要旨
We extended and reanalysed our recent data on motion extrapolation (Sokolov et al, 1997 Perception 26 875 - 889) to examine the timing accuracy as a function of target size, speed, and visible motion path. A target (horizontal pair of dots separated by either 0.2 deg--small, or 0.8 deg--large) moved at a constant speed of either 2.5, 5, or 10 deg s-1 across a horizontal path (2.5 deg--short, or 10 deg--long) and then vanished. Observers pressed a key when they judged that the leading dot of the occluded target had reached one of seven randomly presented positions between 0 and 12 deg.
Similar to earlier findings (eg Yakimoff et al, 1993 Human Factors 35 501 - 510), mean extrapolation response time was found to exceed the arrival time. With long visible paths, the timing accuracy (as indicated by absolute error) was better for moderate and high than for low velocities. This difference increased with increasing extrapolation interval owing to a progressive loss of accuracy for low velocities. However, with short paths, performance was much more accurate for the low than for the moderate velocities even across long extrapolation intervals. The magnitude of errors increased as intervals got longer, with a markedly steeper increase for moderate speeds. Similar effects as those with the short path were also found with different-sized targets; they were stronger for large than for small targets. The findings suggest that the visual system implements different scaling algorithms for motion perception and extrapolation of different-sized stimuli depending on target speed. At higher speeds, processing of visible and occluded motion is likely to share a common scaling mechanism.
