Abstract
Movements become more vigorous when rewards are expected, suggesting that motivational signals influence motor control. Dopaminergic neurons, known to encode reward expectation and prediction error, are prime candidates for linking value and vigor. Here, we show that human reach vigor dynamically tracks canonical dopaminergic learning signals not only at movement onset but also during ongoing motion. Using a reaching task with probabilistic rewards (0, 33, 66, and 100%), we observed that peak velocity scaled with expected value. Crucially, following feedback, reach velocity was transiently invigorated or enervated in proportion to the sign and magnitude of the reward prediction error. Trial-by-trial changes in kinematics reflected value updating, consistent with dopaminergic phasic learning signals. These results demonstrate that movement vigor is modulated by reward-learning signals on rapid timescales, revealing a real-time behavioral readout of motivational computation in the brain.