Stronger Pupil Light Response Is Associated with Fewer Autism Traits in College Students

Background: The pupil light reflex (PLR) is a well-studied index of autonomic nervous system (ANS) functioning, and children with ASD show a slower and weaker PLR as compared to controls (e.g., Daluwatte et al., 2013; Fan et al., 2009). Recent work by DiCriscio and Troiani (2017) extended this to find that pupil adaptation responses to light were related to autism traits in a broad population of children.

Objectives: The present study examined associations between PLR measures and autism-related traits in adults.

Methods: Participants included 51 college students. An SMI RED 120Hz eye-tracker captured pupil data during a PLR paradigm adapted from Nyström et al. (2015). Participants saw nine PLR trials, each starting with a small animated shape in the center of a black screen appearing for 1.6 to 2.4 seconds, followed by a 120-ms white flash. Later in the session, participants completed the Social Responsiveness Scale, Second Edition (SRS-2, Constantino & Gruber, 2012), a measure that has been used to assess autism traits in the broader population (e.g., DiCriscio & Troiani, 2017). Custom Python scripts were used to process each trial's dilation time series to find latency to PLR onset (peak negative acceleration) and relative constriction. Pupil diameter was processed using a degree-2 Savitzky-Golay filter with a window of 11 samples to yield smoothed diameter and acceleration series, which were then further smoothed using a Gaussian convolution with a standard deviation of 5 samples. Relative pupil constriction was calculated from D₀ (baseline diameter) and D_m (minimum diameter) as (D₀ - D_m)²/D₀². All participants had six or more usable trials (i.e., trials with clean data from both eyes), and analyses focused on the PLR variables used in Nyström et al. (2015): mean constriction and median latency.

Results: Correlations were run between PLR measures and SRS-2 Total score. A significant negative correlation was found between mean constriction and SRS-2 Total score, r(49) = -.279, p = .048, with stronger constriction associated with fewer autism traits. The SRS-2 Social Communication and Interaction (SCI) and Restricted Interests and Repetitive Behaviors (RRB) subscales were also examined alongside PLR measures, and results showed that RRB was negatively correlated with mean constriction (r(49) = -.380, p = .006, see Figure 1), and positively correlated with median latency (r(49) = .322, p = .021, see Figure 2), while SCI showed no relations with PLR (ps > .1). This suggests that a stronger and faster PLR is associated with fewer RRB traits.

Conclusions: Past work has found a less robust PLR in individuals with ASD, and Nyström et al. (2018) found that the PLR was highly predictive of later ASD in a group of high-risk infants. The current study is in line with work by DiCriscio and Troiani (2017) in finding that pupil light responses relate to ASD traits in a broad population of adults. The strongest associations in the current study were found with RRB, supporting previous studies that suggest that ANS difficulties could play an important role in RRB (e.g., Condy et al., 2017).