doi:10.1152/japplphysiol.00165.2011
110:1432-1438, 2011. First published 17 March 2011;J Appl Physiol
Achim Pross, Ralph Mager, Anna Wirz-Justice and Oliver Stefani
Christian Cajochen, Sylvia Frey, Doreen Anders, Jakub Späti, Matthias Bues,
physiology and cognitive performance
(LED)-backlit computer screen affects circadian
Evening exposure to a light-emitting diodes
You might find this additional info useful...
31 articles, 10 of which can be accessed free at:This article cites
http://jap.physiology.org/content/110/5/1432.full.html#ref-list-1
including high resolution figures, can be found at:Updated information and services
http://jap.physiology.org/content/110/5/1432.full.html
can be found at:Journal of Applied Physiologyabout Additional material and information
http://www.the-aps.org/publications/jappl
This infomation is current as of September 12, 2011.
ISSN: 0363-6143, ESSN: 1522-1563. Visit our website at http://www.the-aps.org/.
Physiological Society, 9650 Rockville Pike, Bethesda MD 20814-3991. Copyright © 2011 by the American Physiological Society.
those papers emphasizing adaptive and integrative mechanisms. It is published 12 times a year (monthly) by the American
publishes original papers that deal with diverse areas of research in applied physiology, especiallyJournal of Applied Physiology
on September 12, 2011jap.physiology.orgDownloaded from
Evening exposure to a light-emitting diodes (LED)-backlit computer screen
affects circadian physiology and cognitive performance
Christian Cajochen,
1
Sylvia Frey,
1
Doreen Anders,
1
Jakub Späti,
1
Matthias Bues,
2
Achim Pross,
2
Ralph Mager,
3
Anna Wirz-Justice,
1
and Oliver Stefani
2
Centres for
1
Chronobiology and
3
Applied Technologies in Neuroscience, Psychiatric Hospitals of the University of Basel,
Basel, Switzerland; and
2
Competence Team Visual Technologies, Fraunhofer IAO/University Stuttgart IAT,
Stuttgart, Germany
Submitted 7 February 2011; accepted in final form 14 March 2011
Cajochen C, Frey S, Anders D, Späti J, Bues M, Pross A, Mager R,
Wirz-Justice A, Stefani O. Evening exposure to a light-emitting diodes
(LED)-backlit computer screen affects circadian physiology and cognitive
performance. J Appl Physiol 110: 1432–1438, 2011. First published March
17, 2011; doi:10.1152/japplphysiol.00165.2011.—Many people spend an
increasing amount of time in front of computer screens equipped with
light-emitting diodes (LED) with a short wavelength (blue range).
Thus we investigated the repercussions on melatonin (a marker of the
circadian clock), alertness, and cognitive performance levels in 13
young male volunteers under controlled laboratory conditions in a
balanced crossover design. A 5-h evening exposure to a white LED-
backlit screen with more than twice as much 464 nm light emission
{irradiance of 0,241 Watt/(steradian ⫻ m
2
) [W/(sr ⫻ m
2
)], 2.1 ⫻ 10
13
photons/(cm
2
⫻ s), in the wavelength range of 454 and 474 nm} than
a white non-LED-backlit screen [irradiance of 0,099 W/(sr ⫻ m
2
),
0.7 ⫻ 10
13
photons/(cm
2
⫻ s), in the wavelength range of 454 and
474 nm] elicited a significant suppression of the evening rise in
endogenous melatonin and subjective as well as objective sleepiness,
as indexed by a reduced incidence of slow eye movements and EEG
low-frequency activity (1–7 Hz) in frontal brain regions. Concomi-
tantly, sustained attention, as determined by the GO/NOGO task;
working memory/attention, as assessed by “explicit timing”; and
declarative memory performance in a word-learning paradigm were
significantly enhanced in the LED-backlit screen compared with the
non-LED condition. Screen quality and visual comfort were rated the
same in both screen conditions, whereas the non-LED screen tended
to be considered brighter. Our data indicate that the spectral profile of
light emitted by computer screens impacts on circadian physiology,
alertness, and cognitive performance levels. The challenge will be to
design a computer screen with a spectral profile that can be individ-
ually programmed to add timed, essential light information to the
circadian system in humans.
nonvisual effects of light; spectral analysis; shift work; melatonin;
alertness
THE WORLD IS ONLINE. Over 2 billion people use the internet, and
this number is rapidly increasing. In 2010, 1.6 billion comput-
ers, television sets, and cellular phones were sold globally
(www.worldometers.info), which illustrates the numbers of
individuals who spend time in front of computer screens, video
game consoles, or other video monitors. Newer computers and
TV screens are now frequently equipped with light-emitting
diodes (LED), which peak in the short-wavelength region (i.e.,
the blue range at ⬃460 nm). There is ample evidence that a
novel, short-wavelength-sensitive photoreceptor system is pri-
marily responsible for a variety of nonvisual light responses, in
particular, resetting the timing of the circadian pacemaker,
suppressing melatonin production, improving alertness and
performance, and elevating brain activation, as assessed from
EEG-derived correlates of arousal (5, 6, 8, 17, 18, 24, 28, 31).
Furthermore, bright light exposure and exposure to monochro-
matic blue light in the evening lengthens sleep latency and
reduces initial EEG delta activity, a marker of slow-wave sleep
(7, 20). Thus the frequent use of LED sources could have
ramifications on human behavior, since light is the most
important synchronizer of our biological clock. The circadian
pacemaker responds differentially to the resetting effects of
light, depending on the circadian phase of light exposure.
Phase delays occur when light exposure is centered prior to the
core body temperature minimum, whereas circadian-phase ad-
vances can be elicited by light exposures centered after the core
body temperature minimum, which normally occurs in the
second half of the biological night (14). This means that
exposure to artificial light in the evening, when our circadian
timing system is most vulnerable to light, has the capacity to
modify rhythms and thus sleep and neurobehavioral function.
While acute light exposure in the evening may, for instance,
help night workers to become more alert and perform better,
the repercussions of chronic, inappropriate-timed exposure
could lead to circadian misalignment and thus eventually to
sleep problems (23), depression (19), and even the cardiovas-
cular diseases seen in shift workers (27).
Here, we investigated the impact of a LED-backlit computer
screen (enhanced in the short-wavelength region, i.e., 460 nm)
in comparison with a LED-free computer screen on a wide
range of measures in human physiology and behavior, such as
melatonin levels, cognitive performance, and the EEG during
wakefulness. Our main prediction was that a 5-h evening
exposure to a LED-backlit computer screen, in comparison
with a non-LED computer screen, would suppress the evening
increase in melatonin levels and evoke an alerting response
with concomitant improvement in cognitive performance.
METHODS
Healthy, young male volunteers (19 –35 years) were recruited via
advertisements at the University of Basel (Switzerland). Potential study
participants filled out questionnaires about their general health, sleep
quality [Pittsburgh sleep quality index (PSQI)], and sleep-wake behavior
[Munich chronotype questionnaire (MCTQ) (34)]. Volunteers with good
sleep quality (PSQI score ⬍5), no extreme chronotype (⬎3 and ⬍6
points on the MCTQ questionnaire), and good general health underwent
a medical examination carried out by the physician in charge and an
ophthalmologic examination by a certified optometrist to exclude volun-
teers with visual impairments, such as color blindness, diminished pupil
Address for reprint requests and other correspondence: C. Cajochen,
Centre for Chronobiology, Psychiatric Hospital of the University of Basel,
Wilhelm Kleinstr. 27, CH-4012 Basel, Switzerland (e-mail: christian.
cajochen@upkbs.ch).
J Appl Physiol 110: 1432–1438, 2011.
First published March 17, 2011; doi:10.1152/japplphysiol.00165.2011.
8750-7587/11 Copyright
©
2011 the American Physiological Society http://www.jap.org1432
on September 12, 2011jap.physiology.orgDownloaded from
reaction to light, and a reduced visual field. Participants were not
excluded if they wore glasses or contact lenses. Exclusion criteria were
smoking, medication or drug consumption, shift work within the last 3
mo, and transmeridian flights up to 3 mo prior to the study. Thirteen
volunteers (mean age: 23.8 years ⫾ 5.0 SD; mean body mass index:
22.6 ⫾ 1.7 SD) were then selected for the study. All subjects gave
written, informed consent. The study protocol, screening questionnaires,
and consent form were approved by the local ethics committee and
conformed to the Declaration of Helsinki.
During the entire study protocol, which comprised a total of 2 wk,
participants were instructed to keep a regular sleep-wake schedule
(bed times and wake times within ⫾30 min of self-selected target
time). Compliance was verified by sleep logs and ambulatory activity
measurements (Actiwatch-L, Cambridge Neurotechnology, Cam-
bridge, UK). The “in laboratory” part of the study was carried out in
Switzerland between the end of September and beginning of Novem-
ber. In a 25-m
2
room, two cubicles were installed in such a way that
they were completely light shielded, and only the light emitted by the
computer screen fell onto the volunteers’ eyes at a distance of ⬃60
cm. Two different computer screens were compared: a LED-illumi-
nated liquid crystal display screen (HP LP2480zx) and a cold cathode
fluorescent lamp (CCFL)-illuminated screen (HP LP2475w), both
with a screen diagonal of 24 in. and a resolution of 1920 ⫻ 1200
pixels adjusted to the identical luminance of 250 nits (nits as 1 cd/m
2
).
Spectral measurements were carried out using a Konica Minolta
CS-1000 (Konica Minolta Sensing, Osaka, Japan). Both computer
screens were set to a white background with a color temperature of
6,953 K for the LED-illuminated and 4,775 K for the CCFL-illumi-
nated screen, thus reducing the amount of blue light from one-half to
approximately one-third in the LED compared with the non-LED,
CCFL-illuminated screen. The irradiance between 400 nm and 480
nm of the LED-illuminated computer screen was 0,241 Watt/(stera-
dian ⫻ m
2
) [W/(sr ⫻ m
2
)] and 0,099 W/(sr ⫻ m
2
) for the non-LED,
CCFL-illuminated computer screen (Fig. 1). Although the difference
Fig. 1. Top, left: photograph of the non-light-emitting diodes (non-LED) computer screen [HP LP2475w cold cathode fluorescent lamp (CCFL)]; right:
photograph of the LED computer screen (HP LP2480zx LED). Lower: spectral composition {light wavelength by irradiance ⫽ Watt/(steradian ⫻ m
2
⫻ nm)
[W/(sr ⫻ m
2
⫻ nm)]} of light emitted from the LED computer screen (blue line) and the non-LED screen (red line). Inset: blow-up of the spectral composition
in the wavelength range of 410 –500 nm. The photon flux for the LED-backlit screen was 2.1 ⫻ 10
13
photons/(cm
2
⫻ s) in the wavelength range of 454 and
474 nm and 0.7 ⫻ 10
13
photons/(cm
2
⫻ s) in the wavelength range of 454 and 474 nm for the non-LED-backlit screen.
1433COMPUTER SCREEN AND CIRCADIAN AND COGNITIVE PHYSIOLOGY
J Appl Physiol • VOL 110 • MAY 2011 • www.jap.org
on September 12, 2011jap.physiology.orgDownloaded from
in color temperature was visible, the study volunteers did not notice
this difference after 1 wk when they changed to the other computer
screen, because the two displays were arranged in such a way that the
participants could only view one (their “own”) monitor at a time.
During the entire study protocol, the study volunteers were in a seated
position in front of the computer screen with an ambient temperature
of 22°C, air humidity of 60%, and ambient lighting conditions ⬍4 lux.
The volunteers reported 6 h (on average, at ⬃17.30 h) prior to usual
bed time, which was on average, 23.35 h ⫾ 22 min, to the Chrono-
biology Laboratory of the Psychiatric Hospitals of the University of
Basel, where they were equipped with electrodes and sensors for the
physiological recordings. Afterward, volunteers were trained on the
different cognitive tasks and were acquainted with the study room.
Four and one-half hours prior to usual bed time (on average, at 19:00
h), volunteers were dark adapted for 30 min and thus sat in a very
dim-light (⬍4 lux, red light) environment. After dark adaptation (on
average, at ⬃20:00 h), they were asked to sit in front of their computer
screen in their cubicles and to start the 5-h screen exposure episode.
During these 5 h, the study participants were asked to complete the
following tasks: in half-hourly intervals, saliva collection and the
Karolinska Sleepiness Scale (12); and in hourly intervals, the Karo-
linska Drowsiness Test (KDT) (1). Every hour before and after the
relaxing movie (see below in this paragraph), the GO/NOGO task (3),
time estimation task (30), word-pair learning task (26), and visual
comfort and effort scale (4) were completed. Every 50 min, the
volunteers were asked to take a short break for 10 min under dim-light
red conditions in the same room. Furthermore, after the first2hof
sitting in the cubicle, a relaxing, 20-min movie was displayed on the
computer screen, which contained scenes with snowy environments
(i.e., white light). The volunteers were instructed to watch the movie
at a distance of ⬃1 m to ensure constant exposure to the computer
screen light without other ongoing activities (which accentuates
light’s effects on alertness and attention). One hour after the usual bed
time (on average, at 00:30 h), the 5-h laboratory protocol ended, and
the volunteers were allowed to go home. One week later, the entire
study procedure was repeated with the other computer screen type.
The order of the computer screens was balanced and crossed over to
avoid potential sequence effects.
Saliva collections were scheduled every 30 min. A direct double-
antibody radioimmunoassay was used for the melatonin assay (vali-
dated by GC-MS with an analytical least detectable dose of 0.65
pg/ml; Bühlmann Laboratory, Schönenbuch, Switzerland) (32). The
minimum detectable dose of melatonin (analytical sensitivity) was
determined to be 0.2 pg/ml.
To objectively quantify sleepiness, 3-min KDT (1) artifact-free
EEG samples were recorded, once during dim light and hourly during
the5hoflight exposure. The Visual Comfort Scale (4), a 100-mm
visual analogue scale, comprises: 1) screen quality (to read, see
patterns, and optical reflection); 2) visual well-being and comfort; and
3) glare and brightness. Glare and brightness are probed as, respec-
tively, “Does the light have less glare or more?” and “Is the light too
dark or too bright?” More glare and brightness are conceived as
helping to visualize patterns and/or to read, although high levels of
glare and brightness can point to potentially less comfortable light
perception in a given environmental light setting (9).
The GO/NOGO task (3) was used to measure the capacity for
sustained attention and response control. Participants had to press the
space bar within 0.5 s if the letter “M” were shown on the screen. If
the letter “W” were shown, participants were instructed not to press
any buttons. A total of 80% of M letters were shown in a quasi-
random sequence. Approximately 200 M letters were shown during 8
min.
Interval timing was sampled via the concurrent use of two standard
methods of timing research, temporal production, and temporal re-
production. For duration estimations, production target durations were
displayed in conventional units (number of seconds to be produced)
centrally on a computer display using black Arabic digits on a gray
background. The participant’s task was to identify the target duration
and immediately begin holding down the space bar on the computer
keyboard, stopping to depress the space bar after a duration that
subjectively matched the defined target duration. Reproduction target
durations were given via a “carrier stimulus,” i.e., via temporally
delimited display of a black square on gray background centrally on
a computer display. Participants were instructed to hold down the
space bar on the computer keyboard as soon as possible upon the
extinction of the target stimulus and to release the space bar after a
duration, subjectively corresponding to the target duration, had
elapsed. Interval timing sessions consisted of either 15 (production;
three target durations, each presented five times in random order) or
25 (reproduction; five target durations, each presented five times in
random order) (30).
Declarative memory performance was tested via a word-pair learn-
ing task, which consisted of 60 word pairs of semantically unrelated
words. For each of the four test sessions, a new set of 120 words or
60 word pairs, respectively, was used. To allow the creation of
multiple word-pair lists with different words but similar psycholin-
guistic properties, the software EQUIWORD (16) was used. Each pair
of words was displayed on the screen for 6 s, followed by a white-
centered fixation cross for 5 s, during which subjects were instructed
to visually imagine a relationship between the two words of the pair
in the aim to render mnemonic strategies more comparable across
volunteers (11, 26). Immediately after the end of the encoding session,
the recall of the word pairs was conducted. Thereby, 50% of the
previously learned word pairs (⫽30 word pairs) were shown again,
although in a different order, and the remaining 60 words were newly
arranged to 30 word pairs. Hence, similar to the encoding session, the
recall session comprised 60 word pairs, but 30 of them were newly
arranged. For each word pair, the volunteers were asked to answer in
the following manner: 1) it was a known (old) word pair (100% sure),
2) it was never displayed before (new; 100% sure), or 3) it is likely but
not 100% sure to be a known (old) word pair. The assessment of
declarative memory performance was based on the percentage of
correctly remembered “old” word pairs and correctly identified “new”
word pairs.
EEGs were calculated offline from a continuous, 6-referential EEG
recording. All signals were online digitized (16 bit analog-to-digital
converter, 0.021 V/bit; storage sampling rate at 512 Hz Varioport
digital recorder, Becker Meditec, Karlsruhe Germany). The raw sig-
nals were stored online on a memory card (SanDisk, Milpitas, CA)
and downloaded offline to a personal computer hard drive. EEG data
collected during the 3-min KDT were scored for artifacts and sub-
jected to a Fast Fourier Transform (FFT) routine (Vitaport paperless
sleep-scoring software). Two-second epochs were offline subjected to
spectral analysis using a FFT (10% cosine window), resulting in a
0.5-Hz bin resolution. For data reduction, artifact-free, 2-s epochs
were averaged over 20-s epochs. Next, the 20-s epochs were further
reduced by averaging them over each 3-min KDT. EEG power spectra
during each 3-min KDT were calculated for the derivations Fz, FCz,
Cz, CPz, Pz, and Oz in the range of 0.5 to 25 Hz. The electrodes for
the electrooculogram (EOG) were placed at the outer canthi of each
eye, one slightly above the canthomeatal plane and the other slightly
below. All EOG recordings were inspected visually, and slow eye
movements (SEMs) were scored in 20-s epochs. Other eye move-
ments (i.e., saccadic and mixed patterns) were not considered for
analysis. Each 20-s epoch during the study protocol was scored as to
whether at least one SEM occurred, and the presence of more than one
SEM in an epoch did not influence the scoring criteria. SEMs were
scored regardless of their amplitude, but SEMs that occurred during
body movements were not included in the analysis.
For all analyses, the statistical package SAS (Version 9.1, SAS
Institute, Cary, NC) was used. Statistical analyses were carried out for
each variable (subjective sleepiness, GO/NOGO, declarative memory,
time estimation, wake-EEG activity, and salivary melatonin) with a
repeated measure ANOVA (rANOVA) using a general linear model.
1434 COMPUTER SCREEN AND CIRCADIAN AND COGNITIVE PHYSIOLOGY
J Appl Physiol • VOL 110 • MAY 2011 • www.jap.org
on September 12, 2011jap.physiology.orgDownloaded from
Factors in this model included “screen type” (LED vs. non-LED-
backlit computer screen), “time of day,” and for the wake-EEG
activity, it included factor “derivation” (frontal, central, parietal, and
occipital derivations).
P values were based on corrected degrees of freedom, but the
original degrees of freedom are reported. Post hoc comparisons were
performed using two-sided Duncan’s multiple range tests or paired
t-tests. Since salivary melatonin and subjective sleepiness were also
collected during baseline and dark adaptation, these data were in-
cluded in the analyses. For all of the cognitive tasks, data from the 5-h
computer light exposure were included in the analysis. For the
analysis of visual comfort, the five time points when it was carried out
were averaged to provide a global comparison between the two light
settings.
RESULTS
Salivary melatonin levels followed during baseline, dark
adaptation, and a 5-h screen exposure episode yielded a sig-
nificant effect for screen (F
1,11
⫽ 5.9; P ⫽ 0.045), time of day
(F
12,132
⫽ 137.5; P ⬍ 0.0001), and the interaction screen
versus time of day (F
12,132
⫽ 3.0; P ⫽ 0.041; Fig. 2,
left). The
evening increase in endogenous melatonin levels was sup-
pressed and rose later under exposure to the LED screen
compared with the non-LED screen, significant at the follow-
ing time points: 21:15 h, 22:15 h, 22:45 h, and 23:15 h (post
hoc comparisons;Pat least ⬍0.04). Subjective sleepiness
ratings taken at the same time intervals as for the salivary
melatonin assessments yielded a significant effect of time of
day (F
12,132
⫽ 25.9; P ⬍ 0.0001; Fig. 2, right) but no
significant effect for screen or for the interaction screen versus
time of day. However, a separate analysis of subjective sleep-
iness confined to the period when the participants were asked
to take a break and watch the movie (see
METHODS) revealed
significantly lower sleepiness levels when the movie was
displayed on the LED screen compared with the non-LED
screen (Fig. 2, inset, right, P ⬍ 0.04). Analysis of the incidence
of SEMs, an objective marker for sleepiness derived from the
EOGs, revealed significant differences for main factors
“screen” and “night,” although the interaction was not signif-
icant (screen: F
1,11
⫽ 26.2; P ⬍ 0.0004; time of day: F
11,44
⫽
7.8; P ⬍ 0.0001; screen vs. time of day: not significant; Fig. 3,
left). A two-way rANOVA for spectral EEG power density
during the KDTs revealed a significant interaction between the
factors screen and EEG derivation in the frequency bins rang-
ingfrom1to7Hz(P at least 0.05). Thus EEG power density
in these frequency bins were collapsed into a frequency band of
1–7 Hz and further analyzed with a three-way rANOVA,
which yielded a significant factor for screen (F
1,20
⫽ 6.7; P ⬍
0.02) and derivation (F
5,50
⫽ 124.2; P ⬍ 0.0001), a significant
interaction screen versus EEG derivation (F
5,50
⫽ 2.6; P ⬍
0.05), and a significant interaction EEG derivation versus time
of day (F
20,200
⫽ 2.9; P ⬍ 0.02). Accordingly, exposure to the
LED screen resulted in an attenuation of frontal EEG activity
in the range 1–7 Hz (Fig. 3, right), which was not observed in
other derivations.
Fig. 2. Time course of salivary melatonin (left) and
subjective sleepiness levels (right) during baseline,
dark adaptation, and the screen exposure episode (30
min; mean values ⫾ SE; n ⫽ 13). Inset, right: Karo-
linska Sleepiness Scale (KSS) levels during the pre-
sentation of the movie from 21:45–22:15 h. Results of
the LED computer screen condition (); data of the
non-LED computer screen condition (Œ). *Significant
post hoc comparisons when the interaction screen ⫻
time of day yielded significance.
Fig. 3. Time course of the incidence of slow
rolling eye movements derived from the elec-
trooculogram (left) and frontal low-frequency
EEG activity in the range of 1–7 Hz (right)
during dark adaptation and the screen expo-
sure episode (20:00 – 00:15 h; mean values ⫾
SE; n ⫽ 13). Results of the LED computer
screen condition (); data of the non-LED
computer screen condition (Œ).
1435COMPUTER SCREEN AND CIRCADIAN AND COGNITIVE PHYSIOLOGY
J Appl Physiol • VOL 110 • MAY 2011 • www.jap.org
on September 12, 2011jap.physiology.orgDownloaded from
![Fig. 1. Top, left: photograph of the non-light-emitting diodes (non-LED) computer screen [HP LP2475w cold cathode fluorescent lamp (CCFL)]; right: photograph of the LED computer screen (HP LP2480zx LED). Lower: spectral composition {light wavelength by irradiance Watt/(steradian m2 nm) [W/(sr m2 nm)]} of light emitted from the LED computer screen (blue line) and the non-LED screen (red line). Inset: blow-up of the spectral composition in the wavelength range of 410–500 nm. The photon flux for the LED-backlit screen was 2.1 1013 photons/(cm2 s) in the wavelength range of 454 and 474 nm and 0.7 1013 photons/(cm2 s) in the wavelength range of 454 and 474 nm for the non-LED-backlit screen.](/figures/fig-1-top-left-photograph-of-the-non-light-emitting-diodes-tfp6s9a7.webp)
