Musicality and social cognition in dementia: clinical and anatomical associations.

Human musicality might have co-evolved with social cognition abilities, but common neuroanatomical substrates remain largely unclear. In behavioural variant frontotemporal dementia, social cognitive abilities are profoundly impaired, whereas these are typically spared in Alzheimer's disease. If musicality indeed shares a neuroanatomical basis with social cognition, it could be hypothesized that clinical and neuroanatomical associations of musicality and social cognition should differ between these causes of dementia. We recruited 73 participants from the Amsterdam Dementia Cohort (n = 30 female; aged 50-78), of whom 23 had behavioural variant frontotemporal dementia, 22 Alzheimer's disease and 28 were healthy controls. Musicality was assessed using a music-emotion recognition test, melody, tempo, accent and tuning subscores, a musicality summed score, the identification of auditory hedonic phenotypes and music emotion induction using skin conductance responses. Social cognition was assessed across multiple levels, including emotion recognition, theory of mind, socio-emotional sensitivity and understanding of social norms. We used ANCOVA to investigate subgroup differences in musicality and social cognition and linear regressions to investigate associations between musicality and social cognition. All analyses were adjusted for age, sex, musical training and mini mental state examination. Finally, we performed voxel-based morphometry analyses on T₁-weighted MRI to study whether regions for musicality and social cognition overlapped anatomically. We found that patients with behavioural variant frontotemporal dementia performed worse on music-emotion recognition (all P < 0.001) and tempo recognition (all P < 0.05) compared with Alzheimer's disease and on musicality summed score (all P = 0.02) compared to controls only. Furthermore, patients with behavioural variant frontotemporal dementia had lower mean skin conductance responses during emotion-inducing music, compared to Alzheimer's disease (all P < 0.045). Worse music emotion recognition scores were associated with worse facial emotion recognition (P < 0.0001), worse theory of mind (P = 0.0005) and worse understanding of social norms (P = 0.01). Melody and tempo recognition were associated with facial emotion recognition and theory of mind, and accent recognition was associated with the theory of mind. Music emotion recognition and tempo recognition were also associated with executive functions. Worse music emotion recognition, melody recognition, tempo recognition, facial emotion recognition and theory of mind scores were all related to atrophy in the anterior temporal regions and the fusiform gyri, which play a role in multisensory integration, and worse tempo recognition was associated with atrophy of the anterior cingulate cortex. These results support the idea that musicality and social cognition may share a neurobiological basis, which may be vulnerable in behavioural variant frontotemporal dementia.