Open Access
Add to BibSonomy
Abstract
A large number of studies have shown the effect of melanopsin-dependent retinal ganglion cells on humans performing brightness discrimination tasks. These studies often utilized targets that only differ in their melanopsin activation levels, and not in their luminance or hue, which are both factors that make large contributions to brightness discrimination. The purpose of the present study was to evaluate the relative contribution of melanopsin activation to brightness discrimination when luminance and hue are also varying in addition to melanopsin activation. Using an apparatus consisting of three separate high luminance projectors, we were able to manipulate melanopsin-isolating stimulation, and L-, M-, and S-cone stimulation separately, thus allowing us to vary stimuli in their melanopsin activation, luminance, and hue category independently. We constructed three sets of target stimuli with three different levels of melanopsin activation (100%, 131%, and 167% relative melanopsin excitation) and five levels of luminance. We then had subjects do a two-alternative forced choice task where they compared the previously described target stimuli set to a set of four comparison stimuli that varied in their hue category but had identical luminances. We found that in our stimuli set the overall contribution of melanopsin activity to brightness discrimination was small (an average of 6% increase in likelihood to call a high melanopsin activity stimulus brighter compared to a low melanopsin activity stimulus) when luminance and hue also varied. However, a significant interaction showed that when the comparison was between stimuli differing only in melanopsin stimulation (with luminance and hue unchanged) the contribution of melanopsin to brightness judgments was about 3 times larger (an average of 18% increase in likelihood to call a high melanopsin activity stimulus brighter compared to a low melanopsin activity stimulus). This suggests that although luminance and hue have large effects on brightness discrimination such that the melanopsin contribution can become hard to detect, when there are minimal cone-dependent signals available, melanopsin can make a large contribution to brightness discrimination.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. INTRODUCTION
Color includes three primary perceptual categories: hue, saturation, and brightness. Brightness has a long and rich history in color science, and the contribution of luminance to brightness is of such great magnitude that luminance and brightness are often erroneously referred to as being one and the same outside of specialized texts. More appropriately, many texts will refer to luminance as the physical correlate to perceived brightness, but their actual relationship is more complex; luminance is derived from units that depend on the viewing conditions such as the angle of view as well as a subject’s adaptive state, and brightness in a psychophysical task can be defined in many different ways depending on the experimental task. As such, it is not a simple thing to directly relate brightness and luminance, and the specific measures used for each will vary between experiments. Furthermore, hue (which people more readily associate with color) has a profound effect on brightness, and it has been well documented in the literature [1–4]. However, more recently, another contribution to brightness perception that has been recognized is that of intrinsically photosensitive retinal ganglion cells (ipRGCs), also called melanopsin-dependent retinal ganglion cells (mRGCs). Melanopsin contributions to brightness have been studied in animal models and in human psychophysical studies, where it has been demonstrated that under various circumstances these melanopsin-based signals can contribute to brightness perception [5–10].
It is relatively difficult to stimulate these mRGCs while only changing the melanopsin signal (most light sources will also differentially vary cone and rod signals to these cells), but there is wide interest in investigating the extent, when isolated from rod and cone signals, to which these melanopsin signals contribute to color vision in humans. It has also raised the question of how much one should be concerned that many studies in color vision only focus on cone (and rod) contributions and do not also measure what may be contributed by melanopsin activity. Other studies have been done to try to estimate relative brightness contribution by using magnitude estimations. In one such study, a large contribution to brightness from melanopsin was found using magnitude estimations of sequentially presented equiluminant achromatic stimuli [10]. It is common to use stimuli that only differ in their melanopsin stimulation in order to maximize the ability to detect the contribution of the melanopsin-based signal. In order to try to contribute some novel insight to the issues raised here, we have designed a study using one of the oldest methods of brightness discrimination (the two-alternative forced choice task, 2AFC) to see if we can estimate the relative contribution of melanopsin stimulation when presented alongside variations in luminance and hue. We made use of an apparatus that allows independent control of melanopsin signals as well as L, M, and S cones. This approach presents a number of unique issues when studying melanopsin signals, but it allows us to study their contribution to brightness discrimination in ways that would not otherwise be possible. By using this type of arrangement, we hope to provide a broader view of melanopsin signal contributions to stimuli that also vary in cone contributions. These types of stimuli are similar to those that make up the majority of stimuli in the real world as well as many other color-vision experiments.
We designed our experiment to make brightness comparisons between stimuli that have three different levels of melanopsin stimulation, in addition to five levels of luminance, and four separate hue categories. These decisions were made after the collection of pilot data on a much larger set of melanopsin signal levels, luminances, and hues, in order to provide a practical and sufficiently broad array of stimuli. We hypothesized that under conditions similar to those seen in previous experiments (where only the melanopsin signal differs) we would see a large significant effect of melanopsin signals on subjects’ brightness judgments. However, when the stimuli being compared also differed in hue or luminance, we proposed two possible outcomes: one would be that the contribution would be equivalent across all levels (no significant interaction between the melanopsin condition and either hue or luminance) or, conversely, that the contribution to comparisons where luminance or hue varied would make a smaller contribution (a significant interaction such that the difference between melanopsin conditions is less when other factors also vary). These were not preregistered predictions, but were based on our pilot data results.
Fig. 1. Spectra for the four primaries of the three projectors. Blue and red (top row) are both controlled by a single projector with a mid-wavelength filter. Yellow and green (bottom row) are controlled by individual projectors with filters blocking higher and lower wavelengths. These spectra measurements are representative of actual presentation conditions (the other projectors are still turned on and projecting “black” [0, 0, 0] during all measurements).
Fig. 2. Actual image of apparatus used in this experiment, with the right side uncovered. On the left side one can see the chinrest, pinhole cutout, and the white diffuser plate that subjects would view the stimuli on. The right side shows the three projectors and their respective filters, which would not be visible during an experiment, as a black cloth would be covering the entire side.
2. METHODS
A. Apparatus
We used three NEC 5500-lumen widescreen advanced professional installation projectors. The arrangement of the apparatus was identical to that used in Yang (2018) when measuring the effect of ipRGC stimulation on time perception [11]. Each of the projectors has a filter applied such that four primaries (red, blue, yellow, and green) are created that can be varied independently (Fig. 1). The three filtered light sources are then projected onto a white diffuser plate (Fig. 2). In order to align the projectors, an identical circular central stimulus was projected by all three projectors, and then they were manually adjusted until visible edge inhomogeneities of the circle stimuli were eliminated. Due to the nature of the apparatus, visible inhomogeneities were created on the screen, especially near the outer edges. We used a Topcon SR-5000 2D spectroradiometer to measure a central area of the screen that would be used for the experiment (20° visual angle in total). This area appeared homogeneous, and we confirmed the light in the 20° experimental area did not possess inhomogeneities greater than 1% in luminance or Commission Internationale de l’Éclairage (CIE) color space $xyz$ coordinates when a full-field white stimulus was displayed. We then measured the combined output of the three projectors with a Minolta CS-1000 spectroradiometer. In front of the diffuser, a chinrest was utilized with a pinhole cutout placed in front of it. The pinhole cutout was set to 2 mm, and a black square with a circular cutout with a corrective lens attached to the center was placed after it. The total visible area for the experiment was approximately 20° of visual angle.
Table 1. Values of Stimuli Used in Experiment
Table 2. Variation of Medium Orange Stimuli over Time
B. Stimulus Set
The precise measurements of the stimuli used in this experiment can be seen in Table 1. The values seen in Table 1 are an average of values used during experimental sessions (24 in total). Due to the nature of the apparatus, values would vary over time. To compensate for this, measurements were taken before each session, and adjustments made if values deviated more than 1% from the reported values in luminance, CIE, $x$, or $y$ values. Although variation of stimulus parameters between sessions could be substantial, variation within a session was relatively small. Table 2 shows the variation of the CIE $x$ and $y$ values, along with luminance of the medium orange stimulus from Table 1 during a mock experimental session conducted over a 60-min period. For stimuli that varied in luminance, it was set to scale in increments of ${50}\;{{\rm cd/m}^2}$, again accepting 1% deviations in either direction. Greater than 1% variation due to projector instability was unlikely to be seen over the course of a session, as seen in Table 2, which was why we chose 1% as an acceptable level of deviation. All experimental conditions change the relevant stimulus parameters by at least 8%. Reported CIE $xy$ values are based on the 1964 10° standard observer (1964 values were used for ease of use with our spectrometer). The melanopsin signal values reported here represent relative contrasts scaled to “low orange” being 1, as it has the least melanopsin stimulation of any stimulus used in the experiment. The calculation of melanopsin stimulation was done in accordance with that reported in Tsujimura (2011) [12]. Accordingly. we used a pigment template nomogram with a peak wavelength of 480 nm and a peak axial optical density of .4 to calculate a peak spectral sensitivity function for melanopsin stimulation of 489 nm.
We used a total of four hue categories in this experiment: orange, magenta, yellow, and green. These were selected for their diversity and practicality for use with the apparatus (for example, it would not be possible to make an equiluminant blue on this apparatus). The orange stimulus could create the widest variety of melanopsin-varying metamers, and as such was chosen as the basis for the comparison stimuli; every comparison would feature one of the three orange melanopsin activity-varying stimuli compared to either another orange stimulus or a stimulus of another color. Additionally, these three melanopsin activity-varying orange stimuli could vary among five luminance levels (${\sim}{500}$, 550, 600, 650, and ${700}\;{{\rm cd/m}^2}$) that varied within a range that would avoid ceiling or floor effects (if luminance disparity is too great, subjects will always pick the higher luminance target) for a 2AFC. These light levels were also selected for their ease of use with the apparatus (the need for characterization before each session and the relatively large step sizes of some primaries necessitates having a range below that of the maximum output of the apparatus). Other colored stimuli were made equiluminant with the orange median value (${600}\;{{\rm cd/m}^2}$) of luminance to the best of our ability with the apparatus at hand. The reasoning behind selecting the three additional colors in this experiment is as follows: the magenta stimulus used in this experiment represents a stimulus with a relatively low amount of melanopsin signal but a great amount of S-cone activity. Yellow represents both a very low amount of S-cone activity and low melanopsin signal. Finally, green represents a hue with a large amount of S-cone and melanopsin stimulation. It was not possible to make an equiluminant hue with low S-cone activity, but high melanopsin signal, for the present experiment.
The total set of comparisons would be the three melanopsin-varying orange stimuli at five luminance levels [as seen by the left sides of the stimuli in Figs. 3(A) and 3(B)] compared to either a medium melanopsin orange stimulus at the median luminance level (${\sim}{600}\;{{\rm cd/m}^2}$) or one of the three equiluminant (${\sim}{600}\;{{\rm cd/m}^2}$) colored stimuli [four possible comparison stimuli, the right sides of the stimuli in Fig. 3(B)]. This meant the total number of pairs of comparisons was 60 [three melanopsin activity-varying stimuli at five luminance levels each (15 combinations) compared to four hues], with each comparison pair appearing 10 times per session across two sessions, for a grand total of 20 trials per comparison. Pilot data originally utilized every possible comparison, but the length of time required led us to compromise and only focus on a subset of possible comparisons.
Fig. 3. Representation of the task as it would appear to subjects looking through the apparatus. (A) shows a condition where only melanopsin signal varies between the two sides (not actually visible in this paper); (B) shows conditions where the luminance varies (two brighter and two darker than that seen in (A), as well as the four possible comparison hues of orange, magenta, green, and yellow. The melanopsin activity-varying stimulus (low, medium, and high) could be any of the luminance-varying stimuli seen on the left sides of the black line in (A) and (B), whereas the comparison stimuli were any of the four equiluminant stimuli seen on the right side of the black line in (B) [the top left stimulus’s right side is identical to the right side of (A)]. During the actual experiment, both sets of stimuli could appear on either side.
The fixation point was chosen to be a 2.5° central circle. This was chosen for its ability to cover the fovea and its minimal contribution to border induction effects. Similarly, a 2° black vertical line separated the left and right stimuli during a comparison. This was also done to avoid contiguous border induction effects, as previous research has shown 2° and smaller gaps between stimuli can have a powerful effect on perceived brightness and hue [13,14].
All stimuli were generated in MATLAB with Psychtoolbox-3 [15] on a computer running Windows 10. In order to reduce inhomogeneities present at the edges, the stimuli were drawn to continue past the visible borders of the experiment (beyond the 20° circle visible through the optics of the apparatus).
C. Procedure
A total of 12 subjects, between the ages of 19 and 23 (${\rm M}={20.5}$, ${\rm SD}={1.26}$) participated in the experiment. All subjects were monetarily compensated for their participation and were recruited voluntarily from undergraduate and graduate students at the Kochi University of Technology. Eight of the subjects were native Japanese speakers and four were native Chinese speakers. All instructions were given in written Japanese with verbal instructions in either English or Japanese, depending on subject preference. All subjects were confirmed to be color-normal observers using the Ishihara Pseudo-Isochromatic Plate Test. No subjects were informed of the total combinations of the stimulus presentation or the ways they varied before the experiment; they were simply told they would be judging the brightness of two stimuli.
Subjects were presented with two stimuli separated from each other by a black 2° vertical line with a central fixation point consisting of a black circle 2.5° in diameter (see Fig. 3). Stimuli were viewed through a pinhole cutout set to 2 mm, behind which nonfiltering corrective lenses could be placed if a subject required them. The pinhole cutout was moved to be directly in front of the subject’s pupil during the experiment. They were told to respond to the stimuli by pressing 1 if the left side was perceived as brighter or 2 if the right side was perceived as brighter. Stimuli randomly alternated between the two sides during the experiment. Subjects were informed that even if the two sides appeared identical, they must still make a decision. Subjects completed a total of 603 measurements per session spread across two days for a grand total of 1206 judgments, but the first three judgments from each day were excluded, leaving 1200 data points per subject. All judgments were made with the right eye viewing the stimuli while the left eye was covered by an eyepatch. This was done because the placement of stimuli was based on viewing through the right eye.
During stimulus presentation, a comparison set flashed on for 1 s followed by 3 s of blackness. This was done to reduce the probability of chromatic adaptation occurring [16–18]. Although most subjects could make a judgment based on one presentation, if they waited, the same stimuli would flash on again for 1 s. The average amount of time required per session for each subject was around 55 min.
3. RESULTS
Subject data were transformed into a percentage of times a given luminance-varying and melanopsin signal-varying orange stimulus (low, medium, and high at each of their five luminance levels) was judged as being brighter than either the equiluminant orange, magenta, yellow, or green comparison stimulus. The values were recorded as a decimal ranging from 0 to 1. These data were then analyzed using a three-factor repeated measures ANOVA. The three factors were melanopsin condition, with three levels, hue with four levels, and luminance, with five levels. For the output from the repeated measures ANOVA analysis and post hoc pairwise comparisons, please see Appendix A.
The main effect of the melanopsin condition, displayed in Fig. 4, showed significant increases in brightness with increases in melanopsin stimulation at ${\rm p}=.{034}$ (${\rm F}\;({2},\;{22})\; = \;{3.97}$) with a small effect size (${\eta ^2}=.{006}$). However, the average percent brighter response for the low melanopsin signal (42%) was not significantly lower than the medium melanopsin signal (44%), which was not significantly lower than the high melanopsin signal (48%), according to post hoc comparisons.
Fig. 4. Main effect of the melanopsin condition for the three levels of low, medium, and high. The $y$ axis shows the percentage of time an observer said the given melanopsin signal-varying orange stimulus was brighter than an equiluminant (${\sim}{600}\;{{\rm cd/m}^2}$) orange, magenta, yellow, or green stimulus regardless of the luminance of the melanopsin-varying stimulus. Error bars represent the standard error of the mean (SEM) adjusted for within-subjects data [19].
The main effect of luminance, displayed in Fig. 5, also showed significant increases in brightness ratings, with increases in luminance at ${\rm p}\; \lt \;.{001}$ (${\rm F}\;({4},\;{44})\; = \;{26.6}$) with a moderate effect size (${\eta ^{2\:}} = \;.{097}$). Lower luminances were significantly less likely to be rated brighter than higher luminances for all post hoc comparisons.
Fig. 5. Main effect of the luminance condition for the five levels of 500, 550, 600, 650, and ${700}\;{{\rm cd/m}^2}$. The $y$ axis shows the percentage of time an observer said any of the melanopsin signal-varying orange stimuli at the given luminance level were brighter than an equiluminant (${\sim}{600}\;{{\rm cd/m}^2}$) orange, magenta, yellow, or green stimulus. Error bars represent SEM, adjusted for within-subjects data [19].
Finally, the main effect of hue on brightness judgments, displayed in Fig. 6, was also significant at ${\rm p}\; \lt \;.{001}$ (${\rm F}\;({3},\;{33})\; = \;{11.5})$, with a large effect size (${\eta ^{2\:}} = \;.{29}$). Orange melanopsin signal-varying and luminance-varying stimuli were significantly less likely to be rated brighter when compared to yellow (28%), but significantly more likely to be rated brighter than magenta (65%). Although they were also rated brighter than green (39%) comparison stimuli on average, the post hoc comparison was not significant. When orange melanopsin signal-varying stimuli were compared to other orange melanopsin signal-varying stimuli, they were rated brighter approximately half the time (51%), as would be expected since they were equally likely to be brighter or darker and have higher or lower melanopsin signal levels.
Fig. 6. Main effect for hue category for each of the four comparison hues used in this experiment. The $y$ axis shows the percentage of time an observer said any of the melanopsin signal-varying and luminance-varying orange stimuli were brighter than the given equiluminant (${\sim}{600}\;{{\rm cd/m}^2}$) orange, magenta, yellow, or green stimulus. Error bars represent SEM, adjusted for within-subjects data [19].
There were significant interactions for the melanopsin condition and luminance at ${\rm p}=.{015}$ (${\rm F}\;({8},\;{88})\; = \;{2.54}$), as well as luminance and hue at ${\rm p}\; \lt \;.{001}$ (${\rm F}\;({12},\;{132})\; = \;{3.76}$). However, the interaction between melanopsin condition and hue was not significant at ${\rm p}={0.862}$ (${\rm F}\;({6},\;{66})\; = \;{0.42}$). The interaction between the melanopsin condition and luminance is more readily understood by viewing Fig. 7. It shows that when the luminances between the comparison stimuli are identical, the effect of melanopsin stimulation on brightness judgments is significantly greater (the average difference between high and low is .076) than that seen when the luminance levels are different (average difference of .028).
Fig. 7. Interaction between melanopsin stimulation and luminance. The blue line is a low melanopsin signal, the red line is a medium melanopsin signal, and the green line is a high melanopsin signal. Luminance of the melanopsin signal-varying orange stimulus at the time of comparison is seen on the $x$ axis. The $y$ axis shows the percentage of time an observer said the melanopsin signal-varying and luminance-varying orange stimuli were brighter than an equiluminant (${\sim}{600}\;{{\rm cd/m}^2}$) orange, magenta, yellow, or green stimulus. Error bars represent SEM, adjusted for within-subjects data [19].
Fig. 8. Axes and labels are identical to Fig. 7. Here the interaction among all three factors of melanopsin, luminance, and hue is made apparent by eliminating the non-orange comparison hues from the data displayed in Fig. 7. The magnitude of the melanopsin activity effect on brightness judgments between low and high conditions for equiluminant orange (${600}\;{{\rm cd/m}^2}$) is larger than that of the luminance change to ${550}\;{{\rm cd/m}^2}$ or ${650}\;{{\rm cd/m}^2}$. Error bars represent SEM, adjusted for within-subjects data [19].
Fig. 9. Axes and labels are identical to Fig. 8. This is the equivalent graph for magenta comparisons only. Error bars represent SEM, adjusted for within-subjects data [19].
Fig. 10. Axes and labels are identical to Fig. 8. This is the equivalent graph for yellow comparisons only. Error bars represent SEM, adjusted for within-subjects data [19].
Fig. 11. Axes and labels are identical to Fig. 8. This is the equivalent graph for green comparisons only. Error bars represent SEM, adjusted for within-subjects data [19].
Additionally, there was a significant three-way interaction among the melanopsin condition, luminance, and hue at ${\rm p}={0.026}$ (${\rm F}\;({24},\;{264})\; = \;{1.69}$). This is shown in Figs. 8, 9, 10, and 11, where it is clear that the effect of the melanopsin signal is greatest at enhancing perceived brightness when the hue and luminance are identical (when it is compared to another orange stimulus that is the same luminance) as seen in Fig. 8 at the 600 luminance (equiluminant) point. At this point, the average difference between the high and low conditions is .181, whereas magenta is .063 (Fig. 9), yellow is .071 (Fig. 10), and green is .075 (Fig. 11). It is worth noting that for the orange hue seen in Fig. 8, the average effect of the ${50}\;{{\rm cd/m}^2}$ increase (.093) and decrease (.098) have less of an influence on brightness judgments than the change from low to high melanopsin activity (.181).
4. DISCUSSION
This study sought to evaluate the contribution of melanopsin stimulation to brightness judgments under a classical 2AFC task in order to provide insight into their role in a more naturalistic brightness judgment scenario. The results here indicate that melanopsin signals can play a large role in brightness judgment when other aspects of a stimulus (such as hue or luminance) are minimal. When stimuli vary in their other parameters (as is often the case in the real world), the degree to which melanopsin signals can contribute to brightness appears to be greatly diminished. The significant main effects, and interaction of luminance and hue on brightness are well established and need no further explanation here [1–4]. It is worth noting, though, that the brightness induction of yellow and green were both very great despite being the opposite in terms of S-cone and melanopsin stimulation. That yellow should be perceived as the overall brightest hue at equiluminant levels is in accordance with the previously mentioned literature.
One may make a good argument that the way we have created our stimuli here and presented them is not ideal for maximizing melanopsin stimulation. The stimulus set we used does not feature as large a variation in melanopsin stimulation as seen in other experiments where levels up to 400% contrast have been used [20]; instead, at most we have 167% contrast. Our design also did not compensate for individual differences that would require stimuli to be adjusted to maximize stimulation for a given observer’s melanopsin spectral sensitivity curve. A given observer’s maximal stimulation for melanopsin signals as well as their peak wavelength sensitivities for cones in silence substitution vary, such that the assumptions that these stimuli were true silence substitutions for all observers cannot be assumed [21]. There are also concerns of both biological variability and device instability contributing to a level of cone excitation variability in the nominally melanopsin activity-varying conditions [20]. With the design we used here, it would not be practical to make stimuli specific to each observer, and instead, we have utilized a relatively large sample size of 12 individuals, but we were not able to collect equivalent data to those used in Spitchan et al. (2017) for a splatter analysis. We cannot assume that this data set of 12 individuals would balance out to a true population mean of spectral sensitivities, nor can we assume our apparatus did not present differential activation in cone excitations between melanopsin conditions. Despite these issues, the magnitude of the influence of melanopsin activity on brightness shown by the difference in response between the low and high melanopsin conditions for equiluminant orange comparisons was greater than that of an increase or decrease in luminance of 50 cd/m (${\sim}{8.3}\% $) for either, as seen in Fig. 8. Additionally, the overall consistency of the effects shown in Table 3, which has individual subjects’ difference scores for the melanopsin conditions when nothing else varied (Fig. 8, 600 luminance condition), suggests most subjects were able to use the melanopsin signal alone to make brightness judgments. However, the great variability in effect size between subjects suggests that subject characteristics are a factor that should be accounted for in future experiments.
This study also used hues that are known to have large effects on brightness perception (as demonstrated in our results), and varied luminance to a degree which, while remaining difficult, subjects could readily use the provided luminance differences to make brightness judgments as shown by the moderate effect size for luminance. It could be argued we are not giving enough strength to melanopsin-signaling stimuli in this experiment to properly evaluate their potential role in brightness judgments. However, we think we have constructed a rather realistic representation of the types of brightness judgments people would make in real life (or at least in a typical 2AFC experiment) and shown that there are scenarios, where despite using relatively weakly differing melanopsin stimulation stimuli, one can still significantly impact brightness judgments. Additionally, the analysis of natural images used in Allen et al. (2019) showed melanopsin contrast separate from luminance contrast rarely exceeds 20% and found significant effects for grating discrimination at 20% melanopsin signal contrast [22]; this suggests our contrasts of 130% (medium) and 167% (high) may actually be on the high end for a natural scene. This is important to note, as one purpose of this experiment was evaluating the impact melanopsin stimulation can have on brightness judgments in experiments or real-world tasks that are not intended to use melanopsin-isolating stimulation as a variable.
Table 3. Individual Subject Difference Scores for Melanopsin Conditions for Equiluminant Orange Comparisons
Melanopsin activity can also influence hue perception [8] and research by Zele et al. (2018) has proposed a model of brightness estimation that takes into account melanopsin and cone contributions [9]. In this experiment, the hue condition also changes the amount of ${\rm L}/({\rm L}+{\rm M})$ and S-cone activity in a target. Perceived hue category and ${\rm L}/({\rm L}+{\rm M})$ and S-cone activity do not necessarily have the same effect on a target’s brightness. As we did not systematically manipulate the amount of cone activity, and instead picked the hue categories for their melanopsin activity properties, it is hard to isolate these effects from this data set; what we call hue category is therefore confounded with cone excitation. We hope to use this apparatus in a follow-up experiment to look at melanopsin signal effects on hue perception, taking into account the model proposed by Zele et al. and separating out cone excitation effects. We plan to evaluate the potential influence of melanopsin activity and luminance in an inducting surround on the red/green balance of foveal chromatic stimuli, as it has been shown that surround S-cone activity greatly influences red/green balance of foveal stimuli, which could overlap with changes in melanopsin stimulation [23,24]. This is part of a larger body of research looking at the effects of luminance as well as ${\rm L}/({\rm L}+{\rm M})$ and S-cone activity in a surround as an influence on perceived R/G balance of central targets [25].
This experiment is an example of how different tasks and different measures of brightness can influence the relative contribution of melanopsin activation. A given experiment may find large or small effect sizes for a given brightness judgment in that experiment’s confines, but will not necessarily translate to a size of a similar magnitude when the task and/or brightness measurement changes. We have quantified our brightness measure here as the percentage an observer picks one stimulus as being brighter than another in a 2AFC. Were we to use a different measure of brightness on the same apparatus with a similar stimulus set, we would find a different effect size to correspond with that measure of brightness. As such, it is our hope that this experiment can serve as a starting point for further studies looking into the relationship of melanopsin stimulation on brightness judgments under a variety of brightness tasks and where other brightness influencing variables are also present, in addition to serving as a reference for experiments where melanopsin stimulation is likely to impact results.
APPENDIX A: THE OUTPUT FROM THE REPEATED MEASURES ANOVA
Within-Subjects Effects
Post Hoc Comparisons: Mela
Post Hoc Comparisons: Luma
Post Hoc Comparisons: Huea
Funding
Japan Society for the Promotion of Science (KAKENHI (B) 18H03323 to K. S.); Grant-in-Aid for Specific Research to Vision and Affective Science Integrated Research Laboratory, Research Institute from Kochi University of Technology to K. S.
Disclosures
The authors declare no conflicts of interest.
REFERENCES
1. A. Chapanis and R. M. Halsey, “Luminance of equally bright colors,” J. Opt. Soc. Am. 45, 1–6 (1955). [CrossRef]
2. R. J. Ball and S. Howard Bartley, “Changes in brightness index, saturation, and hue produced by luminance–wavelength–temporal interactions,” J. Opt. Soc. Am. 56, 695–699 (1966). [CrossRef]
3. S. A. Burns, V. C. Smith, J. Pokorny, and A. E. Elsner, “Brightness of equal-luminance lights,” J. Opt. Soc. Am. 72, 1225–1231 (1982). [CrossRef]
4. K. Uchikawa, K. Koida, T. Meguro, Y. Yamauchi, and I. Kuriki, “Brightness, not luminance, determines transition from the surface-color to the aperture-color mode for colored lights,” J. Opt. Soc. Am. A 18, 737–746 (2001). [CrossRef]
5. T. M. Schmidt, M. T. Do, D. Dacey, R. Lucas, S. Hattar, and A. Matynia, “Melanopsin-positive intrinsically photosensitive retinal ganglion cells: from form to function,” J. Neurosci. 31, 16094–16101 (2011). [CrossRef]
6. T. M. Brown, C. Gias, M. Hatori, S. R. Keding, M. Semo, P. J. Coffey, J. Gigg, H. D. Piggins, S. Panda, and R. J. Lucas, “Melanopsin contributions to irradiance coding in the thalamo-cortical visual system,” PLoS Biol. 8, e1000558 (2010). [CrossRef]
7. T. M. Brown, S.-I. Tsujimura, A. E. Allen, J. Wynne, R. Bedford, G. Vickery, A. Vugler, and R. J. Lucas, “Melanopsin-based brightness discrimination in mice and humans,” Curr. Biol. 22, 1134–1141 (2012). [CrossRef]
8. D. Cao, A. Chang, and S. Gai, “Evidence for an impact of melanopsin activation on unique white perception,” J. Opt. Soc. Am. A 35, B287–B291 (2018). [CrossRef]
9. A. J. Zele, P. Adhikari, B. Feigl, and D. Cao, “Cone and melanopsin contributions to human brightness estimation,” J. Opt. Soc. Am. A 35, B19–B25 (2018). [CrossRef]
10. M. Yamakawa, S.-I. Tsujimura, and K. Okajima, “A quantitative analysis of the contribution of melanopsin to brightness perception,” Sci. Rep. 9, 7568 (2019). [CrossRef]
11. P. L. Yang, S. I. Tsujimura, A. Matsumoto, W. Yamashita, and S. L. Yeh, “Subjective time expansion with increased stimulation of intrinsically photosensitive retinal ganglion cells,” Sci. Rep. 8, 11693 (2018). [CrossRef]
12. S.-I. Tsujimura and Y. Tokuda, “Delayed response of human melanopsin retinal ganglion cells on the pupillary light reflex,” Ophthalmic Physiolog. Opt. 31, 469–479 (2011). [CrossRef]
13. S. L. Buck, A. Shelton, B. Stoehr, V. Hadyanto, M. Tang, T. Morimoto, and T. DeLawyer, “Influence of surround proximity on induction of brown and darkness,” J. Opt. Soc. Am. A 33, A12–A21 (2016). [CrossRef]
14. D. L. Bimler, G. V. Paramei, and C. A. Izmailov, “Hue and saturation shifts from spatially induced blackness,” J. Opt. Soc. Am. A 26, 163–172 (2009). [CrossRef]
15. M. Kleiner, D. Brainard, and D. Pelli, “What’s new in Psychtoolbox-3?” Perception 36, 1–16 (2007). [CrossRef]
16. J. M. Loomis and T. Berger, “Effects of chromatic adaptation on color discrimination and color appearance,” Vision Res. 19, 891–901 (1979). [CrossRef]
17. M. D. Fairchild and L. Reniff, “Time course of chromatic adaptation for color-appearance judgments,” J. Opt. Soc. Am. A 12, 824–833 (1995). [CrossRef]
18. O. Rinner and K. R. Gegenfurtner, “Time course of chromatic adaptation for color appearance and discrimination,” Vision Res. 40, 1813–1826 (2000). [CrossRef]
19. G. R. Loftus and E. J. M. Michael, “Using confidence intervals in within-subject designs,” Psychonomic Bull. Rev. 1, 476–490 (1994). [CrossRef]
20. M. Spitschan, A. S. Bock, J. Ryan, G. Frazzetta, D. H. Brainard, and G. K. Aguirre, “The human visual cortex response to melanopsin-directed stimulation is accompanied by a distinct perceptual experience,” Proc. Natl. Acad. Sci. USA 114, 12291–12296 (2017). [CrossRef]
21. M. Spitschan and T. Woelders, “The method of silent substitution for examining melanopsin contributions to pupil control,” Front. Neurol. 9, 941 (2018). [CrossRef]
22. A. E. Allen, F. P. Martial, and R. J. Lucas, “Form vision from melanopsin in humans,” Nat. Commun. 10, 2274 (2019). [CrossRef]
23. T. DeLawyer, M. Tayon, C.-L. Yu, and S. L. Buck, “Contrast-dependent red-green balance shifts depend on S-cone activity,” J. Opt. Soc. Am. A 35, B114–B121 (2018). [CrossRef]
24. S. L. Buck, F. Rieke, and T. DeLawyer, “Contrast-dependent red-green hue shift,” J. Opt. Soc. Am. A 35, B136–B143 (2018). [CrossRef]
25. T. De Lawyer, “Brown induction and red-green hue shifts,” Ph.D. dissertation (University of Washington, 2017).
