Abstract
To discover what heuristics observers use in identifying materials across illuminants, we placed randomly crumpled 3-D objects with precisely controlled spectral-reflectances, inside two adjacent compartments with a yellowish (Sun) illuminant on one compartment and a bluish (Sky) illuminant on the other. The compartments were lined with similar multicolored patterns. Three of the objects (Standards) were made from the same paper, while the fourth (Test) varied from the Standard along different color lines. On each trial, the observer was instructed to identify the object with the odd material. Along each color line, the threshold for discrimination within illuminants was obtained from the frequencies of correctly picking the Test compartment. The threshold for reflectance identification across illuminants was obtained from the frequencies of correctly picking the Test object.
For almost all Tests, whenever the Test was reliably discriminated from the Standard, it was also reliably identified as the odd material. However, for certain Tests, the Standard in the Test's compartment was reliably misidentified as the odd reflectance. In the Macleod-Boynton chromaticity diagram, these Tests plot near the line that passes through the Standard under the other illuminant and is approximately parallel to the line joining the means of the backgrounds across the illuminants. It seems that in separating reflectance changes from illuminant changes, observers use the heuristic that the color of a material should shift across illuminants along a line parallel to the color change for the backgrounds. This heuristic is an approximation to, but systematically different from, the generic physical change.
NEI Grants EY07556 & EY13312 to QZ