Multisensory Perceptual Training in Youth with Autism Spectrum Disorder: A Randomized Controlled Trial

Abstract

Differences in sensory functioning — and audiovisual multisensory integration in particular — are commonly observed in individuals with autism. Difficulty integrating auditory and visual speech signals, in particular the temporal processing of these audiovisual cues, is commonly observed in this clinical population. The extent literature suggests that temporal binding of audiovisual speech can be trained (i.e., temporal binding windows [TBWs] can be narrowed) in typically developed adults. The purpose of this study was to evaluate a computer-based perceptual training in youth with autism to determine whether the perceptual training results in (a) narrower TBWs within the training task, (b) narrower TBWs on trained stimuli, (c) narrower TBWs on untrained speech stimuli and/or speakers, (d) improved audiovisual integration on untrained tasks (i.e., the McGurk illusion). We also assessed whether treatment outcomes varied according to individual characteristics (i.e., chronological age, language ability, non-verbal cognitive ability). Thirty youth with autism matched on the group level on chronological age, sex, gender, and pre-training temporal binding windows were randomly assigned to a perceptual training group (n = 15) or a control group (n = 15). Participants in the perceptual training group completed eight days of the training. At post-test, groups were compared on TBWs for trained and untrained stimuli and perception of the McGurk illusion. Language and nonverbal cognition were assessed prior to the onset of the study using standardized assessments. Data were analyzed using multiple regression models. The brief computer-based perceptual training for temporal binding of audiovisual speech resulted in small but non-significant changes in the TBW for the perceptual training group on average compared to the control group. Effects of the training program on TBW outcomes varied according to participant profiles, with significant effects in favor of the training for participants with at least above-average nonverbal IQs and at least average language abilities. However, participants with below-average nonverbal IQs were likely to experience counter-therapeutic effects.