Aging Minds and Social Ties: Neural Signatures of Theory of Mind Predict Social Network Size, But Not Loneliness or Perceived Social Support, in Older Adults

Abstract

Social disconnection in later life has emerged as a significant public health concern, with both structural (e.g., reduced social network size) and experiential (e.g., loneliness) dimensions linked to poorer health and well-being. Prior work has identified multivariate brain signatures associated with Theory of Mind (ToM)—a socio-cognitive ability to infer others’ beliefs and intentions—as predictive of social network size in young adults. However, it remains unclear whether similar neural markers are evident in older adults and whether they differentially predict structural and experiential aspects of social connectedness. The present study addresses this gap by applying multivoxel pattern analysis (MVPA) to fMRI data collected from 41 older adults (aged 65–77 years) while they completed the EmpaToM task, a validated paradigm for probing Theory of Mind (ToM) and empathy. A leave-one-participant-out cross-participant support vector regression (SVR) decoding approach was used to determine whether activation patterns during ToM (vs. non-ToM) conditions predicted individual differences in social network size, perceived loneliness, and perceived social support. ToM-related neural activation patterns in the right temporoparietal junction (rTPJ) significantly predicted two indicators of structural social connectivity (support and sympathy clique sizes), despite age-related declines in behavioral ToM performance. However, these patterns did not predict experiential measures such as perceived loneliness or support. Supplemental analyses using empathy-related activation patterns from the EmpaToM task did not yield significant associations with social network size, pointing to the specificity of ToM processes. However, empathy-related neural signatures in a cluster in the inferior frontal gyrus extending to the right anterior insula (rIFG/AI) marginally predicted variance in loneliness, These findings extend previous research by demonstrating that ToM-based neural representations retain predictive utility for real-world social behavior in older adulthood, despite age-related declines in ToM capacities. Moreover, they underscore a dissociation between the neural correlates of social network size and those of more affective, emotional experiences of connection, pointing to distinct cognitive-affective mechanisms. Specifically, while ToM-related processes may support the maintenance of larger social networks, experiential aspects of connectedness may be more closely tied to socio-affective processes such as empathy, which warrant targeted investigation in future research.