## Revision to Preregistration Document 2 ## ---------- 12/9/2015 ---------- Revisions made to the Preregisration Document: We studied two subjects and discovered that the S-cone screening test was too difficult. We therefore adopted S-cone contrast steps of 4% and 2%, and started data collection again. We also modified the subject instructions to include a description of the overall experiment and to explain the Computerized voice. Instead of "Null the color red - green" and "Null the color blue - yellow" the computer now says: "Please match the color to the background" and "Please match the flicker to the background". ---------- 12/8/2015 ---------- We have not begun subject recruitment at the time of this revision. We have a few parameters that we failed to note in the original preregistration document. We are adding those items here. List of changes to the preregistration document (bold items are new/changed): 1. **The silent substitution modulations are obtained using numerical optimization methods described in Spitschan et al. (2014, 2015a, b). In order to maximize contrast for our targeted modulations, we first search over the device settings yielding a background which maximizes simultaneously the contrast on the cone-silent melanopsin modulation and the yoked L+M+S contrast in the L+M+S modulation. The background that satisfies these constraints may vary slightly in chromaticity between repeat calibrations but we expect it to have an approximate chromaticity of x = ~0.5, y = ~0.4.** 2. **Both the test modulations and the nulling modulations are tailored to account for the individual’s observer age, field size and light-level-dependent bleaching as described extensively in Spitschan et al. (2014, 2015a, b).** 3. All stimuli are generated with a digital light synthesis engine (OneLight Spectra) under computer control, and presented through a custom-made eyepiece with a ​**6 mm artificial pupil**​, as described previously (Spitschan et al., 2014; Spitschan et al., 2015a, b). 4. To detect drift, we will take validation measurements such as those described above after every 5 subjects. We will re-calibrate the OneLight if the luminance of the background has dimmed by more than 5% from the initial calibration measurement or if the contrast in the validation measurements of a +42% cone-silent melanopsin ​**modulation exceeds 2.5% on any cone class, or if the contrast of the targeted deviates by more than 2.5% contrast from the target contrast**​.