Whole-brain network dynamics underlying intolerance of uncertainty

Intolerance of uncertainty (IU) is a transdiagnostic risk factor of psychopathology, but its neural underpinnings remain poorly established. Here, we combined connectome-based predictive modeling (CPM) and edge timeseries in a functional magnetic resonance imaging task design to model the network dynamics of IU (N = 85). First, CPM during passive viewing of surprised faces identified an IU-predictive network, characterized by positive Visual-Salience and negative Subcortical-Motor couplings. Prospective IU was related to both the positive and negative networks, but inhibitory IU was only related to the negative network. IU was better predicted when the uncertainty signal embedded within the facial features of surprise was maximized. Using edge timeseries, we found that individuals with higher IU showed sustained IU-related network engagement during post-task resting-state. Finally, we tested the generalizability of the IU-related network dynamics in an independent large-scale dataset with a sample size greater by tenfold (N = 878). The same network dynamics were associated with internalizing symptoms during a gambling task, but not during pre-task resting-state. Together, these findings suggest that an IU-related network, sensitive to the uncertainty evoked, exhibits sustained temporal dynamics generalizable to clinical symptoms in an independent dataset.