Successful social interactions depend on our ability to understand others' mental states and conform to social norms. While functional neuroimaging has identified a 'social brain' network, these methods lack the temporal resolution to study the fast dynamics of neuromodulators like dopamine and serotonin, key regulators of social behavior. Prior research using pharmacology and economic games like the ultimatum game (a 'take-it-or-leave-it' game assessing fairness norms) has shown causal links between these neuromodulators and social behavior. Lowering serotonin leads to increased rejection of unfair offers, while increasing dopamine reduces aversion to harming others for financial gain. However, pharmacology lacks the temporal precision to capture the fast fluctuations of these neurotransmitters. Animal studies using invasive methods have demonstrated that rapid dopamine fluctuations reflect reward prediction errors (RPEs) – the difference between expected and actual rewards – crucial for learning and behavioral adaptation. Similarly, serotonin fluctuations are involved in coding both social and non-social rewards. Yet, animal models cannot fully replicate the complexity of human social interaction. This study aimed to bridge this gap by using a novel technique called human electrochemistry to measure fast dopamine and serotonin fluctuations in the human brain during a social task.

The roles of dopamine and serotonin in human social interaction have been explored primarily through pharmacological manipulations combined with economic games, particularly the ultimatum game. Studies manipulating serotonin levels during the ultimatum game have demonstrated that reduced serotonin levels increase rejection of unfair offers, while increased serotonin levels decrease rejection. Similarly, studies involving dopamine manipulations have shown that increased dopamine levels can lead to increased selfishness and reduced aversion to causing harm to others for monetary gain. These studies, while informative, have limitations in temporal resolution. Animal studies have provided more detailed insights into the rapid dynamics of dopamine and serotonin, showing that fast dopamine fluctuations reflect reward prediction errors (RPEs), which are crucial for adaptive behavior in social contexts. Transient serotonin changes have been linked to coding of social and non-social rewards. However, translating findings from animal models to human social interaction remains a challenge.

This study used human electrochemistry, a novel technique that allows sub-second measurements of neuromodulator fluctuations in conscious human brains during awake brain surgery. Parkinson's disease patients (n=4) undergoing deep brain stimulation (DBS) surgery for movement disorders participated. Electrochemical recordings of dopamine and serotonin were obtained from the substantia nigra pars reticulata (SNr), a key basal ganglia output structure. Participants played the ultimatum game as responders, interacting with either human or computer avatars as proposers. Each session included 30 trials per condition (human/computer), blocked and counterbalanced across two surgical sessions. Offers ranged from $1 to $9. On approximately one-third of trials, participants rated their mood. A carbon-fiber electrode was inserted into the SNr to measure electrochemical signals. These signals were analyzed using a deep convolutional neural network signal prediction model trained on extensive in vitro datasets, providing sub-second estimates of dopamine and serotonin concentrations. Behavioral data (acceptance/rejection of offers, reaction time, mood ratings) and electrochemical data were analyzed using linear and logistic mixed-effects models at the trial level.

Participants rejected significantly more offers from human than computer avatars, replicating previous findings. This behavioral effect was associated with higher overall levels of dopamine in the human condition, but not serotonin. Crucially, analysis of relative changes in dopamine (compared to a local baseline) revealed that dopamine tracked trial-by-trial changes in offer value, consistent with reward prediction error (RPE) signaling. This RPE signaling was not modulated by the social context (human vs. computer). In contrast, relative changes in serotonin tracked the current offer value regardless of the previous offer or the social context, indicating a role in current value evaluation. The statistical analyses employed mixed-effects models which included fixed effects (population-level) and random effects varying by dataset (session) with a free covariance matrix. The reported effects remained significant even when removing all random effects except for the intercept. The logistic mixed-effects model analysis of choices (accept/reject) showed that offer value positively influenced choices, with participants more likely to accept higher offers. Condition (human/computer) negatively affected choices, with more rejections of human offers. There was no interaction effect between value and condition. Reaction times were longer for rejected than accepted offers. Emotion ratings did not reveal any significant effects. The linear mixed-effects model analysis of overall dopamine/serotonin levels revealed a significant effect of condition on dopamine, with higher dopamine levels in the human condition. However, there were no effects of choice or the interaction between choice and condition on dopamine. There were no effects of condition, choice, or the interaction between condition and choice on serotonin. The analysis of relative dopamine/serotonin changes demonstrated that dopamine tracked the difference in value between current and previous offers (RPE), while serotonin tracked the current offer value.

This study provides direct evidence from the human brain that fast dopamine and serotonin fluctuations reflect distinct aspects of social decision-making. The elevated dopamine levels in the human condition support the hypothesis that dopamine contributes to the overall motivational context influencing fairness judgments. Dopamine's role in RPE signaling during the ultimatum game aligns with previous animal research showing a similar role in value-based learning. Serotonin's focus on current offer value suggests a role in immediate value evaluation. The findings underscore the complementary roles of dopamine and serotonin in value-based decision-making, with dopamine involved in comparing current and past values and serotonin focused on the immediate value. The results generalize across social contexts, suggesting that the observed computational roles extend beyond specific social interactions. The study was conducted in Parkinson’s disease patients undergoing DBS surgery; this raises the question of generalizability to healthy brains, but several factors suggest that the results are likely to generalize.

This research demonstrates, using human electrochemistry, that dopamine and serotonin in the human SNr reflect distinct social context and value signals during economic exchange. Dopamine tracks reward prediction errors, while serotonin tracks current value. These findings highlight the complementary roles of these neuromodulators in social decision-making and suggest avenues for future research investigating their interplay in more complex social interactions, using repeated interactions and sophisticated inference tasks.

The study used a relatively small sample size (n=4 patients) due to the invasive nature of the recording method. The recording site was limited to the SNr, which might not fully capture the neural processes involved in social decision-making. The study population consisted of Parkinson's disease patients undergoing DBS surgery, potentially limiting the generalizability of the findings to healthy individuals. Future studies should address these limitations by using larger, more diverse populations and recording from multiple brain regions.