Shared internal models for feedforward and feedback control.

Area of Science:

• Motor control
• Neuroscience
• Robotics

Background:

• Motor learning often involves adapting to new dynamics.
• Online error correction is crucial for robust motor performance.
• Task-specific adaptation of feedback responses is hypothesized but not directly shown.

Purpose of the Study:

• To investigate if motor feedback responses adapt to new task dynamics.
• To determine if these adapted responses are task-appropriate for unanticipated errors.
• To provide evidence for a smart feedback controller in motor adaptation.

Main Methods:

• Subjects performed reaching arm movements with novel velocity-dependent dynamics.
• Occasional, unanticipated force-pulse perturbations were introduced.
• Changes in online feedback responses to these pulses after adaptation were analyzed.

Main Results:

• Feedback responses to force pulses precisely compensated for the altered velocity-dependent dynamics.
• This compensation occurred even though the specific errors from pulses were never experienced during training.
• Adapted pulse responses were task-appropriate, reflecting the learned dynamics.

Conclusions:

• Motor adaptation leads to task-appropriate feedback responses to novel errors.
• This suggests a 'smart' feedback controller that utilizes learned dynamics.
• Neural processes for feedback control likely involve real-time state prediction and access to updated internal models.