Oculomotor Hide and Seek: Pursuing an Accelerating Target Behind an Occluder. Focus on “Target Acceleration Can Be Extracted and Represented Within the Predictive Drive to Ocular Pursuit”

Abstract

Smooth pursuit eye movements are used to maintain the image of a moving object on the fovea. In the laboratory, most smooth pursuit studies are conducted with a target that travels at a constant velocity in clear view. In this situation, visual feedback created by the target’s motion on the retina allows the pursuit system to keep the fovea directed toward it. However, most objects moving in a natural scene accelerate or decelerate, and often their path is occluded by other objects. The problem of pursuing a real object becomes even more vexing for the pursuit system because of the well-known difficulty of humans to perceive acceleration, and evidence that smooth pursuit suffers from the same limitation. In the paper by Bennett and colleagues (2007), in this issue of Journal of Neurophysiology (p. 1405–1414) the authors investigate pursuit of accelerating targets that disappear behind a virtual occluder. Here, they present novel data showing that the pursuit system can estimate where an accelerating target will appear and how fast it will be moving after the occlusion. In the paper, the authors entertain three hypotheses. The first is a “final velocity” hypothesis, in which observers use the velocity of the target immediately before it disappears to estimate its velocity when it reappears. This hypothesis is based on the prediction of pursuit models that an efference copy signal generated by recent pursuit of a target can maintain eye velocity for a short duration. The second is an “average velocity” hypothesis in which average target velocity over the preocclusion viewing period is used to estimate the speed of the target on reappearance. The third “acceleration” hypothesis states that the acceleration of the target is computed by the pursuit system and that this computation guides the eyes after the occlusion. To test between these hypotheses, observers were presented with a target that moved at different initial speeds, but maintained a constant acceleration throughout each trial. The target was visible for either a brief interval (200 ms), or a longer one (500 or 800 ms), before disappearing from view behind a virtual occluder for 500 ms. The task was designed so that the different target velocities were independent of target acceleration at the time of target disappearance, and the mean target velocity during any interval of the initial visible ramp (except the final 50 ms) had an inverse relationship with target acceleration. Design of the experiment is clever, in that it disambiguates target acceleration, average target velocity, and final target velocity before target disappearance, allowing a clean test of the three hypotheses. When preocclusion target viewing was limited to 200 ms, subjects were unsuccessful at matching