Evaluation of the effects of blue-enriched white light on cognitive performance, arousal, and overall appreciation of lighting

This crossover repeated-measures experimental study design involved 30 healthy adults (19 females, mean age = 23.4 years, range 19–33 years). The sample size was representative of the university population, where recruitment was easier. The young age groups helped to reduce age-related yellowing of the lens of the eye which could increase the variability in terms of retinal light response to enhanced blue white light (29).

The first experimental condition was randomly drawn and though participants were informed on the objectives of the study, they had no intake on which room corresponded to the blue-enriched lighting. Exclusion criteria were: Dyslexia—Presence of a major psychiatric diagnosis (schizophrenia or other psychotic affective disorder, bipolar, depression)—Being diagnosed with retinal disease (uncorrectable 20/20 vision, macular degeneration, cataracts, glaucoma, diabetic retinopathy, etc.), while wearing glasses was not an exclusion factor.—Any disability that may prevent completion of the procedure (brain injury within the last three months or head trauma requiring consultation)—Pregnant women—Night shift workers (unless inactive for the last two months)—Recent travel to a location more than two time zones within the last month—Taking medication (including antihistamines) that may impact alertness levels. Participants who consume caffeine, alcohol, cannabis, and tobacco were not excluded, but were instructed not to use any on the day of the assessments. This approach aimed to isolate the effects of lighting stimulation from the potential influence of other exogenous substances. While regular caffeine drinkers might experience an impact on their performance, replicating this criterion on both examination days enabled us to assess the effects without the confounding influence of other substances. This strategy contributes to a more focused examination of the specific impact of light exposure on cognitive performance and perception. Any other drug use, however, was an exclusion criterion. Each participant signed a consent form approved by the Centre Intégré Universitaire de Santé et des Services Sociaux de la Capitale-Nationale (CIUSSS-CN) Neuroscience and Mental Health Research Ethics Committee.

The data collection spanned 6 weeks from mid-November to end of December 2021 in Quebec City, with sessions conducted between 8 am and 6 pm through the day. All participants were exposed to the two lighting conditions (experimental, LED blue-enriched white light at 5000 K and control current office fluorescent white light at 3,500 K) in a counterbalanced order. Each exposure was performed on a different day to avoid any carry-over effect but at the same time of day, to counteract any circadian influence on variables such as mood and arousal (30). Participants were asked to have a minimum of 1 day and a maximum of 31 days between the two exhibitions. Light exposure took place in two adjacent rooms, which were freshly painted in white prior to the experiment. Each session began with a 20-min period in which the participant was exposed to very dim (about 25 lux, 3,000 K) non-stimulating lighting while completing descriptive questionnaires. This period was considered a pre-lighting condition, allowing for a reset of the ipRGC stimulation and thus creating a similar baseline for each trial day. The rationale behind the 20-min period is in line with the brain activation theory where 20 min of exposure would be sufficient to activate both cortical and subcortical brain regions (3, 31). We therefore assumed that 20 min in dim light would be sufficient to deactivate the retinal melanopsin cells. Following the pre-condition period, the participants were exposed to one of the two lighting conditions for 50 min.

The dimensions of the control condition room were 2.83 m × 3.00 m, which was very similar to the experimental room 2.83 m × 2.62 m. None of the rooms had windows. The office furniture (brand new) in each room was the same and positioned in the same way, so that the exposure to light was similar and the walls remained bare, to limit the qualitative influence of a certain type of decor on the overall assessment of the lighting environment evaluated by the questionnaire. Finally, the wall color of each room was white (reflectance about 80%) to limit the absorption of light by a particular color. Even though the arrangement of the desks and the lack of windows does not comply with architectural and ergonomic standards, this arrangement permitted to have a perfect control over the lighting with no interference from natural light.

Lighting was provided by two ceiling panels consisting of four 4-foot light tubes. In the control condition room, the current light fixtures were 3,500 K E-Lume^® (32) fluorescent tube that yielded to 897 lux (horizontal plane), and 335 lux and 159 mel-EDI (vertical plane). In the experimental room, the conventional fluorescents were replaced with 5,000 K Ecosmart^® (33) LED retro-fit tubes that yielded to 837 lux (horizontal plane) and 345 photopic lux and 263 mel-EDI (vertical plane). Therefore, each room provided roughly the same amount of photopic lux, but different mel-EDI levels, with one about 100 mel-EDI below the recommended standard and the other meeting the standards of 250 mel-EDI. Light spectra irradiance were assessed (vertical plane) with the Ocean Insight © spectrometer HR4000 High Resolution (Peabody, MA) (34) and then entered in the CIE toolbox to derive α-opic EDI (lx) illuminances (25). Photopic lux was measured with the ILT5000 research radiometer (International Light technologies, Peabody, MA). Measurements were taken 1.2 m from the ground (4 feet) in the vertical plane for the spectra and photopic lux (35), and in the horizontal plane at 0.76 m from the ground for the photopic measurement only (36). The LED lighting was chosen on the basis that it was commercially available from various hardware stores. Table 1 depicts the α-opic EDI illuminance (lx) for each light condition and Figure 1 represents the α-opic spectra of each room, both obtained with the CIE toolbox (25). For the pre-condition lighting, we used an auxiliary lamp attached to the desk (DAZZNE®, Longhua District, Schenzen) programmed at 3000 K and 50% light power to expose participants to about 25 photopic lux and 10 mel-EDI. The low light intensity used for the neutral condition period was chosen to be sufficient to complete questionnaires without stimulating the melanopsin ipRGCs (17). The advantage of LED light is that it produces a peak in the blue spectrum that closely match the melanopsin cells peak sensitivity and should therefore be very effective in stimulating the melanopsin system while maintaining a whiter looking color. See Figure 2 for the visual impact of both lightings in the different rooms. These lights also consume less energy (20 watts versus 32 watts) and have a longer life span (36,000 h versus 30,000 h for conventional fluorescents). These characteristics make them a wise choice in an energy-efficient context for future applications in the Far North [for instance (37)].

The independent variable corresponds to the measurement in mel-EDI of each lighting. The primary dependent variables correspond to the measures of concentration and arousal. These are indeed the two variables of primary interest that we hypothesized would improve significantly more in the experimental lighting when compared to conventional lighting. The secondary variables concern visual acuity, contrast perception and general appreciation of the environment. The vision related variables, while being of interest, were not expected to yield to any difference between the control and experimental lighting. In addition, these tests contributed to fill-up the time during light exposition without increasing the mental load. Information on the appreciation of the room was a relevant variable to collect to verify if the experimental environment with blue-enriched white light receives the same appreciation as the conventional fluorescent lighting room.

Each session was conducted in two phases. The first was the 20-min pre-condition period when the descriptive questionnaires were administered. Then, the 50-min light exposure was used to administer the various tests. After the first few minutes of light exposure, the participants took the KSS test once to obtain a baseline value of their sleepiness. Then, they performed the d2-R test to obtain a baseline value of their concentration performance. Afterwards, their visual performance was evaluated with the FrACT test followed by a 20-min period during which they were allowed to read magazines. They were then given the two appreciation questionnaires. The session ended with a second evaluation of the d2-R test and KSS test. Figure 3 provides a visual summary of the protocol.

Three fixed factors were considered in the analyses: Group: Gr 1: participants who began with condition A (n1 = 15); Gr 2: participants who began with condition B (n2 = 15). Condition: A: Control lighting with conventional fluorescent light (3,500 K, ~150 mel-EDI); B: Experimental lighting with LED blue-enriched white light (5,000 K, ~260 mel-EDI). Time: Pre: at the beginning of the exposure to the condition; Post: at the end of the 50-min exposure to the condition. Table 2 summarizes the items (mean and standard error) found in each of the questionnaires.

The factors time and condition were used as repeated measures to establish a possible correlation between the observations made on the same participant. Thus, a 3-factor ANOVA with repeated measures was used. We had to remove two participant’s data since they were considered as extreme values ((i.e., x > percentile 75 + 1.5(percentile 75-percentile 25) OR x < percentile 25–1.5 (percentile 75-percentile 25)) to allow a normal distribution.

The factors condition and time were again used in this 3-factor ANOVA with repeated measures. Both scores, concentration capacity (CC) as well as error rate (E%) were analyzed.

A 2-factor repeated measures ANOVA was used, where the condition factor was considered to establish a potential effect on the same participant. Contrast Sensitivity (in LogCS) and visual acuity (logMAR) were analyzed. One participant’s results were removed from the visual acuity analysis, as they were extreme values, which was calculated in the same way as for the subjective alertness state.

The condition factor was used as a repeated measure in the 2-factor ANOVA with repeated measures. The total preference score along with the comfort score were analyzed.

Again, the condition represented the factor used as a repeated measure in this 2-factor ANOVA with repeated measures. Each of the Visual Analogue Scales (VAS) was analyzed individually.

All the ANOVAs with repeated measures were performed using the MIXED procedure of SAS/STAT software (Version 9.4; SAS Institute Inc., Cary, NC). The graphics in this work were generated using GraphPad Prism 9.3.0 (345) in Viewer mode. GraphPad Prism is a product of GraphPad Software, LLC.

The average number of days between trials was 3 days with a range of 1–9 days. Sex was not equally distributed with almost twice as many women (N = 19) than men (N = 11), but there was no difference in age between sexes (t-test, p = 0.22). The results of the circadian typology show that 63% of the participants belonged to the “neither morning nor evening” category and none of them were in the frankly morning or frankly evening categories. Seasonal affective disorder (SAD) vulnerability was present in 23% of our participants while 33% had excessive daytime sleepiness on the Epworth Sleepiness Scale and 20% had sleep disturbance on the PSQI. None of these findings, however, had a significant impact on the test scores (KSS, d2-R, FrACT, OLS-modified and VAS) in the ANOVA analysis with repeated measures adjusted for co-variables.

Both lighting conditions yielded to some improvement in alertness over the 50-min light exposure, but reach significance only in the experimental light condition (p = 0.03, Cohen’s d = 0.42) (Figure 4).

Subjects significantly improved their concentration performance at the end of the 50-min exposure in both conditions (p < 0.0001) with no significant difference between conditions (p = 0.79), as shown in Figure 5.

One participant’s results had to be removed because all data were outliers. No factor demonstrated a statistically significant change over the 50-min exposure. Thus, participants did not improve by time nor condition (p = 0.40 and p = 0.97, respectively).

No statistically significant difference was shown between the two conditions during the 50-min light exposure (p = 0.61).

When exposed to the experimental, blue-enhanced light (5,000 K, ~260 mel-EDI), participants performed significantly worse on the contrast perception test when compared to their performance in the control lighting (3,500 K, ~150 mel-EDI) (p = 0.01, Cohen’s d = 0.50).

Participants significantly preferred the control room (3,500 K, ~150 mel-EDI) compared to the experimental lighting (5,000 K, ~ 260 mel-EDI) (p = 0.004, Cohen’s d = 0.56), as shown in Figure 6. However, a positive interaction was noted regarding the order of exposure. Thus, participants who started with the experimental condition tended to increase the preference score when they were asked to evaluate the control room to which they were exposed secondly (p = 0.07). Figure 7 illustrates this interaction.

The last question of the questionnaire was not considered since it involved difficulty reading items on the screen. Since the only test required on-screen was visual performance, this question was inapplicable with the setup of our study. Overall, there was no significant difference between the two conditions for visual comfort (p = 0.17).

The control fluorescent-lit room (3,500 K, ~150 mel-EDI) was found to be significantly more comfortable (p = 0.002, Cohen’s d = 0.62), more cheerful (p = 0.02, Cohen’s d = 0.46) as well as more pleasant (p = 0.005, Cohen’s d = 0.55). The experimental LED room (5,000 K, ~260 mel-EDI) was perceived as brighter (p = 0.004, Cohen’s d = 0.58) and tended to be perceived as more stimulating (p = 0.10). In terms of the degree of pleasantness as well as the degree of cheerfulness rated by participants, a positive interaction was noted in relation to the order in which they were exposed. Similarly, to the lighting preference noted in the modified OLS questionnaire, participants who started with the experimental condition tended to rate a higher level of pleasantness for the control room on their second encounter (p = 0.03). The same goes for the level of cheerfulness (p = 0.01).

Although our short exposure study seems to favor the use of conventional fluorescent light, it would be interesting to evaluate the general appreciation of an environment lit by a blue-enriched LED light over a longer period as done by Viola et al. (19). In this way, participants would have more time to accustom themselves to the lighting and potentially improving appreciation for it, as was shown in their study. Moreover, a more recent study comparing a 130-min exposure to 5,700 K and 2,700 K demonstrated improved arousal and attention performance under 5,700 K compared to 2,700 K, though without a significant impact on visual comfort. In fact, participants reported fewer eye-related symptoms under 5,700 K lighting (56). In addition, it is possible that the high reflectance of the white walls (about 80%) had an impact on the overall appreciation of the lighting. Indeed, it seems that white walls allow a better appreciation of warm colored lighting (lower CCT), while blue painted walls allow a better appreciation of cold colored lighting (higher CCT), which would explain why participants preferred the control (3,500 K, ~150 mel-EDI) lighting when compared to the experimental (5,000 K, ~260 mel-EDI) lighting (60). Perhaps a similar study with blue walls for cooler-looking lighting would balance the preferences between the two rooms, although more light would be needed to reach the 250 lux mel-EDI threshold since colored walls do not reflect as much light as white walls.

Another consideration is that there were no windows in the rooms studied, which allowed for total control of melanopic and photopic intensities. On the other hand, these conditions do not represent the full reality of contemporary architecture, where biophilia associated with daylighting is an integral part of building design (61). It may therefore be that the preference for blue-enriched lighting in the Viola et al. (19) study is a combination of chronic exposure time associated with more biophilic components such as natural light.

An important consideration to bear in mind is that our study is currently centered on the Quebec population. It is conceivable that Far North populations might not respond in the same manner due to distinct customs and lifestyle habits that may result in more outdoor activities. Additionally, environmental factors need to be considered, such as the unique conditions in Quebec City where cloud cover might scatter sunlight differently compared to other regions. These cultural and environmental variables could potentially influence how individuals from Far North regions and Quebec City respond to the lighting conditions studied in our research. Further investigation and studies encompassing diverse populations and geographical locations may provide a more comprehensive understanding of the impacts of lighting on different communities. Furthermore, the study population comprised predominantly of university students, limiting its applicability to the working population with potentially higher age demographics. A more precise representation of the working population’s age would be desirable for future replications.

Although the rooms used do not meet the ergonomic and architectural recommendations for offices, they allowed this study to be conducted in a controlled environment. In fact, we were able to recreate an environment with a positioning that allowed participants to be exposed to the 250 mela-EDI standard under one light, while the other light was short of 100 lux below the mel-EDI exposure standard. For example, the desk was positioned so that participants had their backs to the light source and received their exposure primarily through the wall (by reflection), whereas in actual practice this exposure should be more directly in the field of view with a desk positioned under the light fixture. This positioning could also give the impression that the work surface is not sufficiently illuminated, thus reversing the illuminance ratio for performing a task. In addition, this study did not account for the effect of partial light masking of participants when seated at a desk. In addition, because participants did not work on screens except for the very brief FrACT test, the reflection of light off a screen (which would also contribute to enhanced blue light exposure) is not accounted for and should be considered in future studies involving computer work.

This study did not include baseline measures. As indicated in a study by Smolders and De Kort (62), participants may exhibit variations in baseline responses on different days. The absence of baseline measurements in our study leaves us unable to account for or assess the potential impact of this day-to-day variability on participants’ responses to different lighting conditions.