Abstract
Whether or not visual perception is penetrable by high-level cognition has been controversial. Categorical color perception is an example of how cognition modulates perception, which claims that colors across categories (e.g. blue, green) are more discriminable than within category (e.g. green). Most studies on categorical color perception have been conducted in a static visual scene. Here, we investigated the effect of color categories on performance in multiple object tracking (MOT) with two experiments. In MOT, all the targets were assigned a hue varying from blue to green (hue angle: 205° to 170°, step size: 5°) according to condition, while all the distractors were assigned one blue hue (210°) across all conditions. After the MOT task, subjects were required to complete a color naming task to determine the category membership of the stimuli. We fitted the categorization performance with a Sigmoid function and calculated the green-blue border for each subject. The tracking performance was then fitted as a function of the target hue angle. If there was categorical effect, we should expect that the greatest discriminability of the hue angle in MOT would occur around the green-blue border. That is, the hue angle along the tracking curve with the maximum slope should be near the green-blue border. Analysis of correlation between the hue angle with the greatest discriminability in MOT and the green-blue border was then conducted and no significant correlation was found (Experiment 1), including no correlation in the right visual field (Experiment 2). The non-significant correlation suggested that the discriminability of hue differences in MOT is free from the categorization of hues. Yet, Experiment 2 revealed a significant negative correlation in the left visual field, suggesting that color categorization may derive from indiscriminability of chromaticity. Overall, our study supported the noncategorical color perception in MOT.
Meeting abstract presented at VSS 2018