Abstract
Cortisol is known to promote memory for emotionally arousing experiences, yet the neural networks involved in enhancing these memories are unknown. Here, we combine pharmacological fMRI with an analysis approach to determine the dynamic brain network processes by which cortisol enhances the formation of emotional memories. We introduce dynamic connectome-based predictive modeling (dCPM) to identify high temporal resolution, whole-brain functional connectivity networks, and show that distinct networks can successfully predict trial-level arousal (involving salience network) and subsequent memory (involving visual, frontoparietal, and default mode networks). We find that, under cortisol, brain networks associated with both processing arousal and forming memories have more structural and functional overlap. Neural networks that predict arousal are more consistent and strongly engaged under cortisol, whereas networks that predict memory are more specialized to emotional memoranda. Together, these results reveal a dynamic process by which cortisol shifts whole-brain mechanisms to amplify emotional memories: Cortisol amplifies arousal-predictive networks while tuning memory-predictive networks to prioritize emotionally salient information.