(A) We presented uniform surfaces of 6 deg at maximum possible luminance levels, sampling from the available spectrum of wavelengths (Dataset 4, see Materials and methods; n = 5 sessions, 15.240.1 trials per condition in each session ). In addition, we presented black and white surfaces. Shown is the change in LFP power relative to the baseline (gray screen), expressed as a fold-change. (B) Three hues (red, green and blue) were presented at three different luminance levels (approximately 2.5, 5 and 10 cd/m, Dataset 1, see Materials and methods; n = 5 sessions, 19.80.45 trials per condition in each session). Shown are LFP power spectra. LFP power spectrum estimated and normalized as in Figure 1C. (C) Average change in LFP power, expressed as a fold-change based on the fitting procedure, relative to the baseline (gray screen). The dependence of gamma amplitude on stimulus luminance was greater for G than for R or B (difference between high versus low: p0.05, bootstrap test). Gamma oscillations amplitude RB or G across all three luminance conditions, and BG surface stimuli for low and intermediate luminance conditions (P0.05, bootstrap test). (D) Modulation of firing rate relative to baseline, expressed as (stim/base). Horizontal bars at bottom of panel represent significant differences between stimuli at p0.05 (permutation test, multiple comparison corrected for time bins). (B–D) Color hues were adjusted for better discriminability, panel A of Figure 4—figure supplement 2 shows actual hues.