Non-feature tracking in stereoscopic motion discrimination at short durations

In this study we tested the hypothesis that stereoscopic (cyclopean) motion is processed by a low-level mechanism that does not track features (Patterson, 2002). We conducted a motion discrimination experiment using stereo and luminance stimuli while we measured eye movements. The stimuli were luminance and stereo gratings with spatial frequency of 0.2 and 0.4 c/deg, drifting at temporal frequencies of 1 and 4 Hz and two orientations, vertical and horizontal. The horizontal disparity of stereo stimuli was 2.4 arcsec and for luminance stimuli the contrast was 0.8. Proportion of correct responses were estimated using the method of constant stimuli testing five durations between 63 and 398 ms. To measure eye movements, we used the EyeLink 1000 in binocular mode with a temporal resolution of 2000 Hz. Results showed that for all conditions and durations higher than 100 msec the proportion of correct responses was higher than 0.66 for stereo, and for durations higher than 63 ms the proportions were higher than 0.85 for luminance. For vertical luminance stimuli, we found that eyes only moved in the direction of stimulus motion for durations higher than 150 ms; and for stereo, we found small eye movements (<0.2 deg) in the direction of motion only for the longest duration (398 ms). Thus, no feature tracking was found for direction discriminations at short durations in both luminance and stereo. For horizontal luminance stimuli we found the same pattern of eye movements, however, for stereo we found that independently of the stimulus motion (upwards or downwards), eyes always moved downwards about 0.2 deg. Thus, considering all conditions, our results suggest that stereoscopic motion is processed by a low- level mechanism that does not track features.