Abstract
A well-known asymmetry exists between brightness induction for incremental and decremental targets, which is often characterized by the dogma that darkness induction is stronger than lightness induction. We explored this asymmetry with a simple brightness matching paradigm and discovered some novel quantitative properties. Observers adjusted the luminance of a matching disk to match its brightness to that of a fixed test disk surrounded by an annulus. The annulus luminance was experimentally varied to induce a brightness change in the test. In four experimental conditions, the test disk was either a luminance increment or decrement with respect to its annulus, and the background field (against which both the matching disk and the test disk/annulus pair were presented) was either white (highest display luminance) or black (lowest luminance). Seven naïve observers ran in all four conditions. When their average brightness matches were plotted against the annulus luminance on a log-log axis, the results from each condition were well-fit with a 2nd-order polynomial regression model. Increasing the annulus luminance increased the test brightness over one range (assimilation) and decreased it over another (contrast). We examined the parameters of this regression model and discovered that the ratio of the 1st- and 2nd-order coefficients of the model assumes one value for incremental stimuli and another for decremental stimuli (but is the same for white and black backgrounds). We hypothesize that this pattern reflects the different underlying neural gains of ON and OFF cells in primate: a factor that plays a key role in a recent neural edge integration model of dynamic range compression in the staircase Gelb effect (Rudd, this meeting). Although the individual observers in our study made different matches that may have been influenced by top-down factors, we propose that the statistical properties of their brightness matches were thus constrained by their physiology.