Generalization of motor learning based on multiple field exposures and local adaptation

Area of Science:

• Motor control
• Computational neuroscience
• Human motor learning

Background:

• Previous studies suggest motor learning involves encoding dynamics in either extrinsic or intrinsic coordinates.
• Transfer of learning is observed when torque patterns are similar, supporting intrinsic (joint-based) encoding over extrinsic (space-based) encoding.

Purpose of the Study:

• To investigate the generality of intrinsic coding of dynamics.
• To examine how motor learning generalizes when torque patterns vary across the workspace.

Main Methods:

• Experiments involved examining transfer of learning after adaptation to force fields with fixed or varying torque patterns.
• Simulations were used to assess generalization characteristics, particularly after adaptation to opposing force subfields.

Main Results:

• No beneficial transfer effects were found after single-location training.
• Excellent performance generalization was observed at a central workspace location after training at two lateral locations.
• Data support the hypothesis that dynamics information is encoded in intrinsic coordinates.

Conclusions:

• The nervous system encodes dynamics in intrinsic coordinates.
• Generalization of dynamics learning occurs through interpolation between locally adapted control signals.
• This mechanism allows the brain to adapt to varying dynamic environments.