How does our motor system determine its learning rate?

Area of Science:

• Neuroscience
• Motor Control
• Computational Neuroscience

Background:

• Motor learning is fundamentally driven by movement errors, with the learning rate quantifying error correction in subsequent movements.
• Previous research indicated learning rates depend on error signal reliability and motor system state uncertainty, aligning with Kalman filter predictions for optimal error minimization.

Purpose of the Study:

• To investigate whether the learning rate is optimized for every individual movement, not just on average.
• To test the applicability of the Kalman filter model in predicting motor learning behavior at the individual movement level.

Main Methods:

• Participants performed repeated movements to visual targets with their non-dominant hand.
• Endpoint errors were provided as immediate visual feedback after each movement.
• The reliability of error signals was manipulated across three experimental conditions.

Main Results:

• Observed learning rates were inconsistent with Kalman filter predictions, showing slower-than-predicted correction for initial large errors.
• The learning rates for movement extent and direction did not vary as predicted by the Kalman filter.
• A simpler model, employing a fixed learning rate for movements with consistent error-signal reliability, adequately explained the experimental data.

Conclusions:

• The brain does not employ complex state estimation for optimizing motor planning corrections on a per-movement basis.
• A more straightforward strategy is utilized, involving a fixed learning rate determined by the reliability of the error signal.