Opposing BOLD responses to reciprocated and unreciprocated altruism in putative reward pathways

James K. Rilling, Alan G. Sanfey, Jessica A. Aronson, Leigh E. Nystrom, Jonathan D. CohenNeuroreport 2004 · DOI

TL;DR

This study uses fMRI to investigate how the brain processes social cooperation and betrayal. Researchers found that the brain's reward system shows opposing responses: a positive signal for reciprocated cooperation and a negative signal for unreciprocated cooperation. This suggests our brains are wired to learn who to trust, guiding future social interactions.

What Problem Does This Paper Address?

Humans are social creatures who often engage in reciprocal altruism – helping others with the expectation of being helped in return. However, navigating these social exchanges is complex because we can't always predict if our partners will reciprocate. This raises a fundamental question: how does our brain learn to distinguish between reliable social partners and "cheaters" to optimize future interactions? Previous research suggested that areas rich in dopamine, known for reward processing and learning, are involved in mutual cooperation. This paper aims to specifically test the hypothesis that these dopamine-rich regions signal errors in predicting social reciprocity, akin to how they signal errors in predicting monetary rewards.

Methods

The researchers employed functional Magnetic Resonance Imaging (fMRI) to measure brain activity in 19 participants. Participants played a series of single-shot Prisoner's Dilemma games with partners who were, unbeknownst to the participants, computer-controlled to either reciprocate cooperation or not. The Prisoner's Dilemma is a classic game theory scenario where players can choose to cooperate or defect, with payoffs structured to create a tension between individual self-interest and mutual benefit.

The fMRI data was analyzed using a general linear model to identify brain regions showing differential Blood-Oxygen-Level-Dependent (BOLD) responses to two key conditions:

  1. Reciprocated Cooperation (CC): The participant cooperated, and their partner also cooperated.
  2. Unreciprocated Cooperation (CD): The participant cooperated, but their partner defected.

The study focused on specific regions of interest (ROIs) known to receive mesencephalic dopamine projections: the ventral striatum and the ventromedial prefrontal cortex (VMPFC).

Datasets

Not discussed in the paper. The study used data collected from 19 human participants during an fMRI experiment.

Key Insights and Ideas

  • Opposing Signals for Trust and Betrayal: The most striking finding is the distinct neural response to reciprocated versus unreciprocated cooperation. The brain's reward pathways (ventral striatum and VMPFC) show a positive BOLD response when cooperation is met with cooperation, and a negative BOLD response when cooperation is met with defection.
  • Dopamine as a Social Prediction Error Signal: This pattern aligns with the theory that midbrain dopamine neurons signal "reward prediction errors." In this social context, a positive error occurs when a cooperative act is reciprocated (a better-than-expected outcome), and a negative error occurs when it's not (a worse-than-expected outcome).
  • Learning Who to Trust: The study suggests that these opposing neural signals are crucial for learning social contingencies. By experiencing positive feedback for trusting reliable partners and negative feedback for trusting unreliable ones, individuals can adapt their behavior to seek out beneficial social interactions and avoid detrimental ones.
  • The Uniqueness of Human Interaction: Interestingly, the positive BOLD response to reciprocated cooperation was specifically observed when the partner was perceived as human, not when a computer reciprocated. This highlights the special value placed on genuine human social connection.

Key Results

  • Ventral Striatum and VMPFC Activation: Both the ventral striatum and VMPFC showed a significant positive BOLD response to reciprocated cooperation (CC) from a human partner.
  • Ventral Striatum and VMPFC Deactivation: Both regions showed a significant negative BOLD response to unreciprocated cooperation (CD) from a human partner.
  • Specificity of Positive Response: The positive BOLD response to CC was unique to interactions with perceived human partners. CC with computer partners or a $5 reward in a non-social control task did not elicit a significant positive response.
  • Negative Response to Non-Reward: Negative BOLD responses were also observed in the striatum for CD outcomes with computer partners and for $0 control outcomes, suggesting a general response to the absence of an expected reward, not solely tied to social betrayal.

Limitations and Future Work

  • Single-Shot Game: The study used a single-shot Prisoner's Dilemma, meaning participants played each game only once with each partner. This limits the ability to observe how repeated interactions and learning might shape these neural responses over time. Future work could explore iterated versions of the game.
  • Perceived vs. Actual Partner: While participants believed they were playing with humans, the partners were computer-controlled. The study acknowledges this but suggests the perceived social interaction is key. Future research could directly compare brain responses when participants know they are playing with a computer versus a human.
  • Nature of "Error": The paper focuses on prediction errors related to reciprocity. Further research could explore how other social factors, like fairness or intentionality, might modulate these reward signals.
  • Generalizability: The study involved a relatively small sample size (n=19). Replicating these findings with larger and more diverse populations would strengthen the conclusions.

Summary updated

Abstract

Mesencephalic dopamine neurons are believed to facilitate reward-dependent learning by computing errors in reward predictions. We used fMRI to test whether this system was activated as expected in response to errors in predictions about whether a social partner would reciprocate an act of altruism. Nineteen subjects received fMRI scans as they played a series of single-shot Prisoner's Dilemma games with partners who were outside the scanner. In both ventromedial prefrontal cortex and ventral striatum, reciprocated and unreciprocated cooperation were associated with positive and negative BOLD responses, respectively. Our results are consistent with the hypothesis that mesencephalic dopamine projection sites carry information about errors in reward prediction that allow us to learn who can and cannot be trusted to reciprocate favors.
BibTeX
@inproceedings{rilling2004opposing,
  author = {James K. Rilling and Alan G. Sanfey and Jessica A. Aronson and Leigh E. Nystrom and Jonathan D. Cohen},
  title = {{Opposing BOLD responses to reciprocated and unreciprocated altruism in putative reward pathways}},
  booktitle = {Neuroreport},
  year = {2004},
  doi = {10.1097/00001756-200411150-00022},
}
Figure