ARCHIVAL REPORT

Three-Week Bright-Light Intervention Has Dose-

Related Effects on Threat-Related Corticolimbic
Reactivity and Functional Coupling
Patrick M. Fisher, Martin K. Madsen, Brenda Mc Mahon, Klaus K. Holst, Sofie B. Andersen,
Helle R. Laursen, Lis F. Hasholt, Hartwig R. Siebner, and Gitte M. Knudsen
Background: Bright-light intervention is reported to successfully treat depression, in particular seasonal affective disorder, but the
neural pathways and molecular mechanisms mediating its effects are unclear. An amygdala-prefrontal cortex corticolimbic circuit
regulates responses to salient environmental stimuli (e.g., threat) and may underlie these effects. Serotonin signaling modulates this
circuit and is implicated in the pathophysiology of seasonal and other affective disorders.

Methods: We evaluated the effects of a bright-light intervention protocol on threat-related corticolimbic reactivity and functional
coupling, assessed with an emotional faces functional magnetic resonance imaging paradigm at preintervention and postintervention. In
a double-blind study conducted in the winter, 30 healthy male subjects received bright-light intervention (dose range between
participants: .1–11.0 kilolux) for 30 minutes daily over a period of 3 weeks. Additionally, we considered serotonin transporter-linked
polymorphic region (5-HTTLPR) genotype status as a model for differences in serotonin signaling and moderator of intervention
effects.
Results: Bright-light dose significantly negatively affected threat-related amygdala and prefrontal reactivity in a dose-dependent
manner. Conversely, amygdala-prefrontal and intraprefrontal functional coupling increased significantly in a dose-dependent manner.
Genotype status significantly moderated bright-light intervention effects on intraprefrontal functional coupling.

Conclusions: This is the first study to evaluate the effects of clinically relevant bright-light intervention on threat-related brain function.
We show that amygdala-prefrontal reactivity and communication are significantly affected by bright-light intervention, an effect partly
moderated by genotype. These novel findings support that this threat-related corticolimbic circuit is sensitive to light intervention and
may mediate the therapeutic effects of bright-light intervention.

generally critical for regulating mood and processing emotionally

Key Words: Amygdala, bright-light therapy, fMRI, 5-HTTLPR, salient information within the environment.
prefrontal cortex, serotonin Our understanding of the neurobiological mechanisms sensi-
tive to bright-light exposure is limited. A neuroimaging study in

E
xposure to bright light has been used to successfully treat
depression and other affective disorders (1,2). Most notably,
it is used to treat seasonal affective disorder (SAD), a
depressive disorder characterized by recurring, seasonally asso-
ciated depressive symptoms during fall and winter months, as
day length decreases, that abate with the spring and summer
months, as day length increases (3). Considering the overlap in
symptoms characterizing SAD and other depressive disorders
(e.g., anhedonia, increased negative mood, alterations in sleep
alterations, etc.) it is reasonable to hypothesize that these
disorders share common neurobiological underpinnings and that
the clinical effects of bright-light therapy on SAD and related
disorders may be through modulation of neural pathways
From the Center for Integrated Molecular Brain Imaging (PMF, MKM, BM,
KKH, SBA, HRL, HRS, GMK) and Neurobiology Research Unit (PMF,
MKM, BM, KKH, SBA, GMK), Copenhagen University Hospital Rigshos-
pitalet, Copenhagen O; Department of Biostatistics (KKH), University of
Copenhagen, Copenhagen K; Danish Research Centre for Magnetic
Resonance (HRL, HRS), Copenhagen University Hospital Hvidovre,
Hvidovre; and Department of Cellular and Molecular Medicine (LFH),
Copenhagen University, Copenhagen, Denmark.
Address correspondence to Gitte M. Knudsen, M.D., D.M.Sc., Copenhagen
University Hospital Rigshospitalet, Neurobiology Research Unit,
NRU 6931, Rigshospitalet, Blegdamsvej 9, Copenhagen O DK-2100,
Denmark; E-mail: gmk@nru.dk.
Received Jul 23, 2013; revised Nov 22, 2013; accepted Nov 30, 2013.
healthy individuals reported that 1-minute exposure periods to
blue light increased brain responses to auditory stimuli delivered
with angry, negative prosody and functional connectivity
between auditory cortex and the amygdala (4). Such studies of
intervention effects in healthy populations can provide key
insights into the neurobiological mechanisms responsive to
intervention and serve as a useful benchmark for evaluating
effects in clinical cohorts, but such studies are also limited by the
obvious lack of clinical symptoms. However, clinical use of bright-
light therapy typically involves exposure to full-spectrum white
light for 30 to 120 minutes daily over a period of weeks (5). Thus,
although these findings suggest that bright-light exposure affects
the neural processing of negative stimuli, they do not provide
insight into the effects of bright-light exposure as it is used
clinically.
A corticolimbic circuit comprising the amygdala and medial
prefrontal cortex (mPFC), including anterior cingulate cortex,
critically modulates reactivity to emotionally salient environ-
mental stimuli, particularly indices of threat (6,7). A role for this
circuit in the pathophysiology of affective disorders is supported
by numerous neuroimaging studies reporting differences in
threat-related corticolimbic reactivity and circuit function
between healthy and clinical cohorts, some of which normalize
following antidepressant treatment (8). A broad literature in
animal models and humans implicates serotonin signaling in
modulating threat-related corticolimbic circuit function (9,10). In
one of the first descriptions of SAD and bright-light therapy,

0006-3223/$36.00 BIOL PSYCHIATRY 2014;76:332–339

http://dx.doi.org/10.1016/j.biopsych.2013.11.031 & 2014 Society of Biological Psychiatry
P.M. Fisher et al. BIOL PSYCHIATRY 2014;76:332–339 333

Rosenthal et al. (3) identified serotonin signaling as a candidate in
its pathophysiology. Though not all studies have confirmed this,
genetic association, imaging genetics, and neuroreceptor posi-
tron emission tomography studies, including from our own lab,
have identified links between seasonal variation and serotonin
signaling (11,12). These findings support a model wherein the
serotonin system is sensitive to seasonal variation and may
modulate the effects of light on mood and affect, in part by
affecting threat-related corticolimbic circuit function. Despite this,
no studies have evaluated the effects of bright-light intervention
on reactivity of this circuit or how serotonin may moderate such
effects.
Within the current study, carried out during winter, we evaluated
the effects of bright-light intervention, at levels similar to those used
clinically, on threat-related corticolimbic circuit function within a
cohort of 30 healthy male subjects. We also considered interactive
effects of bright-light dose with the serotonin transporter-linked
polymorphic region (5-HTTLPR) polymorphism (including rs25531),
located in the promoter region of the gene (SLC6A4) coding for the
serotonin transporter (5-HTT), as an exploratory model for differ-
ences in serotonin signaling based on evidence that it interacts with
seasonal variation to affect 5-HTT levels (12–14). Participants were
randomly assigned to receive 30 minutes of bright-light intervention
(dose range: .1–11.0 kilolux [klx]) at home every morning for 3
weeks. Participants completed an emotional faces functional mag-
netic resonance imaging (fMRI) paradigm preintervention and post-
intervention, wherein they viewed individual angry, fearful, or
neutral faces (15). Based on previously reported effects of pharma-
cologic intervention studies using antidepressants (16,17), we
hypothesized that bright-light intervention would decrease threat-
related amygdala reactivity and increase amygdala-prefrontal func-
tional coupling. Additionally, we evaluated whether 5-HTTLPR status
moderated effects of bright-light intervention on corticolimbic circuit
reactivity and functional coupling.
Methods and Materials
Participants
Healthy male participants were recruited from the Copenhagen
region via online advertisements for a research protocol approved
by the Ethics Committee of Copenhagen and Frederiksberg, Den-
mark (H-1-2010-091; amendments: 28633,30043). Written informed
consent was acquired from all participants. Inclusion criteria
included: 1) 18 to 45 years old, 2) 18 to 30 body mass index, 3)
no history of alcohol/drug abuse, 4) no history of psychiatric/
neurological illness, 5) no excessive light exposure during study
participation and preceding autumn (e.g., sun tanning), 6) no retinal
pathology, 7) not taking photosensitizing medication, and 8)
Seasonality Pattern Assessment Questionnaire score ⬍ 11 (18).
Status for the 5-HTTLPR polymorphism (i.e., LA, LG, S alleles) within
the SLC6A4 gene was determined for the purpose of study inclusion.
A total of 70 participants were genotyped and underwent a medical
examination to verify meeting inclusion criteria. Of these, 16 LA/LA
and 16 non-LA/LA participants (i.e., 1–2 LG or S alleles) were included
in the study. Due to magnetic resonance imaging (MRI)-related
artifacts, two datasets were excluded from analysis. Thus, our final
sample size was 30 (Table 1). The study was conducted from
November to February, 2011 and 2012.
Genotyping
The triallelic 5-HTTLPR polymorphism, including rs25531 (i.e.,
LA, LG, and S alleles), was determined as previously described (12).
Bright-Light Intervention Protocol
Participants randomized to bright-light intervention received
such that 5-HTTLPR status was evenly represented across high
and low doses. From a standard distance of 50 cm, high-dose
lamps emitted 4.9 to 7.8 klx, whereas low-dose lamps emitted .6
to 1.9 klx. Participants and researchers were blinded to bright-
light dose received throughout the study. Bright-light lamps were
acquired from Smifa Healthcare (Solroed Strand, Denmark).
Bright-light lamps emitted full-spectrum white light with peaks
in the blue spectrum (Figure S1 in Supplement 1), at wavelengths
to which intrinsically photosensitive retinal ganglion cells, a key
regulator of circadian phase, are sensitive (19). There were no
identifiable differences between lamps. Participants received
written and oral instructions for light intervention. Participants
were instructed to sit 50 cm from the lamp for 30 minutes
between 7:00 AM and 9:00 AM every morning for 3 weeks.
Intervention criteria were based on recent SAD treatment guide-
lines (1). Bright-light intervention around sunrise is consistent
with an extension of the photoperiod and previous studies
reported that morning administration is particularly effective in
relieving SAD symptoms in clinical cohorts (20,21).

Table 1. Demographic, Bright-Light Intervention, and Task Performance

Baseline Rescan Bright-Light Effect

n 30 – –
5-HTTLPR (LA/LA vs. LG or S Carrier) 14/16 – –
Age (Years) 24.3 ⫾ 3.7 – –
Education (Years) 16.1 ⫾ 1.4 – –
Intervention Length (Days) 21.2 ⫾ 2.0 (19–28) – –
Bright Light Dose (Kilolux) 3.5 ⫾ 2.9 (.1–11.0) – –
Body Mass Index (kg/m2) 23.2 ⫾ 2.8 23.0 ⫾ 2.7 p ¼ .31
Major Depression Index 5.6 ⫾ 5.1 5.1 ⫾ 4.7 p ¼ .98
Cohen0 s Perceived Stress 9.4 ⫾ 5.1 9.2 ⫾ 5.1 p ¼ .69
POMS-TMD ⫹1.7 ⫾ 16.5 ⫹1.2 ⫾ 15.0 p ¼ .48
Pittsburgh Sleep Quality Index 3.6 ⫾ 1.6 3.3 ⫾ 2.0 p ¼ .99
Faces Task, Reaction Time (msec) 731.4 ⫾ 79.4 753.3 ⫾ 85.2 p ¼ .76
Faces Task, Performance (% Correct) 97.5 ⫾ 1.6 97.5 ⫾ 1.7 p ¼ .79
p values reflect an ANCOVA model with light dose predicting measure at rescan, covarying for measure at baseline. Values reflect mean ⫾ SD
(minimum - maximum). Rescan BMI data available for only 23 participants.
ANCOVA, analysis of covariance; BMI, body mass index; 5-HTTLPR, serotonin transporter-linked polymorphic region; POMS-TMD, Profile of Mood States-
Total Mood Disturbance.

www.sobp.org/journal
334 BIOL PSYCHIATRY 2014;76:332–339
The intervention period began following the baseline MRI scan
and continued for 3 weeks. To estimate light dose received, we
measured illuminance for each lamp with a luxmeter (Elma 1335;
Elma Instruments, Farum, Denmark) and measured lamp-to-eye
distance within the participant’s home environment, where light
intervention was administered (distance from lamp: mean ⫾ SD)
= 71.4 cm ⫾ 30.8; median, range = 58.5 cm, 36–170). The angle
between the person and lamp is relevant for estimating light
dose. Based on participant description, we estimated this angle as
01 or 451 (i.e., directly facing or angled, respectively) for each
participant. From this information, we estimated illuminance
received for each participant (mean ⫾ SD = 3.5 klx ⫾ 2.9;
median, range = 2.8 klx, .1–11.0, see Supplement 1).
To facilitate compliance, participants completed log books and
received daily mobile phone short message service reminders.
Participants reported average time outside during the daytime of
the intervention period (7, 16, and 7 participants reported, on
average, less than 30 minutes outside daily, 30 to 60 minutes, and
more than 60 minutes, respectively).
Personality Measures
To assess effects of bright-light intervention on affect, states
measures were acquired at baseline and rescan, including Major
Depression Inventory, Profile of Mood States, Cohen’s Perceived
Stress Scale, and Pittsburgh Sleep Quality Index (22–25).
Magnetic Resonance Imaging Data Acquisition
Participants were scanned on a 3T-Trio MRI using an eight-
channel head coil (Siemens, Erlangen, Germany) as described
previously (15). Blood oxygen level-dependent (BOLD) fMRI scans
were acquired using a T2*-weighted gradient-echo spiral echo-
planar imaging sequence (repetition time ¼ 2500 msec, echo time
¼ 26 msec, flip angle ¼ 761, in-plane matrix ¼ 64  64, in-plane
resolution ¼ 3  3 mm, number of slices within a whole-brain
volume ¼ 41, slice thickness ¼ 3 mm, gap ¼ .75 mm). Image
acquisition was optimized for signal recovery within orbital frontal
cortex by 1) tilting slice orientation from a transverse toward a
coronal orientation by approximately 301; and 2) use of a
preparation gradient pulse (26). Within each fMRI session, a total
of 312 whole-brain volumes (156 per run) were acquired. We
acquired a T1-weighted, spin-echo sequence, high-resolution
whole-brain three-dimensional structural magnetic resonance
scan (inversion time ¼ 800 msec, echo time ¼ 3.92 msec,
repetition time ¼ 1540 msec, flip angle ¼ 91, in-plane matrix ¼
256  256, in-plane resolution ¼ 1  1 mm, number of slices ¼
192, slice thickness ¼ 1 mm, no gap).
Emotional Faces Paradigm
Participants completed a gender discrimination task including
alternating blocks of individual neutral, fearful, and angry faces,
each presented in the middle of the screen (15). Faces were taken
from the Karolinska Directed Emotional Faces database (27). Each
trial consisted of the stimulus presented for 1800 milliseconds and
an interstimulus interval of 200 milliseconds during which time a
fixation cross (⫹) was present on the screen. Each block comprised
six trials, including three to five faces trials (mean face trials per
block ¼ 4) and one to three null trials (mean null trials per block ¼
2) consisting of a fixation cross (⫹). Faces and null trials were
randomly mixed within each block. In total, 32 blocks of neutral
faces were interleaved between 16 blocks of fearful and 16 blocks
of angry faces presented across two fMRI runs with a brief break
between runs. Neutral faces were presented twice throughout the
task, whereas fearful and angry faces were presented once. Stimulus
www.sobp.org/journal
P.M. Fisher et al.
presentations and response recordings were performed using E-
prime (Psychological Software Tools, Pittsburgh, Pennsylvania).
Functional Magnetic Resonance Imaging Data Analysis
Functional images were first realigned to a subject-specific
mean functional image. The T1-weighted structural image was co-
registered to the functional images and segmented using VBM5
(http://dbm.neuro.uni-jena.de/vbm/vbm5-for-spm5/). Functional
images were normalized into Montreal Neurological Institute space
using normalization parameters estimated during segmentation of
the T1-weighted image. Functional images were smoothed with an
8 mm Gaussian kernel to minimize noise and residual difference in
gyral anatomy. After preprocessing, smoothed functional images
were used within single-subject general linear models employing a
canonical hemodynamic response function to estimate condition-
specific and task-specific BOLD activation (i.e., beta and contrast
images, respectively), including motion parameters and respiratory
measures as covariates.
Single-subject contrast images (i.e., weighted sum of task-
related beta images) were used in second-level random effects
models to determine task-specific reactivity (i.e., fearful and angry
vs. neutral faces). To address the issue of multiple comparisons,
3dClustSim, a software program within AFNI (http://afni.nimh.nih.
gov/afni) that uses a Monte Carlo simulation method was used to
determine cluster extent thresholds for specific regions of interest
unlikely to have occurred by chance (α ⬍ .05). All regions of
interest were defined using the WFU Pickatlas software (http://
fmri.wfubmc.edu/software/PickAtlas), version 3.0.3 (28–30). The
cluster extent threshold required for our amygdala search volume,
given a voxel-level threshold of p ⬍ .01, uncorrected, was k ⬎ 14
voxels. Our mPFC region of interest was defined as Brodmann
areas 24, 25, and 32 (three-dimensional dilation = 1) so as to
include mPFC regions that share anatomical connectivity with the
amygdala (31–33). The cluster extent threshold required for this
search volume was k ⬎ 249 voxels.
Functional Coupling Analysis
We examined corticolimbic functional coupling by evaluating
the correlation in the BOLD time series with one of three seed
regions: left amygdala, right amygdala, and mPFC. Furthermore,
we evaluated the effects of bright-light intervention on functional
coupling and whether 5-HTTLPR moderated this effect. Although
coupling estimates may reflect functionally relevant correlations
between components of neural circuits, they do not establish
causality or directionality (34). We extracted the mean BOLD time
series for each subject from each of our three seeds using MarsBaR
(http://marsbar.sourceforge.net/) (35). We defined the seeds by a 5
mm radius sphere (seed volume: 648 mm3 or 81 voxels) centered
on the voxel maximally responsive to our task. Functional coupling
for each seed was evaluated separately. Extracted time series were
used in functional coupling analysis as has been previously
described (36). Time series were mean-centered and drift-cor-
rected, and individual values exceeding three SD of the mean of
this time series were replaced by the mean of the two adjacent
values. This time series was entered as a regressor in single-subject
design matrices that included as covariates task condition, motion
parameters, and respiratory variables described earlier. Analyses
yielded individual contrast images reflecting the temporal cou-
pling of BOLD signal changes with a given seed.
Data Analysis
We evaluated light dose effects as a continuous predictor using
linear regression, analysis of covariance (ANCOVA), and t tests.
P.M. Fisher et al.
Mean estimates of task-related fMRI reactivity were extracted from
SPM (http://www.fil.ion.ucl.ac.uk/spm/) and analyzed in R, 2.15.1
(http://www.r-project.org/). A threshold of p ⬍ .05 (two-tailed) was
considered statistically significant. We did not observe a significant
association between age or body mass index and task reactivity
estimates; thus, they were excluded from our models. Magnetic
resonance imaging scans were acquired at varied times through-
out the day (range: 7:00 AM to 6:00 PM). Magnetic resonance
imaging scan time was not predictive of fMRI estimates and thus
were excluded from our models (Supplement 1). We evaluated
effects of bright-light intervention using an ANCOVA model (37).
This approach modeled a given fMRI estimate at rescan as a linear
function of that measure at baseline, light dose, and other relevant
covariates (e.g., 5-HTTLPR status). Thus, the estimated association
between light dose and rescan is determined while accounting for
variation at baseline. This approach is advantageous in random-
ized test-retest designs because it accounts for interindividual
variability in baseline response when determining effects of
intervention. The same approach was applied at a voxel level for
evaluating effects of intervention on functional coupling. Voxel-
level functional coupling estimates at baseline and rescan were
extracted from SPM and we applied this ANCOVA model for each
voxel using R. Results were then written back into images that
were visualized using SPM.
Results
Bright-Light Intervention Decreases Threat-Related
Corticolimbic Reactivity
At baseline, we observed bilateral threat-related amygdala
reactivity (i.e., angry and fear vs. neutral faces) across all
participants (left amygdala: [30, 4, 18], z ¼ 4.77, k ¼ 47
voxels, p ⬍ .05, corrected; right amygdala: [30, 4, 18], z ¼ 3.71,
k ¼ 39 voxels, p ⬍ .05, corrected; Figure 1A). Bright-light dose was
statistically significantly negatively associated with threat-related
amygdala reactivity at rescan, controlling for reactivity at baseline
(bright-light effect: left amygdala [95% confidence interval (CI)]:
.26 [.46 to .062], p ¼ .01; right amygdala: .22 [.36 to
.083], p ¼ .003; Figure 1B,C).
At baseline, an mPFC cluster was statistically significantly less
reactive to angry and fear faces compared with neutral faces across
all participants (mPFC cluster: [10, 52, 0], z ¼ 4.38, k ¼ 1304
voxels, p ⬍ .05, corrected; Figure 2A). Bright-light dose was
Figure 1. Effects of bright-light intervention on amygdala reactivity. (A) Statistical parametric map highlighting amygdala response to task at baseline
across all participants (blood oxygen level-dependent contrast: angry and fear – neutral faces). Color bar represents t scores. Bright-light dose received was
negatively associated with (B) left and (C) right amygdala reactivity at rescan, accounting for baseline reactivity values. Thirty individual data points shown
in orange and blue shading represent 95% confidence limit of regression line.
BIOL PSYCHIATRY 2014;76:332–339 335
statistically significantly negatively associated with mPFC reactivity
at rescan, controlling for reactivity at baseline (bright-light effect:
mPFC [95% CI]: .21 [.36 to .058], p ¼ .009; Figure 2B). That is,
higher bright-light dose was associated with a lower mPFC response
to fear and angry faces compared with neutral faces at rescan.
Next, we evaluated whether effects of bright-light intervention
on amygdala and mPFC reactivity were moderated by 5-HTTLPR
genotype status (LA/LA vs. LG or S carriers). We did not observe a
significant genotype by light-dose interaction effect on reactivity
measures (interaction effect: left amygdala [95% CI]: .32 [.73 to
.10], p ¼ .13; right amygdala: .048 [.35 to .25], p ¼ .74; mPFC:
.26 [.58 to .059], p ¼ .11).
Bright-Light Intervention Increases Corticolimbic Functional
Coupling
Next, we evaluated the effects of bright-light intervention on
an estimate of communication within this corticolimbic circuit
using functional coupling. At baseline, we observed statistically
significant positive functional coupling across participants
between each of our seeds and nearly all of our amygdala-
prefrontal search volume (Table S1 in Supplement 1). No brain
regions exhibited significant negative functional coupling with
any seed. Effects of bright-light dose on functional coupling were
evaluated at a voxel level using the ANCOVA model described
above. Bright-light dose was statistically significantly positively
associated with functional coupling at rescan between our left
amygdala seed and mPFC, controlling for baseline coupling
(Table 2; Figure S2 in Supplement 1). Bright-light intervention
did not affect functional coupling between our right amygdala
seed and mPFC. Bright-light dose was also statistically signifi-
cantly positively associated with functional coupling between our
mPFC seed and surrounding mPFC regions (Table 2; Figure S3 in
Supplement 1).
We evaluated whether 5-HTTLPR status moderated effects of
bright-light intervention on functional coupling. There were no
statistically significant moderation effects on functional coupling
with our left or right amygdala seeds. The effect of bright-light
intervention on intraprefrontal functional coupling (i.e., mPFC
coupling with our mPFC seed) was statistically significantly
moderated by 5-HTTLPR status such that bright-light dose was
positively associated with intraprefrontal functional coupling at
rescan in LG or S carriers, but there was no effect on LA/LA
individuals (Table 2; Figure 3).
www.sobp.org/journal
336 BIOL PSYCHIATRY 2014;76:332–339 P.M. Fisher et al.

Discussion
Here, we present findings from the first study to evaluate the
effects of a bright-light intervention similar to clinically adminis-
tered protocols on threat-related corticolimbic reactivity and
functional coupling. Consistent with our hypothesis, we observed
a significant dose-dependent negative association between
bright-light intervention and threat-related reactivity within
amygdala and mPFC. Also consistent with our hypothesis,
bright-light intervention positively affected prefrontal functional
coupling with our left (but not right) amygdala and mPFC seeds.
Interestingly, the effect of bright-light intervention on prefrontal
functional coupling with our mPFC seed was significantly mod-
erated by 5-HTTLPR status such that intervention increased
functional coupling in LG or S carriers, whereas little effect was
observed in LA/LA individuals. Intriguingly, our finding that
5-HTTLPR status moderates the response to bright-light interven-
tion provides further evidence supporting a link between seroto-
nin signaling, bright-light intervention, and seasonality. These
findings implicate this circuit in mediating the clinical effects of
bright-light intervention and can serve as a relevant benchmark
for future studies evaluating the effects of bright-light interven-
tion on brain function as it relates to treatment.
Bright-light therapy has been used to successfully treat SAD (3)
and more recent studies have reported its effectiveness for
treating other affective disorders (2). Previous studies have
reported alterations in function of this corticolimbic circuit in
patients with affective disorders and acute bright-light exposure
affected amygdala reactivity to negative auditory stimuli (4,38,39).
Taken together with our findings, we hypothesize dysfunction of
this circuit represents a neurobiological mechanism shared across
seasonal and other affective disorders and may represent a critical
neural pathway mediating clinical responsiveness to bright-light
intervention (40). Future studies in patients with SAD are

Table 2. Bright-Light Intervention and 5-HTTLPR Effects on Corticolimbic

Functional Coupling
Figure 2. Effects of bright-light intervention on medial prefrontal
cortex (mPFC) reactivity. (A) Statistical parametric map highlighting x, y, z Z Score Cluster Size
mPFC area where functional magnetic resonance imaging response to
angry and fearful faces was less than that of neutral faces at baseline Bright-Light Intervention Effects on Functional Coupling
across all participants. Color bar represents t scores. (B) Bright-light dose Left amygdala seed
was negatively associated with mPFC reactivity (plotted as angry and Positive association – medial 4, 28, 0 4.11 938
fearful – neutral faces) at rescan. Thirty individual data points shown in
prefrontal cortex
orange and blue shading represent 95% confidence limit of
regression line. Right amygdala seed
No significant clusters – – –
mPFC seed
Positive association – medial 4, 28, 2 4.12 1594
Treatment Compliance and Effects on Behavior and Mood prefrontal cortex
We observed good participant compliance throughout the 5-HTTLPR by Bright-Light Intervention Interaction Effects on Functional
intervention period (reported days missed: mean ¼ .9, range ¼ 0–3). Coupling
Bright-light dose was not associated with significant changes in Left amygdala seed
measures of mood, depressive symptoms, and stress (Table 1). No significant clusters – – –
Participants completed the fMRI paradigm with a high degree of Right amygdala seed
accuracy that was unaffected by bright-light intervention No significant clusters – – –
(Table 1). To estimate effects of natural light exposure, we asked mPFC seed
participants to report average time outside (i.e., less than 30 Medial prefrontal cortex 12, 40, 18 3.02 394
minutes daily, 30 to 60 minutes, or more than 60 minutes). Bright- Effects of bright-light intervention were evaluated in a model that did
light dose received was not significantly different across these not include 5-HTTLPR genotype status. All coordinates are given in MNI
groups, suggesting our findings are not confounded by differ- space. Statistical threshold for mPFC search volume: p ⬍ .01 (voxel-level),
ences in natural light exposure (F2,27 ¼ .57, p ¼ .88). Additionally, cluster-extent ⬎ 249 voxels. Left and right amygdala and mPFC seeds are
5 mm spheres centered on (30, 4, 18), (30, 4, 18), and (10, 52,
natural light exposure was not significantly predictive of reactivity 0), respectively.
or functional coupling when added as a predictor in our ANCOVA 5-HTTLPR, serotonin transporter-linked polymorphic region; MNI,
model (ps ⬎ .20). Montreal Neurological Institute; mPFC, medial prefrontal cortex.

www.sobp.org/journal
P.M. Fisher et al.
Figure 3. Serotonin transporter-linked polymorphic region (5-HTTLPR)
moderates effect of bright-light intervention on medial prefrontal cortex
(mPFC)-prefrontal functional coupling. (A) Statistical parametric map
highlighting mPFC cluster, wherein the positive effect of bright-light
intervention on functional coupling with our mPFC seed was significantly
moderated by 5-HTTLPR genotype status (mPFC seed outlined within
inset). Color bar represents t scores. (B) Plot of bright-light intervention by
5-HTTLPR interaction effect showing mean functional coupling estimate
across 394 voxels and is intended only for visualization of interaction
effect. Thirty individual data points are shown in blue or orange. Blue and
orange shading represents 95% confidence limit of regression lines for LG
or S carriers (non-LA/LA) and LA/LA individuals, respectively.
necessary to evaluate this hypothesis and determine whether
changes in corticolimbic reactivity are associated with treatment
response. Such studies would also provide insight into a link
between bright-light effects on specific neurobiological mechanisms
and the subset of clinical symptoms (e.g., depressive vs. anxiety
symptoms) that may be preferentially benefited by bright-light
therapy.
Neuroimaging studies in humans have reported an association
between bidirectional amygdala-prefrontal connections and anxi-
ety (41,42), but rodent studies suggest this communication can
enhance and diminish responses to threatening environmental
cues (6). We observed that amygdala and prefrontal reactivity
were dose-dependently decreased by bright-light intervention. In
contrast, amygdala-prefrontal and intraprefrontal functional
BIOL PSYCHIATRY 2014;76:332–339 337
coupling were increased following bright-light intervention. These
findings are consistent with a model wherein our task-related
reactivity reflects the bottom-up drive of the amygdala on this
circuit, which is dampened by bright-light intervention. By
contrast, our metric of functional coupling may more closely
relate to the top-down component of this circuit, namely
prefrontal-mediated regulation of the threat-related amygdala
reactivity, which is enhanced by bright-light intervention. These
complementary findings converge to support an effect of bright-
light intervention on this circuit that is consistent with its role
mediating responses to salient environmental stimuli. We focused
on a very well studied circuit for which there is extensive
evidence implicating it in the pathophysiology of mood and
affective disorders. Evaluating the effects of light intervention on
neural circuits (e.g., reward) or paradigms specifically probing
complementary features of this corticolimbic circuit (e.g., emotion
regulation paradigms) would benefit our understanding of the
pathways sensitive to bright-light exposure.
Serotonin signaling has been implicated in the pathophysiol-
ogy of SAD and modulating this circuit (10,11). Interestingly, a
recent study in rodents reported that direct retino-raphe projec-
tions modulate serotonergic signaling and depressive-like behav-
ioral phenotypes, providing a direct pathway through which
bright-light exposure can modulate serotonin signaling (43). In
the current study, we oversampled for individuals homozygous
for the 5-HTTLPR LA-allele, as a model for differences in serotonin
signaling and to evaluate possible effects of serotonin signaling
on bright-light intervention. Bright-light dose was positively
associated with intraprefrontal functional coupling in LG or S
carriers but not in LA/LA individuals. These findings are consistent
with a model of heightened sensitivity of LG or S carriers to
environmental input, including seasonal variation (12,44). We
have reported greater seasonal variability in 5-HTT binding in
S-allele carriers compared with LL homozygotes (12) [but see also
(45)]. Though 5-HTTLPR status did not significantly moderate
bright-light effects on task-related reactivity, effects neared
significance for the left amygdala and mPFC (p ¼ .13 and p ¼
.11, respectively). Thus, there may be an effect that we are
underpowered to detect with our sample size. Additionally, we
suggest caution when interpreting our findings related to the
5-HTTLPR, as they, too, may be underpowered, which can increase
risk for spurious findings (46). This emphasizes the importance of
additional studies probing how serotonin signaling interacts with
light intervention. Regardless, these findings provide intriguing
evidence that the response to bright-light intervention of this
corticolimbic circuit may be moderated, in part, by serotonin
signaling. These associations could be more directly evaluated by
future studies pairing a bright-light intervention protocol with a
pharmacologic challenge of the serotonin system.
Our current study is not without its limitations. There is no
specific clinical intervention threshold for bright-light therapy,
but a 30-minute administration period is recommended for a
light dose of 10 klx (1). Thus, our study included intervention
intensities at and below suggested clinical doses. Interestingly,
our findings indicate a dose-dependent association between light
intervention and effects on corticolimbic function across this
range. Though we feel that we reasonably estimated light dose
by measuring aspects of the environment in which intervention
was administered, it is important to note that maintaining
consistent light-intervention dose is more difficult than, e.g., a
pharmacologic challenge using tablets. As such, we are limited in
our ability to precisely measure light dose received by each
participant. We suggest that future studies consider the use of
www.sobp.org/journal
338 BIOL PSYCHIATRY 2014;76:332–339
equipment that may more precisely measure bright-light expo-
sure and other potentially relevant variables. That notwithstand-
ing, a related benefit of our study design is that empirically
estimating light dose as a continuous metric closely mimics real
variation that is to be expected when light intervention is
administered in a clinical context. Unlike pharmacologic inter-
vention studies, bright-light intervention lacks an easily acquired
molecular marker for confirming treatment compliance. Despite
this, we encouraged compliance through daily mobile phone
short message service reminders and the completion of log
books. We did not observe significant effects of bright-light
intervention on mood or other state measures. Although this is
consistent with a previous study, it underscores a limitation of
evaluating intervention effects in healthy individuals (47). Our
study included only male subjects. Though this limits the general-
izability of our findings, our study design is precluded from
confounds such as fluctuations in sex hormones. Future studies
evaluating the effects of bright-light intervention on brain
function in women would complement our current findings and
speak to possible sex interaction effects.
In summary, we show significant evidence for a dose-
dependent effect of 3-week bright-light intervention on
threat-related corticolimbic reactivity and functional coupling.
Furthermore, the effect of bright-light intervention on intrapre-
frontal functional coupling was moderated by 5-HTTLPR genotype
status, further implicating serotonin in modulating this circuit
and sensitivity to light. These findings represent an intriguing
step toward understanding neurobiological mechanisms affected
by bright-light intervention. This may, in turn, benefit our
understanding of neurobiological mechanisms that bias risk
for SAD and mediate responsiveness to bright-light interven-
tion as a therapeutic strategy in SAD and other affective
disorders.
This study was funded by an unrestricted center grant to Cimbi
from the Lundbeck Foundation. Support for the purchase of bright-
light lamp materials was kindly provided by A.P. Møller and Hustru
Chastine Mc-Kinney Møllers Fond til alemene Formaal. BM was
funded by a scholarship from Brain, Mind and Medicines. HRS was
supported by a Grant of Excellence ContAct from the Lundbeck
Foundation (R59 A5399). PMF was funded by a grant from the
Danish Council for Independent Research.
We thank the Danish Research Centre for Magnetic Resonance
for the magnetic resonance imaging facilities; Sussi Larsen, Julian
Macoveanu, and Pernille Iversen for assistance in data collection
and management; and Lone Freyr and Dorthe Givard for their
coordinated effort in maintaining blinding. We thank Anders Brusch
from Danish Fundamental Metrology and Professor Paul Michael
Petersen from the Department of Photonics Engineering Diode
Lasers and LED Systems at the Technical University of Denmark
for their help in measuring lamp characteristics.
HRS declares that over the past 3 years he has received honoraria
as reviewing editor for Neuroimage, as speaker for Biogen Idec
Denmark A/S, and scientific Advisor for Lundbeck A/S, Valby,
Denmark. All other authors report no biomedical financial interests
or potential conflicts of interest.
Supplementary material cited in this article is available online at
http://dx.doi.org/10.1016/j.biopsych.2013.11.031.
1. Pail G, Huf W, Pjrek E, Winkler D, Willeit M, Praschak-Rieder N, Kasper S
(2011): Bright-light therapy in the treatment of mood disorders.
Neuropsychobiology 64:152–162.
www.sobp.org/journal
P.M. Fisher et al.
2. Terman M (2007): Evolving applications of light therapy. Sleep Med Rev
11:497–507.
3. Rosenthal NE, Sack DA, Gillin JC, Lewy AJ, Goodwin FK, Davenport Y,
et al. (1984): Seasonal affective disorder. A description of the
syndrome and preliminary findings with light therapy. Arch Gen
Psychiatry 41:72–80.
4. Vandewalle G, Schwartz S, Grandjean D, Wuillaume C, Balteau E,
Degueldre C, et al. (2010): Spectral quality of light modulates
emotional brain responses in humans. Proc Natl Acad Sci U S A 107:
19549–19554.
5. Praschak-Rieder N, Willeit M (2003): Treatment of seasonal affective
disorders. Dialogues Clin Neurosci 5:389–398.
6. Quirk GJ, Mueller D (2008): Neural mechanisms of extinction learning
and retrieval. Neuropsychopharmacology 33:56–72.
7. Phillips ML, Drevets WC, Rauch SL, Lane R (2003): Neurobiology of
emotion perception I: The neural basis of normal emotion perception.
Biol Psychiatry 54:504–514.
8. Stuhrmann A, Suslow T, Dannlowski U (2011): Facial emotion process-
ing in major depression: A systematic review of neuroimaging
findings. Biol Mood Anxiety Disord 1:10.
9. Holmes A (2008): Genetic variation in cortico-amygdala serotonin
function and risk for stress-related disease. Neurosci Biobehav Rev 32:
1293–1314.
10. Fisher PM, Hariri AR (2012): Linking variability in brain chemistry and
circuit function through multimodal human neuroimaging. Genes
Brain Behav 11:633–642.
11. Praschak-Rieder N, Willeit M (2012): Imaging of seasonal affective
disorder and seasonality effects on serotonin and dopamine function
in the human brain. Curr Top Behav Neurosci 11:149–167.
12. Kalbitzer J, Erritzoe D, Holst KK, Nielsen FA, Marner L, Lehel S, et al.
(2010): Seasonal changes in brain serotonin transporter binding in
short serotonin transporter linked polymorphic region-allele carriers
but not in long-allele homozygotes. Biol Psychiatry 30:1033–1039.
13. Hu XZ, Lipsky RH, Zhu G, Akhtar LA, Taubman J, Greenberg BD, et al.
(2006): Serotonin transporter promoter gain-of-function genotypes
are linked to obsessive-compulsive disorder. Am J Hum Genet 78:
815–826.
14. Lesch KP, Bengel D, Heils A, Sabol SZ, Greenberg BD, Petri S, et al. (1996):
Association of anxiety-related traits with a polymorphism in the serotonin
transporter gene regulatory region. Science 274:1527–1531.
15. Grady CL, Siebner HR, Hornboll B, Macoveanu J, Paulson OB, Knudsen
GM (2013): Acute pharmacologically induced shifts in serotonin
availability abolish emotion-selective responses to negative face
emotions in distinct brain networks. Eur Neuropsychopharmacol 23:
368–378.
16. Harmer CJ, Mackay CE, Reid CB, Cowen PJ, Goodwin GM (2006):
Antidepressant drug treatment modifies the neural processing of
nonconscious threat cues. Biol Psychiatry 59:816–820.
17. Arce E, Simmons AN, Lovero KL, Stein MB, Paulus MP (2008):
Escitalopram effects on insula and amygdala BOLD activation during
emotional processing. Psychopharmacology (Berl) 196:661–672.
18. Rosenthal NE, Genhart M, Sack DA, Skwerer RG, Wehr TA (1987):
Seasonal affective disorder: Relevance for treatment and research in
bulimia. In: Hudson EL, Pope HG, editors. Psychbioloy of Bulimia.
Washington, DC: American Psychiatric Press, 205–228.
19. Sexton T, Buhr E, Van Gelder RN (2012): Melanopsin and mechanisms
of non-visual ocular photoreception. J Biol Chem 287:1649–1656.
20. Terman JS, Terman M, Lo ES, Cooper TB (2001): Circadian time of
morning light administration and therapeutic response in winter
depression. Arch Gen Psychiatry 58:69–75.
21. Lewy AJ, Bauer VK, Cutler NL, Sack RL, Ahmed S, Thomas KH, et al.
(1998): Morning vs evening light treatment of patients with winter
depression. Arch Gen Psychiatry 55:890–896.
22. Forsell Y (2005): The Major Depression Inventory versus Schedules for
Clinical Assessment in Neuropsychiatry in a population sample. Soc
Psychiatry Psychiatr Epidemiol 40:209–213.
23. McNair DM, Lorr M, Droppleman L (1992): Manual for the Profile of
Mood States. San Diego: EdITS/Educational and Industrial Testing Service.
24. Cohen S, Kamarck T, Mermelstein R (1983): A global measure of
perceived stress. J Health Soc Behav 24:385–396.
25. Buysse DJ, Reynolds CF III, Monk TH, Berman SR, Kupfer DJ (1989): The
Pittsburgh Sleep Quality Index: A new instrument for psychiatric
practice and research. Psychiatry Res 28:193–213.
P.M. Fisher et al.
26. Deichmann R, Gottfried JA, Hutton C, Turner R (2003): Optimized EPI
for fMRI studies of the orbitofrontal cortex. Neuroimage 19:430–441.
27. Lundqvist D, Flykt A, Ohman A (1998): The Karolinska Directed
Emotional Faces-KDEF [CD-ROM]. Stockholm: Department of Clinical
Neuroscience, Psychology Section, Karolinska Institutet.
28. Lancaster JL, Woldorff MG, Parsons LM, Liotti M, Freitas CS, Rainey L,
et al. (2000): Automated Talairach atlas labels for functional brain
mapping. Hum Brain Mapp 10:120–131.
29. Lancaster JL, Summerln JL, Rainey L, Freitas CS, Fox PT (1997): The
Talairach Daemon, a database server for Talairach Atlas Labels.
Neuroimage 5:S633.
30. Maldjian JA, Laurienti PJ, Kraft RA, Burdette JH (2003): An automated
method for neuroanatomic and cytoarchitectonic atlas-based inter-
rogation of fMRI data sets. Neuroimage 19:1233–1239.
31. Pandya DN, Van Hoesen GW, Mesulam MM (1981): Efferent connec-
tions of the cingulate gyrus in the rhesus monkey. Exp Brain Res 42:
319–330.
32. Barbas H (1995): Anatomic basis of cognitive-emotional interactions in
the primate prefrontal cortex. Neurosci Biobehav Rev 19:499–510.
33. Ongur D, Price JL (2000): The organization of networks within the
orbital and medial prefrontal cortex of rats, monkeys and humans.
Cereb Cortex 10:206–219.
34. Friston KJ (1994): Functional and effective connectivity in neuorimag-
ing: A synthesis. Hum Brain Mapp 2:56–78.
35. Brett M, Anton J, Valabregue R, Poline J (2002): Region of interest
analysis using an SPM toolbox. Neuroimage 16:S497.
36. Fisher PM, Meltzer CC, Price JC, Coleman RL, Ziolko SK, Becker C, et al.
(2009): Medial prefrontal cortex 5-HT2A density is correlated with
amygdala reactivity, response habituation, and functional coupling.
Cereb Cortex 19:2499–2507.
37. Vickers AJ, Altman DG (2001): Statistics notes: Analysing controlled
trials with baseline and follow up measurements. BMJ 323:1123–1124.
BIOL PSYCHIATRY 2014;76:332–339 339
38. Mayberg HS (2003): Modulating dysfunctional limbic-cortical circuits
in depression: Towards development of brain-based algorithms for
diagnosis and optimised treatment. Br Med Bull 65:193–207.
39. Phillips ML, Drevets WC, Rauch SL, Lane R (2003): Neurobiology of
emotion perception II: Implications for major psychiatric disorders. Biol
Psychiatry 54:515–528.
40. Golden RN, Gaynes BN, Ekstrom RD, Hamer RM, Jacobsen FM, Suppes
T, et al. (2005): The efficacy of light therapy in the treatment of mood
disorders: A review and meta-analysis of the evidence. Am J Psychiatry
162:656–662.
41. Kim MJ, Gee DG, Loucks RA, Davis FC, Whalen PJ (2010): Anxiety
dissociates dorsal and ventral medial prefrontal cortex functional
connectivity with the amygdala at rest. Cereb Cortex 21:1667–1673.
42. Kim MJ, Whalen PJ (2009): The structural integrity of an amygdala-
prefrontal pathway predicts trait anxiety. J Neurosci 29:11614–11618.
43. Ren C, Luan L, Wui-Man LB, Huang X, Yang J, Zhou Y, et al. (2013):
Direct retino-raphe projection alters serotonergic tone and affective
behavior. Neuropsychopharmacology 38:1163–1175.
44. Caspi A, Hariri AR, Holmes A, Uher R, Moffitt TE (2010): Genetic
sensitivity to the environment: The case of the serotonin transporter
gene and its implications for studying complex diseases and traits. Am
J Psychiatry 167:509–527.
45. Murthy NV, Selvaraj S, Cowen PJ, Bhagwagar Z, Riedel WJ, Peers P,
et al. (2010): Serotonin transporter polymorphisms (SLC6A4 insertion/
deletion and rs25531) do not affect the availability of 5-HTT to [11C]
DASB binding in the living human brain. Neuroimage 52:50–54.
46. Button KS, Ioannidis JP, Mokrysz C, Nosek BA, Flint J, Robinson ES,
Munafò MR (2013): Power failure: Why small sample size undermines
the reliability of neuroscience. Nat Rev Neurosci 14:365–376.
47. Rosenthal NE, Rotter A, Jacobsen FM, Skwerer RG (1987): No mood-
altering effects found after treatment of normal subjects with bright
light in the morning. Psychiatry Res 22:1–9.
www.sobp.org/journal