Decoding Hand Kinematics and Neural States Using Gaussian Process Model

Probabilistic modeling of correlated neural population firing activity is central to understanding the neural code and building practical decoding algorithms, the accurate reconstruction of a continuous motion signal is necessary for the control of devices such as computer cursors, robots, or a patient's own paralyzed limbs. For such applications we developed a realtime system that uses Gaussian process techniques to estimate hand motion from the firing rates of multiple neurons. Gaussian Processes for Machine Learning presents one of the most important Bayesian machine learning approaches based on a particularly effective method for placing a prior distribution over the space of functions. Decoding was performed using Gaussian processes model which gives an efficient method for Bayesian inference when the firing rates and hand kinematics are nonlinear. Gaussian processes provide a principled, practical, probabilistic approach to learning in noisy measurements. In off-line experiments, the Gaussian processes model reconstructions of hand trajectory were more accurate than previously reported results. The resulting decoding algorithm provides a principled probabilistic model of motor-cortical coding, decodes hand motion in real time, provides an estimate of uncertainty, and is straight to implement. Additionally the formulation unifies and extends previous models of neural coding while providing insights into the motor-cortical code.