In recent years, neuroscientists have made breakthroughs in understanding behaviour of individual neurons in the motor cortex. Their discoveries have, in theory, paved the way for a new generation of prosthetic devices that could be controlled directly from the brain. The idea would be to collect measurements of neuronal activity from electrodes implanted in the brain, perform real-time statistical signal processing on these data to make inferences about intended motion, and then use the results to drive (for example) a robotic arm. We discuss current laboratory work on a prototype of such a device, with particular emphasis on the statistical signal processing module. Other biological and engineering aspects of the project will also be described briefly. As proof of concept, we show how a trained monkey in the laboratory has been able to use the device to feed itself.
Group for Research in Decision Analysis