Using Eye-Tracking to Investigate Social Information Processing (SIP) in Autism Spectrum Disorders (ASD)

Background: A common characteristic of autism spectrum disorder (ASD) is deficits in social information processing. For effective interactions and communication, it is crucial to assess socio-emotional information from faces and to correctly interpret facial expressions in real-time. Tracking an individual’s eye movements as they view a facial expression can offer insight into how an individual is processing relevant information. Using eye-tracking, we can measure location and duration of gaze, which allows us to assess subcomponents of social information processing in individuals.

Objectives: This study investigated differences in accuracy of emotion identification and eye-tracking metrics (locations and durations) during the Dynamic Affect Recognition Evaluation task (DARE; Bal et al, 2010; affects: happy, sad, anger, fear, surprise, disgust) in both individuals with and without ASD.

Methods: Two groups of participants were recruited to perform DARE. Fifteen individuals (mean age: 13.64±2.33y; IQ>70) diagnosed with ASD (ADOS and most had ADI-R) and 19 individuals (mean age: 12.74±1.97y) without diagnoses. To examine differences in accuracy, trial duration, and fixation between and within groups, different size ANOVAs (5x2, 1x6, and 3x2) followed by post hoc analysis with corrections were performed. Kendall’s tau-b was used to note correlations between accuracy and eye metrics.

Results: Preliminary results suggest that ASD versus control differences (between group analysis) in accuracy of emotion recognition during adolescence are not significant. Significant differences in trial duration between the groups were observed. Individuals with ASD had longer trial durations for happy, surprise, fear, and anger (all: p<0.001) as compared to controls. Examining the eye-tracking data, we noted that controls on average had longer fixation durations on areas of interest (AOIs) in comparison to the ASD group across all emotions (all: p<0.001).

Within group analysis noted a negative correlation between accuracy on disgust and fixation duration on the mouth (τb= -0.579, p=0.009) in the ASD group. Analysis also yielded significant differences in accuracy between: happy > anger (p=0.005), happy > disgust (p=0.002), surprise > anger (p=0.038), and surprise > disgust (p=0.014). Within the group, trial duration was significantly shorter for happy versus anger (p=0.009). This group also had a longer mean fixation duration on the eyes in disgust versus happy (p=0.029). Analysis within the control group yielded a significant difference in accuracy, happy > fear (p=0.016). Trial duration did not significantly differ between emotions in the control group.

Conclusions: While there were no significant differences between groups in the accuracy of identifying emotion, there were differences in the strategies between the groups – where individuals looked (location) and for how long (duration). This suggests that by adolescence individuals with ASD establish a strategy that may be different from non-ASD individuals, but still allows for accurate emotion recognition. Going forward, we will increase sample size to ensure that the results are robust, examine eye movements between AOIs by analyzing frequency of saccades, and investigate the effect of age on task strategy.