Abstract
Purpose: To guide behavior usefully, color perception should predict color memory. A fruit's perceived color, for example, indicates ripeness more reliably if color memory of ripe fruit is accurate. However, some previous research suggests that color memory is poor relative to perception. Why? In everyday tasks, we often must perceive objects in one scene and recall them in a another. Scene changes affect color perception. Can imperfect color memory be explained by imperfect color perception? Methods: Observers viewed real painted wooden cubes in a study booth under a yellowish illuminant and selected a matching color in an adjacent test booth under a bluish illuminant. Eight cubes composed two groups (G1: pink, blue, green, orange, G2: brown, red, yellow, purple). Observers made simultaneous matches (perception) for one group, viewed the other group, waited 15 minutes, and made memory matches. To make matches, observers selected the best-matching paint from a commercial booklet of ∼1000 paint samples. In the test booth, we measured chromaticity (PR 655, u′v′) of each cube and each selected paint sample. Results: Observers selected many different paint samples for each cube in both perception and memory conditions; thus, paint samples were discretized with sufficient resolution to detect failures of constancy. The calculated difference in chromaticity (u′v′) between each painted cube and the selected paint sample (both measured in the test booth) served as a measure of constancy failure. As expected, perceptual constancy for each cube was partial but not complete, and the degree of constancy varied substantially between cubes. Notably, the extent of color constancy in memory was very similar to perceptual color constancy for almost all cubes, as was the variability of color constancy. Conclusions: The data suggest that imperfect color memory for real illuminated objects can be largely explained by a propagation of imperfect perceptual constancy into memory.