Under many conditions below the Nyquist limit, a drifting grating briefly displayed on a CRT is consistently perceived as moving in the opposite direction. Taking into account the sample-and-hold operation of CRT’s, we have derived the temporal-frequency spectra of the displayed gratings and found that they have a broad range of components moving in either direction. We have measured the perceived direction of motion of over 150 different short-duration stimuli, and we have studied the relation that performance bears to narrow-band power imbalance—the normalized difference between power in the positive- and negative-frequency half-lines within a specific band. Perceived direction of motion is highly related to power imbalance in 1-Hz-wide bands centered between 10 and 15 Hz, but none of these bands alone can account for more than 84–91% of the variance of the data, and each band ostensibly fails to explain data from a subset of the stimuli. When broadband power imbalance is determined by weighting the spectrum with an inverted-U-shaped function peaking at ∼12 Hz, the explained variance increases to 91–97%. Our results suggest that the imputation of direction of motion to stimuli with complex spectra is based on broadband power imbalance determined after weighting the temporal-frequency spectrum with an inverted-U-shaped function.
© 1999 Optical Society of AmericaFull Article | PDF Article