Humans use forward thinking to exploit social controllability

Soojung Na, Dongil Chung, Andreas Hula, Ofer Perl, Jennifer Jung, Matthew Heflin, Sylvia Blackmore, Vincenzo G. Fiore, Peter Dayan, Xiaosi GueLife 2021 · DOI

TL;DR

Humans actively use "forward thinking" – mentally simulating future outcomes of their current actions – to understand and leverage social controllability, even when that control is illusory. This cognitive strategy is implemented in the ventromedial prefrontal cortex (vmPFC), suggesting a broader role for this brain region in social decision-making than previously understood.

What Problem Does This Paper Address?

We all experience varying degrees of control over our environments, and this sense of agency profoundly impacts our mental health and behavior. While the importance of controllability is well-established in non-social contexts (like learning to avoid a shock), the specific neurocomputational mechanisms by which humans understand and exploit social controllability remain unclear. This paper aims to bridge that gap by investigating how people use their ability to predict and influence the actions of others in social interactions.

Methods

The researchers employed a multi-pronged approach:

  • Behavioral Experiments: Participants engaged in a modified ultimatum game. In one condition ("Controllable"), their decisions (accept/reject an offer) directly influenced the partner's subsequent offers. In another ("Uncontrollable"), offers were randomized, and participants had no influence. This was conducted in both a lab setting (fMRI) and a large-scale online study.
  • Computational Modeling: They developed and tested "forward-thinking" (FT) models that simulated future interactions up to four steps ahead. These models aimed to capture how participants might estimate the value of their choices by considering their downstream effects. A key parameter, "expected influence" ($\delta$), quantified how much participants believed their actions would change future offers.
  • Neuroimaging (fMRI): Functional magnetic resonance imaging was used to identify brain regions involved in processing these simulated values. Specifically, they looked for neural correlates of the FT model's predicted values, contrasting it with simpler models that didn't incorporate future planning.

Datasets

  • fMRI Sample: 48 healthy adult participants (33 female, age 27.3±9.2 years) from the Dallas-Fort Worth metropolitan area.
  • Online Sample: 1342 participants recruited via Prolific, a more demographically diverse group.

Key Insights and Ideas

  • Forward Thinking is Key to Social Control: The most striking finding is that people don't just react to immediate social cues; they actively simulate future interactions to gauge their influence. This "forward thinking" is crucial for understanding and exploiting social controllability.
  • The vmPFC as a Social Value Integrator: The ventromedial prefrontal cortex (vmPFC) appears to be the neural hub for this process. It doesn't just represent current rewards but integrates the projected future values of actions, effectively computing a "total value" for social decisions.
  • Illusory Control is Real and Persistent: Even when participants were explicitly in an "Uncontrollable" condition, their behavior and computational models suggested they believed they had some influence. This highlights a tendency to perceive control, even when it's not objectively present, and this illusory belief is linked to vmPFC activity.
  • Social Context Matters for Subjective Beliefs: While the forward-thinking mechanism was similar in social and non-social tasks, participants' subjective reports of controllability differed significantly between the two. This suggests that the social nature of an interaction uniquely shapes our perception of control.

Key Results

  • Behavioral Differences: In the "Controllable" condition, participants successfully increased offers over time and were more likely to reject medium to high offers, strategically leveraging their influence.
  • Model Fit: A 2-step forward-thinking model provided a significantly better account of participants' choices than simpler models that didn't incorporate future planning, even in the "Uncontrollable" condition.
  • Neural Evidence: The vmPFC showed significantly greater activity when tracking the predicted total values from the 2-step FT model compared to a model without future planning, in both controllable and uncontrollable conditions.
  • Replication: The core behavioral and computational findings were robustly replicated in a large-scale online study (n=1342).
  • Expected Influence: Participants' estimated "expected influence" ($\delta$) was higher in the Controllable condition and positively correlated between conditions, suggesting a stable individual trait.

Limitations and Future Work

  • Instructional Ambiguity: The authors acknowledge that the lack of explicit instructions about controllability might have influenced the extent to which individuals exploited control or developed illusions of control. Future studies could use explicit instructions to further probe these effects.
  • Uncertainty vs. Controllability: The study design inherently links uncertainty reduction with controllability. Future experiments could aim to dissociate these factors to better understand their independent effects on behavior and neural responses.
  • Learning Mechanisms: The paper suggests that understanding how individuals learn the "expected influence" parameter ($\delta$) in early trials, or build more complex models of others' minds (cognitive hierarchy), would be valuable future directions.
  • Clinical Implications: The findings suggest that aberrant computation of social controllability could contribute to mental health issues and difficulties in social cooperation, opening avenues for clinical research.

Summary updated

Abstract

The controllability of our social environment has a profound impact on our behavior and mental health. Nevertheless, neurocomputational mechanisms underlying social controllability remain elusive. Here, 48 participants performed a task where their current choices either did (Controllable), or did not (Uncontrollable), influence partners' future proposals. Computational modeling revealed that people engaged a mental model of forward thinking (FT; i.e., calculating the downstream effects of current actions) to estimate social controllability in both Controllable and Uncontrollable conditions. A large-scale online replication study (n=1342) supported this finding. Using functional magnetic resonance imaging (n=48), we further demonstrated that the ventromedial prefrontal cortex (vmPFC) computed the projected total values of current actions during forward planning, supporting the neural realization of the forward-thinking model. These findings demonstrate that humans use vmPFC-dependent FT to estimate and exploit social controllability, expanding the role of this neurocomputational mechanism beyond spatial and cognitive contexts.
BibTeX
@inproceedings{na2021humans,
  author = {Soojung Na and Dongil Chung and Andreas Hula and Ofer Perl and Jennifer Jung and Matthew Heflin and Sylvia Blackmore and Vincenzo G. Fiore and Peter Dayan and Xiaosi Gu},
  title = {{Humans use forward thinking to exploit social controllability}},
  booktitle = {eLife},
  year = {2021},
  doi = {10.7554/elife.64983},
}
Figure