Learning and adaptation in cortical control of arm movement

We investigated the motor adaptation and learning in cortical neuron populations as they respond to external perturbations and reinforcement training. We implanted 64 fine-wire electrodes chronically in the sensory-motor cortex of rhesus monkeys. The monkeys were trained to perform a reaching movement in 3D space before we investigated their cortical neurons: (1) adaptation to repeated perturbations; (2) control of a robot; and (3) responses to training. The issues addressed included: (1) the time course of neuronal response changes as the monkey learned to anticipate an external perturbation; (2) the control strategies used by the CNS to compensate for this perturbation; and (3) the relationship between modified cortical activity and adaptation in arm movement. The perturbation evoked large (excitatory) reflex responses in each area of cortex, and was the dominant feature in the perturbation response. The latency of the reflex response peak did not change significantly during adaptation. However, the response gain (reflex peak/volitional peak) decreased, because the volitional component of the response increased during adaptation. The reduction in the reflex gain resulted in a negative correlation between the perturbation induced displacement (PID) which increased, and the reflex gain which decreased during adaptation. Conversely, the response in sensory areas of cortex scaled linearly with the PID, which indicates that the activity in sensory cortex provided a stable measure of the trajectory error introduced by the perturbation. After adaptation, neurons in the motor cortex showed changes in activity that preceded the onset of the stimulus (perturbation). Concurrently, the perturbation response in motor cortex became more target-specific after learning. These results reflect the transition to a feedforward, goal-directed strategy for perturbation compensation.