Abstract

Objective: Neurolinguistic research on social concepts (expressions evoking socio-interactive events, behaviors or traits) mainly employs (pseudo)randomized isolated words, failing to achieve ecological validity. Also, few studies have combined atrophy models with multidimensional measures, a core strategy to reveal direct brain-behavior links. Here I will introduce a multimodal (MRI-fMRI-EEG) framework and two studies tapping comprehension of a social text (ST, rich in interpersonal events) and a matched non-social text (nST, describing a single person’s actions) in cerebellar ataxia (CA) and behavioral variant frontotemporal dementia (bvFTD). Methods: Study 1 involved 15 CA patients and 29 controls. We established the patients’ atrophy pattern, compared ST and nST outcomes between groups, and examined their association with cerebellar connectivity. Study 2 involved 20 bvFTD patients, 41 controls, 23 persons with Alzheimer’s disease, and 25 with Parkinson’s disease. We administered an ST, an nST, and additional texts. Comprehension of each text was correlated with brain volume, fMRI connectivity, and EEG connectivity measures. Results: In Study 1, CA patients showed selective ST deficits and focal cerebellar atrophy. Also, while ST outcomes in controls selectively correlated with connectivity between the cerebellum and key socio-cognitive regions (superior and medial temporal gyri, temporal pole, insula), such associations were absent in CA patients. In Study 2, bvFTD patients also showed selective ST deficits, associated with the volume and fMRI connectivity of social cognition regions (orbitofrontal and temporal gyri, amygdala, cerebellum), as well as reduced frontotemporal EEG connectivity. These patterns were exclusive to the bvFTD group. Conclusions: Text-level social concepts seem linked to cortico-limbic-cerebellar circuits subserving social cognition at large. These findings support ecological embodied views of the domain, paving the way for disease-specific markers.