Thickness of ventromedial prefrontal cortex in
humans is correlated with extinction memory
Mohammed R. Milad*, Brian T. Quinn†, Roger K. Pitman*, Scott P. Orr*‡, Bruce Fischl†, and Scott L. Rauch*§
*Department of Psychiatry and†Nuclear Magnetic Resonance Center, Massachusetts General Hospital and Harvard Medical School, Charlestown, MA 02129;
and‡Research Service, Veterans Affairs Medical Center, Manchester, NH 03104
Edited by Marcus E. Raichle, Washington University School of Medicine, St. Louis, MO, and approved June 3, 2005 (received for review March 24, 2005)
The ventromedial prefrontal cortex (vmPFC) has been implicated in
fear extinction [Phelps, E. A., Delgado, M. R., Nearing, K. I. &
cortical thickness of vmPFC regions is associated with how well
healthy humans retain their extinction memory a day after having
been conditioned and then extinguished. Fourteen participants
underwent a 2-day fear conditioning and extinction protocol. The
conditioned stimuli (CSs) were pictures of virtual lights, and the
unconditioned stimulus (US) was an electric shock. On day 1,
participants received 5 CS?US pairings (conditioning), followed by
10 CS trials with no US (extinction). On day 2, the CS was presented
alone to test for extinction memory. Skin conductance response
(SCR) was the behavioral index of conditioning and extinction.
which cortical thickness was measured. We performed a vertex-
based analysis across the entire cortical surface and a region-of-
interest analysis of a priori hypothesized territories to measure
cortical thickness and map correlations between this measure and
SCR. We found significant, direct correlation between thickness of
the vmPFC, specifically medial orbitofrontal cortex, and extinction
retention. That is, thicker medial orbitofrontal cortex was associ-
ated with lower SCR to the conditioned stimulus during extinction
recall (i.e., greater extinction memory). These results suggest that
the size of the vmPFC might explain individual differences in the
ability to modulate fear among humans.
cortical thickness ? fear conditioning ? orbitofrontal cortex
stress disorder (PTSD), the Diagnostic and Statistical Manual of
Mental Disorders (1) gives the example of a woman who is raped
in an elevator and subsequently comes to fear all elevators.
Recovery from PTSD entails, among other things, learning not
to fear situations associated with the traumatic event (i.e., to
extinguish conditioned fear responses) (2). It has been reported
that 2 weeks after a rape, 92% of victims met symptom criteria
for PTSD, but 3 months later only 47% did (3). Such individual
differences in recovery from PTSD are likely related to genet-
ically influenced individual differences in fear extinction and its
retention (4). It is quite possible that these differences are
mediated by variance in regional brain structures.
Under a Pavlovian (classical) conditioning model, a once-
neutral conditioned stimulus (CS) (e.g., a light) is paired with an
aversive unconditioned stimulus (US) (e.g., a shock). After a few
pairings, the CS comes to elicit various manifestations of a fear
conditioned response, including freezing in rodents (5) and
increased skin conductance in humans (6). When the CS is then
repeatedly presented in the absence of the shock, the condi-
tioned response is extinguished. There is substantial evidence
indicating that fear extinction results in the formation of a new
memory that coexists with, but opposes, the initial conditioning
memory (7, 8). Under favorable circumstances, the extinction
memory is recalled when the CS is later presented.
xtinction of conditioned fear is of substantial basic and
clinical neuroscientific interest. To describe posttraumatic
Convergent data from the animal literature implicate the
ventral medial prefrontal cortex (vmPFC) in the recall and
expression of extinction memory (5, 9). Lesion and pharmaco-
logical manipulation studies show that the vmPFC is critical for
extinction recall after a delay (10, 11). Activity in single neurons
recorded from the infralimbic (IL) region of the vmPFC (12),
mPFC-evoked potentials (13), and vmPFC metabolism (14) are
all inversely correlated with fear expression during extinction
recall. Furthermore, microstimulation of IL neurons reduces
conditioned freezing in rats that have been conditioned but not
extinguished, simulating a postextinction state (12, 15). Thus, in
rodents, the vmPFC appears to inhibit conditioned fear re-
sponses and mediate extinction recall.
Human neuroimaging studies of fear conditioning have tra-
ditionally focused on acquisition (16). However, two recent
functional MRI studies investigated the neural circuitry of fear
extinction. Gottfried and Dolan (17) showed increased activa-
tion in the vmPFC, including medial orbitofrontal cortex
(mOFC), during extinction of aversive olfactory conditioning.
Phelps et al. (18) found that response in vmPFC during day 2
extinction recall significantly correlated with the success of
extinction training on day 1, as measured by skin conductance
response (SCR), consistent with a role for the vmPFC in the
retention of extinction learning. Extinction of eye-blink condi-
tioning has also been shown to activate the mPFC (19). In
addition, several imaging studies have shown that during differ-
ential conditioning, mPFC activity increases to the CS? (a
stimulus that is not paired with the shock), consistent with the
role of the mPFC in signaling safety (20, 21). Thus, human
(v)mPFC regions appear to play a role similar to that in rodents.
Could the size of the vmPFC explain why some people have
better control of their fear and, therefore, perhaps are more
resilient to emotional trauma? Preliminary data from animals
show that superior extinction memory performance is associated
with larger IL.¶Furthermore, morphometric studies in humans
have shown reduced mPFC volume in PTSD relative to non-
PTSD patients (23). The goal of the present study was to
investigate whether individual differences in fear extinction
relate to the size of the vmPFC in healthy humans. Therefore, we
examined the relationship between brain morphometry and the
results of a 2-day differential conditioning experiment during
which SCRs were measured. The psychophysiological experi-
ment we performed was originally designed to investigate the
effect of context manipulations on extinction retention (24). In
the present study, an automated method (25) was used to
This paper was submitted directly (Track II) to the PNAS office.
Abbreviations: vmPFC, ventromedial prefrontal cortex; CS, conditioned stimulus; US, un-
conditioned stimulus; SCR, skin conductance response; mOFC, medial orbitofrontal cortex;
PTSD, posttraumatic stress disorder; IL, infralimbic; ROI, region of interest; rACC, rostral
anterior cingulate cortex; SC, subcallosal cortex; dACC, dorsal anterior cingulate cortex;
CXs, visual contexts.
§To whom correspondence should be addressed at: Department of Psychiatry, 149 13th
Street, CNY 2618, Charlestown, MA 02129. E-mail: email@example.com.
¶Cintron, B. & Quirk, G. J. (2004) Soc. Neurosci. Abstr. 328, 13.
© 2005 by The National Academy of Sciences of the USA
July 26, 2005 ?
vol. 102 ?
measure cortical thickness from the MRIs. We investigated
three vmPFC a priori regions of interest (ROIs) (based on
literature reviewed): rostral anterior cingulate cortex (rACC),
subcallosal cortex (SC), and mOFC. We hypothesized that the
thickness of one or more of these regions would be positively
dorsal anterior cingulate cortex (dACC), selected as a control
thickness were first performed by using manually defined ROIs.
Subsequently, a vertex-based method was used to map the
Participants. Fourteen healthy volunteers (eight men and six
women), 21–34 years of age and recruited from the local
community by means of an advertisement, participated in this
study. Written informed consent was obtained from all partic-
ipants in accordance with the requirements of the Partners
Health Service Human Research Committee at Massachusetts
Conditioning and Extinction Procedure and Psychophysiological Mea-
sures. The fear conditioning and extinction procedures used in
this study were identical to those described in ref. 24. Digital
photographs of two different rooms (a conference room con-
taining books on a bookshelf and an office containing a com-
puter on a desk) constituted the visual contexts (CXs). Each
room contained a lamp, and the two different colors (blue and
red) of the lighted lampshade constituted the CSs. The selection
of the CS? and CS? colors and the CX? and CX? rooms was
randomly determined and counterbalanced across participants.
Contexts and CSs were displayed on a computer monitor ?2 feet
behind the participants while they were in a mock scanner. The
US was a 500-ms electric shock delivered through electrodes
attached to the second and third fingers of the dominant hand.
The intensity of the shock was previously determined by each
participant to be ‘‘highly annoying but not painful (6).’’
The experimental protocol was administered over 2 days (see
Fig. 1). On day 1, the habituation phase consisted of eight trials,
in which the to-be CS? and to-be CS? (four of each) were
presented in a counterbalanced manner within either the to-be
conditioning context (CX?) or the to-be extinction context
(CX?). The conditioning phase consisted of five CS? and five
CS? trials, all presented within CX?. The US occurred imme-
diately after each CS? offset. The extinction phase was divided
into two subphases: early and late, which were separated by an
?1-min rest period. Each subphase consisted of five CS? and
five CS? trials, all presented within the CX?. No shocks were
delivered during the extinction phase. Note that the shock
electrodes remained attached to the participant’s fingers during
the extinction phase and all subsequent phases of the experi-
ment. On day 2, the recall phase was identical to an extinction
five CS? were presented within the CX?, and again no US was
delivered. In half of the participants, the renewal phase preceded
the recall phase, whereas the order was reversed for the other
half of the participants.
For each trial during the experiment, the context picture was
presented for 18 s: 6 s alone, followed by 12 s in combination with
the CS? or CS?. The mean intertrial interval was 16 s (range,
12–21 s). A SCR score was calculated for each CS trial by
subtracting the mean skin conductance level during the 2 s
immediately before CS onset (during which the context alone
was being presented) from the highest skin conductance level
recorded during the 12-s CS duration. Thus, all SCRs to the CS?
and CS? reported herein reflect changes in the skin conduc-
tance level above and beyond any changes in skin conductance
level produced by the context. Although many studies examining
SCRs have used the first-interval or second-interval response,
limiting our scoring of skin conductance to the first or second
intervals would leave the last 4 s of the CS presentation
unscored. Furthermore, it is not clear that the 12-s CS interval
used in the present study can simply be divided into 6-s first- and
second-response intervals. The present scoring method allows
for the detection of the maximal increase in skin conductance
method in previous, published human psychophysiological re-
search, which has supported its validity (6, 26). Furthermore, we
have previously shown that, by using the exact protocol used
herein, a second-by-second analysis of change in skin conduc-
tance levels during fear conditioning for the CS? and CS? trials
shows peak responsivity occurring at ?4–5 s after CS onset,
remained high throughout the 12 s CS presentation (see Fig. 2
in ref. 24). Each skin conductance response was square-root
transformed before analysis. Unless specified, all data are pre-
sented as means ? SE.
Image Acquisition and Analysis Procedures. Acquisition. A Sonata
1.5-T whole-body high-speed imaging device (Siemens, Iselin,
NJ) was used with a 3-axis head coil. Structural MRI data were
gathered in replicate by using a high-resolution 3D MPRAGE
sequence (repetition time?time to echo?flip angle ? 7.25 ms?3
ms?7°) with an in-plane resolution of 1.3 mm and a slice
thickness of 1 mm.
Measurement of cortical thickness in individual participants. These
methods have been described in detail in refs. 25, 28, and 29 but
are briefly summarized herein. The two structural scans for each
participant were averaged to create a single high signal-to-noise
average volume. The resulting volume was used to segment
cerebral white matter (30) and to estimate the gray?white
interface. Topological defects in the gray?white estimate were
fixed (31), and this gray?white estimate was used as the starting
point for a deformable surface algorithm designed to find the
pial surface with submillimeter precision (25). The entire cortex
in each individual was then visually inspected, and any inaccu-
racies in segmentation were manually corrected. All of these
measurement procedures were carried out by investigators who
were blind to the nature of the hypotheses to be tested and blind
to the corresponding SCR data.
For each participant, thickness measures across the cortex
were computed by finding the point on the gray?white boundary
surface that was closest to a given point on the estimated pial
surface (and vice versa) and averaging between these two values
(25). The accuracy of the thickness measures derived from this
technique has been validated by direct comparisons with manual
measures on postmortem brains (28) as well as direct compar-
isons with manual measures on MRI data (29).
Inflation, registration, and interparticipant averaging. The surface rep-
resenting the gray?white border was ‘‘inflated’’ (32), differences
Schematic of the experimental protocol. (Adapted from ref. 24).
Milad et al.
July 26, 2005 ?
vol. 102 ?
no. 30 ?
between individuals in the depth of gyri?sulci were normalized,
and each participant’s reconstructed brain was then morphed?
registered to an average spherical surface representation that
optimally aligned sulcal and gyral features across participants
(32, 33). This spherical morphing procedure was used to con-
struct the cortical thickness difference brain maps.
Computation of mean and statistical cortical thickness difference maps.
The spherical transform was used to map the thickness mea-
surements at each vertex on each participant’s cortical surface
into a common spherical coordinate system (32, 33). The data
were smoothed on the surface tessellation by using an iterative
nearest-neighbor averaging procedure (100 iterations were
applied, equivalent to applying a 2D Gaussian smoothing ker-
nel along the cortical surface with a full width half maximum of
Defining boundaries of ROIs. Boundary definition of the ROIs
selected in the present study was performed as described in ref.
34. Briefly, the dACC and the rACC were divided by a coronal
plane at the genu of the corpus callosum, corresponding to plane
I of the methodology of Caviness et al. (35). More precisely,
plane I contributed the anterior border of the dACC, as well as
the anterior border of the SC territory. The mOFC spanned the
interval between the SC and the frontal pole; the posterior
boundary of the frontal pole marked the anterior border of the
Statistical analysis. Interparticipant group averaging and inference
was performed by using the spherical registration to align the
surfaces, smoothing the surfaces for 100 iterations, and fitting a
general linear model at each vertex of the surface. The model
consisted of % extinction retention in the CX? (recall) and
CX? (renewal) contexts and cortical thickness. Each phase
measurement was analyzed separately from the other because
they were hypothesized to have no interaction.
and Renewal. The full results of the psychophysiology experiment
are reported in ref. 24. For illustrative purposes, the mean SCRs
to the five CS? presentations for each phase in the 14 partici-
pants who also underwent MRI are presented in Fig. 2A.
For the purpose of morphometric correlation, the % extinc-
tion retention index was calculated as follows: each participant’s
SCR to the first CS? trial during the recall and renewal phases
was divided by the highest SCR to a CS? trial during condi-
tioning. This was then multiplied by 100 to yield a % of CR
recovered. In turn, the % CR recovered was subtracted from
100% to yield an extinction retention index. Fig. 2B shows that
71% extinction retention was observed in the CX? during the
recall phase, and 20% extinction retention was observed in the
CX? during the renewal phase, thereby supporting the context
dependency of extinction retention (see ref. 24).
Correlational Analyses of ROIs. We hypothesized that the cortical
thickness of our ROIs would be positively correlated with the
extinction retention index. Because only a positive correlation
was predicted, one-tailed tests were used. Participants under-
went MRI to obtain structural images from which cortical
thickness was measured. The boundaries of the ROIs were
defined as described in ref. 34 and are shown in Fig. 3A. The
average cortical thickness for each ROI was then correlated with
(recall phase), the extinction retention index was significantly
correlated with the cortical thickness of mOFC (r ? 0.52; P ?
conditioning; CX?, conditioning context; CX?, extinction context. (B) Extinction retention index (see text for definition) in each context. CX- corresponding to
recall phase and CX? corresponding to renewal phase.*, P ? 0.05.
Illustration of psychophysiological data. (A) Mean SCRs collapsed across trials for each experimental phase for the CS?. Hbt, habituation; Cond,
cortex; mOFC, medial orbitofrontal cortex; rACC, rostral anterior cingulate cortex; dACC, dorsal anterior cingulate cortex. (B) Regression plots for extinction
retention index separately in CX? and CX? versus cortical thickness in mOFC.
Illustration of brain regions of interest and correlational plots between mOFC thickness and extinction retention. (A) Boundaries of ROI: SC, subcallosal
www.pnas.org?cgi?doi?10.1073?pnas.0502441102 Milad et al.
0.028) but not with SC (r ? 0.18; P ? 0.27), rACC (r ? 0.16; P ?
0.29), or dACC (r ? 0.32; P ? 0.13). In the CX? context
(renewal phase), the extinction retention index was again sig-
nificantly correlated with the cortical thickness of mOFC (r ?
0.59; P ? 0.012) but not SC (r ? 0.14; P ? 0.32), rACC (r ? 0.10;
P ? 0.37), or dACC (r ? 0.11; P ? 0.34). The regression plots
for the mOFC correlations are shown in Fig. 3B.
Automated Vertex-Based Analyses. We also used recently devel-
oped automated methods to measure thickness across the
entire cortical surface and to map correlations between this
measure and extinction retention index. Vertex-based, corre-
lational maps across the entire cortical surfaces depicting the
topography of significant correlations between cortical thick-
ness and extinction retention index at a statistical threshold of
P ? 0.001 (uncorrected, two-tailed) are shown in Fig. 4A Left
for CX? and Fig. 4B Left for CX?. For the CX?, the
scatterplot for the significant correlations between percent
extinction retention and cortical thickness in the mOFC
portion of the vmPFC [(peak vertex: r ? 0.82, P ? 0.0003;
Talairach coordinates 4, 30, ?12); ref. 22] is presented in Fig.
4A Upper Right. For the CX?, the scatterplot for the signif-
icant correlations between extinction retention index and
cortical thickness in the mOFC portion of the vmPFC (peak
vertex: r ? 0.80, P ? 0.0006; Talairach coordinates 6, 28, and
?15) is presented in Fig. 4B Lower Right. Note that the average
r value for the entire volume displaying significant correlation
was ?0.75 in both contexts. Beyond the a priori search
territories, only one additional region in the entire brain,
namely the left superior parietal cortex, exhibited a correlation
with extinction retention (P ? 0.0006) that is of comparable
statistical magnitude to those observed within our a priori
ROIs, and this was in the CX? only (region is outlined in Fig.
4B). Thus, this unbiased, post hoc approach supported the
specificity of our ROI findings to the vmPFC.
To rule out the possibility that the correlations we observed
performed correlational analyses between the mOFC cortical
thickness and SCRs to the CS? during extinction recall and
renewal. We found that responses to the CS? did not correlate
with the cortical thickness of the mOFC (r ? 0.40, P ? 0.15 and
r ? ?0.28, P ? 0.32 in the CX? and CX?, respectively). In
addition, to more specifically test whether the correlations
described in Fig. 4 are due to associative processes, SCRs to the
CS? were subtracted from those to the CS?. These differential
responses were in turn correlated with the cortical thickness of
the mOFC. Once more, we found significant correlations be-
tween vmPFC cortical thickness and the differential SCRs in the
CX? (r ? 0.65, P ? 0.011) and in the CX? (r ? 0.57, P ? 0.035).
Thus, the correlations between the extinction retention index
and mOFC cortical thickness most likely reflect associative
processes mediated during and?or after extinction training. Note
that no correlations were observed between mOFC cortical
thickness and any other phase of the experiment [acquisition of
fear conditioning or extinction training on day 1 (data not
The results show that participants with greater vmPFC thickness,
specifically thicker right mOFC, exhibited smaller SCRs to a
correlations between percent extinction retention and cortical thickness in the vmPFC. Threshold is set at P ? 0.01 (dark blue) to P ? 0.001 (cyan blue). Red areas
show incidental negative correlations. Arrow indicates vmPFC region, whereas circle indicates superior parietal correlation.
Regions with positive correlations between cortical thickness and extinction retention in CX? (A) as well as CX? (B) and regression plots for the
Milad et al.
July 26, 2005 ?
vol. 102 ?
no. 30 ?
showed better extinction retention). Not only do these findings
support the role of the vmPFC in the recall of extinction
memory, they also provide evidence that the cortical thickness
of the vmPFC, as one index of size, may explain a substantial
proportion of the individual differences observed in extinction
retention in humans. The implied comparable role for vmPFC in
the extinction context (CX?, recall phase) as well as in the
original conditioning context (CX?, renewal phase) suggests
that brain regions other than vmPFC are involved in the con-
textual gating of extinction retention.
The correspondence between the structural vmPFC region
that correlated with extinction retention in the present study,
namely, the mOFC, and brain regions activated during func-
tional extinction retention testing in humans, as well as in
rodents, is striking. As noted, Phelps et al. (18) have recently
shown that conditioned fear responses during extinction recall
(as measured by skin conductance) were inversely correlated
with CS?-induced activity in the vmPFC in healthy humans,
indicating a direct relationship between vmPFC function and
level of extinction expressed during recall. The Talairach
coordinates of the vmPFC region shown to be functionally
correlated with extinction recall in the study of Phelps et al. [4,
31, ?6] are remarkably close to those shown to be structurally
correlated with extinction retention in the recall and renewal
phases of the present study [4, 30, ?12 and 6, 28, ?15,
respectively]. It is important to point out that, in the present
study, extinction memory was expressed best in the context in
which extinction learning took place, whereas some extinction
memory was still manifested in the renewal context. Thus,
whereas the overall average of responses in the extinction
context was lower relative to those in the conditioning context,
we did observe correlations between the variance in responses
to the CS? in both contexts. Phelps et al. did not manipulate
context during the test for extinction retention, which there-
fore may suggest that indeed the best comparison between our
data and Phelps’s study is the renewal phase. However, the
coordinates for both loci of correlations were very similar.
Note that these coordinates correspond to the peak vertex
correlation, and that the location of the entire cluster is very
comparable for both tests and also very comparable to that
identified by Phelps et al. Other functional MRI studies have
also shown activation of mOFC during extinction (17, 19, 20).
As previously suggested (9), the IL region of the rat appears
to be homologous to the human vmPFC. It has been shown that
IL activity is inversely correlated with fear responses during
extinction testing after a delay, suggesting that IL is involved
in the recall of extinction memory (12). Recent preliminary
data from rodents show that rats have a high behavioral
variability in their recall of extinction memory 24 h after
training, even when their brain function and structure are not
manipulated¶. Interestingly, measures of IL in these animals
indicated that during extinction recall, rats with larger IL
cross-sectional area exhibited less freezing behavior. Thus, it
appears that the size of the vmPFC may influence or predict
how well fear is inhibited during extinction recall in both
rodents and humans.
The thickness of the rACC, which was one of our a priori
vmPFC brain regions, did not correlate with the recall of
extinction memory. The rACC is thought to be homologous to
the prelimbic (PL) region in rats (36). Phelps et al. (18) did not
report any association between functional activity in this brain
region and extinction recall in humans. Interestingly, in rats,
neural activity in PL also was not correlated with extinction
recall (12). Moreover, whereas IL stimulation reduced condi-
tioned freezing, PL stimulation did not (12, 15). Barrett et al.
(14) showed that glucose metabolism in PL did not correlate
with conditioned freezing during extinction recall in mice.
Thus, although the rACC is consistently implicated in emo-
tional processing in humans, and its dysfunction is linked to
various anxiety disorders (37, 38), this region does not appear
to be involved in the recall of extinction memory. However, the
rACC may be involved in other cognitive components of fear
extinction (39). It was surprising that the thickness of the SC,
another a priori vmPFC region, did not correlate with extinc-
tion retention. Although SC has been suggested to be a human
cortical region that is homologous to that involved in fear
extinction in animals, some anatomical studies suggest that the
neighboring, ventral region of mPFC corresponding to the
mOFC region in the present study bears an even greater
homology to the IL in rats (40).
The involvement of the vmPFC in mediation of autonomic
responses is well documented in rodents as well as in humans.
For example, stimulation of the vmPFC in rats dampened
amygdala-induced increases in blood pressure and defensive
behavior (41). Human neuroimaging studies have shown that
mPFC areas are functionally associated with skin conductance
(42). Nagai et al. (43) have recently shown that activity of the
vmPFC was inversely correlated with skin conductance levels
during a biofeedback relaxation task. It is important to note that
an association between vmPFC and skin conductance per se
cannot account for the correlations we observed between cor-
tical thickness and the index of extinction retention that we used,
because that index is adjusted for individual differences in SC
level. Specifically the extinction retention index during recall (or
renewal) was calculated based on a percentage of skin conduc-
tance conditioned response during conditioning (acquisition).
The proposition that the size of a given brain region may partly
explain the behavioral variance in a particular task is appealing.
of neurons (44) or an increase in the neuropil (45). Alternatively,
an increase in cortical thickness may be due to an increased
number of glial cells, because recent studies have shown that
cortical volume reduction observed in schizophrenic patients in
the PFC is not due to the reduction of the total number of
neurons but rather to the loss of glia (46). Any of these cellular
differences in a brain region might better implement a function
that the region subserves, e.g., in the case of the vmPFC,
dampening the output of other brain regions involved in con-
ditioned fear expression, such as the amygdala. However, it
remains to be established whether in fact there is any corre-
spondence between cortical thickness and increased (or de-
creased) brain function.
Several studies have suggested that larger brain regions reflect a
better behavioral outcome. For example, a recent study has shown
that a larger hippocampus is correlated with better recall of
show that the decrease in the capacity to recall verbal memory is
associated with smaller hippocampal volumes (48). There is con-
of cortical and subcortical regions and a variety of mental disorders
(49). However, larger cortical volumes have also been associated
with some mental disorders. For example, increased cortical thick-
ness in insula and rACC has been observed in patients with animal
phobia (50). Larger orbitofrontal cortex has been found to be
associated with decreased performance of working memory (51).
be compensatory (50).
The data presented herein suggest that the size of the
vmPFC might explain individual differences in the ability to
modulate fear among humans. Variability in the thickness of
the vmPFC across the human population may account for risk
(or resilience) factors for anxiety disorders. It is therefore of
substantial interest that vmPFC appears to be functionally
deficient in PTSD (reviewed in ref. 52). Specifically, neuro-
imaging studies have shown that when exposed to trauma-
www.pnas.org?cgi?doi?10.1073?pnas.0502441102Milad et al.
related cues, PTSD patients show a relative failure of activa-
tion in the mPFC (37, 38, 53, 54, 55). These neuroimaging
studies support the hypothesis that dysfunctional mPFC activ-
ity may underlie the exaggerated fear responses commonly
observed in PTSD. It is also important to note that PTSD
patients have been shown to be deficient in behavioral extinc-
tion (6). A recent morphometric study found reduced vmPFC
volume in persons with PTSD compared with trauma-exposed
persons without PTSD (34). Decreased PFC cortical volumes
have also been observed in panic disorder (56) and obsessive–
compulsive disorder (27).
Recall of extinction learning is also likely to be germane to
the success of behavioral therapies, which are theoretically
extinction-based. Thus, future studies should investigate
whether measures of cortical thickness within the vmPFC
predict therapeutic response to behavioral therapies for anx-
We thank Dr. Gregory J. Quirk for helpful comments on the manuscript
and Michelle Wedig for technical assistance. The work was supported in
part by a grant from the National Institute of Mental Health (to S.L.R.)
and the Massachusetts General Hospital Tosteson Fellowship (to
M.R.M.). In addition, support for this research was provided in part by
the National Center for Research Resources, the National Institute for
Biomedical Imaging and Bioengineering, and the Mental Illness and
Neuroscience Discovery Institute (to B.F.).
1. American Psychiatric Association (1994) Diagnostic and Statistical Manual of
Mental Disorders (American Psychiatric Association, Washington, DC), 4th Ed.
2. Rothbaum, B. O. & Davis, M. (2003) Ann. N.Y. Acad. Sci. 1008, 112–121.
3. Foa, E. B. (1997) Ann. N.Y. Acad. Sci. 821, 410–424.
4. Fredrikson, M., Annas, P., Georgiades, A., Hursti, T. & Tersman, Z. (1993)
Biol. Psychol. 35, 153–163.
5. Maren, S. & Quirk, G. J. (2004) Nat. Rev. Neurosci. 5, 844–852.
6. Orr, S. P., Metzger, L. J., Lasko, N. B., Macklin, M. L., Peri, T. & Pitman, R. K.
(2000) J. Abnorm. Psychol. 109, 290–298.
7. Quirk, G. J. (2004) Learn. Mem. 11, 125–126.
8. Myers, K. M. & Davis, M. (2002) Neuron 36, 567–584.
9. Milad, M. R., Rauch, S. L., Pitman, R. K. & Quirk, G. J. (2005) Biol. Psychol.,
10. Lebron, K., Milad, M. R. & Quirk, G. J. (2004) Learn. Mem. 11, 544–548.
11. Morgan, M. A., Romanski, L. M. & Ledoux, J. E. (1993) Neurosci. Lett. 163,
12. Milad, M. R. & Quirk, G. J. (2002) Nature 420, 70–74.
13. Herry, C. & Garcia, R. (2002) J. Neurosci. 22, 577–583.
14. Barrett, D., Shumake, J., Jones, D. & Gonzalez-Lima, F. (2003) J. Neurosci. 23,
15. Milad, M. R., Vidal-Gonzalez, I. & Quirk, G. J. (2004) Behav. Neurosci. 118,
16. Knight, D. C., Smith, C. N., Cheng, D. T., Stein, E. A. & Helmstetter, F. J.
(2004) Cognit. Affect. Behav. Neurosci. 4, 317–325.
17. Gottfried, J. A. & Dolan, R. J. (2004) Nat. Neurosci. 7, 1144–1152.
18. Phelps, E. A., Delgado, M. R., Nearing, K. I. & Ledoux, J. E. (2004) Neuron
19. Molchan, S. E., Sunderland, T., McIntosh, A. R., Herscovitch, P. & Schreurs,
B. G. (1994) Proc. Natl. Acad. Sci. USA 91, 8122–8126.
20. Morris, J. S. & Dolan, R. J. (2004) NeuroImage 22, 372–380.
21. McIntosh, A. R., Rajah, M. N. & Lobaugh, N. J. (1999) Science 284, 1531–1533.
22. Talairach, J. & Tournoux, P. (1988) A Co-Planar Stereotaxic Atlas of a Human
Brain (Thieme, New York).
M. A., Caviness, V. S., Jr., & Kennedy, D. N. (2003) Psychiatry Res. 122, 13–19.
24. Milad, M. R., Orr, S. P., Pitman, R. K. & Rauch, S. L. (2005) Psychophysiology
25. Fischl, B. & Dale, A. M. (2000) Proc. Natl. Acad. Sci. USA 97, 11050–11055.
26. Pitman, R. K. & Orr, S. P. (1986) J. Abnorm. Psychol. 95, 208–213.
27. Szeszko, P. R., Robinson, D., Alvir, J. M., Bilder, R. M., Lencz, T., Ashtari, M.,
Wu, H. & Bogerts, B. (1999) Arch. Gen. Psychiatry 56, 913–919.
28. Rosas, H. D., Liu, A. K., Hersch, S., Glessner, M., Ferrante, R. J., Salat, D. H.,
van der, K. A., Jenkins, B. G., Dale, A. M. & Fischl, B. (2002) Neurology 58,
29. Kuperberg, G. R., Broome, M. R., McGuire, P. K., David, A. S., Eddy, M.,
Ozawa, F., Goff, D., West, W. C., Williams, S. C., van der Kouwe, A. J., et al.
(2003) Arch. Gen. Psychiatry 60, 878–888.
30. Dale, A. M., Fischl, B. & Sereno, M. I. (1999) NeuroImage 9, 179–194.
31. Fischl, B., Liu, A. & Dale, A. M. (2001) IEEE Trans. Med. Imaging 20, 70–80.
32. Fischl, B., Sereno, M. I. & Dale, A. M. (1999) NeuroImage 9, 195–207.
33. Fischl, B., Sereno, M. I., Tootell, R. B. & Dale, A. M. (1999) Hum. Brain Mapp.
34. Rauch, S. L., Shin, L. M., Segal, E., Pitman, R. K., Carson, M. A., McMullin,
K., Whalen, P. J. & Makris, N. (2003) NeuroReport 14, 913–916.
35. Caviness, V. S., Jr., Kennedy, D. N., Richelme, C., Rademacher, J. & Filipek,
P. A. (1996) Cereb. Cortex 6, 726–736.
36. Ongur, D. & Price, J. L. (2000) Cereb. Cortex 10, 206–219.
37. Bremner, J. D., Vermetten, E., Vythilingam, M., Afzal, N., Schmahl, C.,
Elzinga, B. & Charney, D. S. (2004) Biol. Psychiatry 55, 612–620.
Orr, S. P., McInerney, S. C. & Rauch, S. L. (2001) Biol. Psychiatry 50, 932–942.
39. Herry, C. & Garcia, R. (2003) Behav. Brain Res. 146, 89–96.
40. Ongur, D., Ferry, A. T. & Price, J. L. (2003) J. Comp. Neurol. 460, 425–449.
41. al Maskati, H. A. & Zbrozyna, A. W. (1989) J. Auton. Nerv. Syst. 28, 117–125.
42. Patterson, J. C., Ungerleider, L. G. & Bandettini, P. A. (2002) NeuroImage 17,
43. Nagai, Y., Critchley, H. D., Featherstone, E., Trimble, M. R. & Dolan, R. J.
(2004) NeuroImage 22, 243–251.
44. Pakkenberg, B. & Gundersen, H. J. (1997) J. Comp. Neurol. 384, 312–320.
45. Bourgeois, J. P., Goldman-Rakic, P. S. & Rakic, P. (1994) Cereb. Cortex 4,
46. Stark, A. K., Uylings, H. B., Sanz-Arigita, E. & Pakkenberg, B. (2004) Am. J.
Psychiatry 161, 882–888.
47. Walhovd, K. B., Fjell, A. M., Reinvang, I., Lundervold, A., Fischl, B., Quinn,
B. T. & Dale, A. M. (2004) Neurology 63, 1193–1197.
48. Hackert, V. H., den Heijer, T., Oudkerk, M., Koudstaal, P. J., Hofman, A. &
Breteler, M. M. (2002) NeuroImage 17, 1365–1372.
49. Sheline, Y. I. (2003) Biol. Psychiatry 54, 338–352.
50. Rauch, S. L., Wright, C. I., Martis, B., Busa, E., McMullin, K. G., Shin, L. M.,
Dale, A. M. & Fischl, B. (2004) Biol. Psychiatry 55, 946–952.
51. Salat, D. H., Kaye, J. A. & Janowsky, J. S. (2002) Cereb. Cortex 12, 494–505.
52. Bremner, J. D. (2003) Psychopharmacol. Bull. 37, 6–25.
Peters, P. M., Metzger, L. J., Dougherty, D. D., Cannistraro, P. A., et al. (2004)
Arch. Gen. Psychiatry 61, 168–176.
54. Bremner, J. D., Staib, L. H., Kaloupek, D., Southwick, S. M., Soufer, R. &
Charney, D. S. (1999) Biol. Psychiatry 45, 806–816.
55. Lanius, R. A., Williamson, P. C., Boksman, K., Densmore, M., Gupta, M.,
Neufeld, R. W., Gati, J. S. & Menon, R. S. (2002) Biol. Psychiatry 52,
56. Vythilingam, M., Anderson, E. R., Goddard, A., Woods, S. W., Staib, L. H.,
Charney, D. S. & Bremner, J. D. (2000) Psychiatry Res. 99, 75–82.
Milad et al.
July 26, 2005 ?
vol. 102 ?
no. 30 ?