Neurocognitive characterization of behaviour and mental illness through time-varying brain network analysis

Cognitive processing in humans relies on regional brain activity and large-scale, whole-brain network interactions which evolve in a time-dependent manner. Traditional task-based neuroimaging analysis does not sufficiently consider this temporal dimension, thus restricting the characterisation of networks and resulting in limited brain representation of cognition and behaviour. Time-varying, task-based neuroimaging analysis provides a unique opportunity to characterize brain functional network reconfiguration with ongoing task stimuli and may represent a better neural proxy for cognition and behaviour. In this study, we characterized functional network connectivity (FNC) fluctuations in resting-state and three reinforcement-related tasks, to illustrate network reconfiguration across multiple tasks. We further determined behavioural relevance of the whole-brain and regional FNC by relating network measurements with task performances and psychopathology. We found that several consistent whole-brain functional networks (FNC states) exist across resting-state and task fMRI sessions, and FNC state occurrences are sensitive to the most prominent task stimuli within a task session. In contrast, pair-wise, regional FNC could distinguish specific task conditions. When tested for correlation with psychopathology symptoms as well as clinical depression and alcohol use disorder in independent samples, we showed that FNC derived from time-varying analysis could account for a much higher amount of variance compared to FNC derived from static connectivity. Our results suggest that cognitive processing modulates task-relevant, regional FNC and changes whole-brain functional network connectivity, thus broadly affecting brain network architecture. By considering the different FNC states, time-varying connectivity provides a more comprehensive representation of brain interactions and thus may serve as a more precise neural correlate in quantifying an individual's risk for psychopathology.