Abstract
A luminance-defined blob orbits fixation. Observers judge the blob's position at the time of a change in color of the fixation point or an auditory tone. In addition to the “flash-lag” error, we find over a range of speeds that the standard deviation of position judgments corresponds to a ∼100-ms portion of the blob's trajectory (Murakami 2001), which yields near-chance performance when blob speed reaches ~3 rps. Visual encoding at early stages is temporally very precise, and we seek to understand the source of the large variability in position judgments. The constant temporal variability suggests the positional uncertainty is imposed by a stage of low temporal resolution. To rule out low-level temporal smearing (Newton 1730), observers view two blobs, one orbiting fixation and the other orbiting a point in the periphery, and judge whether they move in phase or out of phase. Observers are nearly 100% accurate at revolution rates (3 rps) for which they are near chance judging position, inconsistent with the blur explanation. In the original task, if the number of blobs simultaneously sharing the orbit is increased, the blur theory predicts chance performance should occur at a lower speed, as a complete blur circle is reached at lower speed. However, we find that the limit is similar for 1, 2, and 4 blobs. Finally, as the radius of the orbit increases from 2 to 6 deg, a variety of low-level visual factors change, but again we find that temporal imprecision of judgments remains constant. Each result suggests that the low temporal resolution of position judgments is not imposed at early stages of visual processing. We speculate that the large temporal variability instead is mostly caused by the process of binding the temporal marker with the corresponding position of the moving object.
supported by Australian Research Council DP to AH