Orthogonalizing the Activity of Two Neural Units for 2D Cursor Movement Control

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computational Neuroscience

Background:

• Brain-machine interfaces (BMIs) require decoding neural signals for device control.
• Reducing the number of electrodes is crucial for practical BMI design.
• Understanding neural plasticity during BMI learning is essential.

Purpose of the Study:

• To investigate if two neurons can effectively control a two-dimensional cursor in a BMI.
• To analyze changes in neural tuning during BMI training.
• To explore the mechanisms of motor learning in BMIs.

Main Methods:

• Training a monkey to perform a 2D center-out task using only two neurons' spiking activity.
• Monitoring and analyzing the preferred direction tuning of these neurons during training.
• Assessing the relationship between neural activity and cursor movement.

Main Results:

• Two neurons were sufficient to achieve smooth 2D cursor control.
• The preferred directions of the two neurons gradually shifted during BMI learning.
• The angle between the preferred directions of the neuron pairs converged towards 90 degrees, indicating increased independence.

Conclusions:

• This research demonstrates the feasibility of controlling a 2D cursor with only two neurons in a BMI.
• BMI learning promotes neural independence, suggesting an efficient motor control strategy.
• Motor cortex plasticity underlies the adaptation of neural signals for effective BMI operation.