Brain Machine Interfaces have witnessed a revolutionary progress in the last few years, providing a proof of concept that brain signals can be effectively used to actuate artificial devices, such as computer cursors on 2D screens or robotic arms moving in 3D space. Their potential use in clinical applications, however, has been hindered by numerous challenges, ranging from diminished information yield over chronic use, to highly nonstationary and heterogeneous characteristics of neuronal activity patterns. For this technology to be clinically viable, it needs to address many of these challenges to pave the way for people with severe sensory, cognitive and motor deficits to significantly improve their lifestyles

This project specifically addresses neural decoding - the problem of translating information in neural signals, collected at cellular and population levels, to control signals that actuate the artificial device. The objective is to provide the most effective means of controlling a robotic arm and hand using a population of cortical neurons recorded through implanted microelectrode arrays in primary motor, premotor and supplementary motor areas of non-human primates. The outcome of these studies is expected to significantly improve the ability of neural decoders to generalize to novel tasks under a variety of contexts, and provide dexterous motor control of artificial limbs, such as reaching to grasp and manipulate external objects.