Audiovisual multisensory integration in individuals with autism spectrum disorder: A systematic review and meta-analysis

Jacob I Feldman,Kacie Dunham,Margaret Cassidy,Mark T Wallace,Yupeng Liu,Tiffany G Woynaroski
DOI: https://doi.org/10.1016/j.neubiorev.2018.09.020
Abstract:An ever-growing literature has aimed to determine how individuals with autism spectrum disorder (ASD) differ from their typically developing (TD) peers on measures of multisensory integration (MSI) and to ascertain the degree to which differences in MSI are associated with the broad range of symptoms associated with ASD. Findings, however, have been highly variable across the studies carried out to date. The present work systematically reviews and quantitatively synthesizes the large literature on audiovisual MSI in individuals with ASD to evaluate the cumulative evidence for (a) group differences between individuals with ASD and TD peers, (b) correlations between MSI and autism symptoms in individuals with ASD and (c) study level factors that may moderate findings (i.e., explain differential effects) observed across studies. To identify eligible studies, a comprehensive search strategy was employed using the ProQuest search engine, PubMed database, forwards and backwards citation searches, direct author contact, and hand-searching of select conference proceedings. A significant between-group difference in MSI was evident in the literature, with individuals with ASD demonstrating worse audiovisual integration on average across studies compared to TD controls. This effect was moderated by mean participant age, such that between-group differences were more pronounced in younger samples. The mean correlation between MSI and autism and related symptomatology was also significant, indicating that increased audiovisual integration in individuals with ASD is associated with better language/communication abilities and/or reduced autism symptom severity in the extant literature. This effect was moderated by whether the stimuli were linguistic versus non-linguistic in nature, such that correlation magnitudes tended to be significantly greater when linguistic stimuli were utilized in the measure of MSI. Limitations and future directions for primary and meta-analytic research are discussed.
What problem does this paper attempt to address?