Neural circuits underlying the pathophysiology of mood disorders

Review
Special Issue: Cognition in Neuropsychiatric Disorders
Neural circuits underlying the
pathophysiology of mood disorders
Joseph L. Price1 and Wayne C. Drevets2
1 Department of Anatomy and Neurobiology Washington University School of Medicine St. Louis MO 63110, USA
2 Laureate Institute for Brain Research, and the Department of Psychiatry, University of Oklahoma College of Medicine, Tulsa, OK,
74136, USA
Although mood disorders constitute leading causes of
and other parts of the brainstem. A large body of human
disability, until recently little was known about their
data from functional and structural imaging, as well as
pathogenesis. The delineation of anatomical networks
analysis of lesions and histological material, indicates that
that support emotional behavior (mainly derived from
this system is centrally involved in mood disorders.
animal studies) and the development of neuroimaging
In this review we discuss the neuroanatomy of the
technologies that allow in vivo characterization of anat-
neural circuits implicated in mood disorders, synthesizing
omy, physiology, and neurochemistry in human subjects
findings from studies of non-human primates and observa-
with mood disorders have enabled significant advances
tions in humans, largely taken from clinical studies. The
towards elucidating the pathophysiology of major de-
results of these studies, conducted using neuroimaging,
pressive disorder (MDD) and bipolar disorder (BD). In this
lesion analysis, and post mortem methodologies, support
review, we integrate insights from human and animal
models in which the pathophysiology of depression
studies, which collectively suggest that MDD and BD
involves dysfunction in an extended network involving
involve dysfunction within an extended network includ-
the mPFC and anatomically-related limbic, striatal, tha-
ing the medial prefrontal cortex and anatomically-relat-
lamic and basal forebrain structures. The abnormalities of
ed
limbic,
striatal,
thalamic
and
basal
forebrain
structure and function evident within the extended `viscer-
structures.
omotor' network putatively impair this network's roles in
cognitive processes such as reward learning and autobio-
Animal and human studies of mood disorders
graphical memory, and may dysregulate visceral, behav-
Major depressive disorder (MDD) and bipolar disorder
ioral and cognitive responses to emotional stimuli and
(BD) constitute the first and fifth leading causes of years
stress [2], potentially accounting for the disturbances man-
lived with disability, respectively [1]. Yet, until recently,
ifest within these domains in mood disorders.
little was known about their pathogenesis, as these con-
ditions are not associated with gross brain pathology or
Cognitive and emotional disturbances in MDD and BD
clear animal models for spontaneous recurrent mood epi-
The clinical phenomenology of major depression implicates
sodes. The development of neuroimaging technologies that
brain systems involved in the regulation of mood, anxiety,
allow in vivo characterization of anatomy, physiology, and
fear (e.g., panic attacks, phobias and post-traumatic stress
neurochemistry in human subjects with mood disorders
syndromes commonly occur co-morbidly with depression),
has enabled significant advances toward elucidating their
reward processing, attention, motivation, stress responses,
pathophysiology. Crucially, the interpretation of the ab-
social interaction, and neurovegetative function (i.e., sleep,
normalities found using these technologies has depended
appetite, energy, weight, libido) [1]. In BD, episodes of
upon by the concomitant delineation of anatomical net-
depression occur alternately with manic or hypomanic
works that support emotional behavior.
episodes, during which mood can become euphoric and
Early studies identified the amygdala, hippocampus,
labile, motivated and reward seeking behavior increases,
and other parts of what was termed the `limbic' system
and psychomotor activity and self-esteem become elevated.
as central parts of the emotional brain. Beginning in the
Thus, the same domains are implicated in depression and
1970s and 1980s and continuing through the last 15 years,
mania, although the characteristic disturbance appears
neuroanatomical techniques based on axonal transport
opposite with respect to emotional valence.
have been applied extensively to the limbic system and
Pathological changes in hedonic capacity and motiva-
prefrontal cortex of monkeys. With these methods, a sys-
tion figure prominently in the clinical phenomenology of
tem has been described that links the medial prefrontal
depression, as well. For example, either depressed mood or
cortex (mPFC) and a few related cortical areas to the
anhedonia can manifest as the cardinal mood symptom
amygdala, the ventral striatum and pallidum, the medial
required to establish the diagnosis of a major depressive
thalamus, the hypothalamus, and the periaqueductal gray
episode (MDE) [1]. The term `anhedonia' was initially
described as the inability to experience pleasure [3]. Since
pleasure is a complex construct, however, the Diagnostic
Corresponding authors: Price, J.L. (pricej@wustl.edu); Drevets, W.C.
(wdrevets@laureateinstitute.org).
and Statistical Manual of Mental Disorders IV (DSM-IV)
1364-6613/$ - see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.tics.2011.12.011 Trends in Cognitive Sciences, January 2012, Vol. 16, No. 1
61

---

Review
Trends in Cognitive Sciences January 2012, Vol. 16, No. 1
[1] operationally defines anhedonia as diminished interest
depressed BD subjects, and in the unaffected twins and
or pleasure in response to stimuli that were perceived as
non-twin siblings of BD subjects (reviewed in [12]). Simi-
rewarding during the premorbid state. Between one-third
larly, autobiographical memory retrieval is impaired in
and one-half of individuals diagnosed with MDD experi-
unmedicated MDD subjects, irrespective of their current
ence clinically significant anhedonia (see [3], for a review).
mood state, insofar as they generate memories that are
Empirical work has emphasized hedonic experience in
over-general, particularly when asked to generate specific
depression, whereas studies of motivation are relatively
memories to emotionally positive cue words (e.g., [13]).
absent. Some researchers found that individuals with de-
This latter finding is consistent with evidence that
pression rate positively-valenced stimuli as being less posi-
depressed subjects exhibit a mood-congruent processing
tive, less arousing, or less able to affect their mood versus
bias, defined as a tendency to bias stimulus processing
controls, although a larger number of studies reported no
towards negative information as compared to positive or
group differences in these ratings (reviewed in [3]). A meta-
neutral information [14-16]. Within the context of atten-
analysis of studies that measured physiological or subjective
tion or memory studies, depressed individuals bias stimu-
affective responses found that depression was associated
lus processing towards sad information, as evidenced by
with blunted reactivity to both positively and negatively-
enhanced recall for negatively versus positively valenced
valenced stimuli [4], suggesting that part of the decline in
information on memory tests [16,17], greater interference
hedonic responses may be due to a generalized affective
from depression-related negative words versus happy or
blunting. Nevertheless, in studies using the `sweet taste
neutral words on emotional stroop tasks [18,19], faster
test', individuals with depression do not differ from controls
responses to sad versus happy words on affective attention
in reported hedonic impact, suggesting that depression is
shifting tasks [15,20], preferential attentiveness to faces
not associated with a deficit in the capacity to feel pleasure
with sad versus neutral expressions on face dot-probe tasks
at the level of basic sensory experience [3].
[21,22], and more negative interpretation of ambiguous
Studies using reinforcement paradigms to explore an-
words or situations (reviewed in [12,23]). Depressed indi-
hedonia in MDD report that individuals with depressive
viduals also show oversensitivity to negative feedback in
symptoms fail to develop a response bias towards rewarded
probabilistic reward tasks [10,11].
stimuli [5-7], providing evidence for insensitivity to re-
ward-relevant information. It is unclear whether this defi-
The neural substrates of mood disorders
cit is driven by reduced hedonic capacity, diminished
Observations in experimental animals
motivation, or both. One study that compared ratings of
There is considerable evidence that the amygdala and re-
experienced emotion in depressed MDD, remitted MDD
lated medial prefrontal cortical areas are centrally involved
and healthy controls across four conditions (anticipating
in
[
(
F
i
g
u
r
e
_
1
)
T
D
$
I
G
] mood disorders, based on imaging and histopathological
monetary rewards, anticipating an unpleasant sensory
stimulus, no change, and avoiding an unpleasant sensory
Human
Dorsal prefrontal system
stimulus) demonstrated that a deficit in experienced emo-
Medial prefrontal network
Orbital prefrontal network
tion was specific to reward anticipation [8].
10p
Cognitive and neuropsychological impairments are
10o
characteristic of major depression, and are reflected in
11m
the diagnostic criteria for an MDE as `an impaired ability
11I
to think or concentrate' [1]. The literature is in disagree-
47r
9
ment, however, regarding the specific nature of cognitive
13b
45
47I
symptoms in depressed patients. Some studies report wide
14r
47m
pgACC
32h
24
ranging deficits that include impairments in early infor-
13m
13I
sgACC
mation processing, attention, memory, and executive func-
10p
25
24sg
47s 1ai
10mc
tions, whereas other studies fail to identify such deficits
32m
10mr
1am
[9,10]. Factors that likely contribute to the discrepancies
1apm
14c
across studies include heterogeneity within patient sam-
14r
11m
ples and medication status. For example, impairments of
Dorsal PFC
spatial recognition memory and delayed matching to sam-
Medial network
Monkey
9
Orbital network PrCO
ple tasks have been reported in medicated subjects with
24c
G
24b
12o
MDD and BD but such impairments generally have not
1ai
12m
1ai
24a
13I
10m
12r
been evident in unmedicated samples with MDD or BD
32
1apm
11I
13m 1am
(see [11,12], for a review).
11m
13b
25
10o
13a
Nevertheless, several specific cognitive impairments
14r
14r
14c
14c
are demonstrable in unmedicated subjects, and in some
TRENDS in Cognitive Sciences
cases, these deficits extend to subjects who have a history
of MDD in remission or to otherwise healthy individuals at
Figure 1. Maps of the orbital and medial surfaces of a human brain (above) and a
high familial risk for a mood disorder. For example, impair-
macaque monkey brain (below), showing architectonic areas as defined in [2]
(human) and [43] (monkey). The medial and orbital prefrontal networks are colored
ments in early information processing are manifest in
yellow and green, respectively. Note that these networks have been defined based
unmedicated MDD samples, as exemplified by longer in-
on connectional data and that the medial network includes some areas on the
spection times in depressed patients than in healthy con-
orbital surface. In addition, the regions referred to as the subgenual and pregenual
anterior cingulate cortex (sgACC and pgACC) are indicated on the human brain
trols, and deficits in verbal memory are apparent in
with orange and red stripes.
62

---

Review
Trends in Cognitive Sciences January 2012, Vol. 16, No. 1
studies in humans, as well as deep brain stimulation in
these neurological disorders affect synaptic transmission
intractable patients [24]. Converging with this is a body of
through limbic-cortico-striato-pallido-thalamic circuits in
experimental anatomical evidence from animals, especially
diverse ways, it is likely that multiple types of dysfunction
monkeys, that there is a connectional network that involves
within these circuits can produce depressive symptoms.
the amygdala and several areas in the mPFC, the adjacent
Patients with early-onset mood disorders manifest neu-
medial edge of the orbital cortex, and a small region in
romorphometric abnormalities that appear relatively selec-
posterolateral orbital cortex (these cortical areas are togeth-
tive for areas within the orbital and medial prefrontal cortex
er known as the medial prefrontal network) [24,25]
(OMPFC) and anatomically related structures within the
(Figure 1). These regions are also involved in a wider circuit
temporal lobe, striatum, thalamus and posterior cingulate
that connects the amygdala and the medial network with
[24]. Cases with affective psychoses have also been differ-
other cortical areas in the anterior and medial temporal
entiated from controls by gray matter loss within the
cortex and the posterior cingulate cortex (Figure 2), as well
OMPFC [28,29]. In contrast, elderly subjects with late-onset
as with subcortical structures in the ventral striatum and
depression show a higher prevalence of neuroimaging
pallidum, the medial thalamus, and the hypothalamus
correlates of cerebrovascular disease relative both to
and brainstem (Figure 3). The data from animals indicates
age-matched healthy controls and to elderly depressives
that this system is involved in forebrain modulation of
with an early age at depression-onset [12]. MDD and BD
visceral function in response to sensory or emotive stimuli,
cases that have either late-life illness onset or psychotic
while the human evidence implicates it in emotion and
features show nonspecific signs of atrophy, such as
mood disorders [24].
lateral ventricle enlargement [12] (see Box 1 for a discussion
Distinct from the medial prefrontal related system, but
of
neuropathological
correlations
with
neuroimaging
adjacent and closely related to it, is a network of areas in
abnormalities).
the central orbital cortex (orbital prefrontal network;
Figure 1). Unlike the medial network, the orbital network
Limbic structures
has connections with several sensory-related cortical
Anatomy and connectivity
areas, and appears to be critical for assessing objects
The connections of the amygdala are a good starting point
and anticipating reward [25]. Depressed subjects show
for understanding emotion related circuitry. Experiments
abnormalities in both networks during functional MRI
in rats, cats, and more recently monkeys have shown that
studies involving reward and emotional processing tasks,
the basal and lateral amygdala have reciprocal connections
although the medial network and related structures are
to the medial prefrontal network and, to a lesser extent, to
more specifically related to mood disorders [26,27].
the orbital network, as well as to related insular and
temporal cortical areas, the mediodorsal thalamic nucleus
Observations in humans
and the ventromedial striatum [30] (Figure 2 and
In humans, the neuroimaging abnormalities found in mood
Figure 3). Outputs to hypothalamic and brainstem areas
disorders generally corroborated hypotheses regarding the
involved directly in visceral control arise from the central
neural circuitry of depression that were based initially on
and medial nuclei, but also from the basal accessory and
observations from the behavioral effects of lesions. For
basal nuclei, especially in monkeys [30]. These projections
example, degenerative basal ganglia diseases and lesions
terminate not just in the medial and lateral hypothalamus,
of the striatum and orbitofrontal cortex increase the risk
but also in the periaqueductal gray, parabrachial nuclei,
for
[
(
F
i
g
u
r
e
_
2
)
T
D
$
I
G
]
developing major depressive episodes [24]. Because
and autonomic nuclei in the caudal medulla. As such, they
Default mode network - Self reference
Rostral
Retrosplenial Ctx.,
Parahippocampal Ctx.
temporal Ctx.
Post. cingulate Ctx.
?
Dorsal prefrontal Ctx.
(Temporoparietal
junction)
Medial prefrontal network
Stimulus assessment,
Fear, Anxiety
Visceral modulation
reward
Amygdala &
Orbital & VL
Hypothalamus
other limbic
prefrontal
PAG
structures
network
TRENDS in Cognitive Sciences
Figure 2. A diagrammatic illustration of connections between the medial prefrontal network and other cortical areas, as well as with the amygdala, hypothalamus and
periaqueductal gray (PAG). Note that the medial network is part of the system that has been defined in humans as the `default mode network' (DMN) and that connections
have been defined in monkeys between all the components of the DMN, except the temporoparietal junction.
63

---

Review
[
(
F
i
g
u
r
e
_
3
)
T
D
$
I
G
]
Trends in Cognitive Sciences January 2012, Vol. 16, No. 1
network in the central orbital cortex relatively free), the
Medial prefrontal network
rostroventral insula, and the temporal pole and inferior
temporal cortex, extending caudally to the primary visual
cortex (V1) [2]. Other amygdaloid connections are with the
Amygdala &
MDm Thalamus
entorhinal and perirhinal cortex and the hippocampus
other limbic
(subiculum), and the posterior cingulate cortex [32].
structures
In the striatum, the amygdaloid projection terminates
Ventral, rostral pallidum
not only in the nucleus accumbens, but also in the adjacent
medial caudate nucleus and ventral putamen. These stria-
tal areas in turn project to the ventral and rostral palli-
Ventromedial
Paraventricular
dum, which itself sends GABAergic axons to the
striatum

thalamus
Acc Nu, Med. caudate
mediodorsal thalamic nucleus (MD; [2]). The pre- and
sub-genual prefrontal cortex (PFC) also projects to the
same ventromedial part of the striatum and is intercon-
Hypothalamus
& brainstem
nected to the same region of MD [25]. The connections
TRENDS in Cognitive Sciences
constitute essentially overlapping cortico-striato-pallido-
thalamic and amygdalo-striato-pallido-thalamic loops [2].
Figure 3. A diagrammatic illustration of the cortical-striatal-pallidal-thalamic loop
As discussed below, these circuits form the core of the
circuit, involving the medial prefrontal network and the amygdala, together with
the ventromedial striatum (nucleus accumbens and medial caudate nuclei), the
neural system implicated in mood disorders.
ventral and rostral globus pallidus, and the medial part of the mediodorsal
thalamic nucleus (MDm). The midline paraventricular thalamic nucleus also has
prominent connections with the medial network, the ventromedial striatum, and
Observations in humans
the amygdala, as well as with the hypothalamus and brainstem. The substantial
In the amygdala, glucose metabolic abnormalities appear
connections from the medial network to the hypothalamus and brainstem are not
more selective for depressive subgroups. In the left amyg-
shown in this figure.
dala, resting metabolism is abnormally elevated specifical-
ly in depressed subjects classified as BD-depressed,
can modulate autonomic and endocrine mechanisms af-
familial pure depressive disease (FPDD), or melancholic
fecting a wide variety of visceral functions [31].
subtype [12]. These subgroups also share the manifesta-
The strongest cortical connections of the large-celled
tion of hypersecreting cortisol under stress [33]. In con-
basal amygdaloid nucleus are with the mPFC rostral
trast, task-based hemodynamic responses of the amygdala
and ventral to the genu of the corpus callosum, but there
show a pattern of abnormalities reflecting negative emo-
are also amygdaloid projections to the medial, caudal,
tional processing biases in depression that appear more
and lateral orbital cortex (leaving the core of the orbital
generalizable across depressed samples (see below).
Discrepant results exist across studies, conceivably
reflecting clinical and etiological heterogeneity extant
Box 1. Neuropathological correlates of neuroimaging
within the MDD and BD syndromes. For example, in the
abnormalities
hippocampus, one study reported that reduced volume was
The structural imaging abnormalities found in mood disorders have
limited to depressed women who suffered early-life trauma
been associated with histopathological abnormalities in post
[34], whereas other studies reported that hippocampal
mortem studies. Such studies report reductions in gray matter
volume, thickness or wet weight in the sgACC, posterior orbital
volume correlated inversely with time spent depressed
cortex, and accumbens, and greater decrements in volume follow-
and unmedicated (e.g., [35]). In addition, the amygdala
ing fixation (implying a deficit in neuropil) in the hippocampus in
volume appears abnormally smaller in unmedicated BD
MDD and/or BD subjects relative to controls [12,92]. The histopatho-
subjects, but larger in BD subjects receiving mood stabi-
logical correlates of these abnormalities include reductions in glia
lizing treatments that are associated with neurotrophic
with no equivalent loss of neurons, reductions in synapses or
synaptic proteins, and elevations in neuronal density in MDD and/or
effects in experimental animals [36].
BD samples in the sgACC, and of glial cell counts and density in the
Depressed subjects show exaggerated hemodynamic
pgACC, dorsal anterolateral PFC (BA9), and amygdala (reviewed in
responses of the amygdala to sad words, explicitly or
[24]). The density of non-pyramidal neurons also appears decreased
implicitly presented sad faces, and backwardly-masked
in the ACC and hippocampus in BD and in the dorsal anterolateral
fearful faces, but blunted responses to masked happy faces
PFC (BA9) in MDD; reductions in synapses and synaptic proteins are
evident in BD subjects in the hippocampal subiculum/ventral CA1
(see [23,37], for a review). A similar pattern of abnormal
region, which receives abundant projections from the sgACC (see
amygdala responses to masked sad and happy faces is
[12], for a review).
observed in unmedicated-remitted subjects with MDD
The glial cell type implicated most consistently is the oligoden-
[23,38], suggesting this abnormality is trait-like. Converse-
drocyte. The concentrations of oligodendroglial gene products,
including myelin basic protein, are decreased in frontal polar cortex
ly, antidepressant drug administration shifts emotional
(BA 10) and middle temporal gyrus in MDD subjects versus controls
processing biases toward the positive direction in both
[12,93]. Compatible with these data the white matter volume of the
healthy and MDD samples [23,39]. This shift is in the
corpus callosum genu and splenium are abnormally reduced in MDD
normative direction, as healthy subjects show a positive
and BD. Perineuronal oligodendrocytes also are implicated in mood
attentional bias, as well as greater amygdala responses to
disorders by electron microscopy [94,95] and reduced gene expres-
sion levels [12,96] in PFC tissue. Perineuronal oligodendrocytes are
masked happy versus sad faces [23,40].
immunohistochemically reactive for glutamine synthetase, suggest-
Taken together, these data indicate that both the nor-
ing they function like astrocytes to take up synaptically released
mative positive processing bias in healthy individuals and
glutamate for conversion to glutamine and cycling back into neurons.
the pathological negative processing bias in MDD occur
64

---

Review
Trends in Cognitive Sciences January 2012, Vol. 16, No. 1
automatically, below the level of conscious awareness, and
`medial PFC' extensively overlaps with the region often
are mediated, at least partly, by rapid, non-conscious
referred to as the anterior cingulate cortex, especially its
processing networks involving the amygdala [23]. The
pre- and sub-genual parts (pgACC and sgACC; Figure 1).
differential response to masked-sad versus masked-happy
The principal difference is that the medial PFC also
faces in the amygdala is associated with concomitant
includes closely related areas rostral and ventral to the
alterations in the hemodynamic responses of the pregenual
ACC, such as the medial part of area 10 and the cortex on
anterior cingulate cortex (ACC), hippocampus, anterior
the gyrus rectus. These areas are more closely connected to
inferotemporal cortex, dorsolateral PFC, posterior cingu-
each other than they are to other parts of the cingulate
late and pulvinar. These regions may participate in setting
cortex. In particular, the dorsal ACC, located superior to
a context or in altering reinforcement contingencies that
the rostral corpus callosum in humans, is part of a different
conceivably may underlie these biases in MDD.
cortical system, more closely related to the orbital network
and the VLPFC [2]. The terms `medial PFC (mPFC)' or
PFC
`ventromedial PFC (vmPFC)' will be used here for the
Anatomy and connectivity
region as a whole, but the terms sgACC or pgACC will
In the 1990s, a series of axonal tracing experiments in
be used in describing some human studies that have
macaque monkeys more completely defined the cortical
particularly focused on the areas immediately around
and sub-cortical circuits related to the OMPFC and the
the genu of the corpus callosum.
amygdala. More specifically, two networks of intercon-
Whereas the orbital network is a sensory-related sys-
nected regions that also have common connections with
tem, the medial network in the OMPFC is an output
other cortical regions were recognized within the OMPFC.
system that can modulate visceral function in relation to
These networks have been referred to as [25] (Figure 1).
emotion or other factors. Importantly, the most prominent
Recently, a similar analysis of the organization and
amygdaloid and other limbic connections are with areas of
connections of the lateral PFC (LPFC) has been performed.
the medial network in the OMPFC [33]. The network
Based on architectonic and connectional analysis, three
comprises areas in the vmPFC, rostral and ventral to
regions can be recognized in monkeys: a dorsal prefrontal
the genu, areas along the medial edge of the orbital cortex,
region dorsal to the principal sulcus (DPFC), which is
and a small caudolateral orbital region at the rostral end of
similar to the medial prefrontal network, a ventrolateral
the insula [43] (Figure 1). The medial network receives few
region ventral to the principal sulcus (VLPFC), which is
direct sensory inputs, but is characterized by outputs to
related to the orbital prefrontal network, and a caudolat-
visceral control areas in the hypothalamus (both medial
eral region just rostral to the arcuate sulcus (CLPFC),
and lateral) and periaqueductal gray (PAG). It is also
which includes the frontal eye fields and is probably part
connected to other cortical regions (the rostral part of
of an attention system (Figure 1). As with the orbital and
the superior temporal gyrus (STGr) and dorsal bank of
medial networks, areas in each region are preferentially
the superior temporal sulcus (STSd), the posterior cingu-
connected to other areas in the same region and are con-
late cortex, and the entorhinal and parahippocampal cor-
nected to a specific set of areas in other parts of the cortex
tex), forming a cortico-cortical circuit that is distinct from
[41,42].
and complementary to the circuit related to the orbital
Orbital prefrontal network and VLPFC. Areas in the
network.
central and caudal part of the orbital cortex and the adja-
The medial network is closely related to the areas dorsal
cent anterior agranular insular cortex comprise the orbital
to it on the medial wall and those around the dorsomedial
prefrontal network (Figure 1). This network is closely
convexity, dorsal to the principal sulcus in monkeys, which
related to a similar system of cortical areas on the ventro-
constitute the dorsal prefrontal cortex (DPFC). This sys-
lateral convexity of the PFC, which can be referred to as the
tem includes areas 8b, 9, the dorsal part of area 46, as well
ventrolateral PFC system (VLPFC) [2]. Both the orbital
as the polar part of area 10 (Figure 1). The DPFC is
network and the VLPFC are characterized by connections
interconnected with itself and with the medial prefrontal
with several sensory areas, including visual areas in the
network, and is also connected to the same set of other
inferior temporal cortex (TEa) and the ventral bank of the
cortical areas as the medial network, including the poste-
superior temporal sulcus (STSv), somatic sensory areas in
rior cingulate cortex, the rostral superior temporal gyrus,
the dysgranular insula (Id), the frontal operculum (Opf)
and the entorhinal and parahippocampal cortex. Further-
and parietal cortex (area 7b), and, for the orbital network,
more, there are outputs from areas of the DPFC to the
the primary olfactory and gustatory cortex [2]. Neurophys-
hypothalamus and PAG, so this system can also modulate
iological studies have also indicated that neurons in orbital
visceral functions.
network areas respond to multimodal sensory stimuli (for
Notably, the medial network, the DPFC, and the other
example, the sight, flavor, and texture of food stimuli).
cortical areas that they are connected to (the posterior
Remarkably, however, the neurons also alter their re-
cingulate cortex, rostral temporal cortex, and medial tem-
sponse in relation to the rewarding or aversive qualities
poral cortex) closely resemble the `default mode network'
of the stimuli. In addition to its role as a system for the
(DMN), which has been defined by fMRI and functional
integration of multimodal stimuli, therefore, the orbital
connectivity MRI (fcMRI) in humans [44] (Figure 2). The
network also functions as a system for the assessment of
DMN is characterized as interconnected areas that are
the affective value of these stimuli [2].
active in a resting, introspective state but decrease activity
Medial prefrontal network and DPFC. For clarity, it is
during externally directed tasks (Figure 2). It has been
important to note that the cortical region termed the
linked to a variety of self-referential functions such as
65

---

Review
Trends in Cognitive Sciences January 2012, Vol. 16, No. 1
understanding others' mental state, recollection and imag-
[56,57]. In contrast, the VLPFC and lateral orbital regions
ination [45].
that include BA45a and BA47 s (Figure 1) and the lateral
frontal polar cortex show inverse correlations with depres-
Observations in humans
sion severity, which suggests that they play an adaptive or
Emotional processing in the medial prefrontal network. The
compensatory role in depression [12].
caudomedial PFC, especially the sgACC, participates gen-
Within the OMPFC a relatively consistent abnormality
erally in the experience and/or regulation of dysphoric
reported in early onset MDD and BD has been a reduction
emotion. In non-depressed subjects, hemodynamic activity
in gray matter in left sgACC [12]. This volumetric reduc-
increases in the sgACC during sadness induction, exposure
tion exists early in illness and in young adults at high
to traumatic reminders, selection of sad or happy targets in
familial risk for MDD [12,28], yet, longitudinal studies
an emotional go-no go study [14], and extinction of fear-
show progression of the abnormality in samples with
conditioned stimuli (see [24], for a review). Moreover,
psychotic mood disorders [29,58] and individuals with
enhanced sgACC activity during emotional face processing
more chronic or highly recurrent illness show greater
predicted inflammation-associated mood deterioration in
volume loss than those who manifest sustained remission
healthy subjects under typhoid vaccine immune challenge
[59]. The abnormal reduction in sgACC volume has been
[46]. Finally, remitted MDD subjects show decreased cou-
primarily identified in mood disordered subjects with evi-
pling between the hemodynamic responses of the sgACC,
dence for familial aggregation of illness [28,58,60].
rostral superior temporal gyrus, hippocampus, and medial
The co-occurrence of increased glucose metabolism and
frontopolar cortex during guilt (self-blame) versus indig-
decreased gray matter within the same regions in mood
nation (blaming others) [47].
disorders has been demonstrated most consistently by
The ventral pgACC and vmPFC situated anterior to the
comparing neuroimaging data from depressed patients
sgACC have been implicated in healthy subjects in reward
before versus after treatment and from remitted patients
processing, and conversely in depressed subjects, in anhe-
scanned before versus during depressive relapse [56,57]. In
donia. In healthy humans, blood oxygen level-dependent
resting metabolic neuroimaging studies, in contrast, the
(BOLD) activity in this region correlates positively with
reduction in gray matter volume in some structures
ratings of pleasure or subjective pleasantness in response
appears sufficiently prominent to produce partial volume
to odors, tastes, water in fluid-deprived subjects, and warm
effects because of their relatively low spatial resolution of
or cool stimuli applied to the hand (see [48], for a review).
functional brain images, such that in some depressed
The ventral pgACC shows activity increases in response to
samples resting metabolism appears reduced in the sgACC
dopamine-inducing drugs and during preference judg-
relative to healthy controls (see [12], for a review). In
ments [48]. Conversely, in depressed subjects this region
contrast, other studies report increased metabolic activity
shows reduced BOLD activity during reward-learning and
in the sgACC in primary or secondary depression, suggest-
higher resting electroencephalographic (EEG) delta cur-
ing that these apparent discrepancies may be explained by
rent density (putatively corresponding to decreased resting
differing magnitudes of gray matter loss across samples
neural activity) in association with anhedonia ratings [48].
[12]. Consistent with this hypothesis, in MDD and BD
More dorsal regions of the pgACC show physiological
individuals who show abnormal reductions of both gray
responses to diverse types of emotionally valenced or
matter volume and metabolism in the sgACC, correction of
autonomically arousing stimuli [49-51]. Higher activity
the metabolic data for partial volume (atrophy) effects
in the pgACC holds positive prognostic significance in
reveals that metabolism is increased in the sgACC in
MDD, as depressives who improve during antidepressant
the depressed phase, and decreases to normative levels
treatment show abnormally elevated pgACC metabolism
with antidepressant treatment. This finding is consistent
and magnetoencephalographic (MEG) and EEG activity
with evidence that sgACC metabolism decreases during
prior to treatment relative to treatment-nonresponsive
symptom remission induced by a variety of antidepressant
cases or healthy controls [52-54].
treatments, including electroconvulsive therapy and deep
Moreover, in the supragenual ACC, depressed subjects
brain stimulation [12,61].
show attenuated BOLD responses versus controls while
Similarly, during probabilistic reversal learning, de-
recalling autobiographical memories [13], associated with
pressed MDD subjects show exaggerated behavioral sen-
lower subjective arousal ratings experienced during mem-
sitivity to negative feedback versus controls, in association
ory recall. Behaviorally, MDD subjects are impaired at
with blunted BOLD activity in the dorsomedial and ven-
generating specific autobiographical memories, particular-
trolateral PFC during reversal shifting and absence of the
ly when cued by positive words. These deficits have been
normative deactivitation of the amygdala in response to
associated with reduced activity in the hippocampus and
negative feedback [11]. Disrupted top-down control by the
parahippocampus [13].
PFC over the amygdala thus may result in the abnormal
Notably, preclinical evidence indicates that distinct
response to negative feedback consistently observed in
medial prefrontal network structures are involved in op-
MDD [15].
ponent processes with respect to emotional behavior [55].
Regions where metabolism correlates positively with de-
Cortical projections to hypothalamus and brainstem
pression severity include the sgACC and ventromedial
Anatomy and connectivity
frontal polar cortex: metabolism increases in these regions
Substantial outputs exist from the medial prefrontal net-
in remitted MDD individuals who experience depressive
work to the hypothalamus, PAG, and other visceral control
relapse under catecholamine or serotonin depletion
centers [25]. The subgenual cortex provides the heaviest
66

---

Review
Trends in Cognitive Sciences January 2012, Vol. 16, No. 1
projection, which terminates in both the medial and lateral
also send direct (non-thalamic) projections to the OMPFC
hypothalamus, and in both dorsolateral and ventrolateral
[24].
columns of the PAG. The origin of the projection extends
In addition to these inputs from limbic structures, which
beyond the medial prefrontal network to include the ros-
are excitatory and probably glutamatergic, MDm also
tral superior temporal gyrus and area 9 in the DPFC, both
receives GABAergic inputs from the ventral pallidum
of which are strongly related to the medial network. Elec-
and rostral globus pallidus (see [24], for a review), which
trical stimulation of the medial network areas in monkeys
is part of the cortico-striato-pallido-thalamic loop involving
produces disturbances in functions such as heart rate and
the OMPFC (Figure 3). In MDm, the GABAergic terminals
respiration [62].
of afferent pallidal fibers synapse on the same dendrites as
the excitatory terminals from the amygdala and other
Observations in humans
limbic structures [71]. It can be expected that these con-
Functional MRI studies in humans show that activity in
vergent but antagonistic inputs would interact to modulate
the mPFC correlates with visceral activation in response to
the reciprocal thalamo-cortical interactions between the
emotional [63,64] or even non-emotional stimuli [65].
OMPFC and MDm. While the limbic inputs are dominant,
Humans with lesions of the vmPFC, centered on the medial
ongoing patterns of thalamo-cortical and cortico-thalamic
prefrontal network, show complete or severe deficits in
activity would be maintained, allowing for consistent be-
visceral responses to emotionally competent stimuli [66],
havior. When pallidal inputs become more prominent,
as would be expected from the connectional data above.
ongoing patterns would be interrupted, allowing a switch
That is, they do not appear to be able to link cortical
from one behavior to another. The affected `behaviors'
analysis of the stimulus or situation to the appropriate
would presumably include those that have been associated
visceral response. In addition, they show striking deficits
with the OMPFC: mood, value assessment of objects, and
in experiencing emotion and in social functioning, includ-
stimulus-reward association. In support of this hypothesis,
ing the ability to make appropriate choices and to control
lesions of the ventral striatum and pallidum, MD, or the
impulse behavior. This deficit has been linked to the
OMPFC have been shown to cause perseverative deficits in
absence of a `somatic marker' provided by the visceral
stimulus-reward reversal tasks in rats and monkeys, such
activation (or the neural signal that produces visceral
that the animals have difficulty switching away from
activation) that normally assists non-conscious cognitive
previously rewarded, but now unrewarded stimuli [72-
processes in controlling behavior [66].
75]. A similar deficit in subjects with mood disorders might
From the viewpoint of mood disorders, over-activation of
be the difficulty of `letting go' of a negative mood or mind-
this visceromotor system (e.g., due to excessive activity in
set long after the resolution of any traumatic events that
the mPFC or sgACC, evoking visceral disturbance) may
might have justified it.
contribute to the chronic sense of `unease' that is a common
Prefrontal projections to the striatum. The OMPFC pro-
component of depression. William James, who fell into a
jects principally to the rostral, ventromedial part of the
severe depressive episode when torn between his religious
striatum. The orbital network areas connect to a relatively
and scientific beliefs, gave one of the most striking self-
central region that spans the internal capsule and includes
descriptions of this aspect of depression. He wrote: ``I
parts of both the caudate nucleus and the putamen. Of
awoke morning after morning with a horrible dread at
more significance for mood disorders, however, the medial
the pit of my stomach, and with a sense of the insecurity of
network areas project to the nucleus accumbens and the
life that I never knew before. . . It gradually faded, but for
adjacent medial edge of the caudate nucleus bordering the
months I was unable to go out into the dark alone. In
lateral ventricle [69] (Figure 3). The amygdala input to the
general I dreaded to be left alone.'' [67].
striatum is essentially coextensive with that of the medial
network. These striatal regions, in turn, project to the
Cortico-striatal-thalamic circuits related to OMPFC
ventral pallidum, which projects to the portion of MDm
Anatomy and connectivity
that is connected to the medial network areas [2].
The PFC has specific connections with the striatum and
Midline `intralaminar' nuclei of the thalamus. In addi-
thalamus and several circuits can be identified. The first
tion to the prefrontal connections with MD, there are also
are the reciprocal thalamo-cortical connections that relay
major connections with the midline-intralaminar nuclei of
subcortical input to the cortex through principal thalamic
the thalamus, which include the paraventricular thalamic
nuclei. The well-known cortico-striato-pallido-thalamic
nucleus (PVT), as well as other nuclei that extend ventrally
loops are closely related to these (Figure 3). The medial
on the midline between the anterior and mediodorsal
prefrontal network, in particular, is connected to both the
thalamic nuclei. These nuclei are reciprocally connected
medial segment of the mediodorsal thalamic nucleus
to the medial prefrontal network areas, with little connec-
(MDm) and the ventromedial part of the striatum
tion to the orbital network, and they have a substantial
[68,69] (Figure 3). Finally, there are circuits that involve
projection to the same areas of the ventromedial striatum
the intralaminar and midline thalamic nuclei, which proj-
that receives input from the medial network areas [76,77]
ect to both the striatum and the cortex.
(Figure 3). They also have important connections with the
Medial segment of mediodorsal thalamic nucleus. The
amygdala, hypothalamus, and brainstem areas, including
MDm receives substantial subcortical inputs from many
the PAG. As such, they are situated to relay information
limbic structures, including the amygdala, olfactory cor-
about visceral and emotional activity to both the medial
tex, entorhinal cortex, perirhinal cortex, parahippocam-
prefrontal network and the cortico-striato-pallido-thalam-
pal cortex, and subiculum [70]. Notably, all of these areas
ic loop with which it interacts.
67

---

Review
Trends in Cognitive Sciences January 2012, Vol. 16, No. 1
Considerable evidence links the PVT to the stress re-
and BNST to the hypothalamus in depression accounting,
sponse [78,79]. In particular, lesions of the PVT block the
for increased stressed cortisol secretion [24]. Notably, cor-
habituation to chronic stress in rats [80], via a mechanism
tisol hypersecretion in mood disorders has been associated
that involves corticosterone action in the PVT [81]. In
with increased metabolic activity in the amygdala and with
humans, a similar habituation to the neuroendocrine
reduced gray matter in rostral ACC [12,33,86].
stress response caused by chronic hypoglycemia [65] is
Furthermore, dysfunction of the medial prefrontal net-
associated with activity in the midline thalamus [82]. It
work and the adjacent orbital network may impair reward
is likely that the role of the PVT is general across many
learning, potentially contributing to the anhedonia and
types of stressors.
amotivation manifest in depression. The ACC receives
extensive dopaminergic innervation from the ventral teg-
Observations in humans
mental area (VTA) and sends projections to the VTA that
The neurophysiological activity of subcortical structures
regulate phasic dopamine (DA) release. In rats, stimula-
that share extensive connections with the medial prefron-
tion of these mPFC areas elicits burst firing patterns in the
tal network show correlations with depressive symptoms.
VTA-DA neurons while inactivation of the mPFC converts
In the accumbens area, the elevation of metabolism under
burst firing patterns to pacemaker-like firing activity [24].
catecholamine depletion correlates positively with the cor-
The burst firing patterns increase DA release in the accum-
responding increment in anhedonia ratings [56]. In addi-
bens, which is thought to encode information about reward
tion, fMRI studies show that hemodynamic responses of
prediction [3]. The mesolimbic DA projections from the
the ventral striatum to rewarding stimuli are decreased in
VTA to the nucleus accumbens shell and the mPFC thus
depressives versus controls and that higher levels of anhe-
play major roles in learning associations between operant
donia are associated with blunted ventral striatal
behaviors or sensory stimuli and reward. If the neuropath-
responses to rewarding stimuli in both healthy [48] and
ological changes in the ACC in mood disorders interfere
depressed subjects (see [3], for a review). Furthermore,
with the cortical drive on VTA-DA neuronal burst firing
during probabilistic reversal learning depressed subjects
activity, they may impair reward learning. Compatible
show impaired reward (but not punishment) reversal ac-
with this hypothesis, depressives show attenuated DA
curacy in association with attenuated ventral striatal
release relative to controls in response to unpredicted
BOLD response to unexpected reward [83].
monetary reward [87] and fail to develop a response bias
towards rewarded stimuli during reinforcement para-
Implications for neurocircuitry-based models of
digms [5-7].
depression
Finally, the patterns of physiological activity within the
Within the larger context of the limbic-cortical-striato-
DMN that involves the medial network are hypothesized to
pallido-thalamic circuits implicated in the pathophysiology
relate to self-absorption or obsessive ruminations [88,89].
of depression, the functional implications of some limbic-
In depressed subjects, increasing levels of DMN dominance
cortical circuits involving the medial prefrontal network
over the putative `task-positive network' (TPN; a group of
merit comment in light of the abundant basic science
structures that are consistently activated during volitional
literature available to guide translational models. The
attention and thought) are associated with higher levels of
anatomical projections from the medial prefrontal network
maladaptive, depressive rumination and lower levels of
to the amygdala, hypothalamus, PAG, locus coeruleus,
adaptive, reflective rumination [90]. Crucially, hemody-
raphe, and brainstem autonomic nuclei play major roles
namic activity increased in the anterior insular region
in organizing the visceral and behavioral responses to
corresponding to intrasulcal BA47 in depressed subjects
stressors and emotional stimuli (see [24], for a review).
at the onset of increases in TPN activity. Hamilton et al.
In rats, lesions of the mPFC enhance behavioral, sympa-
[90] thus hypothesized that the DMN undergirds the re-
thetic, and endocrine responses to stressors or fear-condi-
presentation of negative, self-referential information in
tioned stimuli organized by the amygdala [84,85].
depression and that the intrasulcal BA 47, when prompted
Nevertheless, these relationships are complex, as stimula-
by increased levels of DMN activity, initiates an adaptive
tion of the amygdala inhibits neuronal activity in the
engagement of the TPN. Notably, interpersonal psycho-
mPFC, and stimulation of infralimbic and prelimbic pro-
therapy, which can reduce depressive symptoms in MDD,
jections to the amygdala excites intra-amygdaloid GABA-
enhances activity in the same anterior insular region (i.e.,
ergic cells that respectively inhibit or excite neuronal
intrasulcal BA 47) [24]. Cognitive-therapeutic strategies
activity in the central amygdaloid nucleus (ACe) [55].
for depression thus may depend upon enhancing the func-
For example, the amygdala is reported to mediate the
tion of prefrontal systems that serve as convergence zones
stressed component of glucocorticoid hormone secretion, at
between multiple prefrontal networks, such as BA 47s [91]
least in part, by disinhibiting corticotropin-releasing factor
(Figure 1).
(CRF) release from the hypothalamic paraventricular nu-
cleus; the glucocorticoid response to stress is inhibited by
Concluding remarks
stimulation of glucocorticoid receptors (GR) in the ventral
The experimental observations described here provide an
ACC and lesions in this cortical region thus increase
indication of the circuitry and structures that are involved
adrenocorticotropic hormone (ACTH) and corticosterone
in mood disorders and related conditions. The description
(CORT) secretion during stress in rats [24]. It is conceiv-
is not yet complete and there are many important details
able that dysfunction within the medial prefrontal network
that have yet to be worked out, but important foci such as
would disinhibit the efferent transmission from the ACe
the subgenual prefrontal cortex and the amygdala can be
68

---

Review
Trends in Cognitive Sciences January 2012, Vol. 16, No. 1
5 Henriques, J.B. et al. (1994) Reward fails to alter response bias in
Box 2. Future directions
depression. J. Abnorm. Psychol. 103, 460-466
6 Pizzagalli, D.A. et al. (2008) Reduced hedonic capacity in major
The clinical designation of anhedonia has not heretofore dis-
depressive disorder: evidence from a probabilistic reward task. J.
criminated between decrements in motivation and reductions in
Psychiatr. Res. 43, 76-87
experienced pleasure, despite the significance of this distinction
7 Pizzagalli, D.A. et al. (2005) Toward an objective characterization of an
for translational research. A refined research definition of
anhedonic phenotype: a signal-detection approach. Biol. Psychiatry 57,
anhedonia that parses this symptom cluster into deficits in
319-327
hedonic response to rewards versus diminished motivation to
8 McFarland, B.R. and Klein, D.N. (2009) Emotional reactivity in
pursue rewards is, therefore, needed to delineate between deficits
depression: diminished responsiveness to anticipated reward but not
in the `liking' versus the `wanting' components recognized within
to anticipated punishment or to nonreward or avoidance. Depress.
the preclinical literature [3,97].
Anxiety 26, 117-122
Neurophysiological investigation of the properties of the medial
9 Tavares, J.V. et al. (2003) Cognition in mania and depression. Psychol.
prefrontal areas in animals, especially the areas immediately
Med. 33, 959-967
around the genu of the corpus callosum, are needed. Although
10 Roiser, J.P. et al. (2009) Hot and cold cognition in unmedicated
there is good evidence that cortical areas in this region have an
depressed subjects with bipolar disorder. Bipolar Disord. 11, 178-
important role in modulating visceral reactions to emotive and
189
other stimuli through their connections to the hypothalamus and
11 Taylor Tavares, J.V. et al. (2008) Neural basis of abnormal response to
brainstem, little is known about the mechanisms underlying this
negative feedback in unmedicated mood disorders. Neuroimage 42,
modulation. Several studies are currently investigating the areas
1118-1126
of the orbitofrontal cortex that have begun to elucidate the role of
12 Drevets, W.C. et al. (2008) Brain structural and functional
these areas in reward and stimulus value assessment [98]. To
abnormalities in mood disorders: implications for neurocircuitry
date, however, there has been little or no equivalent investigation
models of depression. Brain Struct. Funct. 213, 93-118
of the medial prefrontal areas.
13 Young, K.D. et al. (2011) Functional anatomy of autobiographical
It is also crucial to investigate the properties of individual cortical
memory recall deficits in depression. Psychol. Med. DOI: 10.1017/
areas within the medial and orbital prefrontal networks. Although
S0033291711001371
these systems are each highly interconnected, they are not fully
14 Elliott, R. et al. (2000) Selective attention to emotional stimuli in a
homogeneous and it may be expected that different architectonic
verbal go/no-go task: an fMRI study. Neuroreport 11, 1739-1744
areas within them have different properties. For example, a recent
15 Murphy, F.C. et al. (1999) Emotional bias and inhibitory control
article indicates that area 10 at the frontal pole has a specific
processes in mania and depression. Psychol. Med. 29, 1307-1321
function in encoding [99]. Little is known, however, about the
16 Murray, L.A. et al. (1999) Mood congruence and depressive deficits in
specific physiological functions of other medial prefrontal areas.
memory: a forced-recall analysis. Memory 7, 175-196
Further human brain mapping studies that compare depressed
17 Bradley, B.P. et al. (1995) Implicit and explicit memory for emotion-
and control samples using high-resolution fMRI are needed to
congruent information in clinical depression and anxiety. Behav. Res.
determine the specific role of the components of the `depression
Ther. 33, 755-770
circuit' on which we have focused (sgACC/pgACC, amygdala,
18 Broomfield, N.M. et al. (2006) Further evidence of attention bias for
ventral striatum, and medial thalamus). This would require careful
negative information in late life depression. Int. J. Geriatr. Psychiatry
development of behavioral testing paradigms. For example,
175-180
would the ventral striatum or medial prefrontal cortex show
19 Gallardo Perez, M. et al. (1999) Attentional biases and vulnerability to
differential activity in relation to the specific function of suppres-
depression. Span. J. Psychol. 2, 11-19
sion of negative thoughts? Similarly, would the subgenual area
20 Erickson, K. et al. (2005) Mood-congruent bias in affective go/no-go
show differential activity that correlates with visceral responses to
performance of unmedicated patients with major depressive disorder.
negative versus positive emotional stimuli?
Am. J. Psychiatry 162, 2171-2173
21 Gotlib, I.H. et al. (2004) Coherence and specificity of information-
identified and their relationships understood. The system
processing biases in depression and social phobia. J. Abnorm.
is complex and, although there are many suggestions, it is
Psychol. 113, 386-398
22 Gotlib, I.H. et al. (2004) Attentional biases for negative interpersonal
not possible yet to identify the specific deficits that result in
stimuli in clinical depression. J. Abnorm. Psychol. 113, 121-135
mood disorders (see also Box 2). The complexity of the
23 Victor, T.A. et al. (2010) Relationship of emotional processing to
system means that there are likely to be multiple factors
masked faces in the amygdala to mood state and treatment in major
that cause different aspects of these heterogeneous disor-
depressive disorder. Arch. Gen. Psychiatry 67, 1128-1138
24 Price, J.L. and Drevets, W.C. (2010) Neurocircuitry of mood disorders.
ders. In spite of such confusion, the progress made in the
Neuropsychopharmacology 35, 192-216
past 15 to 20 years has been impressive. It is now possible
25 O
ngur, D. and Price, J.L. (2000) The organization of networks within
to discuss mood disorders in terms of specific brain sys-
the orbital and medial prefrontal cortex of rats, monkeys and humans.
tems, and there is no question that there is a neurobiologi-
Cereb. Cortex 10, 206-219
cal basis for mood disorders.
26 Murray, E.A. et al. (2010) Localization of Dysfunction in Major
Depressive
Disorder:
Prefrontal
Cortex
and
Amygdala.
Biol.
Psychiatry 69, e43-e54
Acknowledgements
27 Phillips, M.L. et al. (2008) A neural model of voluntary and automatic
J.L.P. is supported by grant R01 MH070941 from the USPHS/NIMH and
emotion
regulation:
implications
for
understanding
the
W.C.D. by funds from The William K. Warren Foundation.
pathophysiology and neurodevelopment of bipolar disorder. Mol.
Psychiatry 13, 829-857
References
28 Hirayasu, Y. et al. (1999) Subgenual cingulate cortex volume in first-
1 American Psychiatric Association (1994) Diagnostic and Statistical
episode psychosis. Am. J. Psychiatry 156, 1091-1093
Manual of Mental Disorders (DSM-IV), APA Press
29 Coryell, W. et al. (2005) Subgenual prefrontal cortex volumes in major
2 O
ngur, D. et al. (2003) Architectonic subdivision of the human orbital
depressive disorder and schizophrenia: diagnostic specificity and
and medial prefrontal cortex. J. Comp. Neurol. 460, 425-449
prognostic implications. Am. J. Psychiatry 162, 1706-1712
3 Treadway, M.T. and Zald, D.H. (2011) Reconsidering anhedonia in
30 Price, J.L. (2003) Comparative aspects of amygdala connectivity. Ann.
depression:
lessons
from
translational
neuroscience.
Neurosci.
N. Y. Acad. Sci. 985, 50-58
Biobehav. Rev. 35, 537-555
31 Price, J.L. (2005) Free will versus survival: brain systems that
4 Bylsma, L.M. et al. (2008) A meta-analysis of emotional reactivity in
underlie intrinsic constraints on behavior. J. Comp. Neurol. 493,
major depressive disorder. Clin. Psychol. Rev. 28, 676-691
132-139
69

---

Review
Trends in Cognitive Sciences January 2012, Vol. 16, No. 1
32 Buckwalter, J.A. et al. (2008) Evidence for direct projections from the
57 Neumeister, A. et al. (2004) Neural and behavioral responses
basal nucleus of the amygdala to retrosplenial cortex in the Macaque
to tryptophan depletion in unmedicated patients with remitted
monkey. Exp. Brain Res. 186, 47-57
major depressive disorder and controls. Arch. Gen. Psychiatry 61,
33 Drevets, W.C. et al. (2002) Glucose metabolism in the amygdala in
765-773
depression: relationship to diagnostic subtype and plasma cortisol
58 Koo, M.S. et al. (2008) A cross-sectional and longitudinal magnetic
levels. Pharmacol. Biochem. Behav. 71, 431-447
resonance imaging study of cingulate gyrus gray matter volume
34 Vythilingam, M. et al. (2002) Childhood trauma associated with
abnormalities
in
first-episode
schizophrenia
and
first-episode
smaller hippocampal volume in women with major depression. Am.
affective psychosis. Arch. G