Abstract
One of the major theoretical views of lightness perception asserts that the highest luminance in an image serves as a standard that is assumed to be white (Gilchrist et al., 1999). We performed a number of experiments that challenge this view. In one set of experiments, five sets of paper Mondrians that spanned four ranges of reflectances (12%−3.1% at one end, and 27.2%−12% at the other) were presented in a black room. Observers matched the reflectance of the highest luminance to a set of standard papers in a separate room. Four separate experiments were run, using two different backgrounds for the matching papers (an articulated random-dot background, and a white background); and two different illumination levels. Although consistently overestimated, the highest luminance was never perceived as white: the highest luminance paper's reflectance could be perceived as low as 39%; the highest perceived value was 74.1%. Both chart type and illumination level had significant effects on perceived lightness: papers in higher illuminants were perceived as lighter, and matches using the white background match pattern was perceived as higher. This experiment was replicated using monitor displays that closely matched the luminance values of the paper Mondrians. We then conducted a more extensive study using monitor displays depicting a Mondrian with a (simulated) 4:1 reflectance ratio range under 13 different simulated illumination levels (ranging from .06–.0015 cd/m2, to 61.8 to 36.05 cd/m2). Lightness matches revealed that the highest luminance continuously increased as simulated illumination increased, exhibiting a compressive nonlinearity similar to Steven's Power law. These results demonstrate that the visual system does not treat the highest luminance as an anchor, but rather, that perceived lightness increases as a monotonic function of luminance.