The Neural Basis of Brain-Computer Interface Control
Despite substantial progress, brain-computer interfaces (BCIs) have yet to fulfil the promise of enabling communication with completely locked-in subjects, e.g. with those in late stages of amyotrophic lateral sclerosis (ALS). In this talk, I argue that one reason for this shortcoming is the prevailing research focus on neural signals that provide information on a subject's intent, neglecting those processes that determine whether a subject is in a state of mind suitable for operating a BCI. I discuss our recent efforts in understanding the neural processes that determine performance in BCIs based on sensorimotor-rhythms (SMRs), and present empirical evidence that fronto-parietal gamma-oscillations can be used to predict a subject's capacity to operate a BCI on a trial-to-trial basis. I further demonstrate that subjects can learn to self-regulate fronto-parietal gamma-oscillations by neurofeedback, and argue that this provides interesting insights into the attentional states involved in BCI control. I conclude my talk by discussing our recent efforts in training a locked-in ALS patient to modulate fronto-parietal gamma-oscillations.