Visual Scanning Patterns of School-Age Children Completing an Adapted Reading the Mind in the Eyes Test

Background: Impairments in mentalizing represent a key deficit in social cognition for children with autism spectrum disorder (ASD). Explicit measures of mentalizing abilities, such as the Reading the Mind in the Eyes Test (RMET), generally show between-group differences with lower performance in individuals with ASD compared to typically-developing controls. However, some individuals with ASD do achieve high accuracy on explicit mentalizing measures, often representing a discrepancy from continued difficulty in applying mentalizing skills within everyday social contexts. One hypothesis is that individuals with ASD may utilize compensatory strategies, such as relying on verbal reasoning skills, to solve mentalizing problems in a different way than typically-developing peers. To help understand the underlying mechanisms of children’s mentalizing performance, we developed a computer-based version of the RMET co-registered with eye-tracking. By examining visual scanning patterns as participants completed the RMET, we were able to measure not just whether individuals were accurate in social attributions, but also how individuals were attempting to make social attributions on a moment-by-moment basis.

Objectives: The aims of the current study were (1) to compare performance on a computerized version of the RMET to autism severity metrics and (2) to identify visual scanning patterns predictive of RMET performance and severity of autism symptomatology.

Methods: Participants included 17 children with ASD (15 male, 2 female) between the ages of 8 and 11 years old. Children were verbally fluent (Full Scale IQ: mean=100.7 [17.5]) and represented a broad range of level of social disability (ADOS-2 Calibrated Severity Score: mean=6.9 [2.4], range=1 to 10). Participants were enrolled in a randomized clinical trial of a combination behavioral-pharmacological treatment targeting social cognition and social behavior; the current study focuses only on data at baseline, pre-treatment.

Eye-tracking data were collected while participants completed an adapted, computerized version of the child RMET (Figure 1). Percentage of visual fixation time on the eyes image and on word response options was calculated for each child. In addition, visual scanning patterns were evaluated using more temporally- and spatially-sensitive measures of saccade frequency and directionality.

Results: Performance on the RMET ranged from 25 to 90% correct responses. Greater accuracy on the RMET was correlated with reduced autism symptomatology as assessed both by clinician rating on the ADOS-2 (r=-0.66, p<0.005) and parent rating on the SRS-2 (r=-0.52, p<0.05). RMET performance was also positively correlated with higher cognitive functioning (r=0.56, p<0.05). Participants who exhibited higher visual attention to the eyes image of RMET stimuli and lower visual attention to word response options tended to have less severe symptomatology (eyes: r=0.62, p<0.05, words: r=-0.53, p<0.05). There was no significant relationship between visual attention measures and cognitive functioning. Ongoing analyses are examining visual scanning patterns that may mediate the relationship between cognitive functioning and RMET performance.

Conclusions: The adapted, computerized RMET yielded a broad range of mentalizing performance and was predictive of social functioning in our pilot sample of school-age children with ASD. By evaluating both mentalizing accuracy and underlying mechanisms, this new approach holds promise as a more sensitive measure of social cognition.