Relative contributions of luminance and chromaticity to the Craik—Cornsweet effect

Abstract

A Craik—Cornsweet (CC) edge resembles the spatial double derivative of a smoothed step edge. Traditionally, such stimuli are presented as black-white profiles and a difference in brightness is observed on either side of the edge which does not directly correspond to the luminance profile. Our stimuli were generated on a fully calibrated colour monitor optically incorporated within a Maxwellian view system. We used a radially symmetric version of the CC stimulus, presented transiently for 200 ms on a uniform steady white background field. The profile constituted a step edge, blurred with a 0.25° Gaussian, which was then differenced with the original step edge. The contrast of the stimulus could have any direction in L, M, S cone contrast space. Detection thresholds were obtained in two theoretically important planes in cone contrast space. The CC effect was measured by determining the contrast of a step which when added to the stimulus cancelled the effect. These cancellation results were obtained in the two planes at 6x threshold for each of three observers. Consistent with the spatial filtering characteristics of the luminance and chromatic systems, we found the CC effect to be predominantly driven by the luminance component of the stimulus but reduced by any chromatic component. The achromatic direction had the greatest effect.