Decoupling simultaneous motor imagination and execution via orthogonal ECoG neural representations.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Brain-Computer Interfaces

Background:

• The brain manages multiple motor programs concurrently, but brain-machine interfaces (BMIs) struggle with simultaneous tasks.
• Latent neural representations are proposed as a solution to improve BMI performance during concurrent movements.

Purpose of the Study:

• To investigate if latent neural representations allow for independent decoding of executed and imagined movements using electrocorticography (ECoG) signals.
• To determine if separable neural subspaces exist for concurrent motor tasks.

Main Methods:

• Utilized neural signals from a tetraplegic individual with a wireless epidural electrocorticography (ECoG) device.
• Applied dimensionality reduction techniques to analyze mesoscale neural signals during motor execution and imagery.
• Identified frequency band contributions shaping neural signal subspaces.

Main Results:

• Motor execution and motor imagery were found to occupy partially overlapping subspaces in neural signals.
• Despite shared variance, orthogonal, condition-specific dimensions were identified.
• Successful decoding of simultaneously executed and imagined movements was achieved.

Conclusions:

• ECoG signals reveal separable neural subspaces for executed and imagined actions.
• Independent control of executed and imagined movements is feasible, even when performed concurrently.
• This research paves the way for advanced BMIs capable of simultaneous control and natural movement integration.