BME PhD Proposal Presentation: Andrew Law

Voluntary Modulation of Primary Motor Cortex Ensemble Activity for Brain-Computer Interface Control

Co-Supervised by Prof. Marc Schieber and Prof. Greg Gdowski

Abstract

For individuals with spinal cord injury, prosthetic devices controlled by neural signals can restore lost motor function and thereby improve quality of life. The conventional approach for neural prosthesis control involves sophisticated algorithms that decode recorded neuronal activity into prosthesis control signals. An alternative approach is to define a fixed transfer function which directly relates recorded neuronal activity to prosthesis movement. Rather than decode neural activity for prosthesis control, this method assigns prosthesis-related encoding responsibilities to recorded neurons. The utility of the prosthetic device then depends on the ability of the user to learn to voluntarily modulate and control the firing rates of these neurons.

Previous studies have shown that, when given direct feedback of neuronal activity, human and non-human primates can learn to voluntarily modulate the firing rates of single primary motor cortex (M1) neurons to control brain-computer interfaces (BCIs). Single-unit recordings, however, are non-stationary and can be unreliable for prosthesis control. The combined activity of M1 neuronal ensembles is less susceptible to non-stationarities in individual single-unit recordings and would be more reliable for controlling prosthetic devices. Whether or not primates can voluntarily modulate ensemble activity for prosthesis control remains an open question. In a preliminary study, we have found that non-human primates can learn to control a BCI by voluntarily modulating the combined activity of M1 ensembles consisting of two neurons.

Based on these initial findings, the aims of this proposal are:

1. Investigate the effect of M1 neuronal ensemble size on voluntary modulation of ensemble activity for BCI control;
2. Investigate the effect of M1 neuronal ensemble composition on voluntary modulation of ensemble activity for BCI control; and
3. Investigate the effect of BCI training on the co-modulation patterns of M1 ensemble neurons during natural motor behavior.

The results of this work will provide valuable insights into:

1. How the number and type of M1 neurons in an ensemble affects neural prosthesis control and
2. How learning to control a neural prosthesis affects the activity of ensemble neurons.