Abstract
Here we provide a unified account of the relationship between perceived speed and signal intensity in random dot kinematograms (RDKs). Previous studies have shown a mixed pattern of results with both increases and decreases in perceived speed as a response to increases in signal intensity (coherence). These differences may be accounted for by looking at the temporal characteristics of the patterns used. We demonstrate that, for speed integration over an arbitrary temporal window, reduction of update rate can have dramatic consequences on the resultant speed distribution. Low update rates result in a bimodal speed distribution. We propose that observers segment such speed distributions, choosing the higher speed dots as signal and treating the remainder as background. Our analysis reveals that this can readily account for the inverse relationship between signal intensity and perceived speed found in patterns with low update rates. In contrast, with high update rates a unimodal distribution is found that can correspondingly account for the increase in perceived speed found with increases in coherence. We support these findings by introducing the notion of “trajectory coherence” in RDKs. This allows us to titrate between RDKs in which signal dots remain as signal dots for their entire trajectory and RDKs in which a dot's allocation to signal or noise is decided on a frame-by-frame basis. This manipulation affects the speed distribution histogram of the RDK whilst holding pattern coherence and update rate constant; increasing trajectory coherence results in an increase in the bimodality in the speed distribution. In support of the analysis presented above we find that increasing trajectory coherence (and thus bimodality of the speed distribution) results in increases in perceived speed. Our results indicate that perceived speed is the result of both temporally integrative and temporally segregative processes.