Abstract
Purpose. To characterize the dynamics of position computation for moving targets in human vision.
Methods. On a dark monitor, observers (N=3) viewed a rotating line that was straddled by two horizontally aligned flashes. The luminance of the rotating line was either 1.0 or 2.5 log units (LU) above its detection threshold. The luminance of the flashes varied from 0.2 to 4 LU above their detection threshold. To probe the steady-state phase of the position computation process of a moving target, we used the continuous motion (CM) paradigm, in which line rotation started long before the occurrence of flashes. To probe the transient phase of the position computation process, we used the flash-initiated cycle (FIC) paradigm, in which line rotation started concurrently with the presentation of the flashes. The observers judged the direction of spatial offset between the rotating line and the flashes at the instant when the flashes were perceived.
Results. For both luminances of the moving line, the perceived misalignment of the line was the same in the FIC and CM paradigms when the flashes were of relatively low detectability, but systematically grew larger in the FIC paradigm as the detectability of the flashes increased. For the brightest flashes, the difference in perceived flash misalignment between the FIC and CM paradigms reached 80 ms for the dimmer rotating line. For the brighter rotating line, the differences between the perceived misalignment in the FIC and CM paradigms were consistently about 25 ms less.
Conclusions. Because the latency of the flashes is expected to decrease as their detectability increases, the systematic luminance-dependent changes in perceived flash misalignment between the FIC and CM paradigms provide a quantitative characterization of differences between transient and steady-state phases of position computation for a moving target. Our findings are consistent with the differential latency hypothesis but not with the postdiction hypothesis.