Figure 4

Stimulus imprecision analysis for Experiment 1. (a) Left panel. LMS content of the low-melanopic and high-melanopic backgrounds in Experiment 1, in cone coordinates. Stimulus spectra were multiplied by cone fundamentals (normalized to peak at 1) to obtain coordinates for each cone class; these were then averaged to get a combined LMS cone coordinate. Right panel. Predicted JNDs if background LMS content were driving LMS detection threshold, under a Weber’s Law assumption for detection (JND proportional to background LMS content). Same format as Fig. 2b, but with median (across sessions) JNDs and ± standard error of the median for each of three participants shown in a single plot. (b) Measured LMS content of the threshold flicker stimuli in Experiment 1. Left panel. Measured LMS content of the flicker stimuli corresponding to detection threshold in the low-melanopic and high-melanopic conditions, expressed as proportion of the nominal LMS contrast of those flicker spectra (i.e., value of 1 indicates LMS contrast was as nominal). Each line plots median ± 1 standard error of the median for one participant, with the median taken across sessions. Right panel. JNDs for Experiment 1, based on measured rather than nominal flicker spectra. (c) L–M chromatic contrast of the LMS flicker stimulus spectra in Experiment 1. Left panel. Measured (L–M)/(LMS) contrast of the flicker stimuli in the low-melanopic and high-melanopic conditions. Each line plots median ± 1 standard error of the median for one participant, with the median taken across sessions. Right panel. Predicted JNDs, assuming that detection of the flicker stimulus was mediated not by LMS contrast, but rather by the absolute magnitude of the L–M chromatic contrast. Under this assumption, greater L–M contrast in the flicker stimulus predicts lower JNDs, under the null-hypothesis that cone-mediated flicker detection does not change with background melanopsin stimulation.