How does the brain learn to control neuroprosthetics?
Numerous studies have now shown the feasibility of direct “brain control” of a neuroprosthetic device, yet there is much to be discovered about the underlying neurophysiological mechanisms that make this possible.
Our thesis is that effective control of a neuroprosthetic must be learned, much in the same way that a new motor skill, like playing the piano, must be learned. Learning to move a limb or an external device both require an interplay between expectations, or intention of movement, and outcomes, or the observed and felt reality of changes in position. Integrated sensory and motor neural pathways must inform each other to control actions.
In closed loop brain-machine interface (BMI) systems, there is the opportunity to create a two-learner system, in which brain and machine adapt to each other to improve performance. We aim to understand how the brain adapts to the introduction of a neuroprosthetic and learns to control it, a type of abstract skill learning. Our approach is described below. In the next section you will see how we develop algorithms to improve decoder adaptation.