Dominant neuroanatomical models hold that humans regulate their movements via loop-like cortico-subcortical networks, including the subthalamic nucleus (STN), thalamus, and sensorimotor cortices (SMC). Inhibitory commands across these networks are purportedly sent via transient, burst-like signals in the β frequency (15-29Hz). However, since human depth-recording studies are typically limited to one recording site, direct evidence for this proposition is hitherto lacking. Here, we present simultaneous multi-site depth-recordings from SMC and either STN or thalamus in humans performing the stop-signal task. In line with their purported function as inhibitory signals, subcortical β-bursts were increased on successful stop-trials and were followed within 50ms by increased β-bursting over SMC. Moreover, between-site comparisons (including in a patient with simultaneous recordings from all three sites) confirmed that β-bursts in STN precede thalamic β-bursts. This provides first empirical evidence for the role of β-bursts in conveying inhibitory commands along long-proposed cortico-subcortical networks underlying movement regulation in humans.

Recently it has been acknowledged that the basal ganglia nuclei play a major role in cognitive control; however, the contribution by their network remains unclear. Previous studies have demonstrated the role of the subthalamic nucleus (STN) in cognitive processing and suggested that its connections to cortical and other associated regions regulate response inhibition during conflict conditions. By contrast, the role of the internal globus pallidus (GPi) as the output nucleus before the thalamic relay has not yet been investigated during cognitive processing. We recorded local field potentials (LFPs) from externalized deep brain stimulation (DBS) electrodes implanted bilaterally in the GPi (n=9 participants with dystonia) and STN (n=8 participants with Parkinson's disease (PD)) during a primed flanker task. Both dystonia (GPi group) and PD participants (STN group) responded faster to the congruent trials than the incongruent trials. Overall, the dystonic GPi group was significantly faster than the PD STN group. LFPs showed elevated cue-triggered theta (3-7 Hz) power in GPi and STN groups in a similar way. Response-triggered LFP beta power (13-25 Hz) was significantly increased in the GPi group compared to the STN group. Results demonstrate that GPi activity appears to be critical in the cognitive processing of action selection and response during the presence of conflict tasks similar to the STN group. As both GPi and STN nuclei are involved in cognitive processing; therefore, these nuclei may be targeted for neuromodulation to improve cognitive control via DBS.