The Relation Between Visual Disengagement and Autistic Traits: An Eye-Tracking Study of 410 Individuals

Background:  Attentional atypicalities are often found in individuals with ASD, and problems with reallocating attention have been proposed as a plausible explanation for the cognitive and social deficits characterizing the autism phenotype. Using the visual disengagement paradigm, where the task is to move one’s gaze away from a central stimulus (CS) when a new stimulus appears in the periphery, several studies suggest that flexibility in visual orienting is impaired in adults and children with autism, as well as infants with a later diagnosis of ASD. ASD is commonly conceived as the extreme end of a phenotypic and etiological continuum. Yet, how visual disengagement relates to autistic traits in the typical population remains unknown. Furthermore, case-control differences have not always been replicated when visual disengagement is assessed in childhood (i.e., beyond infancy).

Objectives:  To test the hypothesis that slower visual disengagement is related to higher ASD traits in a typical population sample of children.

Methods:  The final sample consisted of 410 twins ranging age from 9-14 years recruited from the Child and Adolescent Twin Study in Sweden (CATSS). All twins were included in the analysis below, but all effects remained in a control analysis where one twin from each twin pair was removed. Twin modeling analysis is not reported at this point, but is an upcoming step of the analysis. ASD traits were assessed using the Social Responsiveness Scale (SRS). Visual disengagement was measured using the prosaccade (gap-overlap) paradigm on a Tobii T120 eye tracker. In the current analysis, our dependent measure, the disengagement effect, was defined as the difference in saccadic reaction times between the overlap (CS remained) and the baseline conditions (CS disappeared when peripheral stimuli appeared). Additionally, the relation between SRS and absolute performance in the aforementioned conditions as well as a third, the gap (CS disappears 200ms before the peripheral appears) is analyzed.

Results:  We found the expected effect of condition on saccadic reaction times: gap (M=200.2ms, SD= 56.9ms), baseline (M=235.4ms, SD= 60ms), overlap (M=261.9ms, SD= 75.4ms; all pairwise comparisons p<.001). We found no indication that the disengagement effect was related to severity scores on the SRS (r = .01, p>.25). However, higher ASD traits were associated with slower visual orienting in all conditions (Gap: r = .178, p=.001; Baseline: r = .171, p<.001; Overlap: r = .140, p=.004). Neither age nor number of valid trials were related to the disengagement effect.

Conclusions:  These data indicate that visual disengagement (as typically operationalized in the literature, e.g. Elsabbagh et al 2013) is not related to the level of autistic traits in typically developing children of 9-14 years. Rather, the correlations with absolute latencies (in all conditions) could suggest that higher autistic traits are associated with slower reactive eye movements. With the reservation that this was a study of traits and not a case-control study, these data from several hundreds of individuals do not support that visual disengagement is related to autism in late childhood.