Coupled functional and structural brain features of the default, affective and sensorimotor networks predict antisocial personality traits. A jICA and Transformer-based foundation model approach

Antisocial Personality traits (ASPT) are characterized by emotional detachment, impulsivity, and disregard for social norms, yet their neural underpinnings remain incompletely understood. The present study employed a multimodal neuroimaging approach to identify structural and functional brain patterns associated with antisocial personality traits in a large healthy sample to test the hypotheses that ASPT are associated with reduced default mode and affective networks functionality and structural integrity. Using joint Independent Component Analysis (jICA), we integrated fractional amplitude of low-frequency fluctuations (fALFF), gray matter (GM), and white matter (WM) measures from 205 participants. Three independent components (IC5, IC7, and IC9) jointly predicted antisocial scores on the Personality Styles and Disorders Inventory (PSDI). Specifically, reduced fALFF–GM–WM expression in IC5 (Default Mode Network, DMN) and IC7 (affective network), as well as increased expression in IC9 (sensorimotor–reward network) were associated with higher antisocial traits. A supervised machine learning method known as Transformer-based foundation model pre-trained on millions of synthetic tabular datasets confirmed the generalizability of our findings. Correlation analyses further showed that antisociality was linked to low Agreeableness, low Conscientiousness, and high Trait Anger and Anger-Out, consistent with deficits in empathy, self-control, and emotional regulation. Together, these findings suggest that antisocial tendencies reflect a distributed dysfunction across self-referential, affective, and sensorimotor–reward systems, providing a network-level framework for understanding the neural architecture of ASPT.