Effects of Correlated Colour Temperature and Timing of Light Exposure on Daytime Alertness in Lecture Environments

Transcription

1 J. Light & Vis. Env. Vol.34, No.2, Paper Effects of Correlated Colour Temperature and Timing of Light Exposure on Daytime Alertness in Lecture Environments Emilia RAUTKYLÄ, Marjukka PUOLAKKA, Eino TETRI and Liisa HALONEN Lighting Unit, School of Science and Technology, Aalto University, Finland Paper received October 28, 2008, Accepted January 21, 2010 ABSTRACT There is previous evidence that using high correlated colour temperature (T cp ), white light can improve concentration and alertness, resulting in better productivity in office environments. In this study, it was investigated whether the same applies to students exposed to light, while attending lectures. The study consisted of two sequential field studies, conducted during spring and autumn. The subjects were exposed to light provided either by 4,000 K or 17,000 K T5 fluorescent lamps for 90 minute lecture periods. The subjects were asked to judge their level of alertness on a 9-step scale at the beginning, and then at the end of each period. In spring, no change in alertness during the lecture period was detected. There was neither shown correlation between the T cp of light and subjective alertness. Instead, the study in autumn found that the decrease in alertness during lectures was significant, and post-lunch dip effect appeared strong in the afternoon. The autumn study indicated that exposure of the subjects to a 17,000 K light source in afternoon lectures, potentially assists students to maintain higher levels of alertness, as compared with the results of the 4,000 K light. This study therefore suggests that the colour temperature, and the timing of the light exposure, play important roles in student alertness in lecture room environments. KEYWORDS: light, alertness, colour temperature, lecture environment, dynamic lighting 1. Introduction Since the discovery of intrinsically photosensitive retinal ganglion cell (iprgc) in the human retina, it has been suggested that it is possible to use light to control levels of alertness. The alerting effects of light are often explained by its mediating effect on melatonin suppression 1). Nevertheless, this explanation can only apply at night, because during the daytime, melatonin levels are low. It is likely that the mechanism behind the alerting effects of light involves melanopsin, the photopigment in iprgcs 2). Melanopsin is most sensitive to blue light with wavelengths of around 480 nm 3)4). There is evidence that melanopsin is involved in light-induced alertness during night time 5). It has, however, not yet been established whether melanopsin plays a role in daytime alertness as well. The action spectrum of melanopsin has been defined using quasi-monochromatic light exposure. Therefore there have been many studies examining whether it is possible to enhance daytime alertness, by exposing subjects to blue light of a narrow spectrum 6). The call for new recommendations in general lighting, has, however, raised a need to investigate whether alertness can be affected by light of a continuous spectrum. For this purpose, it was more sensible to setup a test bed with light sources of wider spectral distribution. The search for a light spectrum that is beneficial to both the visual, and the non-visual systems of perception, has turned into a comparison of the effects of light sources of differing correlated colour temperatures. In a study conducted between February and May 2006, Mills and co-workers 7) used five items of the SF-36 questionnaire, and a modification of the Columbia Jet Lag Scale to evaluate the wellbeing and work performance of office employees. They reported that compared to 2,900 K lighting, 17,000 K lighting helped to improve concentration and daytime alertness in office work. These results are consistent with the recent findings of Viola and co-workers 8). They exposed 104 white-collar workers to two lighting conditions (4,000 K and 17,000 K), each lasting 4 weeks, and used questionnaire and rating scales to assess their alertness, mood, sleep quality, performance, mental effort, headache and eye strain throughout the eight-week intervention. According to Viola and co-workers, exposure to blue-enriched white light improved subjective alertness, performance, and evening fatigue. Together these findings lead us to hy- 5 The Illuminating Engineering Institute of Japan

2 60 J. Light & Vis. Env. Vol.34, No.2, 2010 pothesize that high Tcp light might induce a greater effect on alertness during daytime working hours than low Tcp light. However, there are two things that should be investigated before the alerting effect of high Tcp light can be verified, and generalized to apply in other environments. Firstly, in the two experiments 7)8) there was no control over the additional exposure to light outside the working hours. Thus it is not known whether previous light exposure or other environmental and subject dependent factors could have influenced alertness. Considering e.g. that the alerting effect of coffee is wellestablished 9), and exposure to daylight has been proven to enhance alertness among office workers 10), the involvement of other factors in mediating alertness needs to be clarified. Secondly, it is worth investigating the effect of dependency on the time of day. Van Bommel has suggested that the right combination of action and relaxation in a working environment could be achieved by following the human rhythm with artificial dynamic lighting 11). In this scenario, light should be cool-white at wake-up time, and turn more warm-white towards the midday, to create a relaxing atmosphere for lunchtime. After lunch a sharp rise in colour temperature would reactivate the body for work. In this study, the effect of Tcp on daytime alertness was examined when other possible external factors were taken into account. A test setting that exposed the subjects to two light sources, differing in their Tcp, was built in a lecture hall. The objective of the study was to observe the effects of light source colour temperature on alertness, using both high Tcp light and low Tcp light. Moreover, these tests were designed to investigate whether the subjects reaction to light of a certain Tcp is dependent on the time of day or year. 2. Methods and subjects 2.1 Background and basic design The study took place in Helsinki (60 10'N and 24 56'E), where due to its geographic location, the availability of daylight varies between 6 and 19 hours, depending on the time of year. To observe the seasonal effects on the study results, the study was conducted in two stages; first stage in late spring and second in late autumn. The amount of daylight hours was approximately 12 hours in the first, and 9 hours in the second stage. The basis of both experiments was the same. In the latter experiment the study protocol was improved. The improvement was based on feedback from the subjects of the first experiment. All experiments were conducted in lecture environments with university students as subjects. The test environment was designed so that it would be as natural as possible for the students. Therefore the experiments were carried out while the subjects were attending university courses. Lecturers, course content and lecture session topics varied between the different sessions studied. The effect of these variables on alertness was tested by asking how interesting the lecture felt. This was done at the end of each lecture. The illumination levels and illumination distribution were set to be suitable for visual tasks, such as reading the lecture slides and taking notes. No daylight was allowed in the lecture hall. 2.2 Physical setting Both experiments were conducted in a 620 m 2 amfishaped lecture hall equipped with 69 OfficeNova 240TCS luminaires with D6 optics by Philips Oy Luminaires. In each luminaire there were two different 49 W light sources: T5 lamp with Tcp of 4,000 K (Ra>80) and Philips ActiViva lamp with Tcp of 17,000 K (Ra>80). Both lamps were equipped with electronic ballasts and were dimmable. The lighting control system Helvar DIGIDIM used DALI-protocol. In this paper the lighting environments will be referred as 4,000 K (Figure 1) and 17,000 K (Figure 2) environments according to the light sources. Figure 1 4,000 K lighting environment Figure 2 17,000 K lighting environment The Illuminating Engineering Institute of Japan 6

3 J. Light & Vis. Env. Vol.34, No.2, Figure 3 Light spectra measured in the lighting environments under the two light sources. T cp provided by 4,000 K light source was 3,870 K and by 17,000 K was 12,370 K. The spectral power distributions of the two lighting environments were measured with Ocean Optics HR400 High-resolution Spectrometer (Figure 3) in the middle of the hall with maximum luminous output of the lamps (measured at 0.8 m, i.e. table height). Due to the reflections from the dark surface materials the measured Tcp provided by 4,000 K fluorescent light source was actually 3,870 K and the Tcp provided by 17,000 K light source was 12,370 K. Dimming the lights did not change the Tcp significantly. 2.3 Illuminance The two light sources were used randomly in turns so that in half of the lectures of each experiment the lighting was provided by 4,000 K and in half of them by 17,000 K flourescent lamps. The lighting conditions were alike except for the light spectrum. Based on a pilot study the horizontal illuminance in the hall was set to 800 lx (measured at 0.8 m in all studies) in the first study. However, feedback from the subjects indicated the need to increase the illumination level. Hence, in the second study the illuminance was set to 1,000 lx. At the same time the illuminance uniformity (Emin/Eave) improved from 0.7 to Timing and duration of the light exposure Both experiments included tests at two different times of the day, morning and afternoon. In the first stage in spring, the lectures took place between 9:15-10:45 and 12:15-13:45. For the second stage the lectures were chosen so that there would be a longer time period between them. Therefore the lectures in autumn were held between 8:15-9:45 and 14:15-15:45. In both experiments the test period was 6 weeks during which 10 lectures took place. The morning and afternoon lectures were held on different weekdays to avoid order effect with different light sources. Each lecture was of 90 minutes duration with no breaks. 2.5 Protocol The subjects were assigned a personal identifier and they were asked to fill in questionnaires both before and after each lecture. The subjects were not informed that this was a study of lighting. They were told that the study is about environmental factors in a learning situation. The lecturer was aware of the study but did not take part in it. In both experiments each subject filled a prequestionnaire which linked the age, sex and the identifier that the subjects used throughout the entire study. One essential part of the pre-questionnaire was Morningness-Eveningness Questionnaire 12). This questionnaire was used to determine the chronotype of the person. Chronotype is an attribute of human beings reflecting whether they are alert and prefer to have their activities early or late in the day 13). Subjective evaluation forms consisted of 9-step Karolinska Sleepiness Scale (1 referring to very alert and 9 to very sleepy, fighting sleep, an effort to keep awake ) 14). There were also questions about environmental factors such as air draughts, heat, noise levels, and how the lecture felt in terms of the interest in the lecture. In the experiments, alertness was marked down before and after the lecture, while the other questions were only answered at the end. Because of the nature of the study (a random sample of students as subjects, real life test-setting) the subjects were not asked to follow a certain sleep-wake routine throughout the study. Instead, the use of stimulants, such as coffee, energy drinks etc., and eating before the lecture were recorded and taken into account in the data analysis. The room temperature and CO2-concentration in the lecture hall were measured at the beginning and at the end of each lecture in both experiments. The second study also included measuring the illuminance in the corridor and outside the building. This was done to provide data about the lighting conditions the subjects had been exposed to immediately preceding the lecture. For the same reason, the subjects were asked about their movements within the past hour. 2.6 Subjects 16 university students participated in the spring study (all male; age range years; mean 24±1.3 SD). Based on the pre-questionnaires, 6 subjects of the 16 were categorized as morning type and 2 subjects as evening type by their chronotype. Eight subjects were neither evening nor morning type and were thus classified as neither type. In the autumn study there were a total of 138 university students (17 female, 121 male; age range years; mean 22±1.8 SD). By chronotype grouping, 33 were morning type, 22 evening type 7 The Illuminating Engineering Institute of Japan

4 62 J. Light & Vis. Env. Vol.34, No.2, 2010 and 83 neither type. No student was able to participate in any more than one of the studies. Because attending lectures is voluntary, the amount of participants decreased in each study as the course progressed. On average, each subject took part in 46% of the lectures in the first study (73 returned answer sheets) and 22% of the lectures in the second study (309 answer sheets). One answer sheet was provided, per subject, per participated lecture. 2.7 Data analysis To investigate the effect of lighting on students alertness throughout the entire course, all observations were used in the analysis instead of choosing, for example, the lecture with the greatest amount of participants, i.e. the largest sample. Statistical analysis of the repeated measures was carried out with Statistic Package for the Social Science (SPSS). The significance level was set to 0.05 in all comparisons. Comparing the alertness at the end, to the alertness at the beginning of the lecture, was done by using two-tailed paired Student t-test and its probability value Pt. This t-test is a special case of ANOVA that assesses whether the means of two groups are statistically different. The relationship between the Tcp and the change in alertness was measured with the Pearson Chi-Square test, appropriate for assessing associations of categorical variables. This was done for each lighting condition separately. Finally, the factoring of the lecture content, prior lighting conditions, use of stimulants, consumption of food, chronotype of the subject and environmental factors in the results, was analyzed by adding layers in the cross tabulation. In order for the correlation of the Tcp and the change in alertness to apply despite the factors mentioned, the majority of the Pearson Chi-Square values Pcs had to be within the significance level of 0.05 with each added layer. In simple correlations of two continuous variables Pearson s correlation coefficient r was calculated. Unfortunately spring and autumn data could not be compared due to the adjustments in study design (illuminance, lecture times etc.), and also the major difference in sample sizes. 3. Results 3.1 Change in alertness The students were more alert at the beginning of the lecture in the autumn experiment (4.4±1.8 SD, median 4) compared to the spring experiment (5.0±1.5 SD, median 5). Due to the differences in the experiment setting and the subjects, it was however, not possible to determine whether the students had come to the lecture significantly more alert in autumn than in spring. The individuals were of different state of alertness when they arrived to the lecture. To be able to compare the effect of the Tcp between subjects and lectures, the effect was evaluated in terms of the changes in alertness instead of the absolute values of alertness at the beginning and at the end of the lecture. Therefore the change from 1 to 4 was considered the same as the change from 5 to 8; hence in both cases the alertness decreased three steps despite whether it remained in the alert or in the sleepy side on the Karolinska Sleepiness Scale. In spring, students alertness decreased points (Table 1) on average during the lectures. According to the paired t-test the change was not significant in any of the four lighting conditions (Pt>0.05). As for autumn, the change proved to be significant. In all lighting conditions in autumn it was very unlikely that the alertness did not change (Pt=0). As presented in column Mean in Table 1, in the morning lecture the alertness decreased on average 1.3 points with Tcp=4,000 K, and 1.4 points with Tcp=17,000 K. In the afternoon the decreases were 2.3 points and 1.0 points respectively. The change in alertness with each lighting condition is presented in Figures 4a-d. 3.2 Effect of Tcp and timing of exposure on alertness In spring there was shown no difference in the change in alertness that could be explained by the used light source. That applied both the morning and the afternoon lecture as seen in Table 2. As for in the autumn study, the alertness changed significantly differently, depending on whether light of low Tcp or high Tcp was used (Pcs=0.019<0.05). Compared to low Tcp, exposure to Table 1 The changes in alertness in different lighting conditions T cp=correlated colour temperature, SD=standard deviation, t=t statistic; the higher the absolute value, the less similar the means of the two samples are, df=degrees of freedom, P t=two-tailed probability value; if P t<0.05, the change in alertness is significant. Timing T cp (K) Change in Alertness Number of subjects Total Mean SD t df P t Spring - morning 4, , Spring - afternoon 4, , Autumn - morning 4, , Autumn - afternoon 4, , The Illuminating Engineering Institute of Japan 8

5 J. Light & Vis. Env. Vol.34, No.2, Table 2 Effect of T cp on changes in alertness in different lighting conditions Data of alertness is presented in Table 1. P cs=pearson Chi-Square probability value; if P cs<0.05, T cp correlates significantly with the changes in alertness. Timing T cp (K) Total P cs Spring - morning 4, , Spring - afternoon 4, , Autumn - morning 4, , Autumn - afternoon 4, , Figure 4 Changes in alertness in spring (4a-b) and in autumn (4c-d) The vertical bars indicate standard deviation. light of high Tcp helped the students to better retain their alertness in afternoon hours when the post-lunch dip effect is the strongest. In the morning lectures Tcp of the light had no effect on the changes in alertness (Pcs=0.899>0.05). 3.3 Variables factored in the correlation of Tcp and change in alertness in autumn afternoon To verify the results presented in 3.2, it was investigated whether the effect of the Tcp was independent from the lecture content and person s choronotype. Statistical assessment was also carried out to find out whether the Tcp affected differently the alertness of those who had come to the lecture from outside compared to those who had spent the hour prior to the lecture inside. Same kind of assesment was applied to those who had eaten or taken stimulants prior to the study and to those who had not. In addition to these comparisons the environment was factored in. The cross tabulations of these variables possibly influencing the relation of Tcp and change in alertness are presented in Table 3 and Table 4. The analysis of the data showed that on scale from 1 not interesting at all to 5 very interesting, the amount of interest the subjects had in the lecture did not affect the result but the correlation between the Tcp and the change in alertness applied independently from the lecture. This can be seen in Table 3 where the majority of the Pcs-values for lecture factor are within the signifigance level It indicates that there was only a small possibility that the detected correlation between the Tcp and the change in alertness would not have applied then the lecture was factored in. In addition, the Chi-Square test showed that the relation between the Tcp and the change in alertness, did not depend on subject s chronotype. Instead, the reaction to the colour temperature appeared to be similar in the majority of the chorotype cathegories (Pcs=0.020 and 0.049<0.05). Pcs-values for prior exposure to light were found to be below 0.05, indicating that the correlation between the Tcp and the change in alertness, was independent from lighting conditions that the subjects were exposed to prior to the lecture. However, it was not recorded how long each subject had spent time in the previous lighting conditions. Therefore, to be able to fully exclude the effect of prior exposure on the changes in alertness during the test, the lighting conditions should have been investigated more thoroughly. Using stimulants, such as coffee or energy drinks, was shown to have some effect on the results suggested in 3.2. Subjects who had taken stimulants were less likely to respond to the Tcp with changes in alertness compared to those who had not. In fact, the correlation of the Tcp and the change in alertness in autumn afternoon applied only with stimulant-free subjects 9 The Illuminating Engineering Institute of Japan

6 64 J. Light & Vis. Env. Vol.34, No.2, 2010 Table 3 Subject related variables factored in the correlation of T cp and the change in alertness in autumn afternoon. P cs=pearson Chi- Square probability value; if P cs<0.05, the variable does not influence the correlation. In Lecture 1 = not interesting at all, 5 = very interesting Variable Rating T cp (K) Change in Alertness Total P cs Lecture 1 4, , , , , , , , , , Prior exposure Outdoor 4, , Indoor 4, , Stimulants NO 4, , YES 4, , Food NO 4, , YES 4, , Chronotype Morning 4, , Neither 4, , Evening 4, , (Pcs=0.025<0.05). Also eating before the lecture was found to influence the correlation, even so that those subjects who had eaten before the lecture were not found to react to the high Tcp (Pcs=0.123>0.05). As presented by the length of the vertical bars in Figure 5, the room temperature and CO2-concentration varied some during the lectures. The variation of CO2- concentration correlated positively with the number of subjects attending the lecture (r=0.975) and the concentration was the highest in the lecture with the highest number of subjects. The amount of CO2, however, never rose above 800 ppm, a value often considered at which the air starts to feel stuffy 15). The room temperature remained between 20 C and 22 C throughout the test period and its variation was hardly influenced by the number of subjects (r=0.271). According to the statistical analysis the room temperature and CO2- concentration did not have an effect on the relation of Tcp and changes in alertness. In subjective assessment of the environmental factors the air quality, dustiness, smelliness, air dryness, draughtiness, heat and noisiness scored 3 on average, which referred to neutral feeling on a scale from 1 too hot to 5 too cold. As presented in Table 4 the statistical analysis showed no correlation between the environmental factors in question and the subjects reaction to the colour temperature. Therefore the effects that the Tcp had on alertness can be considered independent from the indoor climate. 4. Discussion In autumn alertness decreased significantly during the lectures. This was not seen in spring. One possible explanation to this is the difference in the amount of daylight available outside the test hours. Although in theory the length of the day did not differ much between spring and autumn studies (12h and 9h respectively) the intensity of daylight that the subject was exposed to prior to the test period, depended on the timing of the lecture. In spring the lectures were held on both sides of noon when the Sun is at it s zenith. Outside illuminance was measured to be approximately 4,000 lx prior-to-morning lecture and 30,000 lx prior-to-afternoon lecture (measured vertically at 1.6 m, i.e. eye level in open space). In autumn, the outdoor illuminance prior to the morning lecture, was below 100 lx due to the earlier timing of the lecture and later sun rise. The afternoon lecture was held later than in spring and the outdoor illuminance was approximately 2,000 lx. Therefore, it is clear that the students were exposed to much more light prior to the lectures in the spring study compared to the study in autumn. Because the indoor light conditions were also different in the two stages, it is, unfortunately not possible to verify the role of daylight in the changes in alertness in this study. There is, however, reason to believe that the whole light history plays a role in the determination of the alertness, not only the artificial lighting during the lecture. The Illuminating Engineering Institute of Japan 10

7 J. Light & Vis. Env. Vol.34, No.2, Table 4 Environmental variables factored in the correlation of T cp and the change in alertness in autumn afternoon. N.A=correlation cannot be calculated. P cs=pearson Chi-Square probability value; if P cs<0.05, the variable does not influence the correlation. 11 The Illuminating Engineering Institute of Japan

8 66 J. Light & Vis. Env. Vol.34, No.2, 2010 Temperature ( C) Temperature CO2-concentration Number of subjects in the afternoon lectures in autumn CO2-concentration (ppm) Figure 5 Mean room temperature and CO 2-concentration for each lecture afternoon lecture in autumn. The values were recorded twice during each lecture; at the beginning and at the end. Variation of the values during the lecture is marked with vertical bars. There were also other differences in the results from spring and from autumn that can be explained by the different timings of the lectures. In autumn, the alertness changed differently depending on whether low or high Tcp light was used during the afternoon lecture. Such effect was not seen in the spring study. One likely explanation to that lies in the hormonal activity of human that varies at different times of the day from morning to afternoon. According to the previous studies cortisol levels are high and body temperature low in the morning. In the afternoon the levels are controversal 18)19) and work performance decreases between 1pm and 4pm 20). This dip in performance during the midafternoon hours is often referred to as the post-lunch dip effect. In autumn the first lecture was held early in the morning (8:15-9:45) and the afternoon lecture late in the afternoon (14:15-15:45) when the post-lunch dip appears strong. As for in spring the lectures were held at times (9:15-10:45 and 12:15-13:45) close to the noon before the real dip-time. The earlier timing of the afternoon lecture in spring can well explain why no effect of the Tcp on the changes in alertness was seen in the afternoon lecture in spring; Tcp played a role only during the real post-lunch dip time. Due to the post-lunch dip it was expected that the alertness would decrease more in the late afternoon lecture than in the morning in autumn. That applied however, only to the lecture with the 4,000 K lighting condition. When 17,000 K light source was used the dip did not occur. Instead, using 17,000 K as the colour temperature of the light source helped the students to better retain their alertness in the afternoon lecture. These results are in line with the claim that cool light stimulates and activates the body 21). They indicate that might be possible to control the alertness with the colour temperature of the light. With that they support the scenario of Van Bommel to use dynamic lighting that allows changing the colour of the light at different moments of the day 11). It has been suggested that the phenomenon of postlunch dip can occur even when the individual has not had lunch 22). However, in this study there was found a significant correlation between those who had had lunch prior to the lecture and the way Tcp affected alertness. In fact, eating was found to promote tiredness in afternoon hours decreasing the alerting effect of the light. Also stimulants were found to mask light s effect. It has been proven that e.g. caffeine (the active substance of coffee, tea, energy drinks and many soft drinks) reduces physical fatigue and restores mental alertness 22). In this study it was seen that those subjects who had taken stimulants were less likely to respond to the Tcp with changes in alertness. However, the questionnaire did not provide information about normal caffeine consumption of the subject. Therefore it was not known, whether the subject was e.g. deeply addicted to caffeine and needed a certain amount to gain the stimulating effects, or whether he had perhaps, missed his daily portion and was suffering from its absence. 5. Conclusions The present study broadly investigated how white light of two different correlated colour temperatures (4,000 K and 17,000 K) affects students alertness in a real lecture environment. In the autumn study the correlated colour temperature (Tcp) of the light condition during the lecture correlated with changes in alertness in afternoon hours. By exposing the student to cool white light of 17,000 K, afternoon tiredness caused by diurnal hormonal cycles could be reduced. In spring no correlation between the Tcp of light and changes in alertness was detected. In fact, in the spring study, the students remained better alert during all lectures, supposedly due to the more intense exposure to outdoor daylight outside the lecture hours. Eating was found to promote the post-lunch dip even in the presence of the cool white light. Because the link between eating and the dip in alertness has not been fully established, this should, however, be further investigated. Caffeine induced same kinds of effects as cool white light, indicating that the alerting effect of the 17,000 K light applies only with the stimulant-free subjects. Lecture content and environmental factors, on the other hand, did not affect the relation of Tcp and changes in alertness. The study provides us information about the effects of exposure to blue-enriched white light during autumn afternoon hours in a lecture setting. This information is valuable because it shows that there might not be need to use special light treatment to improve alertness but that positive effects could also be achieved by using The Illuminating Engineering Institute of Japan 12

9 J. Light & Vis. Env. Vol.34, No.2, general light of certain Tcp at certain time of day and year. The results of the study support the idea of dynamic lighting in terms of varying the colour of light from morning to afternoon following the hormonal activity. It is essential to note that this study investigated student groups reactions to different lighting conditions. If the same was done to individuals, more careful analysis of subjects chronotypes and daily rhythms and routines would be needed. It is clear that the richness of the study lies in the real life test setting. However, the findings remind us about the importance of studying the variables that might have a masking effect on the relation of light and alertness. This is important especially in field studies where only a few parameters can be controlled. In future studies that investigate the relation of light and daytime alertness, the seasonal effects should be further considered. Besides determining the amount of daylight that the subjects are exposed to prior to the experiments, the season itself can be a strong psychological factor. The difficulties in the interpretation of the results indicate that subjective evaluation should not be the only measure to analyze alertness. Instead, research methods that objectively assess alertness (i.e. biological markers and performance tests) need to be further developed for studying alertness objectively in real life settings. In studies with prolonged exposure to light of high intensity, it might also become essential to evaluate and to control the possible side effects, such as macula alteration and eye-strain 23). Acknowledgements The work was carried out by the former Lighting Unit of Helsinki University of Technology, today known as Lighting Unit of School of Science and Technology, Aalto University, Finland. The authors acknowledge the Finnish Funding Agency for Technology and Innovation (Tekes), Philips Oy Luminaires, Helvar Oy, Senate Properties and the Finnish Work Environment Fund for their support. References (1) Cajochen, C., Münch, M., Kobialka, S., Kräuchi, K., Steiner, R., Oelhafen, P., Orgül, S. and Wirz-Justice, A.: High sensitivity of human melatonin, alertness, thermoregulation, and heart rate to short wavelength light. J Clin Endocrinol Metab, Vol.90, No.3, pp (2005). (2) Revell, V.L., Arendt, J., Fogg, L.F and Skene, D.J.: Alerting effects of light are sensitive to very short wavelengths. Neurosci Lett, Vol.399, No.1-2, pp (2006). (3) Panda, S., Nayak, S.K., Campo, B., Walker, J.R., Hogenesch, J.B. and Jegla, T.: Illumination of the melanopsin signaling pathway. Science, Vol.307, No.5709, pp (2005). (4) Qiu, X., Kumbalasiri, T., Carison, S.M., Wong, K.Y., Krishna, V., Provencio, I. and Berson, D.M.: Induction of photosensitivity by heterologous expression of melanopsin. Nature, No.433, pp (2005). (5) Gooley, J.J., Lu, J., Fischer, D. and Saper, C.B.: A broad role for melanopsin in nonvisual photoreception. J Neurosci, No.23, pp (2003). (6) Lehr, S., Gerstmeyer, K., Jacob, J.H., Frieling, H., Henkel, A.W., Meyrer, R., Wiltfang, J., Kornhuber, J. and Bleich, S.: Blue light improves cognitive performance. J Neural Transm, Vol. 114, No.4, pp (2007). (7) Mills, P.R., Tomkins, S.C. and Schlangen, J.M.: The effect of high correlated colour temperature office lighting on employee wellbeing and work performance. J Circadian Rhythms, Vol.5, No.2 (2007). (8) Viola, A.U., James, L.M., Schlangen, L.J.M. and Dijk, D.-J.: Blue-enriched white light in the workplace improves self-reported alertness, performance and sleep quality. Scand J Work Environ Health, Vol.34, No.4, pp (2008). (9) Nehlig, A., Daval, J.L. and Debry, G.: Caffeine and the central nervous system: Mechanisms of action, biochemical, metabolic, and psychostimulant effects. Brain Res Rev, Vol. 17, No.2, pp (1992). (10)Figueiro, M.G., Rea, M.S., Stevens, R.G. and Rea, A.C.: Daylight and productivity: a possible link to circadian regulation. 5th international LRO Lighting Research Symposium Light and Human Health, 3-5 Nov Palo Alto, CA. The Lighting Research Office of the Electric Research Institute, pp (2002). (11) Van Bommel, W.: Dynamic Lighting at work both in level and colour. 2nd CIE Expert Symposium on Lighting and Heath, 7-8 Sep Ottawa, Canada. CIE, pp (2002). (12)Horne, J.A. and Östberg, O.: A Self Assessment Questionnaire to Determine Morningness- Eveningness in Human Circadian Rhythms. International J Chronobiol, Vol.4, No.2, pp (1976). (13)Paine, S.-J., Gander, P.H. and Travier, N.: The Epidemiology of Morningness/Eveningness: Influence of Age, Gender, Ethnicity, and Socioeconomic Factors in Adults (30-49 Years). J Biol Rhythms, Vol.21, No.1, pp (2006). (14)Åkerstedt, T. and Gillberg, M.: Subjective and objective sleepiness in the active individual. Int J Neurosci, Vol.52, No.1-2, pp (1990). (15)NIOSH Publication No Criteria for a recommended standard: occupational exposure to car- 13 The Illuminating Engineering Institute of Japan

10 68 J. Light & Vis. Env. Vol.34, No.2, 2010 bon dioxide. National Institute for Occupational Safety and Health. p.179 (1976). (16)Tonello, G.: Seasonal affective disorder: Lighting research and Environmental psychology. Lighting Res. Tech., Vol.40, pp (2008). (17)Wirz-Justice, A., Benedetti, F. and Terman, M.: Individual Chronotherapeutic Elements: Light, Wake Therapy and Sleep Phase Advance. Chronotherapeutics for Affective Disorders, Basel, Switzerland, pp (2009). (18)Ogle, T.F. and Costoff, A.: Endocrinology. In book: Essentials of Human Physiology, ed. by Nosek, T.M. Medical College of Georgia, Augusta, GA (1999). (19)Foster, L. and Kreitzman, L.: Rhythms of life: The biological clocks that control the daily lives of every living thing. Profile Books Ltd., London, U.K. (2004). (20)Hedge, A.: Biological Rhythms: DEA 325. Cornell University, New York, NY (1999). (21)Fleischer, S., Krueger, H. and Schierz, C.: Effects of brightness distribution and light colours on office staff. Results of the Lighting Harmony project. 9th European Lighting Conference Lux Europa 2001, June Reykjavik, Island. National Lighting Societies, Institutes or Committees of the European Countries, pp (22)Monk, T.: The Post-Lunch Dip in Performance. Clinics in Sports Medicine, Vol.24, No.2, pp (2005). (23)Kogan, A.O. and Guilford, P.M.: Side Effects of Short-Term 10,000-Lux Light Therapy. Am J Psychiatry, Vol.155, No.2, pp (1998). *Figure 1 and 2 appear in color on J-STAGE; The Illuminating Engineering Institute of Japan 14