Abstract
<title>Abstract</title> <p>Phasic sympathetic arousal, indexed by electrodermal activity (EDA), is accompanied by distributed cortical activity, yet which cortical rhythms track these autonomic responses, and when, remains poorly understood hemodynamic imaging has mapped the contributing network but lacks the temporal resolution to resolve this coupling. In 85 adults, we combined magnetoencephalography (MEG) with EDA during a sustained-threat task evoking a graded build-up of arousal. A deep-learning model decoded single-trial sympathetic-response magnitude from whole-brain MEG. Time-resolved liner-mixed effect modelling then showed phasic EDA was preceded by early, overlapping beta synchronization in ventral-visual and posterior-cingulate cortex and gamma synchronization with theta suppression in the right anterior insula, followed by later anterior-cingulate alpha desynchronization. Across individuals, the strength of this oscillation–EDA coupling varied markedly. Critically, anxiety tracked not cortical power or autonomic level alone, but their coupling strength identifying neural–autonomic coupling as a dimension of individual differences in anxiety.</p>