Abstract
The visual system contains biases in sensitivity for selective features such as spatial frequencies around 3 cycles per degree and patterns with cardinal orientations. We hypothesized that a similar bias in perceptual sensitivity should exist for a selective combination of luminance- and color-contrast between a stimulus and its surrounding background. To find an optimal trade-off, we measured pupil orienting responses using gaze-contingent flicker pupil perimetry (gcFPP). We presented observers a sequence of stimuli, each flickering at 2 Hz for 5 seconds at a different angle and eccentricity across the visual field. To alter color contrast at the expense of luminance contrast with a stimulus, we systematically varied the color (dark blue or yellow) and luminance (0-45%) of the background. The stimulus-on color was complementary to the background (bright yellow or blue) and fixed at 100% luminance. To test local effects of luminance and color contrast changes, we also varied the stimulus-off color, being either black or displaying the background’s color. The amplitude of constrictions and dilations in response to stimulus on- and off-sets served as a proxy of visual sensitivity. We found an optimal pupil response amplitude for both yellow and blue stimuli presented on a background with 25-35% luminance. The amplitude decreased by 25% when displaying a colorless, black background. Furthermore, a black stimulus-off region amid a colored background evoked 23% stronger responses than a colored stimulus-off region. Modelling pupillary amplitudes as a reflection of the Euclidian distance between background and stimulus color in CIE LAB space suggested a sensitivity ratio of 0.8 for color- versus luminance-contrast. These findings imply (1) that a preferred trade-off between luminance- and color-contrast exists, (2) that global (not local) color contrast almost equally adds to visual sensitivity as luminance contrast does, and (3) that gcFPP efficiently measures color sensitivities.