Learning a novel myoelectric-controlled interface task.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Robotics

Background:

• Effective control of myoelectric prostheses and brain-machine interfaces (BMIs) necessitates learning complex neuromotor transformations.
• Understanding the neural mechanisms for adapting to novel motor control schemes is crucial for advancing assistive technologies.

Purpose of the Study:

• To investigate the human motor system's capacity to learn abstract neuromotor mappings for controlling a myoelectric interface.
• To determine the influence of muscle selection (hand vs. arm) and sensory feedback on learning and performance.

Main Methods:

• Subjects controlled a 2D cursor using electromyogram (EMG) signals from hand and arm muscles with an immobilized limb.
• Interface mappings were either intuitive or nonintuitive, with varying cursor variability.
• Proprioceptive feedback was degraded using vibration in a second experiment.

Main Results:

• Subjects successfully learned even nonintuitive cursor control mappings to high performance levels.
• Muscle-tuning functions adapted to minimize cursor variability, with a preference for hand muscles due to their greater capacity for novel synergies.
• Vibration-induced sensory degradation impaired task performance but did not alter the learning rate.

Conclusions:

• The motor system can acquire internal models of abstract neuromotor mappings without overt movement or accurate proprioception.
• The distal motor system (hand muscles) offers more flexible control signals for neuromotor prostheses compared to proximal arm structures.