A virtual target controls fixation better than a remembered target

Previously we investigated how an eye is controlled during fixation when it is covered and unable to view a visual target that its fellow eye sees. During monocular fixation the two eyes moved differently - the covered eye traversed a larger region (assessed with bivariate contour ellipse area or BCEA) and drifted faster which we attributed to poorer fixation control of the covered eye using a “virtual” target constructed by the brain rather than the visual target. Here we ask how fixation is controlled when neither eye sees the target. Binocular eye movements were recorded simultaneously in a darkened room at 1000 Hz with an EyeLink. Observers fixated a small spot for 500 msec that disappeared while fixation of the remembered target was maintained for 19 sec. Compared to an occluded eye’s behavior during monocular fixation, here when the target’s position had to be remembered, the two eyes continued to move differently. However, relative to the occluded eye in the monocular condition, BCEAs in both eyes were considerably larger, drift was slower, saccade frequency was lower, and saccades were larger. However, as during monocular viewing, saccades remained conjugate. This suggests that the oculomotor strategies for following a remembered target are different from those used to follow a virtual target, and result in poorer control. Our new model of binocular control includes a virtual target to provide position error feedback to an occluded eye. Without the virtual target the eyes wander farther and produce less frequent saccades that apparently do not well correct for the eye drift. The differences between the movements of the two eyes provides further evidence that drift during fixation is controlled independently while microsaccades remain conjugate, as posited in our model.