Autistic Traits Predict Weaker Sensitivity to Reward in Emotion Perception

Background:

When we look at a person’s face and try to infer what she is feeling, our interpretation is usually influenced by context (Barrett et al., 2011). One contextual factor that plays a significant role is utility, i.e. how useful it is to us to detect emotional changes. Recent advances in perception research indicate that individuals with autism spectrum disorder (ASD) show basic differences in how they process sensory cues. A common Bayesian explanation for these differences is that those with ASD rely more on immediate sensory information, while they appear to be less influenced by prior perceptual expectations (Pellicano & Burr, 2014). The forming of prior perceptual expectations could also be the psychological mechanism through which context modulates emotion perception. Yet it is unclear whether ASD is associated with a lower tendency to integrate context into emotion judgments.

Objectives:

The purpose of this study is to examine whether ASD-like traits affect how much individuals rely on personal utility when making emotion judgments. Using computational modeling we estimate how changes in facial cues and utility contribute to response bias in emotion judgments, and how these estimates are represented in brain activation.

Methods:

Forty-five healthy adults completed an emotional signal detection task while undergoing functional magnetic resonance imaging (FMRI). Participants were told a fictional background story in which the protagonist was just saying goodbye to his girlfriend for a day (neutral outcome) or for a year (sad outcome). Their task was to guess the emotional outcome based on facial expressions of the protagonist, morphed to convey varying degrees of emotional intensity. Utility was manipulated by offering rewarding and punishing payoffs for different response strategies, incentivizing in turn more liberal or more conservative response biases. After each trial, participants received feedback on their performance and the amount of money they gained or lost. We developed a computational model that extends a neural network model of criterion learning (Helie, 2014) with signal detection theory and prospect theory. This model estimates individuals’ sensitivity to perceptual information, sensitivity to reward, and reward bias. We regressed participants’ autistic traits as assessed by the autism spectrum quotient (AQ) against these parameter estimates. For FMRI analysis we entered them as parametric modulators into FMRI analysis.

Results:

Autistic traits predicted sensitivity to reward and reward bias, but not sensitivity to perceptual information. This effect was negative, with increasing autistic traits leading to a weaker role of reward. Reward sensitivity modulated brain activation in the striatal region. Stimulus sensitivity modulated brain activation in ventromedial prefrontal regions, striatal regions, and posterior midcingulate cortex.

Conclusions:

In line with our assumptions, ASD-like traits were associated with a lower reliance on utility. This could be a potential route to explaining why individuals with ASD often experience difficulties with perceiving emotional expressions. A weaker integration of reward context could signify that their interpretation of emotional faces is less flexibly adapted to situational needs. Emotional expressions are often embedded into a complex environment, which may make their interpretation particularly challenging for those with ASD.