Abstract
Aside from well-known local contrast effects on brightness, several studies also have suggested a process from the contrast of a nonlocal border affecting brightness (often depicted as a ‘filling-in’ spreading from the border). These include studies on the Craik-Cornsweet-O'Brien illusion, and on brightness induction and perception (e.g., ‘assimilation’, Reid and Shapley, 1988; Paradiso and Nakayama, 1988). We used classification images (Ahumada and Lovell, 1971) to estimate the perceptual filter for contrast detection as a method to assess both local and nonlocal contrast effects. Three observers performed a yes/no detection task of a uniform circular decrement near threshold (radius = 0.68 deg, contrast = 3.1%) against a uniform background (luminance = 30.0 cd/m2) presented for either 100, 200, or 400 ms. Surrounding the signal were ‘rings’ of image noise that varied with radial distance, but were uniform for a given radial distance. This radial noise covered only the surrounding area (out to a radial distance of 1.36 deg), and not the signal itself. For all stimulus durations, evidence for nonlocal effects were found at the border between the surrounding image noise and the uniform background; a positive contrast for the surround (relative to the background) led to judgments of a signal decrement (approximately 0.60 deg from that contrast). Classification images for all stimulus durations also indicated a local contrast effect at the border between the signal and the surrounding image noise, and an intermediate band of the surrounding image noise that had little effect upon the judgment. Lastly, little or no change was found in the classification images across the different stimulus durations. The findings suggest local and nonlocal contrast processes on brightness that are completed within 100 ms; they are consistent with other studies of brightness induction and perception, and with the luminance profile for the Craik-Cornsweet-O'Brien illusion.