proposed by another player; rejection of low, unequal offers is commonly interpreted as altruistic or costly punishment. Prior psychopharmacological work implicates the serotonin (5-HT) system in UG responder behaviour: lowering 5-HT via acute tryptophan depletion increased rejection of moderately unfair offers, while increasing 5-HT with SSRIs reduced such rejections. Psilocybin and MDMA both have prominent serotonergic actions and have shown promise in treating mood and trauma-related disorders, but their effects on high-level social decision-making have not been fully characterised. Gabay and colleagues report two experimental studies that test whether psilocybin and MDMA alter UG behaviour. They hypothesised that both compounds would reduce rejection rates of unfair offers in a first-person (FP) responder condition, that MDMA would not change rejection in a third-party (TP) condition (where equality considerations rather than direct personal harm dominate), and that MDMA would increase offers made when participants acted as proposers. Both studies included a non-social control (random-generated offers) and used repeated-measures designs to probe these effects.

Study One (psilocybin) was embedded within a larger protocol and analysed as an open-label, within-participant comparison of a drug-free screening session versus a psilocybin session. Twenty-three male volunteers with prior psychedelic experience were recruited; 20 completed the study (mean age 26.6 years). The extracted text indicates analyses of UG responses were ultimately based on 19 participants due to one non-completion. On the psilocybin day, 2 mg of psilocybin was administered intravenously over two minutes during neuroimaging, and the UG was performed 60 minutes later when acute perceptual effects had diminished. The UG employed two offer-origin conditions: first-person (offers purportedly from other people) and random-generated (computer-generated offers). Offers (all from a £20 stake) comprised multiple instances of 10%, 20%, 30%, 40%, 50%, 80% and 90%; for analysis the authors grouped 10–20% as 'unfair', 50% as 'fair' and 80–90% as 'hyper-fair', excluding 30% and 40% from main analyses. The task presented 96 offers in total. Test-retest reliability of this UG version was assessed in an independent sample using the rptR package, yielding high repeatability (FP R = 0.82; random-generated R = 0.69). Categorical accept/reject data were analysed using repeated-measures logistic regression implemented with generalized estimating equations (GEE) in SPSS; reported p-values were Bonferroni-corrected. Study Two (MDMA) used a double-blind, placebo-controlled, counterbalanced crossover design. Twenty-one male volunteers with prior MDMA experience were recruited; 20 completed both sessions (mean age 24.8 years). Participants attended two experimental days at least one week apart; 100 mg MDMA or placebo was administered and the UG performed 95 minutes post-dose. This UG variant included three offer-origin conditions (first-person, third-party, random-generated), two runs per session, and varied stake sizes and proportions to examine utility effects. Each run presented 72 offers and asked participants to make five offers themselves; offers were classified for analysis as unfair (10–20%), fair (45–50%) and hyper-fair (80–90%). Test-retest reliability in an independent sample was high (FP R = 0.96; TP R = 0.96; random-generated R = 0.84). In addition to the UG, participants completed a reward sensitivity reaction-time task (to probe sensitivity to monetary reward) and two questionnaires: the Social Value Orientation (SVO) and the Social Reward Questionnaire (SRQ). Categorical UG responses were analysed with GEE logistic regression; offers made by participants were converted to percentages and compared across sessions with paired t-tests. Reaction-time data were analysed with mixed ANOVA. All reported p-values were Bonferroni-corrected.

Psilocybin study: Almost all fair and hyper-fair offers were accepted in both sessions, with one exception who rejected multiple hyper-fair offers in the psilocybin session. Analyses of unfair offers used data from 19 participants. There was a main effect of offer origin: unfair offers attributed to the random-generated source were less likely to be rejected than first-person offers (χ2(1,18) = 5.30, OR = 0.70, p = 0.039). Compared with the drug-free session, psilocybin reduced the probability of rejecting unfair first-person offers (χ2(1,18) = 5.53, OR = 0.48, p = 0.018) and also reduced rejection of unfair random-generated offers (χ2(1,18) = 8.63, OR = 0.30, p = 0.009). MDMA study: There were no effects of run or utility and no interactions of these factors with treatment, so subsequent analyses combined runs and stake sizes. The model showed no overall main effect of treatment, but main effects of offer origin and fairness and a significant three-way interaction. Post-hoc parameter estimates indicated that MDMA lowered the probability of rejecting unfair first-person offers (χ2(1,18) = 11.02, OR = 0.57, 95% CI 0.41–0.80, p = 0.001). The reduction in rejection in the third-party condition did not reach conventional statistical significance (OR = 0.68, 95% CI 0.45–1.01, p = 0.057). No treatment effect was observed for unfair offers from the game-server (random-generated; OR = 0.85, p = 0.413). Treatment order had no significant effect. When acting as proposers, participants offered more on MDMA than on placebo: mean offer was 48.2% on placebo versus 55.7% on MDMA (mean difference 7.5%, SD = 10.25; t(18) = 3.17, p = 0.006), with a large effect size (Cohen's d = 0.82). Reward sensitivity and questionnaires: In the reward sensitivity task participants responded faster on rewarded versus unrewarded trials, but there was no effect of MDMA on reaction times or reward sensitivity. On the SVO there was no evidence MDMA altered prosocial or egoistic orientation. On the SRQ, a treatment-by-subscale interaction was observed and post-hoc tests showed an increase in the prosocial interactions subscale after MDMA (t(19) = -3.19, p = 0.03). A post-hoc regression found that change in SRQ prosociality predicted change in rejection of unfair first-person offers (beta = -0.498, R2 = 0.248, p = 0.035).

Gabay and colleagues report that both psilocybin and MDMA reduced rejection of unfair offers in first-person UG conditions, and that MDMA increased the amount offered when participants acted as proposers. The MDMA-induced reduction in first-person rejections was associated with increases in self-reported prosocial interactions on the SRQ. The authors situate these findings within models that decompose costly punishment into inequality aversion, cost–benefit calculation and social reference frame, and introduce the notion of a 'social reward currency'—the subjective value assigned to punishing norm violations. They argue their data are most consistent with these compounds shifting that exchange rate so that punishing becomes relatively less rewarding while concern for the direct relationship with interaction partners gains salience. The discussion considers alternative accounts: pharmacologically, the findings align with prior serotonergic manipulations of UG behaviour, supporting a role for serotonin in modulating rejection. The broader pharmacology of MDMA—noradrenergic and dopaminergic effects in addition to serotonergic release—is acknowledged, and the authors note that receptor-specific blockade studies would help to clarify mechanisms. They also consider whether changes reflect altered cost sensitivity or loss aversion, but report no evidence from the reward sensitivity task that MDMA altered basic reward sensitivity. The third-party condition pattern broadly supported the hypothesis that MDMA's effects are more marked when participants are directly involved, though the third-party reduction narrowly missed significance and requires replication in larger samples. Key limitations are emphasised: the psilocybin study was open-label and not counterbalanced; even the placebo-controlled MDMA study faces blinding challenges because inert placebos do not effectively mask psychedelic or empathogenic effects. Both studies recruited only male participants with prior positive experience of the study drug, which limits generalisability and may introduce selection bias. Different UG versions and settings were used across studies, and dose–response relationships were not explored. The authors therefore call for replication, investigation of receptor mechanisms, and extension to more diverse samples. They conclude that their results characterise how psilocybin and MDMA alter social decision-making—likely via serotonergic modulation—and suggest relevance for understanding social cognition alterations in psychiatric conditions.