Abstract
Simultaneous contrast adaptation was measured using actual illuminated surfaces in a large experimental chamber (82 cm wide × 112 cm tall × 164 cm deep). Three walls and the floor of the chamber, all formed from identical colored cardboard (gray, pink, or blue), were visible. In some conditions, two ceramic mugs and a piece of cardboard having 24 colored patches were placed in the chamber to add a variety of illuminant cues to the scene. A square cardboard annulus (3 sizes, same or different color to the walls) was attached to the back wall of the chamber. In the center of the annulus was a test square whose chromaticity and luminance could be varied continuously by the observer. Two color-normal observers adjusted the test square to appear perfectly achromatic and so that the achromatic sensation was composed of 50% black and 50% white (three-dimensional RGB adjustment, 32 settings/condition). Illumination was identical in all 21 conditions [CIE (x,y,Y)=(0.34, 0.34, 20.9)]. Results: Simultaneous contrast was weaker for both observers when objects were placed in the chamber, indicating that the objects reduced the influence of the local surround on color perception. Since the objects should not alter the information provided by the local surround, the result also suggests that the objects caused the visual system to depend more on other illumination cues. Although difficult to estimate precisely, changes in more global image statistics (i.e. weighted spatial mean) would seem to be too small to explain the observed reduction in simultaneous contrast. The reduction in simultaneous contrast was much greater for one observer than for the other, implying significant individual differences in the way that information from different illuminant cues is combined. Shifts in achromatic settings due to simultaneous contrast represent failures of color constancy (illumination was constant), but the shifts suggest how the visual system might achieve color constancy when illumination does change.
Supported by EPSRC grant GR/S95558