Reshaping computational neuropsychiatry beyond synaptopathy.

The article focuses on the application and limitations of Active Inference (AI), a computational neuroscience framework, in explaining neuropsychiatric disorders through neuronal message passing and belief updating. It distinguishes between synaptopathies—primarily psychiatric disorders involving dysfunctional synaptic precision weighting—and damaged brain pathologies, such as vascular dementia and severe neurodegenerative diseases, where structural brain damage disrupts free energy minimization processes fundamental to AI. While AI effectively models disorders characterized by synaptic dysfunction and dysconnection, its principles may not fully apply to brain-damaged conditions due to impaired neural circuitry and loss of precision weighting. The authors suggest that integrating this distinction can refine computational models of neuropsychiatric symptoms and emphasize the need for cautious application of AI frameworks in brain injury contexts.