Coming to an Understanding of Fixational Eye Movements

Abstract

Fixational eye movements play a major role in visual perception; without such movements the world would quickly fade away. We are using a mathematical description of fixational eye movements combined with a realistic biophysical model of the primate retina to obtain a better understanding of how fixational movements contribute to visual perception. Physiologically realistic jitter was used to modulate the retinal location of an otherwise stationary image, here consisting of Gaussian-weighted sinusoidal gratings with superimposed 1/f background noise. Images were reconstructed using either 1) the mean or 2) the cross-covariance of the simulated retina output. To assess the quality of the two reconstruction methods, reconstructed images were classified as either vertical or horizontal. Our preliminary findings suggests that the spatial correlations produced by fixational eye-movements can contribute to visual perception. Ultimately, the predictions of the model can be tested in psychophysical studies.