Abstract
Are microsaccades simply random movements that return a drifting eye to a fixation point? Or do microsaccades and drift play a functional role in visual analysis?
Our work on lightness algorithms for computer vision suggests that movements on the temporal and spatial scale of microsaccades and drift might be extremely useful for measuring colorimetric phenomena that are critical for recognizing illumination flux. To test this hypothesis, we created a number of different stimuli that simulate the spatial distribution of color and intensity, i.e., the spatiospectral order, typically created by shadows or, more generically, illumination boundaries. As a control, we generated another set of stimuli constituted of exactly the same colors in the same quantities but that did not contain patterns plausibly created by illumination boundaries, and thus appeared to represent only material boundaries. We then monitored the microsaccades and drift of 40 subjects who were asked to free-view each of the stimuli for 20 seconds each.
Our experiments suggest that microsaccades and drift are indeed substantially controlled by the high-order spatiospectral differences between material and illumination boundaries, often thought to be accessible only at the cortical level. To a significant extent, microsaccades and drift are not random and may be tied to the functional detection of illumination flux and shadows.