Statistical learning for non-social and socially-meaningful stimuli in individuals with high and low levels of autistic traits

Autism spectrum disorder (ASD) is associated with weaknesses in social communication and interaction but potential strengths in perceptual processing of non-social stimuli. It is unknown to what extent such strengths and weaknesses affect statistical learning (SL), which is the ability to learn statistical regularities from environmental input. Rather than focus on individuals with a diagnosis of ASD, we take a spectrum approach to autism and examine undiagnosed Chinese young adults who either have high or low levels of autistic traits (ATs) as assessed by the Autism-Spectrum Quotient. Experiment 1 incorporated non-social and non-linguistic auditory input (pure tones) whereas Experiment 2 used socially-meaningful input (spoken Chinese disyllables). The results showed a striking dissociation between the different SL tasks. For the non-social stimuli (Experiment 1), both individuals with high and low ATs showed evidence of SL of the input regularities, with the individuals having high levels of ATs showing significantly better performance than those with low levels of ATs. On the other hand, when socially-meaningful stimuli were incorporated (Experiment 2), only the individuals with low ATs showed evidence of SL; the performance of the high ATs group was not significantly different from chance. These findings suggest that ATs differentially affect the learning and processing of non-social and socially-meaningful stimuli, which in turn has implications for clinical interventions for ASD and for individuals with high ATs.