Yangyi Shi¹ (), Rhea T. Eskew, Jr.¹; ¹Northeastern University

The short-wavelength (S) cone photoreceptors exhibit unique characteristics compared to the long-wavelength (L) and middle-wavelength (M) cones. To investigate the perceptual representation of S-cone contrasts at suprathreshold levels, we conducted two psychophysical experiments, building on prior studies with achromatic stimuli (Shi & Eskew, 2024; Shooner & Mullen, 2022). In the first experiment, we used maximum likelihood difference scaling (MLDS) (Knoblauch & Maloney, 2008; Maloney & Yang, 2003), with stimuli ranging from twice the detection threshold to the maximum monitor contrast. Observers compared two pairs of contrasts in each trial and judged which pair appeared more similar. In the second experiment, the same observers performed a two alternative forced choice pedestal discrimination task with the same stimuli used in perceptual scaling. Both experiments separately examined S-cone increment (S+) and decrement (S−) stimuli. Our results reveal that, unlike our previous achromatic findings (Shi & Eskew, 2024), the estimated perceptual scales for both S+ and S- are approximately linear and approximately symmetric. However, consistent with Gabree et al. (2018), pedestal discrimination thresholds rise more rapidly for S+ pedestals than for S- pedestals. At higher pedestal contrasts, S+ pedestals produce stronger masking effects than S− pedestals. The original MLDS model, which assumes a constant magnitude of perceptual noise, cannot account for the increasing discrimination thresholds when the scales are linear, as in both S+ and S− conditions. We propose and fit a variable noise MLDS model, where noise grows with the square root of contrast. As Kingdom (2016) showed, whether noise is constant with contrast or not has little effect on perceptual scales generated by a difference-scaling method, but contrast-dependent noise strongly affects discrimination threshold functions. Our results show that with S cone stimuli, the increasing noise allows us to predict pedestal discrimination thresholds from MLDS results.

Acknowledgements: NSF BCS-2239356