Abstract
It is known that a moving stimulus appears to last longer than a static one. To assess how duration overestimation relates to parameters such as spatial frequency (SF), temporal frequency (TF), and speed, we measured the perceived duration of drifting Gabor patches, with various combinations of SF (0.5, 1, 2, 4 c/deg) and TF (1, 2, 4, 8, 16 Hz) of the carrier sinusoid (tapered by a static Gaussian envelope). The physical durations were ranged within 0.3-1 s. In Expt. 1, we used the matching technique to determine the physical duration of the static patch that just appeared the same duration as the moving patch. As a result, we obtained overestimation ranging 110–160%, as an increasing function of TF and a decreasing function of SF. When data were replotted against speed, however, all data collapsed into a logarithmic increasing function of speed irrespective of SF and TF. In Expt. 2, we used the reproduction technique, in which the subject first observed a stimulus and then held a button for the same interval as the perceived duration. Again, overestimation was better described as a logarithmic function of speed, rather than SF or TF. These results apparently contradict the recent finding by Kanai et al. (2006), who noticed overestimation as an increasing function of TF, not speed, for a variety of stimulus configurations. We looked at the Fourier domain of their motion stimuli and ours and found that, whereas our Gabor patches contained a well confined frequency band centered at the desired frequency, Kanai et al.'s motion stimuli had strong DC component in SF power spectra leading to standing flicker. Therefore, we reconcile the apparent discrepancy by the conclusive remark: speed governs the illusion when drifting motion is dominant, whereas TF governs the illusion when flicker is dominant in the stimulus.