RESUMO
BACKGROUND: Though aberrant face processing is a hallmark of autistic spectrum disorder (ASD), findings on accompanying brain activity are divergent. Therefore, we conducted an activation likelihood estimation (ALE) meta-analysis of studies examining brain activity during face processing. METHODS: We searched PubMed and PsycINFO using combinations of terms as 'fMRI', 'Autism Spectrum Disorder', 'Face Perception'. Eligible studies reported on DSM-diagnosed ASD individuals, compared to controls (HC), using face stimuli presented in fMRI and reporting whole-brain analysis coordinates. We compared two approaches: 'convergence of differences' (primary analysis) using study-level coordinates from ASD vs. HC contrasts, and 'differences in convergence' (secondary) pooling coordinates within each group separately, and contrasting the resultant ALE maps. RESULTS: Thirty-five studies (655 ASD and 668 HC) were included. Primary analysis identified a cluster in amygdala/parahippocampus where HC showed greater convergence of activation. Secondary analysis yielded no significant results. CONCLUSIONS: Results suggest that ASD dysfunction in face processing relies on structures involved in emotional processing rather than perception. We also demonstrate that the two ALE methodologies lead to divergent results.