Beyond where: When and how brain stimulation drives state transitions

Abstract Brain stimulation is increasingly used to treat neurological and psychiatric conditions, but stimulation sites and timing are typically chosen based on a priori assumptions from prior studies, not individualised mechanisms. Here, we investigate why some brain regions respond more to stimulation, when they respond best, and whether optimal targets for brain-state transitions are defined by functional anatomy (e.g. an auditory region in a resting-to-listening task transition), by system dynamics, or a combination of both. To do so, we build subject-specific whole-brain Hopf models fitted to MEG data, separating signals into five canonical frequency bands. We find that regions with a smaller oscillation radius and greater temporal variability respond more strongly to stimulation. Moreover, responsiveness was highly dependent on state and frequency band, with slower frequencies exhibiting phase-driven responsiveness, while faster ones depending more on network synchrony. Importantly, optimal nodes for transitioning between brain states are not defined by functional labels but rather by their dynamical regime. These results emphasise the value of mechanistic, personalised models for identifying when and where to stimulate the brain, and may help inform more precise brain-stimulation strategies.