The Neural Substrates of Social Influence on Decision Making

Damon Tomlin, Andrea Nedić, Deborah A. Prentice, Philip Holmes, Jonathan D. CohenPLoS 2013 · DOI

Abstract

The mechanisms that govern human learning and decision making under uncertainty have been the focus of intense behavioral and, more recently, neuroscientific investigation. Substantial progress has been made in building models of the processes involved, and identifying underlying neural mechanisms using simple, two-alternative forced choice decision tasks. However, less attention has been given to how social information influences these processes, and the neural systems that mediate this influence. Here we sought to address these questions by using tasks similar to ones that have been used to study individual decision making behavior, and adding conditions in which participants were given trial-by-trial information about the performance of other individuals (their choices and/or their rewards) simultaneously playing the same tasks. We found that bilateral insula exhibited a parametric relationship to the degree of misalignment of the individual's performance with those of others in the group. Furthermore, activity in the bilateral insula significantly predicted participants' subsequent choices to align their behavior with others in the group when they were misaligned either in their choices (independent of success) or their degree of success (independent of specific choices). These findings add to the growing body of empirical data suggesting that the insula participates in an important way in social information processing and decision making.

Introduction

While neuroscientists have a robust understanding of how individuals learn and make decisions in isolation—often through the lens of reinforcement learning—the mechanisms by which social information alters these processes remain less clear. Previous research into social decision-making often utilized complex economic games or interactions with fictitious agents, making it difficult to isolate the specific neural systems integrating social data into basic choice behavior. This study bridges that gap by adapting a standard two-alternative forced choice (TAFC) task to include real-time social information from actual peers.

Methods and Experimental Design

The researchers recruited 115 participants, with 86 individuals undergoing fMRI scanning. Participants were organized into groups of five, playing a series of decision-making games where they chose between two options ("A" or "B") to earn rewards. The experiment utilized a 6x4 factorial design, crossing six reward functions with four social information conditions:

  1. No Information: Participants only observed their own rewards.
  2. Choices: Participants saw the button choices of their four peers.
  3. Rewards: Participants saw the points earned by their peers.
  4. Both: Participants saw both the choices and rewards of their peers.
The four experimental conditions used to test the influence of social information on decision-making.

To analyze the data, the authors employed Hierarchical Logistic Regression, determining if brain activity (BOLD signal) could predict a participant's decision to "switch" buttons on the next trial beyond what behavioral data alone could predict.

Key Insights: The Insula as a Social Thermostat

The study identified the bilateral insula as a key neural hub for processing social influence. The insula showed a "parametric relationship" with group alignment: as a participant's choice or reward level deviated further from the group (marginality), insula activity increased. This suggests the insula acts as a detector for "social discomfort" or prediction errors regarding one's standing within a group.

Crucially, this activity was not merely a passive reflection of difference. High BOLD signals in the insula on one trial significantly predicted that the participant would "switch" their choice on the following trial to align with the group. Interestingly, the insula responded to misalignment in choices even when those choices did not affect the participant's actual reward, suggesting a neural drive for conformity distinct from profit maximization.

Results and Statistical Significance

  • Negative Correlation with Alignment: In the "Choices" condition, activity in the insula, dorsal anterior cingulate cortex (dACC), and dorsolateral prefrontal cortex (DLPFC) was negatively correlated with group alignment ($p < 0.05$, corrected).
  • Model Improvement: Adding insula BOLD activity to the behavioral model significantly improved the prediction of "switching" behavior. In the "Choices" condition, the model including insula activity accounted for the most variance ($p < 10^{-14}$).
  • Consistency: The insula remained the strongest predictor of behavior across the "Choices," "Rewards," and "Both" conditions, outperforming other regions like the thalamus or dACC.

Conclusion and Future Directions

These findings suggest that the insula plays a vital role in social information processing and decision-making by signaling misalignment with a group. While the study identifies these neural substrates, it does not yet provide a full computational model incorporating social weights, which remains a goal for future research. Additionally, while the TAFC task provides a controlled environment, future work could explore if these same insula-driven mechanisms hold true in more complex, high-stakes social environments.

Summary updated

Abstract

The mechanisms that govern human learning and decision making under uncertainty have been the focus of intense behavioral and, more recently, neuroscientific investigation. Substantial progress has been made in building models of the processes involved, and identifying underlying neural mechanisms using simple, two-alternative forced choice decision tasks. However, less attention has been given to how social information influences these processes, and the neural systems that mediate this influence. Here we sought to address these questions by using tasks similar to ones that have been used to study individual decision making behavior, and adding conditions in which participants were given trial-by-trial information about the performance of other individuals (their choices and/or their rewards) simultaneously playing the same tasks. We asked two questions: How does such information about the behavior of others influence performance in otherwise simple decision tasks, and what neural systems mediate this influence? We found that bilateral insula exhibited a parametric relationship to the degree of misalignment of the individual's performance with those of others in the group. Furthermore, activity in the bilateral insula significantly predicted participants' subsequent choices to align their behavior with others in the group when they were misaligned either in their choices (independent of success) or their degree of success (independent of specific choices). These findings add to the growing body of empirical data suggesting that the insula participates in an important way in social information processing and decision making.
BibTeX
@inproceedings{tomlin2013neural,
  author = {Damon Tomlin and Andrea Nedić and Deborah A. Prentice and Philip Holmes and Jonathan D. Cohen},
  title = {{The Neural Substrates of Social Influence on Decision Making}},
  booktitle = {PLoS},
  year = {2013},
  doi = {10.1371/journal.pone.0052630},
}

Loading related content...

Figure