Prolonged neural encoding of visual information in autism

Autism spectrum disorder (ASD) is associated with a hyper-focused visual attentional style, impacting higher-order social and affective domains. The understanding of such peculiarity can benefit from the use of multivariate pattern analysis (MVPA) of high-resolution electroencephalography (EEG) data, which has proved to be a powerful technique to investigate the hidden neural dynamics orchestrating sensory and cognitive processes. Here, we recorded EEG in typically developing (TD) children and in children with ASD during a visuo-spatial attentional task where attention was exogenously captured by a small (zoom-in) or large (zoom-out) cue in the visual field before the appearance of a target at different eccentricities. MVPA was performed both in the cue-locked period, to reveal potential differences in the modulation of the attentional focus, and in the target-locked period, to reveal potential cascade effects on stimulus processing. Cue-locked MVPA revealed that while in the TD group the pattern of neural activity contained information about the cue mainly before the target appearance, the ASD group showed a temporally sustained and topographically diffuse significant decoding of the cue neural response even after the target onset, suggesting a delayed extinction of cue-related neural activity. Crucially, this delayed extinction positively correlated with behavioral measures of attentional hyperfocusing. Results of target-locked MVPA were coherent with a hyper-focused attentional profile, highlighting an earlier and stronger decoding of target neural responses in small cue trials in the ASD group. The present findings document a spatially and temporally overrepresented encoding of visual information in ASD, which can constitute one of the main reasons behind their peculiar cognitive style.This study unveils how incoming visual input is encoded at the neural level in children with autism spectrum disorder (ASD), and how this encoding impacts on their attentional selection. Using multivariate pattern analysis (MVPA) of high-resolution EEG data, we show that children with ASD exhibit a temporally prolonged and topographically broader overrepresentation of visual information, as compared to their typically developing peers. Such excessive stimulus encoding could in turn lead to difficulties in reorienting attentional resources to subsequent relevant incoming stimuli, leading to their peculiar hyperfocused attentional profile. This study not only provides new evidence about important neural mechanisms of visual information processing in ASD, but it also shows that MVPA approaches can detect neural dynamics that may remain otherwise largely blind to other univariate analyses approaches.