AbstractState-of-the-art therapeutic brain stimulation strategies are delivered open loop, using fixed parameters. However, brain states exhibit spontaneous fluctuations dependent upon different behavioural or disease states. Here, we use a model of the cortico-basal ganglia-thalamic circuit to demonstrate how connectivity underpins changes in subcortical beta oscillations – a commonly used control parameter for deep brain stimulation in Parkinson’s disease. We show that recurrent cortical-subcortical loops involving either the cortico-subthalamic or pallido-subthalamic pathways can act in antagonism to modulate the expression of beta band activity (14-30 Hz). These pathways alter the relative timing of intermittent activity across the network, with increased pallido-subthalamic connectivity increasing the propensity of the circuit to enter a state of autonomous oscillation. We demonstrate that phase-locked stimulation can modulate these oscillations, with an efficacy that ultimately depends upon the connectivity across the circuit. This work outlines critical factors required to implement state-adaptive closed-loop brain stimulation.HighlightsConverging inputs to the subthalamic nucleus arriving via the external segment of globus pallidus and cortex act in antagonism and promote different beta rhythms.Phase locked stimulation has the capacity to selectively enhance or suppress a brain rhythm depending on the stimulation timing.The efficacy of stimulation and the parameters required to deliver it, e.g. stimulation timing, effective sensing and stimulation locations, are functions of network state.