Dennis London

Bottom line: in humans, the pallidothalamic circuit looks less like a pure movement release mechanism and more like a real-time interface that keeps evaluation close to action—linking decision, movement, and feedback in a format that can support adaptive control. 10/10

After feedback, outcome signals emerged most strongly in STN and motor thalamus. And in motor thalamus, confidence and outcome gradually converged into a signal consistent with unsigned reward prediction error—a measure of surprise magnitude. 9/10

One striking feature was persistence: confidence signals didn’t vanish when the movement ended. They persisted into feedback and the inter-trial interval, keeping an “expected outcome” trace available when the actual outcome arrives. 8/10

In other words, the circuit doesn’t just pass signals forward. It appears to reformat them: cognitive and motor variables become increasingly aligned downstream, so that movement-related modulations also carry evaluative information. 7/10

If these signals appear across nodes, what is the circuit doing? Population analyses revealed a transformation along the pathway: upstream representations were more dynamic, while downstream representations—especially in motor thalamus—were more low-dimensional and temporally stable. 6/10

That activity wasn’t “motor-only.” Within these movement-linked responses, neurons and populations encoded choice and confidence continuously—before movement, during movement, and after movement. 5/10

Classic circuit diagrams suggest a tidy push-pull chain across nodes. What we saw was simpler—and more surprising: firing rates rose sharply at movement onset in every node and stayed elevated through movement and feedback. 4/10

We recorded single-neuron activity across multiple nodes of the human pallidothalamic circuit—GPe, GPi, STN, and motor thalamus—while participants made perceptual decisions and reported choice + confidence with a single reaching movement. 3/10

First: a huge thank you to the patients who volunteered for this work while undergoing DBS surgery. This science is only possible because of their generosity.

And thank you to my colleagues Marisol Soula, Ling Pan, Michael Pourfar, Alon Mogilner, and @roozbehkiani.bsky.social! 2/10

The basal ganglia output is often framed as a motor gate: corticostriatal circuits select an action, then GPi/SNr-thalamus helps release it. In humans, we find this same pathway carries cognitive variables embedded in movement signals—and even produces learning-relevant signals after feedback. 1/10