Atypical Conjunctive Visual Processing in ASD: Domain General or Face Specific?

Background: The human face is arguably the most socially-informative visual stimulus available, conveying relevant social information such as identity and emotional expressions. Individuals with autism spectrum disorder (ASD) have demonstrated difficulties with identifying faces in a variety of studies. Eye-tracking studies have suggested that individuals with ASD may process faces atypically, but it is still an open question as to whether these differences are with faces processing per se, or with conjunctive visual processing more generally.

Objectives:

1) Determine whether conjunctive processing in ASD varies across face and object stimulus types similarly to TD.

2) Determine if atypical conjunctive processing in ASD is face specific, or is more broadly observed, in this case across stimulus ambiguity.

Methods: We used eye-tracking to assess visual conjunctive processing in a 2x2 design, varying between face and object stimuli, and high- and low-ambiguity. Pairs of face and object stimuli were presented, while individuals reported whether faces/objects were the same or different. Stimuli were either high or low ambiguity. High-ambiguity objects shared 2/3 features, while low-ambiguity shared 0/3 features. Face identities were morphed, so that low-ambiguity pairs differed by 60%, and high-ambiguity pairs by 10% (Figure 1). Eye-gaze data was analyzed based on within-item saccades relative to between-item saccades. Higher proportions of within-item saccades are representative of greater conjunctive processing (Figure 1).

The paradigm was first conducted in TD individuals (n=20), with data related to autistic traits. Following a significant finding relating ASD traits and visual conjunctive processing, we recruited an additional 33 TD (n=53 total) and 10 ASD individuals (diagnosis confirmed via ADOS, data collection ongoing) ranging in ages from children to young adults to measure between-group differences in patterns of visual conjunctive processing.

Results: The initial study of TD adults (n = 20) showed that visual conjunctive processing was significantly correlated with autism traits severity (r = -0.38, p = 0.05) – the higher the level of autistic traits, the lower the level of conjunctive processing (Figure 2A). In the primary experiment, comparing conjunctive processing with low- and high-ambiguity faces and objects across TD and ASD, we observed a significant 3-way, group-by-stimulus-by- ambiguity interaction (Figure 2B-D; F_(1,61) = 4.33, p = 0.04, ƞ_p² = 0.07). A greater impact of ambiguity level was found for the TD group compared to individuals with ASD (F_(1,61) = 3.88, p = 0.05, ƞ_p² = 0.06). TD participants showed greater conjunctive processing with high- relative to low-ambiguity objects (t = 8.20, p < 0.001, d = 0.89) but showed no difference for faces, as predicted (Figure 2B). Autistic individuals, however, showed no differences across any of the four conditions (Figure 2C).

Conclusions: Results from this on-going study suggest gaze patterns during visual conjunctive processing are being influenced by autism traits. As face and object stimuli were processed similarly, our results indicate that individuals with ASD are employing a more general-purpose visual processing strategy. Together, the results suggest that further investigation is warranted to characterize the relationship between autism traits and visual conjunctive processing for low- and high-ambiguity stimuli.