Toward High-Performance Neural Control of Prosthetic Devices

Neural electrical signals would be processed into electronic control signals.

A program of basic and applied research in neuroscience is dedicated to (1) advancing fundamental understanding of how the human brain plans and executes arm movements and (2) designing and building high-performance neural prostheses for controlling arm prostheses. The basic-research part of the program involves experiments on non-human primates by use of techniques of chronic-electrode-array electrophysiology, computational neuroscience, theoretical neuroscience, and observations of reaching behavior. The appliedresearch part of the program includes, as part of the effort to develop neural prostheses, an effort to decode (that is, to extract scientifically and prosthetically useful signals from) neural activity in real time, use the signals generated in the decoding process to move computer cursors, and utilize the knowledge thus gained to design and validate high-performance neural-prosthetic algorithms.

The main scientific findings thus far concern the preparation (including planning) and execution of arm movements. It was found that the neural activity associated with preparation of an arm movement could be mathematically modelled as an attractor system wherein neural activity becomes more "accurate" as planning proceeds. Another finding is that the speed of a planned arm movement (and not merely the direction and distance) is also planned.

The main accomplishment of the applied part of the program was the design and demonstration of an unprecedentedly fast and accurate neural prosthetic system that included implanted electrodes and electronic brain/computer interface circuitry. The development of this system was part of the first study to demonstrate that neural prosthetic systems that rely on implanted electrodes can substantially outperform systems that rely on such relatively noninvasive neural-signal receptors as surface electrodes used in electroencephalography.

Other accomplishments of the program include the following:

  • Development of low-power analog and analog-to-digital-converter electronic circuits that could be suitable for surgical implantation for processing neural signals; and
  • Investigation of the effects of unavoidable movements of implanted electrodes in relation to neurons and initiation of a concomitant effort to develop both means of correcting for such movements and less-invasive means of sensing neural signals.

This work was done by Krishna Shenoy et. al. of Stanford University for the Naval Research Laboratory.



This Brief includes a Technical Support Package (TSP).
Document cover
Toward High-Performance Neural Control of Prosthetic Devices

(reference NRL-0021) is currently available for download from the TSP library.

Don't have an account? Sign up here.