Neural decoding of movements: From linear to nonlinear trajectory models.
To date, the neural decoding of time-evolving physical state for example, the path of a foraging rat or arm movements has been largely carried out using linear trajectory models, primarily due to their computational efficiency. The possibility of better capturing the statis- tics of the movements using nonlinear trajectory models, thereby yield- ing more accurate decoded trajectories, is enticing. However, nonlinear decoding usually carries a higher computational cost, which is an im- portant consideration in real-time settings. In this paper, we present techniques for nonlinear decoding employing modal Gaussian approxi- mations, expectatation propagation, and Gaussian quadrature. We com- pare their decoding accuracy versus computation time tradeoffs based on high-dimensional simulated neural spike counts.