Brain-machine interface for eye movements

Area of Science:

• Neuroscience
• Neuroprosthetics
• Brain-Computer Interfaces

Background:

• Brain-machine interfaces (BMIs) have successfully decoded reach intentions in humans and nonhuman primates (NHPs).
• Previous research has not explored applying BMIs to decode intended eye movements.

Purpose of the Study:

• To investigate the feasibility of using BMIs to decode intended eye movements from neuronal activity.
• To assess the potential of such BMIs to aid paralyzed patients.

Main Methods:

• Recorded neuronal activity from the lateral intraparietal area (LIP) in NHPs.
• Used Bayesian inference to predict eye movement plans in real time from LIP neuronal ensembles.
• Decoded eye movement plans without requiring the animal to make an actual eye movement.

Main Results:

• Real-time prediction of eye movement plans was achieved using small ensembles of LIP neurons.
• Prediction accuracy improved through learning at the neuronal ensemble level, especially for challenging predictions.
• Population learning involved BMI parameter updates and changes in individual neuron responses.
• Decoded eye movement plans were used to control a computer cursor.

Conclusions:

• Neuronal ensemble responses can be shaped to optimize BMI prediction accuracy.
• BMIs for decoding eye movements are a promising assistive technology for individuals with paralysis.
• This study provides strong evidence for the potential of eye-movement-based BMIs.