Investigation of the Pursuit Eye Movement Under-Registration Hypothesis (Tracking).

Abstract

Three models, which propose that pursuit eye-movement velocity is poorly sensed by human observers, were described and tested. All of the models explain why the speed and extent of the movements of tracked visual targets are underestimated by suggesting that pursuit eye-movement velocity is under-registered. The models differ in the particular way velocity is hypothesized to be under-registered. The constant-velocity under-registration model, suggests a fixed loss of approximately 1(DEGREES)/sec occurs during pursuit. Another model suggests that a proportional loss in perceived velocity occurs. A third model suggests that velocity is not directly registered; instead a fixed-velocity assumption of .1(DEGREES)/sec to 1(DEGREES)/sec is substituted by the perceptual system. Specific predictions that these models make about localization accuracy as a function of pursuit eye-movement velocity and extent were examined in Experiment I. Although localization errors were consistent with the general under-registration hypothesis, differences in localization were observed for conditions that the specific models predicted would produce no differences. Retinal slip of the tracked target, produced at the start of the target's movement, was next examined as a possible reason for the failure to find support for the models. Unlike constant-velocity target motion, harmonically ramped target motion is reputed to be tracked with less retinal slip of the target. Experiment II measured the effects of both constant and harmonically ramped target velocity on localization accuracy. Ramped target velocities were (1) tracked with less retinal slip, and (2) resulted in greater localization errors than did constant-velocity targets. The pattern of localization errors, however, continued to be inconsistent with the models' predictions. A different strategy was next employed to test a property that all three models share: that under-registration effects should add together. Experiment III tested the hypothesis that under-registration errors cancel in the pursuit of reciprocally moving targets. No evidence of cancellation was found. In particular, the reciprocal-tracking data resembled the pattern found when only the second half of the reciprocal trajectory was tracked. Experiment IV directly compared localization after reciprocal and non-reciprocal tracking, matched so that the second half of the reciprocal trajectory was the same as the non-reciprocal trajectory. Little evidence of a difference between the two conditions was found, suggesting that additivity of under-registration does not occur. Four plausible explanations of the non-additivity effect were proposed.