Large-Scale Cerebellar-Cerebral Network Alterations Underlying Impaired Social Predictive Processing in Cerebellar...

The cerebellum is a key node for detecting and adjusting social-prediction internal models. Yet social prediction has been scantly investigated in neurodegenerative cerebellar patients (CB). Thus, we examined CB social-prediction abilities and related cerebellar-cerebral networks, using tailored tasks and resting-state fMRI. Thirteen adult CB and twenty-six matched controls completed three computerized prediction tasks, where contextual priors activate specific internal models setting expectations which may influence recognition of ambiguous targets (emotion expressions, personality-traits, non-social objects). Accuracy and response times were analysed through 2 × 2 × 2 mixed-effects ANOVAs. A congruency index (CI) reflecting whether response was driven by contextual influence was derived from performance efficiency scores and analysed through a 2 × 2 mixed-effects ANOVA. Participants underwent 3T brain MRI, including T1-weighted and resting-state fMRI sequences. Posterior and vermian cerebellar lobules were chosen a-priori as regions of interest and used in a Seed-Based fMRI analysis examining cerebellar-cerebral functional connectivity (FC). FC values from patients' altered connectivity clusters were extracted and correlated with CIs. CB performed worse in the personality-traits task, showing higher reliance on contextual cues. Cerebellar-cerebral underconnectivity was found within networks involved in mentalizing and learning. In CB, higher CI in the personality-traits task, reflecting greater contextual influence and failure to revise behavioural responses, significantly correlated with right lobule VIII-right putamen underconnectivity. These findings indicate reduced flexibility in updating higher-order social predictions when contextual expectations conflict with incoming information, accompanied by altered cerebellar-cerebral FC. Exploratory findings further point to cerebellar-basal ganglia connectivity as a potentially relevant network component of predictive social cognition, warranting further investigation.