Cameron Zhang¹, Kara Wang¹, Jason Chow², Phillip Guan², Alex H. Williams³, David H. Brainard¹, Fangfang Hong¹; ¹Department of Psychology, University of Pennsylvania, ²Reality Lab Research, Meta, ³Center of Neural Science, New York University

Chromatic adaptation to a steady background has a profound effect on color discrimination thresholds. Previous studies that examined this effect typically restricted threshold measurements to cases in which the reference stimulus matched the adapting chromaticity, to a few discrete references, or along a limited set of color-perturbation directions. It remains unknown how chromatic adaptation reshapes full iso-discrimination contours when the reference varies continuously throughout color space. Here, we investigate this question across an isoluminant plane by using adaptive trial placement combined with a semi-parametric Wishart Process Psychophysical Model (WPPM) to characterize thresholds at all references and color-perturbation directions. We studied two adaptation conditions with matched luminance: a neutral gray and a blue daylight (~12,000 K) background. On each trial, participants (N = 2) viewed three matte, blobby objects for 1s, and identified the odd one out based on surface color. The two adaptation conditions were interleaved two runs per session, with condition order randomized within each session. Participants adapted to the background for 120s before the first trial of each run. Each participant completed ~7,500 trials per condition, and we fitted the WPPM separately for each. The results reveal clear changes in the field of iso-discrimination contours across the adapting backgrounds. Under gray adaptation, the smallest contour (highest sensitivity) is centered on the reference whose chromaticity matches the gray background. Under blue adaptation, the smallest contour shifts towards the background, while locations that were most sensitive under gray adaptation have increased thresholds. In both cases, the major axes of the contours tend to point toward the adapting background. These findings indicate that changes in adapting background systematically modulate color discrimination thresholds. This dataset provides a rich benchmark for evaluating mechanistic models of color vision and hypotheses about efficient coding.