Distinct neural alterations in schizophrenia and autism: A meta-analysis of social cognition and emotion processing

Melanie Boisvert,Florence Pilon,Laurent Mottron,Stephane Potvin
DOI: https://doi.org/10.1101/2024.10.11.24315338
2024-10-13
Abstract:Objective: Two conditions alter socio-communicative behaviors in humans: autism and schizophrenia. However, it is not well-known if these disorders share the same neural alterations during socio-emotional tasks. The main objective was to examine neural alterations in autism and schizophrenia during emotional and social cognition tasks. Our second objective was to determine if these alterations were common or distinct between disorders. Methods: Functional neuroimaging studies using an emotional or a social cognition paradigm in schizophrenia or autism were queried from three databases. We selected articles if they reported whole brain coordinates of different activations between autism/schizophrenia participants and non-clinical controls. Using SDM, we analyzed the coordinates of brain activity differences between case and control groups, categorized by diagnosis and paradigm. Results: The meta-analysis aggregated 104 studies in schizophrenia and 80 studies in autism spectrum disorder. During emotional tasks, individuals with autism showed reduced activity in the left amygdala, while those with schizophrenia showed reduced activity in the right inferior frontal gyrus and the median cingulate gyrus. During social cognition tasks, alterations in both conditions did not survive corrected statistical thresholds. No spatial conjunction was observed between the alterations seen in each disorder during both emotional/social cognition tasks at both corrected and uncorrected thresholds. Conclusions: These results suggest that the emotional processing in autism and schizophrenia in adulthood are characterized by alterations of bottom-up and top-down mechanisms of the emotional network, respectively. It should encourage the pursuit of functional neuroimaging studies on emotion processing using machine learning to differentiate these two conditions.
What problem does this paper attempt to address?