Multimodal evidence for a mechanistic model of working memory deficits in schizophrenia

Working memory (WM) deficits are central to schizophrenia (SCZ), yet their mechanistic basis remains unclear. We combined computational modelling with genetic, transcriptomic, behavioural, and fMRI data to construct a mechanistic account of WM impairment in SCZ. Post-mortem RNA expression from prefrontal and anterior cingulate cortex (ACC) was integrated with single-cell, network, and synaptic plasticity models to show how SCZ-related changes in ion channel-encoding and plasticity-regulating genes alter sustained delay-period activity and long-term potentiation, suggesting an impairment of WM. The model predictions were supported by behavioural WM test (letter-number sequencing) results and polygenic risk scores for SCZ based on ion channel and plasticity gene sets. Mendelian randomization, together with nominally significant single-gene risk analyses, implicated specific ion channel genes, particularly CACNA1I, as putatively causal for both SCZ liability and WM deficits. fMRI N-back data supported ACC-specific delay-period impairments. These multimodal findings highlight candidate, druggable mechanisms for cognition-focused interventions in SCZ.