Content uploaded by Claudia Catani
Author content
All content in this area was uploaded by Claudia Catani on Oct 18, 2017
Content may be subject to copyright.
ELEVATED RESPONSE OF HUMAN AMYGDALA TO NEUTRAL
STIMULI IN MILD POST TRAUMATIC STRESS DISORDER: NEURAL
CORRELATES OF GENERALIZED EMOTIONAL RESPONSE
M. BRUNETTI,
a,b
* G. SEPEDE,
a,b
G. MINGOIA,
a,b
C. CATANI,
c
A. FERRETTI,
a,b
A. MERLA,
a,b
C. DEL GRATTA,
a,b
G. L. ROMANI
a,b
AND
C. BABILONI
d,e
a
Institute of Advanced Biomedical Technologies, University of Chieti,
Chieti, Italy
b
Department of Clinical Sciences and Bioimaging, University of Chieti,
Chieti, Italy
c
Department of Clinical Psychology, University of Bielefeld, Bielefeld,
Germany
d
Department of Biomedical Sciences, University of Foggia, Foggia,
Italy
e
Casa di Cura San Raffaele, Cassino, Italy
Abstract—Previous evidence from functional magnetic reso-
nance imaging (fMRI) studies has shown that amygdala re-
sponses to emotionally neutral pictures are exaggerated at a
group level in patients with severe post-traumatic stress dis-
order (PTSD) [Hendler T, Rotshtein P, Yeshurun Y, Weizmann
T, Kahn I, Ben-Bashat D, Malach R, Bleich A (2003) Neuroim-
age 19(3):587– 600]. The present fMRI study tested the hy-
pothesis that amygdala responses are elevated not only in
response to negative pictures but also to neutral pictures as
a function of disease severity in patients with mild symptoms
and in subjects who did not develop symptoms. To this end,
fMRI scans were performed in 10 patients with mild PTSD and
10 healthy controls (both victims of a bank robbery), during
the execution of a visuo-attentional task in which they were
asked to observe emotionally negative or neutral pictures.
Control subjects showed enhanced amygdala responses to
emotionally negative stimuli compared to neutral stimuli. On
the contrary, PTSD patients were characterized by high
amygdala responses to both neutral and emotional pictures,
with no statistically significant difference between the two
classes of stimuli. In the entire group, we found correlations
among the severity of the PTSD symptoms, task perfor-
mance, and amygdala activation during the processing of
neutral stimuli. Results of this study suggest that amygdala
responses and the selectivity of the emotional response to
neutral stimuli are elevated as a function of disease severity
in PTSD patients with mild symptoms. © 2010 IBRO. Pub-
lished by Elsevier Ltd. All rights reserved.
Key words: amygdala, emotion, functional magnetic reso-
nance imaging (fMRI), post-traumatic stress disorder (PTSD).
Intensive fear processing generated by overwhelming ex-
periences such as traumatic episodes has been system-
atically explored to disclose the biological correlates of
stress response. Patients with post-traumatic stress disor-
der (PTSD) experience intensive fear due to the continu-
ous reliving of the past trauma, exhibit exaggerated re-
sponses to emotionally negative stimuli and tend to misin-
terpret innocuous stimuli as potential threats (van der Kolk,
1994).
Several lines of evidence based on clinical functional
magnetic resonance imaging (fMRI) or positron emission
tomography (PET) findings have shown that abnormally
increased brain response as well as elevated physiologi-
cal/behavioral fear responses is a common element of
different anxiety disorders (Etkin and Wager, 2007).
An increased amygdala response to emotional visual
stimuli in PTSD patients compared to control subjects has
been reported (Rauch et al., 2000; Williams et al., 2006).
These responses of the amygdala in patients increased
during the processing of emotionally negative visual stimuli
and symptoms provocation (Liberzon et al., 1999; Rauch
et al., 2000). The same was true when the attention of
patients was devoted to salient (rare) non-threatening au-
ditory stimuli in a typical oddball paradigm. Specifically,
there was an increment of the activation of the dorsal
anterior cingulate cortex (dACC) and left amygdala to sa-
lient stimuli, interpreted as hyper-vigilance (Bryant et al.,
2005). Enhanced responses of the amygdala were also
related to an arousal-driven emotional enhancement of
memory (Sommer et al., 2008). Furthermore, skin conduc-
tance responses (SCR) to subliminal salient emotional
stimuli were delayed in PTSD patients with damage to the
left or right amygdala (Gläscher and Adolphs, 2003).
Different findings lead to different interpretations of the
role of the medial prefrontal cortex-amygdala system in
emotion processing. Studies based on script driven para-
digms compared to neutral stimuli showed a correlation
between the decrease of medial prefrontal activation and
the enhancement of amygdala activation in patients (Shin
et al., 1999, 2004). On the other hand, an fMRI study using
masked facial stimuli with PTSD patients versus control
subjects (Rauch et al., 2000) revealed an original finding of
amygdala hyper-responsivity dissociated from the “top-
down” influences of medial prefrontal cortex; the authors
observed an exaggerated amygdala response to general
(and not trauma-related) negative stimuli in the PTSD
group. Moreover, Liddell and colleagues verified, by
means of fMRI, the hypothesis of an automatic alerting
network engagement during subliminal facial signals of
*Correspondence to: M. Brunetti, Institute for Advanced Biomedical Technol-
ogies, University “G. D’Annunzio” of Chieti, Via dei Vestini, 33, 66013 Chieti
(CH), Italy. Tel: ⫹39-0871-3556935; fax: ⫹39-0871-3556930.
E-mail address: mbrunetti@itab.unich.it (M. Brunetti).
Abbreviations: AC, anterior commissure; ANOVA, analysis of vari-
ance; CAPS, clinician administered PTSD scale; fMRI, functional mag-
netic resonance imaging; PC, posterior commissure; PTSD, post-
traumatic stress disorder; RT, reaction time; SPM, statistical paramet-
ric map.
Neuroscience 168 (2010) 670–679
0306-4522/10 $ - see front matter © 2010 IBRO. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.neuroscience.2010.04.024
670
fear processing. A collateral network attended by brain-
stem structures (superior colliculus), thalamus (pulvinar)
and amygdala was also observed (Liddell et al., 2005).
Several studies have reported conflicting findings regard-
ing the modulation of amygdala activity during the viewing
of neutral and negative visual stimuli (Bremner et al., 1999;
Yang et al., 2004; Britton et al., 2005; Phan et al., 2006).
Most of these studies have adopted designs where PTSD
patients were exposed to trauma-related stimulation; in
these conditions, the amplitude of amygdala activation was
found to increase in these patients (Liberzon et al., 1999;
Pissiota et al., 2002). Script driven paradigms were also
adopted to demonstrate a negative correlation between
the activation of the medial prefrontal cortex and the amyg-
dala (Shin et al., 1999, 2004). On the contrary, other
studies have reported either decreased amygdala activity
(Britton et al., 2005; Phan et al., 2006) or no difference
between patients and controls in amygdala activation
(Bremner et al., 1999; Yang et al., 2004). Noteworthy, most
of these studies used neutral stimuli as control stimulation
and therefore could represent a potential bias in the inves-
tigation of amygdala response. Previous evidence by Hen-
dler and colleagues (2003) showed increased amygdala
activation in PTSD patients irrespective of the emotional
content of the visual stimuli. An exaggerated response of
the PTSD patients’ amygdala for the neutral stimuli could,
then, prevent marked differences between amygdala re-
sponses to negative vs. neutral stimuli when PTSD and
control groups were contrasted. Therefore, in their study it
is not clear if the relationship between PTSD patients’
amygdala responses and severity of the symptoms are
correlated across subjects or whether the difference can
be appreciated only as a mean group difference. The
existence of a correlative relationship would allow the use
of amygdala response to neutral stimuli as an instrumental
surrogate marker in clinical and pharmacological research
in PTSD patients.
To address the above issue, the present study tested
the hypothesis that the responses of the amygdala are
abnormal not only to negative emotional stimuli but also to
neutral stimuli, as a correlative function of disease severity
in PTSD patients. Noteworthy, a methodological obstacle
for the evaluation of this hypothesis is the possibility that
the severity of the PTSD symptoms could cause huge
differences in mean amygdala responses to visual stimuli
between PTSD and healthy subjects, thus inflating the
correlation between amygdala responses and disease se-
verity across subjects. To minimize this eventuality, we
selected a population of PTSD patients with a relatively low
degree of disease symptoms. This clinical feature of the
recruited mild PTSD patients had the additional advantage
of allowing us to investigate the correlative relationship
between amygdala responses and disease severity in the
gray zone in subjects who developed vs. not developed the
disease following the trauma. The importance of this gray
zone is supported by clinical studies reporting evidence
about the features of the response to stress in subjects
with PTSD symptoms at subthreshold degrees compared
to overt PTSD patients (Zlotnick et al., 2002), in line with
the idea that stress symptoms stem upon a continuum
along subthreshold, mild and severe PTSD. Finally, previ-
ous studies have observed some laterality in amygdala
activity during fear processing, thus we also considered
hemisphere as a factor in the present study (Williams et al.,
2006).
EXPERIMENTAL PROCEDURES
Subjects
Ten patients with PTSD and 10 age, sex and education-matched
healthy control subjects were recruited. All participants were bank
clerks, victims of one or more armed bank assaults about 10
months before the experiments (range of 2–18 months). Table 1
reports their demographic information.
All participants underwent an extensive clinical examination
carried out by an expert psychiatrist (GS) and a clinical psychol-
ogist (MB). A broader range of traumatic event types, including car
accidents and criminal attacks, was assessed using the event
checklist of the Clinician Administered PTSD Scale (CAPS) (Blake
et al., 1998). Standardized clinical instruments were used for the
assessment of DSM-IV diagnoses: CAPS for the diagnosis and
quantification of PTSD and related dissociative features and Mini
International Neuropsychiatric Interviews (MINI) (Sheehan et al.,
1994) for diagnoses of DSM-IV axis one disorders. At the time of
the present study, participants met DSM IV diagnostic criteria for
the following current co-morbid diagnoses: dysthymia (n⫽1 PTSD
subject), agoraphobia without history of panic attack disorder
(n⫽1 PTSD subject), and social phobia (n⫽1 control subject).
None of the participants had been previously treated with psycho-
pharmacologic interventions.
For subject recruitment, inclusion criteria included right hand-
edness, assessed by Edinburgh Inventory (Oldfield, 1971), normal
or corrected to normal (soft contact lenses) vision, age from
20–50 years, and victim of an armed bank robbery. Exclusion
criteria included seizure disorder, progressive neurological and/or
systemic disorders, metallic implants, significant unstable concur-
rent medical illness, hormone replacement therapy, electroconvul-
sive or light therapy, administration of concomitant medication that
could alter mood or cerebral metabolism (e.g. benzodiazepines,
antidepressants, mood stabilizers, stimulants, and steroids) within
30 days prior to screening, history of any substance/alcohol abuse
or dependence within the past 6 months (nicotine dependence
was allowed) pregnancy, and not understanding the experimental
procedures. Two PTSD patients were excluded from the study.
One patient was excluded due to co-morbidity (i.e. Panic disor-
der), whereas the other patient asked to interrupt the experiment.
Two control subjects were excluded due to concurrent medical
illness. The clinical features of the two groups and main statistical
comparisons are reported in Table 1.
Table 1. Demographic and clinical characteristics of the two subject
groups
Variable PSTD group
(n⫽10)
Control group
(n⫽10)
Mean age in years(SD) 39.7 (6.34) 36.8 (12.1)
Mean school educ. in years (SD) 16 (2.5) 14 (2.1)
N Females(%) 6 (60) 6 (60)
Trauma load [mean number (range)]
CAPS event types 5 (1–8) 4 (0–5)
CAPS PTSD symptom score,
mean (SD)
30.8* 6.2*
Indices represent the results of ANOVA: (F(1,18)⫽34.9 P⬍⫽0.01).
M. Brunetti et al. / Neuroscience 168 (2010) 670–679 671
All subjects received detailed explanations about study design
and gave written informed consent according to the Declaration of
Helsinki (World Medical Association Declaration of Helsinki, 1997).
The protocol was approved by the local Ethics Committee (School of
Medicine Ethic Committee, University of Chieti, Italy).
Stimuli and experimental design
One hundred and eighty colored pictures were chosen from the
International Affective Picture System (IAPS) on the basis of their
normative ratings (IAPS, Lang et al., 1999; Catani et al., 2009). Of
these, 90 pictures depicted unpleasant scenes (e.g. mutilations,
assaults, dead bodies, etc) (IAPS rating: valence⬍3, 5.3⬍aro-
usal⬍7) and 90 pictures depicted neutral contents (e.g. mushrooms,
mugs, neutral faces, etc) (4.5⬍valence⬍5.5 and arousal⬍4).
1
The pictures were presented using Gaglab, an in-house pro-
gramme based on Matlab (http://www.mathworks.com/). From a
technical point of view, stimuli were projected over a transparent
screen inside the scanner tunnel and were viewed through a
mirror system mounted on the top of the magnetic resonance
imaging (MRI) head coil.
The pictures were presented according to a “boxcar” design.
Thirty-six blocks consisting of 5 pictures each (Stimulus Onset
Asynchrony (SOA)⫽3 s, stimulus duration⫽2 s) were alternated
witha1scontrol state (cross fixation). Each stimulation block
contained five pictures, which were either all unpleasant or all
neutral to induce homogeneous brain responses to stimulus con-
tent. The order of the unpleasant and neutral blocks was random-
ized to pair order effects, tonic arousal, and fatigue across sub-
jects. Every block or control state lasted 15 s. All 180 pictures
were presented only once to avoid habituation effects. The sub-
jects’ task was to observe each picture and to quickly press a
button on a compatible MRI keypad using their right hand.
The subjects had to press left button if the pictures contained
a vegetable item such as plants, flowers, etc. (target pictures),
whereas they had to press right button if the pictures contained no
vegetable item. There was an equal number of targets in the
unpleasant and neutral pictures, corresponding to 25% of the total
number of pictures. Noteworthy, the vegetable items were always
neutral in valence (i.e. there was no firing tree or bloody flower,
etc.) in any target picture, to reproduce typical conditions trigger-
ing dysfunctional behavior in patients with PTSD. These patients
suffer from sudden emotional crisis while they engaged in neutral
daily activities where neutral and emotional stimuli are available in
the environment. During the current experiments, the subjects’
responses were recorded using an MRI compatible response
device with a two button keypad for the right hand. Fig. 1 illustrates
the experimental design.
fMRI recordings
BOLD contrast functional imaging was performed with a SIEMENS
MAGNETOM VISION scanner at 1.5 T (Siemens, Erlangen,
Germany) by means of T2*-weighted echo planar imaging
(EPI), free-induction decay (FID) sequences with the following
parameters: TR 3 s, TE 60 ms, matrix size 64⫻64, FOV 256 mm,
in-plane voxel size 4⫻4 mm, flip angle 90°, slice thickness 4 mm
and no gap. A standard head coil was used and the subject’s head
was fixed with foam pads to reduce involuntary movements. Func-
tional fMRI volumes consisted of 28 bicommissural transaxial
slices including the cortical regions of interest. All stimuli were
presented in one run; 240 volumes were acquired starting with
three dummy volumes and a control period.
A high resolution structural volume was acquired at the end of
the session via a 3D MPRAGE sequence with the following fea-
tures: sagittal, matrix 256⫻256, FoV 256 mm, slice thickness 1
mm, no gap, in-plane voxel size 1⫻1 mm, flip angle 12°, TR⫽9.7
ms, and TE⫽4 ms.
fMRI data analysis
Preprocessing and statistical analysis of the fMRI data were per-
formed using Brain Voyager QX 1.9 software (Brain Innovation,
The Netherlands). Due to T1 saturation effects, the first three
scans of each run were discarded from the analysis. Preprocessing
of functional scans included motion correction and removal of linear
trends from voxel time series. A three-dimensional motion correction
was performed by means of a rigid body transformation to match
each functional volume to the reference volume (the fourth volume)
estimating three translation and three rotation parameters.
These parameters were stored in log-files and inspected to
check that estimated head movement was not larger than approx-
imately half a voxel (⫾2 mm) for the functional run and that no
task-correlated movement had occurred (Friston et al., 1996; Ha-
jnal et al., 1994). Spatial normalization was performed for struc-
tural and functional datasets. The spatial normalization of the
structural volumes was performed in two steps. The first step cons-
isted in aligning the 3D MPRAGE dataset of each subject with the
stereotactic axes. For this step the location of the anterior commiss-
ure (AC), the posterior commissure (PC) and two rotation parameters
for midsagittal alignment were specified manually in the 3D dataset.
In the second step the extreme points of the cerebrum were speci-
fied. These points together with the AC and PC coordinates were
then used to scale the 3D datasets into the dimensions of the
standard brain of the Talairach and Tournaux atlas (Talairach and
Tournoux, 1988) using a piecewise affine and continuous transfor-
mation. All investigated areas were included in this space.
To transform the functional data into Talairach space, the
preprocessed functional time series were first resampled at a
voxel size of 3⫻3⫻3 mm and coregistered with the corresponding
structural dataset. The coregistration transformation in BrainVoy-
ager QX was determined by concatenating an initial alignment
matrix obtained using the siemens position parameters of the
functional and structural images with a fine tuning alignment matrix
obtained by means of an intensity-driven alignment algorithm. The
alignment between functional and anatomical scans was finally
checked by means of a thorough visual inspection. Then the rigid-
body AC-PC transformation matrix and the piecewise affine Ta-
lairach grid scaling transformation performed for the 3D anatomical
dataset were applied to coregistered functional data. This procedure
resulted in a normalized four-dimensional data representation (vol-
ume time course) for each functional run. In order to avoid quality
loss due to successive data sampling, spatial normalization was
actually performed using a single transformation matrix obtained by
combining the different spatial transformations described.
A random effect-group analysis was performed using the
general linear model (GLM) (Friston et al., 1995) with correction
for temporal autocorrelation (Bullmore et al., 1996; Woolrich et al.,
2001). Two predictors of interest (neutral and negative images
blocks) were considered, whereas “baseline” corresponded to the
cross fixation blocks.
To account for hemodynamic delay, the boxcar waveform
representing the rest and task conditions was convolved with an
empirically founded hemodynamic response function (Boynton et al.,
1996). No spatial or temporal smoothing was performed in this anal-
ysis. For each subject we obtained one Statistical Parametric Map
(SPM), comparing each task with the baseline. Thresholding of SPM
was performed at P⬍0.05 corrected for multiple comparisons by
means of the False Discovery Rate (FDR) (Genovese et al., 2002).
In addition, to search for activated areas showing within and
between group effects, a voxel-wise random effect group analysis
was performed as well. In this analysis a percent signal change
normalization of the time series from each subject was performed.
Group SPMs were obtained by comparing each condition with the
control state (“negative pictures vs baseline” and “neutral pictures
vs baseline”), and by comparing the two conditions directly (neg-
ative pictures vs. neutral pictures); thresholding was performed
M. Brunetti et al. / Neuroscience 168 (2010) 670–679672
using the same method as for the individual maps. Activated areas
were obtained from the group activation maps of the whole brain
analysis considering those voxels showing a significant response
(P⬍0.05 corrected) to any experimental condition. The mean time
course of the fMRI signal from voxels belonging to a given acti-
vated area was analyzed for each subject, and the individual BOLD
responses to the different stimulation conditions were characterized
by the BOLD signal intensity variation in each activated area. The
relative signal variation between rest (cross fixation) and task period
was calculated from the fitted parameters of the GLM: BOLD %
change⫽(beta⫻100)/rest, where beta represents the estimated am-
plitude of the variation of the fMRI signal during task.
Statistical analysis
fMRI data. The regional comparison of activation was un-
dertaken by means of the analysis of variance (ANOVA) for re-
peated measures. The dependent variable of the ANOVA analysis
Fig. 1. (top) Examples of negative (unpleasant) and neutral pages respectively, consisting of five pictures each (SOA⫽3 s, stimulus duration⫽2s)
alternated witha1scontrol state (cross fixation). A total of 36 blocks was presented. The subjects’ task was to observe each picture and to press
as quickly as possible the left button in the case of target detection and the right button in the case of no target detection. Targets were vegetable
elements (e.g. plants, flowers) in the pictures (equally presented in unpleasant and neutral images and corresponding to 25% of the total number of
pictures) (bottom) Scheme of the experimental blocks presentation. Each stimulation block was presented in a randomized order alternated with a
control state (cross fixation). The duration of every block and every control state was 15 s each.
M. Brunetti et al. / Neuroscience 168 (2010) 670–679 673
was the relative variation of the BOLD signal between the task and
rest conditions (BOLD % change). The ANOVA factors were the
following:
1. Group: PTSD subjects and control subjects;
2. Valence: neutral and negative valence of the pictures.
A Duncan post hoc test was performed to evaluate the sta-
tistical significance of the effects.
Finally, as additional analysis, hemispheric dominance was
introduced as further factor for bilaterally activated areas only.
Behavioral data. Response accuracy and reaction times in
the MR scanner were examined by means of an ANOVA. The
ANOVA factors (independent variables) were Group (PTSD subjects
vs. control subjects) as a between-group factor and Valence (neutral
vs. negative valence of the pictures) as a within-subject factor. Sig-
nificant effects were dissected by means of the Duncan post hoc test.
Spearman rank order coefficients were computed to examine the
relationship between behavioral data (reaction time; accuracy) and
the BOLD beta values of brain areas showing statistically significant
Group effects in the mentioned ANOVAs. From a technical point of
view, statistical analysis was performed using Statistica 6.1 software
(Statsoft Italia srl, 2003).
RESULTS
Behavioral data
Responses to negative (unpleasant) pictures were less
accurate and had longer mean RTs compared to re-
sponses to neutral pictures.
Accuracy. A significant effect of valence on accuracy
was observed; both groups performed worse during the ob-
servation of negative pictures [F(1–18)⫽166.7, P⬍0.01].
There was also a slight trend towards a Valence; *Group
interaction [F(1–18)⫽3.28, P⬍0.09], indicating a slightly bet-
ter performance for negative pictures in the PTSD group with
respect to the control group. However, there was no remark-
able difference in accuracy between groups.
Mean RT. A significant effect of Valence on RT was
observed; both groups were slower in responding to neg-
ative pictures [F(1–18)⫽21.6, P⬍0.01]. The mean RTs did
not significantly differ between groups; it was just slightly
slower in the PTSD group than in the control group [F(1–
18)⫽3.31, P⬍0.09]. More detailed information about be-
havioural results is reported in Table 2. Spearman rank
order coefficients were computed to examine the relation-
ships between behavioral data and CAPS score in the
whole sample (PTSD and control subjects). The correla-
tion between CAPS subscale B score (re-experiencing
symptoms) and mean RTs in all subjects was statistically
significant during both neutral and negative pictures. The
higher the score of the CAPS B subscale, the slower the
RT (poor performance). See Table 3 for more details.
fMRI data
Between group effects. A statistically significant fMRI
effect for the factor Group was observed only in the bilat-
Table 2. Behavioural results and statistical analysis
Percentage correct response (accuracy)
Performance variable Negative Neutral
Mean SD Mean SD
Controls 76.0 7.3 94.2 2.8
PTSD 81.2 7.7 94.8 4.5
All subjects 78.6 7.8 94.5 3.7
2 group⫻2 valence ANOVA results
Factor Effect df Error df FP Duncan post hoc test
Group 1 18 1.5 0.23 NS
Valence 1 18 166.7 ⬍0.0001 Neutral⬎Negative
Valence⫻group interaction 1 18 3.28 0.09 NS
Mean reaction time (RT)
Performance variable Negative Neutral
Mean SD Mean SD
Controls 998.0 192.1 879.4 163.1
PTSD 1139.2 178.0 1001.9 158.1
All subjects 1068.6 194.3 940.7 198.5
2 group⫻2 valence ANOVA results
Factor Effect df Error df FP Duncan post hoc test
Group 1 18 3.31 0.09 NS
Valence 1 18 21.6 0.0002 Negative⬎Neutral
Valence⫻group interaction 1 18 0.11 0.73 NS
SD, standard deviation; df, degrees of freedom; NS, not significant.
M. Brunetti et al. / Neuroscience 168 (2010) 670–679674
eral amygdala activation: [F(1–18)⫽11.8, P⬍0.01]. Amyg-
dala activated voxels were selected considering those vox-
els showing a significant response (P⬍0.05 corrected) to
any experimental condition (voxels of the clusters⫽243
mm
2
right and 324 mm
2
left; [Right amygdala: x⫽8, y⫽⫺6,
z⫽⫺8; Left amygdala: x⫽⫺18, y⫽⫺6, z⫽⫺11]). Specifi-
cally, there was a higher BOLD activation of bilateral
amygdala in the PTSD group than in the control group.
This statistical main effect was confirmed when behav-
ioural performance (accuracy and mean RTs) was used as
a covariate [F(1–16)⫽15.379, P⬍0.01]. Group x Valence
interaction was also observed [F(1–18)⫽4.32, P⬍0.05]. A
post hoc Duncan test showed the following results: a sta-
tistically significant difference in BOLD activation in the
amygdala was observed between the PTSD and control
groups during the neutral pictures (P⬍0.01; see Fig. 2).
This was not true during the negative pictures.
No between groups effect was observed in other ROIs.
Correlation data. Since the two groups showed a
significant BOLD difference only in neutral condition, we
performed a correlation analysis on the PTSD and control
subjects as whole group during this condition. Spearman
rank order coefficients were used to examine the relation-
ship between behavioral data and BOLD beta values of
activated areas in the whole sample (PTSD and control
subjects). The correlation between the BOLD % signal
change in the amygdala and the mean RTs in all subjects
is represented in Fig. 3a: a statistically significant effect for
the left amygdala during the vision of neutral pictures is
observable (number of voxels in the left amygdala of the
entire group during neutral condition⫽324 mm
2
); the larger
the amygdala activation, the slower the RTs (P⬍0.05). No
statistically significant correlation was observed during the
viewing of negative pictures (P⬎0.05). Fig. 3b shows the
correlation between the global CAPS score and the BOLD
percentage signal change in the amygdala for the neutral
pictures across the whole sample: a statistically significant
effect for the right and left amygdala was observed. The
higher the score in the global CAPS, the higher the BOLD
in the bilateral amygdala. These results motivated a cor-
relation study for the single CAPS subscales B, C and D.
For the neutral pictures, there was a positive correlation
between the brain BOLD response in the bilateral amyg-
dala and CAPS subscales B (re-experiencing symptoms),
C (avoidance symptoms); and D (hyperarousal symp-
toms). The higher the amygdala activation, the higher the
score to these sub-scales that probe symptom severity. No
statistically significant correlation was observed for the
negative pictures (P⬎0.05) [see Table 3 for more details].
Statistical threshold was not reached when the same cor-
relation analyses were conducted on each group sepa-
rately plausibly due to our small sample size.
Overall results in all subjects as whole group
Whole brain voxel-wise analyses, which was performed on
all subjects as a whole group, revealed a statistically sig-
nificant effect during the different contrasts (P⬍0.05 cor-
rected). Main results are reported in Table 4.
Hemispheric dominance
Additional analysis performed on the two groups, revealed
a statistically significant hemisphere effect during the dif-
ferent contrasts (P⬍0.05 corrected). Main results are re-
ported in Table 5.
DISCUSSION
Previous fMRI evidence has shown that amygdala re-
sponses to emotionally neutral pictures are exaggerated at
group levels in patients with severe PTSD (Hendler et al.,
2003). The present fMRI study tested the hypothesis that
elevated amygdala responses are not only linked to neg-
ative pictures but also to neutral pictures as a function of
disease severity in PTSD patients with mild symptoms and
in individuals who did not develop symptoms. To this end,
an fMRI scan was performed on PTSD patients with mild
symptoms and healthy controls (both victims of a bank
assault) during a visuo-motor task designed to reproduce
an environmental situation that typically involves an auto-
matic emotional response. Indeed, the subjects performed
an attentional task that was not related to the emotional
valence of the pictures (to detect vegetable elements),
while processing a potentially emotional stimulus. In this
condition, emotional reactions are automatic and implicit
since the subjects were not required to pay explicit atten-
tion to the emotional valence of the pictures. In addition,
this task allowed us to evaluate the extent to which the
emotional content of pictures affects performance.
The principal finding of the present study is that abnor-
mal amygdala responses following mild trauma could be
best assessed by examining responses to neutral rather
than negative stimuli. Our results, in fact, indicate a signif-
icantly larger amygdala response to negative pictures
compared to neutral pictures in the control group; on the
Fig. 2. BOLD response in the bilateral amygdala. The ANOVA indi-
cates a Group⫻valence interaction [F(1–18)⫽4.32, p⬍0.05] and Post-
Hoc Duncan test confirms a statistical difference in the BOLD activa-
tion during neutral stimulation between PTSD and control group
(** P⬍0.01). No statistical differences between groups in the BOLD
activation during negative stimulation were found. Activated voxels
were selected considering those voxels showing a significant re-
sponse (P⬍0.05 corrected) to any experimental condition (voxels of
the cluster⫽229; [Right amygdala: x⫽8, y⫽⫺6, z⫽⫺8; Left amygdala:
x⫽⫺18, y⫽⫺6, z⫽⫺11]).
M. Brunetti et al. / Neuroscience 168 (2010) 670–679 675
contrary, the amygdala responses in the PTSD patients
were not significantly different between the two classes of
pictures. Moreover, our data demonstrate that amygdala
responses scale with symptom severity across a range of
subjects that either developed or didn’t develop PTSD, as
supported by the correlation of the PTSD symptoms and
the bilateral amygdala responses to neutral pictures in the
whole sample. In other words, although amygdala BOLD
responses to both negative and neutral images are high in
PTSDs, we only evidenced a relationship between amyg-
dala activity for neutral stimuli and symptom severity. This
result could be interpreted as a hyper activation of this
structure in response to neutral stimuli, as an enhanced
threshold generalized from threatening towards unthreat-
ening information processing. On the other hand, results
may also be explained as a coping strategy for emotional
events: PTSD patients may develop a similar generalized
response to both neutral and negative stimuli.
Another interesting result is the correlation between
BOLD signal in the left amygdala during the viewing of
neutral pictures and reaction times to these pictures (the
higher the amygdala activation, the worse the perfor-
mance). We also expected a correlation between the re-
action time and amygdala activity during the observation of
the negative pictures, but the lack of statistical significance
is probably due to the small sample size. The trend was,
however, in line with our hypothesis of a relationship be-
tween amygdala activity and attention. Further data sup-
porting the influence of PTSD symptoms are supplied by
the correlational data highlighting a relation between per-
formance and PTSD score during both neutral and nega-
tive stimulation. On the other hand, our behavioural results
Fig. 3. (a) Scatterplot of the correlation between BOLD response in the left amygdala of the entire group during neutral condition and the mean
Reaction Time during task performance in the same session (neutral pictures stimulation) [Spearman correlation: n⫽20; Rs⫽0.51, P⬍0.05].
(b) Bilateral amygdala mask obtained from the group activation maps (pooling PTSD and control groups) superimposed on the structural image of one
of the subjects (lower). Scatterplot of correlation between BOLD response in the right and left amygdala in neutral vs baseline contrast and the CAPS
score [Right amygdala Spearman correlation: Rs⫽0.65, P⬍0.01; Left amygdala Spearman correlation: Rs⫽0.71, P⬍0.01]; (upper right and left
respectively).
M. Brunetti et al. / Neuroscience 168 (2010) 670–679676
showed how the emotional content of the pictures influ-
ences the whole sample’s performance. The PTSD pa-
tients showed better accuracy and slower reaction times
with respect to the control subjects when viewing the neg-
ative pictures, although the trend was not statistically sig-
nificant possibly due to the small sample size. This trend
could be explained as an attentional-arousal bias triggered
by amygdala-based pre-attentive threat evaluation sys-
tems implied in the homeostatic regulation of the anxiety
level (Bishop et al., 2007, 2008).
These results complement several lines of previous
evidence showing exaggerated amygdala responses in
anxious people (Liberzon et al., 1999; Rauch et al., 2000),
as a reflection of an abnormal engagement of brain fear
circuitry in which the amygdala would play a crucial role by
triggering emotional memory and behavioral manifesta-
tions of fear (LeDoux, 1996, 2000). In the past, it has been
shown that in PTSD patients, exposure to trauma-related
stimuli (vs. neutral stimuli) or emotionally negative pictures
were associated with a high activation of amygdala and
hyper-arousal (Liberzon et al., 1999; Pissiota et al., 2002;
Lanius et al., 2006; Williams et al., 2006). The same was
true when attention was devoted to non-threatening audi-
tory stimuli in conditions of hyper-vigilance (Bryant et al.,
2005). In a recent study, enhanced amygdala responses
were also related to an arousal-driven emotional enhance-
ment of memory (Sommer et al., 2008). A conclusive neu-
rophysiological model of PTSD is premature at the present
early stage of research, even considering that the group of
patients in the present study is relatively small. It is im-
probable that the present results are affected by the fact
that certain populations show exaggerated amygdala re-
sponse to neutral faces, due to the small percentage (less
than 20%) of recognizable face expression in our selected
pictures. These pictures were distributed across several
blocks of five pictures, thus we would have lost too many
blocks to remove the “face” pictures for a control analysis.
Based on the present results, a future study should ma-
nipulate the variable “face” in the stimulation items.
In line with our results and those in literature, it can be
speculated that pathological fear is characterized by an
enhancement of arousal induced by both emotionally neg-
ative and neutral stimuli in PTSD patients. Instead, a dif-
Table 3. Spearman rank order collection between behavioural data, Right and left Amygdala BOLD percentage signal change during neutral picture
stimulation and CAPS score
Right Amygdala Left Amygdala RT during negative
picture stimulation
RT during neutral picture
stimulation
Global CAPS 0.65 0.71 0.49 0.49
(P⬍0.01) (P⬍0.01) (P⫽0.01) (P⬍0.05)
CAPS B 0.58 0.74 — —
(P⬍0.01) (P⬍0.01)
CAPS C 0.52 0.47 — —
(P⬍0.05) (P⬍0.05)
CAPS D 0.58 0.72 — —
(P⬍0.01) (P⬍0.01)
RT during neutral picture stimulation — 0.51 — —
(P⬍0.05)
RT, reaction time; CAPS, clinician administered PTSD scale.
Table 4. Overall results in all subjects obtained with the voxel-wise
whole brain random-effects analysis: brain areas activated during
neutral and negative visual stimulation in PTSD and control subjects
Brain region Hemisphere Ba Talairach
coordinates
xyz
Neutral condition
Medial prefrontal cortex R 10 21 40 13
Middle frontal gyrus L 10 ⫺32 42 13
Posterior insular cortex L — ⫺41 ⫺11 14
Negative condition
Amygdala R — 8 ⫺68
Amygdala L — ⫺18 ⫺611
Dorso-lateral prefrontal
cortex
R923⫺56 40
Middle frontal\precentral g R 6 52 ⫺10 36
Putamen R — 19 11 11
Visula area (V3) L 19 ⫺33 ⫺81 ⫺8
Middle temporal gyrus R 21 50 ⫺14 ⫺7
Negative Vs Neutral
Middle frontal\precentral g R 6 52 ⫺10 36
Visula area (V3) L 19 ⫺16 ⫺52 1
Occipito-temporal\fusiform
gyrus
R3744⫺69 6
Occipito-temporal\fusiform
gyrus
L37⫺38 ⫺63 ⫺16
Table 5. ANOVA results for “hemisphere” (Right vs. Left) factor ob-
tained with the voxel-wise whole-brain random-effects analysis for
bilaterally actived areas only: main effect in all subjects during nega-
tive visual stimulation and interaction (hemisphere * group) for the
amygdala during negative visual stimulation
Brain region Hemispheric effect
Main effect
Dorso-lateral prefrontal cortex R⬎LP⬍.01
Putamen R⬎LP⬍.01
Visual areas (V1,V2,V3) R⬎LP⬍.01
Fusiform gyrus L⬎RP⬍.05
Interaction (hemisphere * group)
Amygdala (PTSD) R⬎LP⬍.05
Amygdala (controls) L⬎RP⬍.05
M. Brunetti et al. / Neuroscience 168 (2010) 670–679 677
ferent pattern would characterize control subjects experi-
encing trauma but not developing PTSD. This speculation
would explain, in terms of enhanced amygdala responses
to neutral stimuli, previous evidence showing a poor incre-
ment of activation of the amygdala during the observation
of traumatic or emotionally negative stimuli in PTSD with
respect to control subjects (Bremner et al., 1999; Yang et
al., 2004). Unspecific reactivity of the amygdala and the
relative hyper-arousal might be at the basis of the poor
selectivity of information processing during stimulus en-
coding and retrieval processes. The same mechanism
would explain the decline of cognitive performance in the
PTSD patients as revealed by reaction time and the se-
verity of symptoms.
Noteworthy, the present results showed that in our
control subjects, the negative stimuli induced a BOLD
response in the amygdala comparable to that of the base-
line condition, contrary to previous studies that showed
larger amygdala responses in similar experimental condi-
tions (Phan et al., 2006). A tentative explanation might be
the fact that control subjects were bank clerk victims of one
or more armed bank assaults about 10 months before the
experiments. When compared to healthy subjects not ex-
periencing traumatic events, the present control subjects
might be characterized by a sort of adaptive “inhibition” of
the amygdala during the processing of emotional stimuli as
a mechanism to prevent PTSD symptoms. This result is in
line with a recent study showing that amygdala response to
stress-related content is positively correlated with changes
in stress symptoms, implying that the more emotionally
significant the context is for the individual after stress, the
more intensely this region (together with the hippocampus)
responds (Admon et al., 2009). The authors suggest a
casual relation between a previous high reactivity of the
amygdala and an individual tendency to develop stress
symptoms following a stressful life event such as a trauma.
Future research should evaluate (ii) possible differ-
ences in amygdala responses to neutral stimuli between
asymptomatic bank clerk victims and subjects not experi-
encing similar traumatic situations; and (ii) differences in
amygdala response to neutral stimuli between mild and
severe PTSD patients for possible clinical applications.
CONCLUSION
In conclusion, the present fMRI study focused on the grey
zone between asymptomatic bank clerk victims and mild
PTSD patients. In the whole participants’ sample, correla-
tions between the severity of the PTSD symptoms, task
performance, and amygdala activation during the process-
ing of neutral pictures were described. Control subjects
showed larger amygdala responses to emotionally negative
pictures compared to neutral pictures response. On the con-
trary, the amygdala responses of the PTSD patients were
high but not statistically different between the two classes of
stimuli. These results suggest that in the PTSD patients with
mild symptoms, the activation of amygdala and the selectivity
of the emotional response to neutral stimuli are abnormal as
a function of disease severity. This would represent a neural
correlate of heightened generalized “alert” threshold that is
typically observed in these patients, which could represent a
warning strategy of the organism towards a dangerous envi-
ronment after emotional traumas. Future research should test
the hypothesis that such a correlative relationship allows the
use of amygdala response to neutral stimuli as an instrumen-
tal surrogate marker in the clinical and pharmacological re-
search in mild and severe PTSD patients.
Acknowledgments—The authors would like to thank Dott. Gianni
Perrucci, ITAB-University “G. D’Annunzio of Chieti, for technical
support, and Dott. Beth Fairfield, Faculty of Psychology-University
“G. D’Annunzio of Chieti, for a careful editing of the manuscript.
The authors would also like to thank local Bank Labour Union for
collaboration in this research.
REFERENCES
Admon R, Lubin G, Stern O, Rosenberg K, Sela L, Ben-Ami H, Hendler
T (2009) Human vulnerability to stress depends on amygdala’s pre-
disposition and hippocampal plasticity. Proc Natl Acad SciUSA
106(33):14120–14125.
Bishop SJ, Jenkins R, Lawrence AD (2007) Neural processing of
fearful faces: effects of anxiety are gated by perceptual capacity
limitations. Cereb Cortex 17:1595–1603.
Bishop SJ (2008) Neural mechanisms underlying selective attention to
threat. Ann N Y Acad Sci 1129:141–152.
Blake D, Weathers F, Nagy L, Kaloupek D, Charney D, Keane T
(1998) Clinician Administered PTSD Scale (CAPS). Boston, VA:
National Center for Post-Traumatic Stress Disorder, Behavioral
Science Division.
Boynton GM, Engel SA, Glover GH, Heeger DJ (1996) Linear systems
analysis of functional magnetic resonance imaging in human.
J Neurosci 116:4207–4241.
Bremner JD, Narayan M, Staib LH, Southwick SM, McGlashan T,
Charney DS (1999) Neural correlates of memories of childhood
sexual abuse in women with and without posttraumatic stress
disorder. Am J Psychiatry 156:1787–1795.
Britton JC, Phan KL, Taylor SF, Fig LM, Liberzon I (2005) Corticolimbic
blood flow in posttraumatic stress disorder during script-driven
imagery. Biol Psychiatry 57(8):832–840.
Bryant RA, Felmingham KL, Kemp AH, Barton M, Peduto AS, Rennie
C, Gordon E, Williams LM (2005) Neural networks of information
processing in posttraumatic stress disorder: a functional magnetic
resonance imaging study. Biol Psychiatry 58(2):111–118.
Bullmore E, Brammer M, Williams S, Rabe-Hesketh S, Janot N, David
A, Mellers J, Howard R, Sham P (1996) Statistical methods of
estimation and inference for functional MR image analysis. Magn
Reson Med 35:261–277.
Catani C, Adenauer H, Keil J, Aichinger H, Neuner F (2009) Pattern of
cortical activation during processing of aversive stimuli in trauma-
tized survivors of war and torture. Eur Arch Psychiatry Clin Neu-
rosci 259(6):340–351.
Etkin A, Wager TD (2007) Functional neuroimaging of anxiety: a
meta-analysis of emotional processing in PTSD, social anxiety
disorder, and specific phobia. Am J Psychiatry 164:1476–1488.
Friston KJ, Holmes AP, Worsley KJ, Poline JP, Frith CD, Frackowiak
RSJ (1995) Statistical parametric maps in functional imaging: a
general linear approach. Hum Brain Mapp 2:189–210.
Friston KJ, Williams S, Howard R, Frackowiak RSJ, Turner R (1996)
Movement related effects in fMRI time series. Magn Reson Med
35:346–355.
Genovese CR, Lazar NA, Nichols T (2002) Thresholding of statistical
maps in functional neuroimaging using the false discovery rate.
Neuroimage 15(4):870–878.
M. Brunetti et al. / Neuroscience 168 (2010) 670–679678
Gläscher J, Adolphs R (2003) Processing of the arousal of subliminal
and supraliminal emotional stimuli by the human amygdala. J Neu-
rosci 23:10274–10282.
Hajnal JV, Myers R, Oatridge A, Schwieso JE, Young IR, Bydder GM
(1994) Artifacts due to stimulus correlated motion in functional
imaging of the brain. Magn Reson Med 31:283–291.
Hendler T, Rotshtein P, Yeshurun Y, Weizmann T, Kahn I, Ben-Bashat
D, Malach R, Bleich A (2003) Sensing the invisible: differential
sensitivity of visual cortex and amygdala to traumatic context.
Neuroimage 19(3):587–600.
Lang PJ, Bradley MM, Cuthbert B (1999) International Affective Pic-
ture System (IAPS): technical Manual and Affective Ratings. The
Center for Research in Psychophysiology, University of Florida,
Gainesville, FL.
Lanius RA, Bluhm R, Lanius U, Pain C (2006) A review of neuroim-
aging studies in PTSD: heterogeneity of response to symptom
provocation. J Psychiatr Res 40(8):709–729.
LeDoux J (1996) Emotional networks and motor control: a fearful view.
Prog Brain Res 107:437–446.
LeDoux JE (2000) Emotion circuits in the brain. Annu Rev Neurosci
23:155–184.
Liberzon I, Taylor SF, Amdur R, Jung TD, Chamberlain KR,
Minoshima S, Koeppe RA, Fig LM (1999) Brain activation in PTSD
in response to trauma-related stimuli. Biol Psychiatry 45:817–826.
Liddell BJ, Brown KJ, Kemp AH, Barton MJ, Das P, Peduto A, Gordon
E, Williams LM (2005) A direct brainstem-amygdala-cortical
“alarm” system for subliminal signals of fear. Neuroimage 24(1):
235–243.
Oldfield RC (1971) The assessment and analysis of handedness: the
Edinburgh inventory. Neuropsychologia 9:97–113.
Phan KL, Fitzgerald DA, Nathan PJ, Tancer ME (2006) Association
between amygdala hyperactivity to harsh faces and severity of
social anxiety in generalized social phobia. Biol Psychiatry 59:
424–429.
Pissiota A, Frans O, Fernandez M, von Knorring L, Fischer H, Fredrik-
son M (2002) Neurofunctional correlates of posttraumatic stress
disorder: a PET symptom provocation study. Eur Arch Psychiatry
Clin Neurosci 252:68–75.
Rauch SL, Whalen PJ, Shin LM, McInerney SC, Macklin ML, Lasko
NB, Orr SP, Pitman RK (2000) Exaggerated amygdala response to
masked facial stimuli in posttraumatic stress disorder: a functional
MRI study. Biol Psychiatry 47(9):769–776.
Sheehan DV, Lecrubier Y, Janavs J, Knapp E, Weiller E, Bonora LI,
Amorim P, Lépine J-P, Sheehan MF, Baker RR, Sheehan KH
(1994) Mini International Neuropsychiatric Interview MINI. Univer-
sity of South Florida Institute for research in psychiatry, Tampa,
Florida and INSERM-Hôpital de la Salpêtrière, Paris, France.
Shin LM, McNally RJ, Kosslyn SM, Thompson WL, Rauch SL, Alpert
NM, Metzger LJ, Lasko NB, Orr SP, Pitman RK (1999) Regional
cerebral blood flow during scriptdriven imagery in childhood sexual
abuse–related PTSD: a PET investigation. Am J Psychiatry
156:575–584.
Shin LM, Orr SP, Carson MA, Rauch SL, Macklin ML, Lasko NB,
Peters PM, Metzger LJ, Dougherty DD, Cannistraro PA, Alpert NM,
Fischman AJ, Pitman RK (2004) Regional cerebral blood flow in
the amygdala and medial prefrontal cortex during traumatic imag-
ery in male and female Vietnam veterans with PTSD. Arch Gen
Psychiatry 61(2):168–176.
Sommer T, Gläscher J, Moritz S, Büchel C (2008) Emotional enhance-
ment effect of memory: removing the influence of cognitive factors.
Learn Mem 15(8):569–573.
Talairach J, Tournoux P (1988) Co-planar stereotaxic Atlas of the
human brain. New York, NY: Thieme.
van der Kolk BA (1994) The body keeps the score: memory and the
evolving psychobiology of PTSD. Harv Rev Psychiatry 1:253–265.
Williams LM, Kemp AH, Felmingham K, Barton M, Olivieri G, Peduto A,
Gordon E, Bryant RA (2006) Trauma modulates amygdala and
medial prefrontal responses to consciously attended fear. Neuro-
image 29(2):347–357.
Woolrich MW, Ripley BD, Brady M, Smith SM (2001) Temporal auto-
correlation in univariate linear modeling of FMRI data. Neuroimage
14:1370–1386.
Yang P, Wu MT, Hsu CC, Ker JH (2004) Evidence of early neurobio-
logical alternations in adolescents with posttraumatic stress disor-
der: a functional MRI study. Neurosci Lett 370:13–18.
Zlotnick C, Franklin CL, Zimmerman M (2002) Does “subthreshold”
posttraumatic stress disorder have any clinical relevance? Compr
Psychiatry 43(6):413–419.
(Accepted 13 April 2010)
(Available online 21 April 2010)
M. Brunetti et al. / Neuroscience 168 (2010) 670–679 679