Exploring visual statistical learning in pre-school verbal and minimally verbal autistic children: an electrophysiological study

Statistical learning refers to our ability to detect regularities in what we see or hear, simply bybeing exposed to them, without trying to learn. This automatic mechanism is essential for languagedevelopment and especially for identifying word boundaries in spoken language by recognizing howsounds follow one another. Statistical learning may be especially relevant in the context of autism, asaround 30% of autistic children do not develop spoken language, for reasons not yet fully understood.Language impairments in autism might be linked to difficulties in statistical learning. In this study,we use EEG to explore visual statistical learning in both verbal and minimally verbal autistic children,in comparison to TD peers. The experiment required children to watch sequences of images (aliens),sometimes with unpredictable images appearing. Results revealed a distinct neural response pattern:autistic children showed a significant N1 component to unpredictable stimuli, whereas TD childrenexhibited a P300 response in a later time window (400–500 ms). We did not observe differencesbetween verbal and minimally verbal autistic children. These findings suggest preserved earlydetection of statistical regularities in autism, even in the absence of a late P300 response, which mayreflect reduced attentional engagement.