Attention Allocation during Visual Search in ASD: Results from the ABC-CT Feasibility Study

Background: Attenuated attention to social stimuli is a hallmark characteristic of autism spectrum disorder (ASD), and visual search tasks have been used to address salience of social stimuli such as faces (Gliga et al., 2009). 7-14-month-old infants who later developed ASD demonstrated a face “pop-out” effect, much like their typically-developing (TD) peers (Elsabbagh et al., 2012). However, older children with ASD attended to faces less than TD controls in the presence of competing non-social stimuli (Sasson et al., 2008). Visual search may potentially monitor changes in social functioning, and restrictive and repetitive behaviors (Sasson et al., 2008). Here, we evaluated the possibility that visual search tasks can aid in the detection of reliable biomarkers of social impairments in children with ASD.

Objectives: The current study aimed to evaluate whether visual search can reveal reliable differences in attention between children with ASD and TD controls, and to relations between visual search performance, and social and repetitive behavioral features of the autism phenotype. This study is part of a larger project on creating objective measures of treatment gains in clinical trials.

Methods: 46 children (23 ASD, 23 TD; 4.01-11.40 years) participated in the visual search task (adapted from Gliga et al., 2009). Stimuli comprised six circular arrays of five colored images (face, cell phone, vehicle, bird, and scrambled face). Participants viewed each array for 20 s while gaze patterns were recorded using a SR Eyelink 1000 Plus eye tracker. The primary dependent variable was percentage of time spent looking at the face; additional dependent variables included on-screen looking time and number of fixations to each target. Participants also completed a battery of behavioral assessments of social and cognitive development. Eye-tracking data were analyzed using linear mixed-effects modeling with repeated effects of trial; diagnosis, trial, and their interaction were included as fixed effects, and full-scale IQ as a covariate. Associations between social and cognitive measures, and visual search performance were determined with Spearman correlations.

Results: On-screen looking time did not differ by diagnosis status (F_{1, 49.689}=1.304, p=0.26) or trial (F_1,44=1.56, p=0.19), suggesting no between-group difference in overall visual engagement. Children with ASD looked less at faces (F_1,43.55 =13.130, p=0.001) than TD controls, over and above effects of full-scale IQ and trial. Similarly, children with ASD made fewer fixations within the face than TD controls (F_{1, 36.381}=6.027, p=0.019at non-face, object targets did not differ by diagnosis status (F_{1, 41.193}=0.255, p=0.616), suggesting a specific attenuation in attention to social stimuli. Greater fixation to the face was associated with higher Vineland Communication Standard Scores (rho=0.374, p=0.011).

Conclusions: Faces captured and held attention less for children with ASD, suggesting that the visual search task is sensitive to differences in social attention between diagnostic groups. In addition, attention to faces was associated with better functional communication skills, implying that visual search performance indexes a relevant domain of social functioning in ASD. Visual search tasks may help to objectively address changes in social functioning or treatment response in clinical trials.