Time-of-day Dependent Light Effects on Subjective Affect and Comfort

Light, one of the key environmental components for both life and work, played signicant role in subjective feelings (e.g. affect and comfort), but the exact effects and mechanisms were still to be determined. The present study screened thirty healthy adults (13 females, 22.45 ± 3.26 years) and examined subjective affect and comfort under short-term white lights with different combination of correlated color temperature (CCT) and illuminance at different times of day (e.g. morning, afternoon, and evening). Our results showed a signicant interaction between illuminance level and time-of-day on subjective comfort. Participants felt more comfortable under 50 lx and 100 lx instead of 500 lx in the evening, and more comfortable under 500 lx in the morning and afternoon. In addition, a positive correlation between illuminance and comfort in the morning and a negative correlation between them in the evening were found. No signicant effect of CCT on any subjective feeling was revealed. Our results necessitate the consideration of time-of-day in understanding lighting effects and application of healthy lighting in daily life.


Introduction
Apart from signi cant role in inducing visual effects, light also plays a critical role in non-visual functions, e.g. subjective affect and comfort 1 . Since Kripke et al. (1983) found that exposure with high-intensity light can reduce the depression symptoms in patients with depressive disorder 2 , light had been widely used to treat affective disorder in clinical practice [3][4][5] . In recent years, researches have also suggested signi cant effects of light on subjective affect and comfort in normal population 1,6-8 , which provided possibilities for using light to improve subjective well-being in daily life and attracted sustainable attention in the past decades.
Illuminance and spectral distribution are two commonly studied properties of light in modulating subjective affect and comfort, dated back to the in uential study by Kruithof who presented a graph to show preferred interior lighting conditions in terms of illuminance and correlated color temperature 9 .
According to the Kruithof's rule, people felt more comfortable and pleasing at low illuminance and low CCT light condition and at high illuminance and high CCT condition. The Kruithof's rule was widely applied in interior lighting design, although subsequent studies were highly controversial [10][11][12][13][14] . For example, Park et al. (2013) found a signi cant effect of illuminance but not CCT on pleasant feeling, at which subjects felt more pleasant at high (600 lx) instead of low illuminance (300 lx) condition 13 . On the contrary, Sunwoo et al. (2017) found similar effect of high and low illuminance (700 lx vs. 300 lx) on affect under short-term light exposure (~several minutes). In the study of the relationship between spatial distribution of light and subjective affect and comfort, lower CCT was endorsed by some 1,15-17 while higher CCT was endorsed by another researches 7,[18][19][20] . For example, Sunwoo et al. (2017) found a signi cant positive effect of lower CCT (3000 K) on subjective feelings (e.g. felt more comfort and happiness), as opposed to 6500 K, but  found that exposure to a higher CCT condition (6500 K) for 2 hours before bedtime strongly suppressed the secretion of melatonin and enhanced subjective well-being (e.g. mood, tension and physical comfort), as opposed to the 2500 K CCT condition. In addition, many studies failed to nd the effect of CCT on subjective comfort and affect 13,21 .
Factors such as lighting duration and timing may contribute to the con icting results. At one hand, lighting duration ranged from several minutes 1 to hours 22 in previous laboratory studies and could last even several weeks in eld studies 18 . On the other hand, subjects can be exposed to light in the morning (e.g. 08:00; 23 , afternoon (e.g. 15:00; 13 , and/or evening (e.g. before sleep; 19 . Anecdotal evidence has suggested different timing of lighting may have reverse effect, e.g. advancing or delaying circadian phase 24 . Also, the majority of researches about the effects of light were carried out together with cognitive tasks, e.g. go/no-go, N-back working memory, Psychomotor Vigilance Task, Flanker task, long-term memory, and emotional judgment task 13,16,17,22,25 . The tasks performed during light exposure and associated (potential) fatigue might also confound the effects of light on subjective affect and comfort 26 .
Previous studies had con rmed the effectiveness of short-term light exposure on modulating physical and psychological processes 8,13,27,28 . In the current study, we examined the effects of short-term light exposure with different combination of illuminance, CCT, and exposed time of day on subjective affect and comfort to minimize the confounding in uence of fatigue on subjective feelings. We also spared subjects during light exposure, aiming to isolate "purer" effects of light on subjective affect and comfort. We found signi cant interaction between the illuminance level, but not CCT, and time-of-day on subjective comfort. Our results suggested that research of lighting effects and design of healthy lighting should take into account the in uence of lighting time of the day.

Results
MANOVA analysis only revealed signi cant main effect of time-of-day on arousal (F(2, 241) = 5.572, p = 0.004, Partial η² = 0.044) and signi cant interaction of illuminance and time-of-day on comfort (F(4, 241) = 3.980, p = 0.004, Partial η² = 0.062). All other interactions were not signi cant (Table 1).   To further explore the relationship between illuminance and subjective comfort, a Spearman correlation analysis was performed. Interestingly, a positive correlation (r = 0.280, p = 0.011) was found between illuminance and comfort in the morning while a signi cantly negative correlation (r = -0.320, p = 0.006) was found between them in the evening (Figure 3).

Discussion
A signi cant main effect of time-of-day on arousal level was found in the current study. We didn't nd any signi cant effect of CCT and illuminance on the affective states (e.g. valence, arousal, dominance) under short-term light exposure (e.g. 3 mins). Interestingly, a signi cant interaction between illuminance and time-of-day on subjective comfort was detected. Further analysis revealed a dissociated relationship between illuminance and comfort in the morning and evening.
Previous studies have also revealed signi cant effect of time-of-day on subject's arousal level 29,30 . For example, the work performed by Parrott who assessed the arousal in different groups (e.g. sedative smokers, stimulant smokers, nicotine deprived smokers and non-smokers) at different times of day showed that feelings of arousal were signi cantly affected by time. The result that all groups started the day after waking with low arousal but peaked at different times suggested that even though the changing patterns over time differed between subgroups, the arousal was signi cantly affected by time of day 29 . Our result that participants felt more arousal in the morning and evening also suggested this point.
In the present study, the effect of illuminance, instead of CCT, on subjective comfort was found to be signi cant. Our results were inconsistent with the Kruithof's rule that suggested the preferred interior lighting conditions with combinations of illuminance and CCT 9 while partially consistent with some other researches that only found signi cant effect of illuminance on subjective comfort. According to Boyce's work, people felt more pleasant and comfortable in light condition with higher illuminance level while the CCT didn't show signi cant effect 10 . Fotios revised the Kruithof's curve and suggested that the variation in CCT has a negligible effect on subjective pleasure and the one condition to avoid is low illuminance less than approximately 300 lx 12 . On the other hand, our results indicated the effects of illuminance on subjective comfort was affected by the lighting time of the day: subjects felt more comfortable in higher illuminance level in the morning, and low illuminance in the evening.
Light is one of the major cues to discriminate day and night and keep normal circadian rhythm 31,32 . Exposure to less light at daytime and bright light in the evening can impact circadian rhythm negatively 33,34 . We argue that light echoed with natural circadian may inherently link to more positive effects 13  To minimize the potential confounding of fatigue on subjective feelings, the present study adopted a short-term light exposure. However, compared to short-term exposure, long-term exposure to light might have a different or even a reverse effect on cerebral activity. For example, a research conducted in 2017 explored the modulation of alerting effects by timing and duration of light exposure and indicated that an alerting effect of short-term exposure to monochromatic blue light was testi ed by a decrease in lowerfrequency electroencephalogram bands while a reverse effect of long-term exposure to blue light was demonstrated by a increases in lower frequency bands 35 . The longer lighting duration should be investigated in future.

Conclusions
In summary, the present study reported a signi cant effect of interaction of light illuminance and time-ofday on subjective comfort. Our results contribute to the understanding of light effect and will be of interest to the design of healthy lighting in daily work and life.

Participants
Thirty healthy adults, aged 18 to 31 years (13 females, 22.45 ± 3.26 years), participated in the study and received monetary compensation. All subjects had normal or corrected-to-normal visual acuity, normal color vision, and were free of ocular diseases. All participants gave their written informed consents before experiment began. The study was approved by the local ethical committee of Institute of Psychology, Chinese Academy of Sciences and all research activities were adhered to the principles of the Declaration of Helsinki.

Study design
The experiment employed a 3 × 3 × 3 mixed-group design with three independent variables: illuminance (50 lx vs. 100 lx vs. 500 lx, at 0.75 m height table level) and CCT (2,700 K vs. 5,000 K vs. 6,500 K) as within-subject factors, and time-of-day (morning vs. afternoon vs. evening) as betweensubjects factor. The relative spectral distribution of the nine light conditions was shown in Figure 4. The morning session started at 9:30 a.m., the afternoon session started at 3:00 p.m., and the evening session started at 7:30 p.m. Constrained by the availableness of subject's schedule, nine, thirteen, and eight subjects were assigned to the morning, afternoon and evening session, respectively. Each participant went through all nine light conditions and every condition was repeated twice. The testing order of all light conditions was counterbalanced across subjects.
For each subject, after 5-min dark adaptation (< 0.01 lx; Figure 5), different lights were presented one after another with a 2-min dark (< 0.01 lx) interval in-between. In each trial, participants were exposed to one light condition for 3 min, during which subjects were asked to sit still in the rst 2 min and rate their current subjective feelings after a voice prompt in the last 1 min. Subjects were informed to sit quietly without closing eyes during exposure.

Environmental Setting
The study was carried out in a windowless room, with a size of 4.36 × 4.03 ×2.70 meters. The walls and ceiling of the room were white. Six identical custom-made LED lighting devices, with spectrum ranged from 380 nm to 780 nm and surface homogenized by scatter plate, were mounted onto the ceiling. Each LED lighting device consisted of 22 classes of monochrome LEDs and two classes of white LEDs. The intensity of each class of the LED can be adjusted for different combinations of spectral intensity with control accuracy better than 1%. The device can simulate most natural lighting conditions, in which the desktop illuminance of typical CIE white illuminant can reach 750 lx and correlated color temperature can cover from 1800 K to 7500 K.

Assessment of subjective feelings
Subjective affect states, comfort, and eye fatigue were evaluated under each light condition. Participants rated their affects in terms of valence, arousal, and dominance with 9-point Self-Assessment Manikin Test 36 , with 1 for unpleasant, calm, and controlled state, and 9 for pleasant, excited, and in-control state. The subjective comfort and eye fatigue under each light condition were rated via a 7-point scale, with 1 for discomfort or non-fatigue to 7 for comfort or eye fatigue.

Statistical analysis
All analyses were performed by SPSS version 20 (IBM, USA). The rating scores of subjective feeling for each light condition were averaged respectively for each subject. The effects of time of day (morning vs. afternoon vs. evening; between-subject factor), CCT (2700 K vs. 5000 K vs. 6500 K; withinsubject factor), and illuminance (50 lx vs. 100 lx vs. 500 lx; within-subject factor), and interactions of the three factors on subjective feelings, e.g. valence, arousal, dominance, and comfort, were computed using multivariate analysis of variance (MANOVA), with eye fatigue as covariate. Post-hoc tests of simple effects were performed with Bonferroni correction 37 . Non-parametric correlation analysis was performed to test the relationship between illuminance and comfort in different times of day.

Data Availability
The data that support the ndings of this study are available from the corresponding author upon reasonable request.