A control-theoretic approach to brain-computer interface design

Brain-computer interfaces (BCIs) have the potential to restore motor abilities to paralyzed individuals. These systems act by reading motor intent signals directly from the brain and using them to control, for example, the movement of a cursor on a computer screen or the motion of a robotic limb. To construct a BCI, a mapping must be specified that dictates how neural activity will actuate the device. How should these mappings be constructed to maximize user performance? Most approaches have focused on this problem from an estimation standpoint, i.e., mappings are designed to implement the best estimate of motor intent possible, under various sets of assumptions about how the recorded neural signals represent motor intent. Here we forward an alternate approach to the BCI design problem, using ideas from optimal control theory. We first argue that the brain can be considered as an optimal controller. We then introduce a mathematical definition of BCI usability, and formulate the BCI design problem as a constrained optimization problem that maximizes this usability.