Predictability Affects Temporal Bias in the Flash-Lag Illusion

Abstract

Abstract: : Purpose: In the flash-lag illusion, a flashed stimulus appears to spatially lag a continuously visible one. Typically, the spatial offset of the flash is manipulated to measure spatial bias. This parameter is often transformed into an equivalent temporal offset without measuring the temporal psychometric function directly. Recently, Murakami (Journal of Vision, 2001) measured temporal psychometric functions and concluded that the flash-lag effect can be explained solely by temporal parameters. If the temporal domain alone accounts for the effect, then flash predictability should not differentially affect performance. The current experiments measured the temporal bias directly by varying the temporal offset of a flash relative to a moving stimulus. The position of the flash was either predictable or not. Methods: Two equally-spaced, radially-aligned .3 deg₂ white dots on a gray background completed three circular clockwise rotations around a central dot of the same size at a speed of 1 Hz. The radius of the circle traveled by the outermost dot was 3 deg. Flashes occurred adjacent and eccentric to the outermost moving dot during the second rotation, and observers did not respond until all rotations were complete. Flashes could occur at one of five clockwise positions from vertical: 90, 135, 180, 225, and 270 deg. In the predictable condition, observers were informed of the flash position; further, positions were blocked over trials and presented in sequential order. In the unpredictable condition, flash positions were randomized and observers were not informed of the flash position. The time of the flash was manipulated in 13.3 ms increments relative to the time that the moving dots appeared adjacent to each position. Observers were instructed to determine if the flash was spatially leading or lagging. The mean (bias) of the resulting psychometric functions was estimated using a probit fit. Results: Unpredictable flashes typically generated a larger illusion than predictable flashes. Means for the predictable and unpredictable conditions differed by as much as 30 ms in some flash positions. Conclusion: The motion flash-lag illusion cannot be due entirely to temporal parameters because predictability differentially affected the magnitude of the illusion.

Keywords: 477 motion-2D • 510 perceptual organization • 596 temporal vision