Dynamic fMRI networks of human emotion.

Abstract

The experience of emotions is that of dynamic, time-changing processes. Yet, many functional MRI (fMRI) studies of emotion average across time to focus on maps of static activations, overlooking the temporal dimension of emotional responses. In this study, we used time-resolved fMRI, group spatial independent component analysis (ICA), dual regression, and Gaussian curve fitting to examine both the spatial and temporal properties of whole-brain networks during a behavioral task. This task included trials that spanned over 25 s of watching short, emotionally evocative movie clips, making emotion-related decisions, and an intertrial rest period. We identified four whole-brain networks with unique spatial and temporal features that mapped onto different stages of the task. A network activated early in the course of the task included perceptual and affective evaluation regions, while two later networks supported semantic interpretation and decision-making, and a final network aligned with default mode activity. Both spatial and temporal properties of all four networks were modulated by the emotional content of the movie clips. Our findings extend current models of emotion by integrating temporal dynamics with large-scale network activity, offering a richer framework for understanding how emotions unfold across distributed circuits. Such temporal-spatial markers of emotional processing may prove valuable for identifying and tracking alterations in clinical populations.