Significance Responsive neurostimulation is an increasingly accessible treatment for medication-resistant epilepsy that aims to suppress seizures using electrical stimulation from implanted intracranial electrodes. However, the optimal cortical location and time point for intervening once a seizure begins are not well understood. Here we represent a seizure as a series of effective connectivity networks over time and compute metrics of network controllability and optimal control energy. Our results allow us to characterize when and where the brain network may be the most responsive to an external stimulus.