Fig 1.
Methods and procedure of the three experiments.
A: backgrounds of the stimuli. B: test chromaticities of the targets in the stimuli. C: targets of the stimuli for Experiments 1, 2, and 3 (highlighted in blue against the background images). D: illuminant conditions of the experiments. E: procedure of the experiments with an example of two trials. F: normalized physical gamuts of illuminants 6500 Kelvin (K) and 3000 K. Background images were adapted from a publicly available dataset published by Foster et al. [21] under a Creative Commons Attribution 4.0 International (CC BY 4.0) license.
Fig 2.
Results from Experiments 1, 2, and 3.
A: Experiment 1 results: mean observers’ settings (N = 8), averaged across all backgrounds, for the 6500 K condition (left) and the 3000 K condition (right). Optimal colors are shown for comparison, and the correlation between the two loci is reported. The filled gray areas represent the 95% confidence intervals of the observers’ settings. B: Experiment 2 results: same format as panel A. The left part shows the averaged results for the 6500 K condition; in addition, the right part shows individual observer results, clustered by judgment strategy. C: Experiment 3 results: same format as panel A. The averaged results are presented for the 6500 K condition. D: results of the three experiments, averaged across observers and backgrounds, plotted together. For Experiment 1, only the 6500 K condition is shown. Asterisks indicate significant differences between the loci: * p and ** p
. The notation n.s. denotes no statistically significant difference. See Analysis section for statistical details.
Fig 3.
Additional analysis of Experiments 2 and 3.
A: additional questionnaire conducted after the main analysis. The figure presents the two questionnaire items together with the individual observers’ responses for Experiment 2 (left) and Experiment 3 (right). Each plot also displays the correlation between the two questions, along with the linear regression line fitted to the individual observers’ responses. B: comparison and correlation between the mean settings (i.e., mean luminosity thresholds) of observers who were able to directly judge the luminosity thresholds through the impression of self-luminosity and those who relied on the notion of naturalness to make their judgments (the clustering is based on the responses to the questions presented in panel A of this figure) in Experiment 2 (on the left) and Experiment 3 (on the right).
Table 1.
Results of first questionnaire for Experiment 2.
Table 2.
Results of first questionnaire for Experiment 3.