Closed-loop error damping in human BCI using pre-error motor cortex activity

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Intracortical brain-computer interfaces (BCIs) decode neural activity for motor intent but struggle with real-time performance compared to able-bodied individuals.
• Restoring function for individuals with motor deficits using BCIs remains a significant challenge, necessitating improved control accuracy and reliability.

Purpose of the Study:

• To investigate the utility of a neural error signal for real-time error detection and correction in closed-loop motor BCIs.
• To enhance the performance and usability of intracortical BCIs for individuals with motor impairments.

Main Methods:

• Analyzed neural data from four individuals with spinal cord injury performing cursor control tasks using intracortical BCIs.
• Developed and implemented a classifier to detect a neural error signal in parallel with motor decoding for online error modulation.
• Assessed performance improvements in cursor kinematics and complex motor tasks with and without error modulation.

Main Results:

• Identified a pre-error component in cortical activity, enabling earlier error detection before kinematic errors occur.
• Demonstrated that error modulation significantly improves online BCI control performance for cursor kinematics.
• Showcased the robustness of error modulation across different motor tasks and environments without task-specific recalibration.

Conclusions:

• Neural error signals can be reliably detected and disentangled from motor intent in cortical activity.
• Error modulation, even with simple classifiers, substantially enhances the accuracy and reliability of BCI control.
• This approach offers a promising strategy for improving the clinical applicability of motor BCIs.