Neurosignal analysis during saccadic eye movements

An inverse method is developed to investigate the oculomotor neural control signal. The oculomotor plant is modeled with a fourth-order linear homeomorphic saccadic eye movement model. Parameter estimation is performed using a conjugate gradient search method which minimizes the integral of the absolute value of the error squared between the model and the data. Derivatives are computed using band-limited differentiation (BLD) techniques and the results are compared to derivatives computed using the two-point central difference method. Using the BLD technique allows for greater flexibility in choosing a cutoff frequency, provides true low-pass filtering above the cutoff frequency, and gives more consistent results than using the two-point central difference method. The input to the oculomotor plant model is computed. The input is a cumulative force consisting of the agonist and antagonist active state tensions and their respective rate-of-change terms. The agonist force is also determined by approximating the antagonist active state tension and removing if from the cumulative input force.