Abstract
In a transparent motion display, with multiple motions and colors presented in the same region, if a new color appears, noticing this change is easy. However, when the color-motion pairing of all the dots is reversed, but no new colors are introduced, we find that detection of the change is difficult. Our results suggest that color-motion pairings are not readily available to change detection.
Transparent motion was created with a field of dots. Half moved leftward, half moved rightward, half were red, and half were green. We manipulated the consistency of color and motion pairing— the proportion of red (green) dots moving in a particular direction. In the 0% consistency condition, half of the red dots moved left and half moved right, and the same for the green dots. Here, swapping the colors of all dots does not change which colors each surface has. In the 100% consistency condition, all dots moving in one direction are red before the swap and green after the swap, such that each surface's color changes completely. To avoid tracking of individual dots, a concurrent secondary task at fixation was included. Observers had great difficulty in detecting a color swap in the zero consistency condition (d'=1.2), and performance improved dramatically as the consistency increased (d'=3.0-for 100% consistency), suggesting that a color swap could be detected only when a surface's color changed. This effect of consistency was observed even when all dots had the same direction (which was perceived as a single surface). But with stationary dots, performance was perfect, suggesting that the impairment was specific to tracking of moving dots. Taken together, these results show that humans are remarkably blind to changes in local color-motion binding, unless these changes lead to a change in the features that belong to a surface.
Support from Global COE (D07 to Kyoto Univ.) and Grant-in-Aid for scientific research (C) (#19500226) from JSPS to JS, and an Australian Research Council DP to AOH.