Individualized prediction of trait self-control from whole-brain functional connectivity

Self-control is a core psychological construct for human beings and it plays a crucial role in the adaptation to society and achievement of success and happiness for individuals. Although progress has been made in behavioral studies examining self-control, its neural mechanisms remain unclear. In this study, we employed a machine-learning approach-relevance vector regression (RVR) to explore the potential predictive power of intrinsic functional connections to trait self-control in a large sample (N = 390). We used resting-state functional MRI (fMRI) to explore whole-brain functional connectivity patterns characteristic of 390 healthy adults and to confirm the effectiveness of RVR in predicting individual trait self-control scores. A set of connections across multiple neural networks that significantly predicted individual differences were identified, including the classic control network (e.g., fronto-parietal network (FPN), salience network (SAL)), the sensorimotor network (Mot), and the medial frontal network (MF). Key nodes that contributed to the predictive model included the dorsolateral prefrontal cortex (dlPFC), middle frontal gyrus (MFG), anterior cingulate and paracingulate gyri, inferior temporal gyrus (ITG) that have been associated with trait self-control. Our findings further assert that self-control is a multidimensional construct rooted in the interactions between multiple neural networks.