Contagion of Temporal Discounting Value Preferences in Neurotypical and Autistic Adults

Louisa Thomas, Patricia Lockwood, Mona M. Garvert, Joshua H. BalstersJournal of Autism and Developmental Disorders 2021 · DOI

Abstract

Neuroeconomics paradigms have demonstrated that learning about another's beliefs can make you more like them (i.e., contagion). Due to social deficits in autism, it is possible that autistic individuals will be immune to contagion. We fit Bayesian computational models to a temporal discounting task, where participants made decisions for themselves before and after learning the distinct preferences of two others. Two independent neurotypical samples (N = 48; N = 98) both showed a significant contagion effect; however the strength of contagion was unrelated to autistic traits. Equivalence tests showed autistic (N = 12) and matched neurotypical (N = 12) samples had similar levels of contagion and accuracy when learning about others. Despite social impairments being at the core of autistic symptomatology, contagion of value preferences appears to be intact.

Introduction

Can you "catch" someone else’s financial habits? In neuroeconomics, "social contagion" refers to the phenomenon where observing another person’s preferences shifts your own internal value representations. A primary area of study is temporal discounting (TD)—the tendency to devalue rewards the further they are in the future (e.g., choosing £10 today over £15 in a month).

Because autism spectrum conditions (ASC) are characterized by difficulties in social cognition and "Theory of Mind," researchers hypothesized that autistic individuals might be less susceptible to this form of social influence. This study investigated whether social differences in autism create an "immunity" to the contagion of value preferences, or if the underlying cognitive mechanisms for integrating social information remain intact.

Methods

The researchers conducted three studies using a Bayesian computational framework to track how participants' internal "discount rates" changed after social observation.

  1. The Task: Participants performed a TD task choosing between small immediate rewards and larger delayed rewards. This was interleaved with blocks where they predicted the choices of two "simulated agents"—one more impulsive and one more patient than the participant.
  2. Computational Modeling: The authors fitted a log-hyperbolic model to the choices, focusing on the $log\ k$ parameter (the discount rate).
  3. Measuring Contagion: Rather than simple choice shifts, the study used Kullback–Leibler divergence ($D_{KL}$) to measure the "distance" between a participant's belief distribution before and after observing an agent. This quantified the change in both preference and certainty.
Experimental task design showing self-choice and other-prediction blocks.

Key Results

Across all three studies, participants showed a clear and significant contagion effect. Notably, this effect was not symmetrical; participants in all groups were significantly more likely to shift their preferences toward a "patient" agent (someone willing to wait for more money) than toward an "impulsive" one. This suggests a inherent social bias toward perceived self-control.

Divergence scores across three studies showing significant contagion, particularly for patient agents, in both neurotypical and autistic samples.

Crucially, the study found that autism is not an immunity to contagion. Autistic adults (Study 3) showed levels of contagion and learning accuracy nearly identical to neurotypical controls. Furthermore, within the neurotypical population, the level of autistic traits (measured by the Autism Spectrum Quotient) did not predict the strength of the contagion effect. This suggests that the mechanism for preference contagion is distinct from the social-communicative traits typically associated with autism.

Discussion and Conclusion

The findings suggest that while autistic individuals may struggle with implicit social cues, they are highly effective at integrating explicit social information into their decision-making processes. The use of explicit feedback (telling participants if their predictions were correct) may have bypassed traditional social intuition difficulties.

While the clinical sample size was small ($N=12$), Bayesian equivalence tests provided strong support for the similarity between groups. This research indicates that the cognitive architecture for social value contagion is robust and remains functional even in populations with significant social-communicative challenges. Future work should explore if these effects persist when social information is more ambiguous or presented without explicit feedback.

Summary updated

Abstract

Neuroeconomics paradigms have demonstrated that learning about another's beliefs can make you more like them (i.e., contagion). Due to social deficits in autism, it is possible that autistic individuals will be immune to contagion. We fit Bayesian computational models to a temporal discounting task, where participants made decisions for themselves before and after learning the distinct preferences of two others. Two independent neurotypical samples (N = 48; N = 98) both showed a significant contagion effect; however the strength of contagion was unrelated to autistic traits. Equivalence tests showed autistic (N = 12) and matched neurotypical N = 12) samples had similar levels of contagion and accuracy when learning about others. Despite social impairments being at the core of autistic symptomatology, contagion of value preferences appears to be intact.
BibTeX
@article{thomas2021contagion,
  author = {Louisa Thomas and Patricia Lockwood and Mona M. Garvert and Joshua H. Balsters},
  title = {{Contagion of Temporal Discounting Value Preferences in Neurotypical and Autistic Adults}},
  journal = {Journal of Autism and Developmental Disorders},
  year = {2021},
  doi = {10.1007/s10803-021-04962-5},
}
Figure