The EEG signals used in this study are inherently quite noisy and low bandwidth and are unlikely to yield a clinically useful neural prosthetic device. There a

The EEG signals used in this study are inherently quite noisy and low bandwidth and are unlikely to yield a clinically useful neural prosthetic device. There are alternatives, but these usually involve (a completely reasonable and clinically justifiable) surgery. Intracortical electrode arrays are used by the Braingate program and other BMI researchers, provide much higher bandwidth, signal to noise, robustness, and are likely necessary for real-world utility for tetra or paraplegic individuals. Since people like disclosing things here, I work in a lab that's one of the Braingate sites and we're biased towards implantable arrays, as no other system can currently in 2015 achieve the same levels of performance.