Abstract
The perceived lightness of a focal stimulus depends on the context in which it is presented. In some cases these effects can be explained parsimoniously in terms of simple adaptive mechanisms, while in other cases such explanations are tenuous. Adelson's checkerboard illusion (http://web.mit.edu/persci/people/adelson/checkershadow_illusion.html) provides an interesting example: two gray squares of identical intensity, one in a shadow and one not, have very different apparent lightness. We performed experiments to determine the nature of the mechanisms underlying this effect. Our working hypothesis is that consistency between context effects revealed by appearance and discrimination data is a signature of early mechanisms, while a dissociation between the two indicates the action of higher-level processes. We have shown that context effects induced by changes of uniform backgrounds (JOSA, 2005) and relatively unstructured contrast are common to both appearance and discrimination.
Ten observers participated. Observers set lightness matches between stimuli presented at two locations in the checkerboard illusion (one in and one out of the shadow). We also measured discrimination thresholds for each observer at the two relevant locations. The same measurements were made for a control configuration where the appearance of shadow was eliminated from the checkerboard. he asymmetric matches revealed a large appearance effect in the illusion configuration, and almost no effect in the control configuration. The threshold measurements were unchanged between illusion and control configurations. In contrast with our results for simple context manipulations, lightness effects revealed by the checkerboard illusion cannot be explained by adaptive processes that mediate discrimination performance.