Predictive coding in psychopathology: mechanistic model or metaphorical re-description?

Predictive coding (PC) has become a central framework in contemporary cognitive neuroscience, proposing that the brain operates as a hierarchical inference system that continuously minimizes the mismatch between predicted and actual sensory input. Its extension into clinical neuroscience has been accompanied by considerable enthusiasm, yet attempts to translate its computational principles into explanations of psychiatric and neurological disorders have yielded uneven results. The present review critically examines the clinical applicability of PC across three diagnostic domains: schizophrenia, autism spectrum disorder (ASD), and mood and anxiety disorders. Drawing on findings from neuroimaging, electrophysiology, and computational modeling, the discussion evaluates how disturbances in prediction error signaling, the precision weighting of sensory evidence relative to prior beliefs, and hierarchical inference have been proposed to relate to core clinical phenomena such as hallucinations, sensory hypersensitivity, and affective dysregulation. Particular attention is given to persistent theoretical tensions, including debates surrounding prior precision, the mapping between neural proxies and behavior, and the inconsistent use of PC terminology across diagnostic contexts. By adopting a structured and comparative approach, this review aims to clarify where predictive coding offers testable mechanistic insight into psychopathology, and where its explanatory scope remains limited or provisional.