Adaptation paths to novel motor tasks are shaped by prior structure learning

Area of Science:

• Motor control
• Neuroscience
• Robotics

Background:

• Motor skills with shared structures are learned faster.
• The "structure learning" hypothesis proposes generalization relies on shared dynamic and kinematic relationships.
• This reduces learning dimensionality and biases adaptation to related tasks.

Purpose of the Study:

• To investigate how learned motor structures influence subsequent adaptation.
• To test predictions of the structure learning hypothesis.
• To determine if structure learning affects involuntary reflexes.

Main Methods:

• Human subjects performed reaching movements in 3D virtual reality.
• Subjects experienced vertical or horizontal plane perturbations (visuomotor rotations or velocity-dependent forces).
• Adaptation to novel, diagonal perturbations was assessed after training.

Main Results:

• Adaptation to unpracticed diagonal perturbations occurred along previously learned vertical or horizontal structures.
• Resulting adaptation trajectories were curved, demonstrating a bias.
• This effect was robust, observed at the single-subject level, and occurred within and across trials.
• Structure learning was shown to alter involuntary visuomotor reflexes.

Conclusions:

• Learned motor structures constrain subsequent adaptation, leading to biased and curved trajectories.
• Structure learning is not solely a high-level cognitive process, as it impacts involuntary reflexes.
• Findings support the structure learning hypothesis and highlight its role in motor skill generalization and adaptation.