Neuron 51, 1–12, September 7, 2006 ª2006 Elsevier Inc.DOI 10.1016/j.neuron.2006.07.029
Resolving Emotional Conflict: A Role
for the Rostral Anterior Cingulate Cortex
in Modulating Activity in the Amygdala
Amit Etkin,1,2,6,* Tobias Egner,3,7Daniel M. Peraza,3
Eric R. Kandel,1,2,4and Joy Hirsch1,2,3,5,*
1Center for Neurobiology and Behavior
2Kavli Institute for Brain Sciences
3Functional MRI Research Center
4Howard Hughes Medical Institute
5Departments of Radiology and Psychology
Columbia University Medical Center
Neurological Institute Box 108
710 West 168th Street
New York, New York 10032
protected from emotional conflict from interference by
task-irrelevant emotionally salient stimuli. The neural
mechanisms by which the brain detects and resolves
emotional conflict are still largely unknown, however.
Drawing on the classic Stroop conflict task, we devel-
oped a protocol that allowed us to dissociate the gen-
eration and monitoring of emotional conflict from its
resolution. Using functional magnetic resonance im-
aging (fMRI), we find that activity in the amygdala
and dorsomedial and dorsolateral prefrontal cortices
reflects the amount of emotional conflict. By contrast,
the resolution of emotional conflict is associated with
activation of the rostral anterior cingulate cortex. Acti-
vation of the rostral cingulate is predicted by the
and is accompanied by a simultaneous and correlated
reduction of amygdalar activity. These data suggest
that emotional conflict is resolved through top-down
As William James first pointed out (James, 1890), we are
exposed, in our everyday life, to a larger number of sen-
sory stimuli than we can dedicate processing resources
oritize processing and diminish distraction by stimuli ir-
optimal performance, the brain is thought to resolve
‘‘conflict’’ by monitoring continuously for distracters
that produce responses that are incompatible with the
current task (Botvinick et al., 2001). Emotionally salient
stimuli, such as those that signal potential danger, are
particularly effective in interfering with ongoing cogni-
tive processing (LeDoux, 2000; Mathews, 1990; Tipples
and Sharma, 2000). However, despite the important
role that control over the effects of emotional distracters
plays in both normal functioning and in clinical mood
and anxiety disorders (Mathews, 1990; Mathews and
MacLeod, 1985), the neural mechanisms by which emo-
tional conflict is monitored and resolved are largely un-
known. This stands in sharp contrast to nonemotional
(cognitive) conflict, where extensive knowledge already
exists. Here, we exploit some of the methodological in-
sights gained from the study ofcognitive conflict resolu-
tion and apply them to emotional conflict.
Assessing Emotional Conflict
Theclassic paradigm for studying nonemotionalconflict
is the color-word Stroop task (MacLeod, 1991; Stroop,
1935). In this task, subjects are asked to indicate the
type-face (ink) color of a word (e.g., ‘‘green’’ or ‘‘red’’)
that is also the name of a color (e.g., ‘‘RED’’ or
‘‘GREEN’’). When the ink color and the word meaning
are not the same (i.e., incongruent), processing of the
word meaning and the ink color lead to different, incom-
ing reaction times. In the Stroop task, therefore, seman-
tic incompatibility between the task-relevant and
task-irrelevant stimulus dimensions produces response
Previous studies of emotional conflict employed a dif-
ferent form of the Stroop task, the ‘‘emotional Stroop’’
task. In this task, subjects are asked to identify the ink
color of words that are either emotionally neutral (e.g.,
‘‘apple’’) or emotionally salient (e.g., ‘‘death’’) (Mathews
and MacLeod, 1985; McKenna, 1986). Slowing of reac-
tion times for color naming of emotional words relative
to neutral words serves as a measure of emotional inter-
ference. The emotional Stroop task, however, is not
thought to assess directly the interference of emotional
processing with cognitive processing. Rather it as-
sesses the ability of emotional stimuli, processed in
parallel, to withdraw attention from the main task. This
is because the meaning of the emotional word stimuli
is neither semantically related to the task-relevant infor-
mation (ink color) nor does it lead to responses that
directly compete with the selection of the correct re-
sponse (Algom et al., 2004). The traditional emotional
Stroop task, therefore, does not provide a measure of
emotional conflict comparable to the measure of cogni-
tive conflict provided by the classic color-word Stroop
task. Moreover, in normal subjects, behavioral interfer-
ence by these emotional distracters is either not de-
tected at all (Williams et al., 1996) or habituates very rap-
idly (Compton et al., 2003; McKenna, 1986). Such lack of
reliable behavioral effects limits the conclusions that
can be drawn from previous imaging studies that used
emotional Stroop-like tasks (Bishop et al., 2004; Comp-
ton et al., 2003; Whalen et al., 1998).
To assess more directly the effects of emotional con-
flict, we developed a paradigm in which emotional con-
*Correspondence: email@example.com (A.E.); jh2155@columbia.
6Present address: Department of Psychiatry and Behavioral Sci-
ences, Stanford University School of Medicine, 401 Quarry road,
Palo Alto, California 94305. E-mail: firstname.lastname@example.org.
7Present address: Cognitive Neurology and Alzheimer’s Disease
Center, Feinberg School of Medicine, Northwestern University,
320 East Superior, Searle 11, Chicago, Illinois 60611. E-mail:
and task-irrelevant emotional dimensions of a stimulus.
We modified the classic color-word Stroop paradigm
and developed an emotional conflict task in which faces
with fearful and happy expressions were presented with
roll and Young, 2005). We thenasked subjects to identify
the emotional expression of the faces while ignoring the
words, which were either congruent or incongruent with
the facial expression (see Experimental Procedures and
Figure 1 for example stimuli). Incongruent stimuli were
thus associated with response conflict that arose from
an emotional incompatibility between task-relevant and
task-irrelevant stimulus dimensions (e.g., a fearful ex-
pression with the word ‘‘happy’’). Our emotional conflict
task therefore represents an appropriate emotional ana-
log to the color-word Stroop task (Algom et al., 2004).
The Neural Circuitry of Conflict Processing
The anterior cingulate cortex has frequently been
thought to play a critical role in executive attention (Bot-
vinick et al., 2004; Bush et al., 2000; Carter et al., 1999;
Ridderinkhof et al., 2004). Examination of cytoarchitec-
ture and connectivity patterns, as well as lesion and im-
aging studies, however, has led some authors to divide
the anterior cingulate cortex into a dorsal ‘‘cognitive’’ di-
vision and a ventral ‘‘affective’’ division (Bush et al.,
2000; Devinsky et al., 1995; Vogt et al., 1992). The dorsal
cingulate and adjacent dorsomedial prefrontal cortex
are connected with ‘‘cognitive’’ regions such as the lat-
eral prefrontal and motor/premotor cortices (van Hoe-
sen et al., 1993). The ventral division, composed of
rostral (pregenual) and subgenual components, is con-
nected with ‘‘affective’’ regions such as the amygdala
(Carmichael and Price, 1995; van Hoesen et al., 1993).
The dorsal and ventral divisions of the anterior cingulate
are also interconnected with each other (Musil and Ol-
son, 1988a, 1988b; van Hoesen et al., 1993).
Response conflict on cognitive tasks commonly acti-
jacent dorsomedial prefrontal cortex (Botvinick et al.,
2004; Bush et al., 2000; Ridderinkhof et al., 2004). It
was uncertain, however, whether these regions were
responsible for monitoring or for resolving conflict, and
thus resolution of this issue required an experimental
paradigm capable of dissociating conflict monitoring
from conflict resolution (Botvinick et al., 2001, 2004;
Kerns et al., 2004; MacDonald et al., 2000; Posner and
DiGirolamo, 1998; Posner et al., 1988). Such a dissocia-
tion can be achieved by examining the behavioral ef-
fects of trial sequence, and these studies reveal that
there is less reaction time interference (i.e., less conflict)
for incongruent trials if they are preceded by an incon-
gruent trial than if they are preceded by a congruent trial
(Botvinick et al., 1999; Egner and Hirsch, 2005a, 2005b;
Gratton et al., 1992; Kerns et al., 2004). These findings
suggest that the conflict generated by an immediately
prior incongruent trial activates an anticipatory mecha-
nism, which leads to improved conflict resolution on
the next trial (Botvinick et al., 2001). Incongruent trials
can thus be separated depending on whether they are
associated with high conflict resolution and conse-
quently less conflict (an incongruent trial preceded by
an incongruent trial) or low conflict resolution and thus
ent trial). Neural activity in regions responsible for either
generating or monitoring conflict should reflect the
amount of behavioral conflict, resulting in higher activity
for low conflict resolution trials in these regions, while in
brain regions implicated in conflict resolution, reduced
conflict should be associated with increased neural
This distinction between otherwise identical incon-
gruent trials allows for a dissociation between conflict
generating/monitoring andconflict resolving processes.
Indeed, a series of recent neuroimaging studies have
found activity in the dorsal cingulate and the dorsome-
dial prefrontal cortex associated with the monitoring of
was associated with the resolution of conflict (Botvinick
et al., 1999; Kerns et al., 2004; Egner and Hirsch 2005a).
The dorsal cingulate most likely does not constitute
a functionally homogenous region. The dorsal cingulate
and adjacent dorsomedial prefrontal cortex, for exam-
ple, have additional roles in general outcome evaluation
Figure 1. Stimuli and Example Timelines
Used in the Emotional Conflict Task
Subjects were asked to identify the affect of
faces with fearful or happy expressions that
had either ‘‘fear’’ or ‘‘happy’’ written across
them. Stimuli were either congruent or incon-
gruent with respect to facial expression and
word, which created emotional conflict.
(Rushworth et al., 2004), volitional processes (Nachev
et al., 2005), attentional selection (Posner and DiGiro-
lamo, 1998; Weissman et al., 2004), and autonomic con-
trol (Critchley et al., 2003).
In contrast to the understanding we now have of cog-
nitive conflict, emotional conflict has not been compara-
bly investigated. A variety of tasks involving emotional
processing in general (Bush et al., 2000), the processing
of emotional distracters (Vuilleumier et al., 2001), and
particularly the processing of negatively valenced stim-
of the ventral ‘‘affective’’ division. By extrapolation from
of cognitive conflict, the rostral and subgenual cingulate
conflict (Bishop et al., 2004; Bush et al., 2000; Whalen
et al., 1998). These studies, however, presented emo-
tional stimuli in blocks (Bishop et al., 2004; Bush et al.,
2000; Compton et al., 2003; Whalen et al., 1998) and
therefore could not distinguish the mechanisms associ-
ated with the monitoring of conflict from those associ-
ated with the resolution of conflict, as has been possible
for cognitive conflict. The exact function of the rostral
cingulate in emotional conflict, therefore, remains un-
clear. We wanted to determine whether the rostral cin-
gulate acts as the emotional analog to the dorsal cingu-
late by monitoring emotional conflict, as has been
suggested by some investigators (Bishop et al., 2004;
Whalen et al., 1998), or whether it plays a different role,
for instance by resolving emotional conflict.
In the current study, we distinguished between trials
of high and low conflict resolution to differentiate the
neural processes that track emotional conflict (because
they are involved in conflict generation or monitoring)
from those associated with emotional conflict resolu-
tion. We expected that the emotional nature of the stim-
uli will recruit regions such as the amygdala and rostral
cingulate. We also reasoned that because emotional
conflict in our task results in response conflict, regions
involved in processing response conflict may also be
After determining which regions tracked emotional
conflict and which regions were involved in emotional
conflict resolution, we turned our attention to their func-
tional interconnectivity. We specifically tested whether
neural conflict resolution would be recruited in a flexible
manneron thecurrent trial,tothedegreeappropriate for
the amount of conflict signaled by conflict-tracking re-
gions on the previous trial (Botvinick et al., 2001; Kerns
ful, the regions underlying conflict resolution must in
turn regulate the source of the emotional conflict.
involved in conflict resolution was associated with de-
creased activity in upstream regions involved in conflict
Detection and Resolution of Emotional Conflict
We carried out the emotional conflict task on a group of
19 healthy volunteers and, based on reaction times,
found robust behavioral interference associated with
emotional conflict in every subject (all incongruent ver-
sus congruent trials, p = 0.0000002; Figure 2A). A similar
effect was seen for accuracy (p = 0.0001, Figure 2B), in-
dicating that the reaction-time effect was not a result of
a speed-accuracy tradeoff.
We next compared reaction times on low conflict res-
olution trials (incongruent trials that followed congruent
ones) with high conflict resolution trials (incongruent tri-
als that followed incongruent ones). The data showed
reduced reaction time interference during high conflict
resolution trials (p = 0.01, Figure 2A), consistent with ac-
congruent trial, which bolstered conflict resolution on
the current trial (previous 3 current trial ANOVA p =
0.03; Figure 2A). A similar effect was evident at a trend
level for accuracy (p = 0.096; Figure 2B). Also, as seen
in Figure 2C, no reaction time differences were seen be-
tween presentation of fearful and happy faces (p = 0.76)
or words (p = 0.54), which were counterbalanced across
congruency conditions, suggesting that behavior in this
task was driven primarily by emotional conflict.
Finally, since high conflict resolution trials entailed
a repetition of an incongruent stimulus, which does not
occur on low conflict resolution trials, it remained a pos-
sibility that repetition priming rather than conflict resolu-
tion could mediate faster reaction times in high conflict
resolution trials (Mayr et al., 2003). In designing the
Figure 2. Behavioral Dissociation between Conflict Monitoring and
Mean reaction times (A) and accuracy (B) 6SEM for incongruent and
congruent trials split by previous-trial and current types. (C) Reac-
tion times in this task are not driven by the emotional content of
the faces or words independent of whether they were congruent
Neural Circuit for Emotional Conflict Resolution
task we avoided direct stimulus repetition. However, we
could not avoid repetition of the stimulus category in
some high conflict resolution trials (e.g., fearful face
and happy word followed by another fearful face and
happy word). We therefore compared reaction times
on high conflict resolution trials in which category repe-
tition occurred to trials in which repetition did not occur
(Egner and Hirsch, 2005a) and found no differences in
reaction times (p = 0.45).
Together, these data confirm that behavioral interfer-
ence arising from emotional conflict can be detected in
healthy volunteers and that conflict leads to enhance-
ment of conflict resolution on the subsequent trial. We
next used these behavioral findings to investigate differ-
ences in regional brain activity associated with emo-
tional conflict monitoring and resolution using fMRI.
Specifically, we focused on the distinction between
high and low conflict resolution trials in order to dissoci-
ate conflict monitoring from conflict resolution (Botvi-
nick et al., 1999; Egner and Hirsch, 2005a; Kerns et al.,
Emotional Conflict Monitoring and Resolution
Effects in the Amygdala and Prefrontal Cortex
An a priori region of interest analysis in the amygdalae
(see Experimental Procedures) revealed a cluster in
the right amygdala where activity was greater in low
than high conflict resolution trials ([18, 2, 216], Z =
2.99; Figures 3A and 3B), suggesting that activity in the
amygdala reflects the amount of emotional conflict. No
voxels in the amygdala were more active in high than
low conflict resolution trials, event at a lenient p < 0.05,
uncorrected. The effect of emotional conflict in the
amygdala also did not differ between fearful and happy
face targets, which were counterbalanced across low
and high conflict resolution trials (all voxels p > 0.1,
data not shown). In addition, no difference was ob-
served in the region of the amygdala sensitive to emo-
tional conflict for the response to fearful versus happy
faces (p = 0.51) or fear versus happy words (p = 0.76;
Figure 3B). This is consistent with previous studies
that have shown comparable amygdala activations to
fearful and happy facial expression (Fitzgerald et al.,
2006; Yang et al., 2002). These results therefore show
that emotional conflict itself leads to activation of the
amygdala, independently of stimulus valence.
Next, we focused on the medial and lateral prefrontal
cortices, including the anterior cingulate, as these fron-
tal regions have been previously implicated in cognitive
conflict monitoring and resolution (Botvinick et al., 1999,
2001, 2004; Egner and Hirsch, 2005b; Kerns et al., 2004;
MacDonaldetal., 2000;Ridderinkhof etal.,2004).Within
the prefrontal cortex, activity tracking the amount of
emotional conflict (low > high conflict resolution trials)
was observed in three areas: midline dorsomedial pre-
frontal and bilateral dorsolateral prefrontal cortices
(DMPFC [22, 38, 38] Z = 4.08, right DLPFC [42, 14, 32]
Z = 3.7 and [44, 18, 52] Z = 3.13, left DLPFC [244, 18,
24] Z = 3.63; blue in Figure 4A). By contrast, activity cor-
responding to the resolution of conflict (high > low con-
flict resolution trials) was observed in only one area: the
rostral anterior cingulate ([210, 48, 0] Z = 4.02 and
[210, 36, 2] Z = 3.47; red in Figure 4A). These data
suggest that the rostral cingulate is engaged during
emotional conflict resolution and not for the monitoring
or generation of such conflict.
Since high conflict resolution trials are associated
with faster reaction times than low conflict resolution tri-
als, greater rostral cingulate activity in high conflict res-
olution trials could in theory simply be a result of high
conflict resolution trials being ‘‘easier.’’ This view is con-
sistent with the involvement of the rostral cingulate in
a ‘‘resting state’’ network, which may be suppressed
during difficult tasks (Raichle et al., 2001). In our task,
Figure 3. The Amygdala Is Involved in Conflict Monitoring
(A) Active right amygdala cluster in the low > high conflict resolution
contrast (displayed at p < 0.005, uncorrected).
(B) Mean right amygdala activity (6SEM) is greater in low than high
conflict resolution trials. No significant differences were seen in re-
Contrast estimates (b weights) correspond to an average of all
Figure 4. CorticalRegionsInvolved inEmotionalConflictMonitoring
(A) Frontal lobe regions whose activity is greater in high than low
conflict resolution trials (red) or in low than high conflict resolution
trials (blue), displayed at p < 0.001, uncorrected.
(B) Conjunction between regions showing both greater activity in
high than low conflict resolution trials and congruent than incongru-
ent trials (red) or greater activity in low than high conflict resolution
trials and incongruent than congruent trials (blue) at a lenient uncor-
rected threshold of p < 0.05.
an even greater reaction time difference was observed
between all congruent and all incongruent trials. The
‘‘trial ease’’ explanation would predict that rostral cingu-
late activity should be greater in high than low conflict
resolution trials and also greater in congruent than in-
congruent trials. We therefore conducted a conjunction
analysis for voxels activated at a very lenient threshold
(uncorrected p < 0.05) in both high > low conflict resolu-
tion trialsand congruent > incongruent trials (Figure 4B).
Even at this lenient threshold we found no significant
voxels in the rostral cingulate, arguing that activity in
the rostral cingulate does not merely reflect the ease
of the trial but rather reflects an active process of con-
flict resolution. By contrast, as would be expected of
a conflict-monitoring region, the dorsomedial and dor-
solateral prefrontal cortices (Figure 4B and data not
shown) were both more active in low than high conflict
Sequential Emotional Conflict Tracking and
Resolution Involves an Interaction between
the Amygdala and the Rostral Cingulate
A critical prediction of a sequential conflict monitoring
and resolution model is that the amount of previous-trial
conflict determines the amount of conflict resolution on
the current trial and, thus, that conflict resolution acts in
an anticipatory fashion. We evaluated activity in regions
dorsomedial and dorsolateral prefrontal cortices) on in-
congruent trials and correlated it with rostral cingulate
To control for nonspecific effects of general brain activ-
ity, we removed common correlations between our
regions of interest and a control region, the left supra-
marginal gyrus, which was the most significant task-
responsive region outside of the frontal lobe (cf. Kerns
et al., 2004). We found that incongruent trial activity in
the dorsomedial prefrontal cortex (r = 0.54, p = 0.02),
the right dorsolateral prefrontal cortex (r = 0.57, p =
0.01), and the amygdala (r = 0.52, p = 0.03) all predicted
rostral cingulate activity on the following trial (Figures
5A–5C). Correlations using the left dorsolateral cortex
were nonsignificant (p > 0.2). Performing the correla-
tions without removing the variance shared with the
left supramarginal gyrus did not change the results
(data not shown).
How might activation of the rostral cingulate mediate
the resolution of emotional conflict? One possibility is
through an interaction with the amygdala, a region we
have found to reflect the amount of emotional conflict.
The rostral cingulate projects to the amygdala, and the
amygdala in turn regulates various sites, including the
hypothalamus and through it the sympathetic nervous
system (Bechara et al., 1995; Paxinos, 1990). In experi-
mental animals, inhibition of the amygdala by the medial
prefrontal cortex (Quirk et al., 2003; Rosenkranz and
of learned fear (Delgado et al., 2006; Quirk et al., 2003).
These data suggest that, in humans, increased activity
in the rostral anterior cingulate may also lead to a reduc-
tion in the activity of the amygdala, which would thereby
reduce emotional responsivity.
To examine interregional connectivity, we carried out
a psychophysiological interaction (PPI) analysis (see
Experimental Procedures), in which we examined the
context-specific relationship between the activity in
two regions during each trial type on an acquisition-
nificant inverse relationship between conflict resolution-
related activity (high versus low conflict resolution trials)
in the rostral cingulate and simultaneous activity in the
right amygdala ([16, 0, 216] Z = 2.87; left amygdala
p > 0.1). Shown in red in Figure 6A is the amygdala clus-
ter negatively coupled to the rostral cingulate, while in
blueistheright amygdala clusteractivatedby emotional
lap in yellow. Thus, the region of the amygdala that re-
sponds to emotional conflict overlaps substantially with
the region negatively coupled to the rostral cingulate.
We next examined activity in this overlapping region
to understand the relationship between the amygdala
and the rostral cingulate on each trial type. As shown
in Figure 6B, during high conflict resolution trials,
greater activity in the rostral cingulate predicted re-
duced activity in the amygdala (p = 0.001). By contrast,
during low conflict resolution trials, activity in the rostral
cingulate did not significantly predict activity in the
amygdala (p = 0.2). Finally, we explored whether regula-
tion of activity in the amygdala may arise from a cortical
site apart from the rostral cingulate. Previous work has
suggested an inverse relationship between activation
in the amygdala and activation in the dorsal cingulate
Figure 5. Recruitment of the Rostral Cingu-
late Is Predicted in an Anticipatory Fashion
by Previous Trial Activity in Regions Tracking
The amount of conflict signal in the dorsome-
dial prefrontal cortex (A), dorsolateral pre-
frontal cortex (B), or amygdala (C) on incon-
gruent trials correlates with activity in the
rostral cingulate (rACC) on the following
(i.e., postincongruent) trial. Regional activity
was defined by the average of all suprathres-
hold voxels described in Figures 3 and 4.
Activity in each region has been adjusted for
shared variance with a fourth, task-respon-
sive region outside the frontal lobe to control
for nonspecific effects of individual differ-
ences in the magnitude of brain activation.
Plotted are the postadjustment Z scores.
Neural Circuit for Emotional Conflict Resolution
and dorsomedial prefrontal cortex (Drevets and Raichle,
1998; Hariri et al., 2003). We therefore carried out an-
other psychophysiological interaction analysis using
a dorsomedial prefrontal cortical seed region. However,
no voxels within either the left or right amygdala showed
coupling with the dorsomedial prefrontal cortex (all p
cingulate may resolve emotional conflict in part by de-
creasing engagement of the amygdala by incongruent
To test this idea further, we correlated the magnitude
of rostral cingulate-predicted reduction of amygdalar
resolution. Those individuals whose rostral cingulate
predicted greater amygdalar activity reduction (in the
psychophysiological interaction) showed significantly
greater conflict resolution as measured by high minus
low conflict resolution trial reaction time differences
(r = 0.48, p = 0.04; Figure 6C).
One important function of the amygdala is to recruit
autonomic responses to emotionally salient stimuli by
activation of the sympathetic nervous system through
the amygdala’s hypothalamic projections (Bechara
et al., 1995; Paxinos, 1990). Totest whether suppression
of autonomic responsivity, like suppression of amygda-
lar activity, was related to the success of conflict resolu-
tion, we recorded skin conductance responses (SCR)
from an additional group of ten subjects performing
the emotional conflict resolution task outside of the
scanner (see Experimental Procedures). We found that
greater blunting of the SCR on high conflict resolution
trials relative to low conflict resolution trials predicted
better emotional conflict resolution (r = 0.71, p = 0.02,
see Figure 6D; reaction time previous 3 current trial
ANOVA p = 0.04). These data further support the idea
that the rostral cingulate resolves emotional conflict by
suppressing amygdalar activity and output, which leads
to a blunting of the sympathetic autonomic response
to incongruent emotional distracters.
Effective Connectivity between the Rostral Cingulate
and the Amygdala Is Modulated by Previous-Trial
Since the analysis of functional connectivity above does
not allow inference about the directionality of the rela-
tionship (Friston et al., 2003), we were interested in fur-
ther characterizing the dynamic changes in the direc-
tional interactions between the rostral cingulate and
the amygdala as a function of whether the previous trial
was incongruent and thus signaled the need for greater
conflict resolution on the current trial. To do so, we car-
ried out a dynamic causal modeling analysis. Dynamic
causal modeling can indicate the directionality of inter-
regional interactions, within thecontext ofamodel com-
termined connections (Friston et al., 2003; Penny et al.,
2004). Effects are divided into ‘‘intrinsic’’ connectivity
in the absence of stimulation and modulatory effects as-
sociated with a particular experimental manipulation,
such as the presence of previous-trial incongruency.
In dynamic causal modeling, a significant positive
modulatory effect implies directional ‘‘activation’’ of the
target by the source region, while a significant negative
modulation implies ‘‘inhibition’’ of the target by the
source region (see equation describing these effects in
the Experimental Procedures section). Because of the
spatial resolution of fMRI, these modulatory effects can-
not be directly interpreted as excitatory or inhibitory ef-
interact with each other as a whole in the context of pre-
specified anatomical connectivity (Friston et al., 2003).
We tested a simple dynamic causal model focused on
the interactions between the rostral cingulate and the
amygdala. Input corresponding to all visual stimuli was
allowed to drive both the rostral cingulate and amyg-
dala, as intracerebral recordings in both regions in hu-
mans reveal similar, fast-latency responses to visual
stimulation that can distinguish between aversive and
nonaversive images (Kawasaki et al., 2001, 2005; Kro-
lak-Salmon et al., 2004; Oya et al., 2002). Current-trial in-
congruency was also allowed to modulate the amygdala
to rostral cingulate pathway (i.e., ‘‘forward’’ pathway).
Previous-trial incongruency, a signal that leads to
Figure 6. The rACC Is Negatively Coupled with the Amygdala, and
the Strength of This Coupling and of the Blunting of Autonomic Re-
sponsivity Predict Successful Conflict Resolution
(A) Overlay of the right amygdala cluster showing significant nega-
olution trials (red, p < 0.005, uncorrected) and the right amygdala
cluster activated by emotional conflict from Figure 3A (blue). Over-
lapping voxels are displayed in yellow.
(B) Psychophysiological interaction (6SEM) between the rACC and
the amygdala separately during low and high conflict resolution
(C) Greater rACC-predicted reduction in amygdalar activity in high
versus low conflict resolution trials (more negative high minus low
olution (more negative high minus low conflict resolution trial RT
differences; r = 0.48, p = 0.04).
(D) Greater dampening of autonomic responsivity (more negative
high minus low conflict resolution trial SCR differences) predicted
better conflict resolution (r = 0.71, p = 0.02).
increased current-trial conflict resolution, was allowed
to modulate the ‘‘forward’’ and ‘‘backward’’ pathways
between the rostral cingulate and the amygdala.
We found that the intrinsic connections between the
rostral cingulate and the amygdala were not significant
at baseline (p > 0.68 for both; Figure 7). Crucially, how-
ever, previous-trial incongruency led to a significant
‘‘inhibitory’’ modulation in the activation of the back-
(p = 0.02). Modulation of the amygdala to rostral cingu-
late pathway was not significant for either current- or
previous-trial incongruency (p > 0.74 for both). Further-
more, previous-trial incongruency modulated the back-
ward connection more robustly than the forward con-
nection (p < 0.05). These results suggest that the
triggering of increased current-trial conflict resolution
by previous-trial incongruency is associated with a spe-
cific enhancement of a top-down inhibitory pathway
from the rostral cingulate to the amygdala. These results
are consistent with previous animal studies of medial
2006; Quirk et al., 2003; Rosenkranz and Grace, 2002).
We here report an approach for analyzing the neural
mechanisms involved in the monitoring and resolution
of emotional conflict. We find that response conflict
arising from emotional incongruence leads torobust be-
havioral interference in healthy subjects and that this in-
terference can be reduced through an anticipatory con-
flict-resolution mechanism that is recruited in response
to conflict on the current trial. Activity in the amygdala
and dorsomedial and dorsolateral prefrontal cortices
tracked the amount of emotional conflict created by
emotionally incompatible stimuli, while activity in the
rostral anterior cingulate cortex was associated with
the resolution of this conflict. We also observed that ac-
tivity in the amygdala and dorsomedial and dorsolat-
eral prefrontal cortices on conflict trials directly
predicted resolution-related activity in the rostral cin-
gulate on the following trial, consistent with the se-
quential, anticipatory nature of conflict monitoring
and resolution processes. Furthermore, we found that
activation of the rostral cingulate during high conflict
resolution trials was accompanied by a concomitant
reduction in amygdalar activity. The degree to which
rostral cingulate activation predicted reduced amygda-
lar activity, as well as the reduction in autonomic re-
sponsivity, a function regulated by the amygdala, was
related to subjects’ behavioral success at emotional
conflict resolution. By analyzing effective connectivity
between the amygdala and rostral cingulate we found
that previous-trial incongruency, which leads to greater
current-trial conflict resolution, was associated specif-
ically with activation of an ‘‘inhibitory’’ top-down path-
way from the rostral cingulate to the amygdala. These
findings advance our understanding of the mecha-
nisms by which amygdala activity is regulated for the
purpose of resolving emotional conflict and, in particu-
lar, the function of the rostral cingulate during emo-
Generation of Emotional Conflict by the Amygdala
In our task, unintended processing of the distracter
word and its emotional significance in an incongruent
stimulus led to activation of an emotional representation
incompatible with that of the intended face target. Acti-
vation of incompatible emotional representations leads
to a competition for some of the same neural resources
in the processing of the face and word. This conflict for
processing resources isrepresented in thebrain bothas
emotional conflict and consequently as response
conflict. Our data indicate that activity tracking of the
amount of conflict (either because regions generate or
monitor conflict) can be seen in the amygdala and dor-
somedial and dorsolateral prefrontal cortices. Because
the amygdala is associated with affective processes
and the dorsomedial and dorsolateral prefrontal corti-
ceshavebeen associatedwithnonemotional attentional
processes, it is tempting to suggest that emotional con-
flict is generated in the amygdala, while the resultant re-
sponse conflict involves the dorsomedial and dorsolat-
eral prefrontal cortices, a hypothesis that can be
tested in future work.
The role of the amygdala in the generation of emo-
tional conflict is also supported by the fact that the
amygdala is sensitive to both emotionally valenced
words (Isenberg et al., 1999) and facial expressions
(Breiter et al., 1996; Fitzgerald et al., 2006; Pessoa
et al., 2003), indicating that both task processes (facial
affect identification) and the task-irrelevant word dis-
ies have also found that the amygdala is particularly
sensitive to ambiguity (a type of conflict), both in the
context of interpreting facial expressions (Kim et al.,
2003, 2004) and in the context of uncertainty during
decision making (Hsu et al., 2005).
Figure 7. Previous Trial Incongruency Increases Negative Effective
Connectivity from the Rostral Cingulate to the Amygdala
Coupling coefficients for directional intrinsic connectivity are shown
as values next to the arrows between the rACC and the amygdala,
while coefficients for the modulatory (bilinear) effects are shown
sic paths. The only significant effect in the model was greater nega-
tive modulation of the rACC to amygdala path by previous-trial in-
congruency, which triggers greater current-trial conflict resolution
(p = 0.02).
Neural Circuit for Emotional Conflict Resolution
Emotional Conflict Resolution through Top-Down
Control of the Amygdala by the Rostral Cingulate
The amygdala regulation model described above posits
further that the rostral cingulate is not the emotional an-
alog to the dorsal cingulate (Bishop et al., 2004; Whalen
et al., 1998), but rather is associated with the resolution
of emotional conflict. Indeed, these results are consis-
tent with a recent study tying rostral/subgenual cingu-
late activation to fear extinction (Phelps et al., 2004),
a process that may be similar to emotional conflict res-
olution. Rostral cingulate activation has also been ob-
served during placebo anxiety reduction, a process in
which control over an emotional stimulus (an aversive
picture) is recruited to diminish the effect of the emo-
tional stimulus (Petrovic et al., 2005).
The top-down directionality of the rostral cingulate-
amygdala interaction that we propose is supported by
an extensive animal literature on the ability of the medial
prefrontal cortex to inhibit the amygdala (Delgado et al.,
2006; Quirk et al., 2003; Rosenkranz and Grace, 2002).
However, one must also consider an alternative ‘‘bot-
tom-up’’ model, in which the amygdala can be thought
to normally exert an inhibitory effect on the medial pre-
frontal cortex, as has been suggested on the basis of
fear-conditioning data in rodents (Garcia et al., 1999).
Were the amygdala to habituate specifically during
high conflict resolution trials (due to the repetition of
incongruent stimuli not seen in low conflict resolution
trials), decreased amygdalar activity would lead to de-
creased rostral cingulate inhibition, and as a conse-
quence the rostral cingulate would be ‘‘disinhibited,’’
thus appearing to be activated. Therefore, with respect
to the amygdala and rostral cingulate, the bottom-up
as a ‘‘top-down’’ model wherein the rostral cingulate
inhibits the amygdala.
An important argument against the bottom-up model
stems from the condition-specific psychophysiological
interaction data. Specifically, were the rostral cingulate
to inhibit the amygdala only when it is activated during
the high conflict resolution trials (i.e., a top-down
model), then negative coupling between the rostral cin-
gulate and the amygdala would only be seen during
these high conflict resolution trials, as we observed
(Figure 6B). The bottom-up model, by contrast, would
predict greater negative coupling in low compared to
high conflict resolution trials. This is because amygdala
inhibitionoftherostral cingulate would likely begreatest
when amygdala activation is greatest (i.e., during low
conflict resolution trials). The bottom-up model is there-
fore inconsistent with our findings (Figure 6B).
To further strengthen a top-down inhibitory model of
rostral cingulate function in regulating the amygdalar re-
sponse to emotional conflict, we sought confirmatory
evidence by examining effective connectivity between
these two regions using dynamic causal modeling. We
found that previous-trial incongruency, which triggers
greater current-trial conflict resolution, led to specific
strengthening of a pathway from the rostral cingulate
to the amygdala. Moreover, this pathway was inhibitory,
as increasesinrostral cingulate activity wereassociated
also in line with a considerable amount of animal work,
dial prefrontal cortex leads to inhibition of both sponta-
neous and evoked activity in the amygdala (Quirk et al.,
2003; Rosenkranz and Grace, 2002).
A Neural Circuit for the Monitoring and Resolution
of Emotional Conflict
In contrast to the conflict resolution function of the ros-
tral cingulate, we found that emotional conflict monitor-
ing was instead associated with a dorsomedial prefron-
tal region. Thus, we suggest that the dorsal cingulate/
dorsomedial prefrontal cortex may have a conserved
role in response conflict monitoring, regardless of
whether the source of response conflict is cognitive or
emotional. These data argue against a strict functional
division of the cingulate/dorsomedial prefrontal cortex
into a ventral affective and a dorsal cognitive compo-
nent (Bush et al., 2000).
Interestingly, we find robust activity tracking the
amount of conflict bilaterally in the dorsolateral prefron-
tal cortex. These data seem to differ qualitatively from
studies of cognitive conflict, which have shown dorso-
lateral prefrontal cortical activity to be associated with
conflict resolution, rather than monitoring (Egner and
Hirsch, 2005a, 2005b; Kerns et al., 2004; MacDonald
et al., 2000). However, recent meta-analyses report
that the dorsolateral prefrontal cortex is activated by ef-
fortful attentional processing associated with incongru-
ent stimuli in the Stroop task, task-switching, and high
working memory loads (Derrfuss et al., 2005; Duncan
and Owen, 2000; Wager et al., 2004). Therefore, the dor-
solateral prefrontal activations associated with conflict
in the current study may simply reflect generic effects
of task difficulty. Whether lateral frontal regions display
dissociable effects in cognitive versus emotional con-
flict processing can be addressed in future work. In
addition, it will be important to explore how the rostral
cingulate-mediated emotional regulation mechanism
described here relates to the dorsolateral prefrontal-
mediated emotion reappraisal mechanism described by
others (Ochsner and Gross, 2005).
ferences intrait anxietypredicted reactiontimesand ac-
tivation of the amygdala only when fearful faces were
processed unconsciously, not when they were pro-
cessed consciously (Etkin et al., 2004). This suggested
that the unconscious biases in activation of the amyg-
dala may be subject to secondary regulation by con-
tial evidence that the rostral cingulate may be a key
regulator of amygdalar activity and autonomic respon-
sivity and that this regulation is related to the behavioral
success of emotional conflict resolution. Taken to-
in healthy subjects involves two distinct stages—an ini-
tial unconscious, anxiety-related bias reflected in amyg-
by a secondary context-responsive suppression of
amygdalar responsiveness by the rostral cingulate.
Relevance to Mood and Anxiety Disorders
Our experiments on healthy subjects were carried out
in order to understand what role the rostral cingulate
normally plays in nonpathological emotional conflict.
But the data also allow us to better understand a variety
aggerated interference from emotional distracters (Wil-
liams et al., 1996). Patients with post-traumatic stress
disorder (PTSD),forexample,consistently showahypo-
active rostral cingulate during trauma recall (Hull, 2002)
and in tasks involving emotional processing or distrac-
tion. In PTSD, the severity of the symptoms also corre-
lates with the degree of rostral cingulate hypoactivation
(Shin et al., 2005). In depression, resistance to treatment
is associated with hypoactivity of the rostral cingulate
(Kumari et al., 2003). Indeed, lower rostral cingulate
activity prior to treatment actually predicts a poor re-
sponse to antidepressant therapy (reviewed in Etkin
et al., 2005). As would be predicted by our results, in
both depression and PTSD, hyperactivation of the
amygdala occurs to both conscious and unconscious
threat (Davidson et al., 2002; Hull, 2002; Rauch et al.,
2000; Sheline et al., 2001). Taken together, these find-
ings suggest that elevated amygdalar activity and exag-
gerated behavioral interference may be due to deficient
amygdalar inhibition by the rostral cingulate, which
leads to an inability to deal with emotional conflict. The
capacity for recruitment of the rostral cingulate may
thus determine how well an individual can cope with
the intrusion of negative emotional stimuli or mental
Nineteen healthy volunteers (ten females, nine males; average age
26.6 [SD 5.2]) took part in the fMRI study, and ten healthy volunteers
(three females, seven males; average age 25.1 [SD 7.2]) took part in
the skin conductance study, all after giving their informed consent
according to institutional guidelines for protection of human sub-
jects (Columbia University).
Emotional conflict resolution task: Stimuli were presented with Pre-
sentation software (Neurobehavioral Systems, http://nbs.neuro-bs.
com) and during fMRI scanning were displayed on VisuaStim XGA
LCD screen goggles (Resonance Technology, Northridge, CA). The
sion photographs drawn from the set of Ekman and Friesen (Ekman
and Friesen, 1976). Faces were cropped and the words ‘‘FEAR’’ or
‘‘HAPPY’’ written in prominent red letters across the face, such
that word and expression were either congruent or incongruent
(Figure 1A). Stimuli were presented for 1000 ms, with a varying inter-
stimulus interval (ISI) of 3000–5000 ms (mean ISI = 4000 ms) during
which a central fixation cross was shown. Stimuli were presented
in pseudorandom order(counterbalanced forequalnumbersof con-
gruent-congruent, congruent-incongruent, incongruent-congruent,
and incongruent-incongruent stimulus pairings). There were neither
direct repetitions of the same face with varying word distracters, in
order to avoid negative priming effects, nor direct repetitions of ex-
act face-word-distracter combinations, in order to avoid repetition
priming effects (Mayr et al., 2003). Genders, identities, and affects
on the faces were randomized throughout the task, and stimulus oc-
currences were counterbalanced across trial types and response
buttons. Subjects were instructed to respond as fast and accurately
as possible, by pushing response buttons corresponding to ‘‘fear’’
(right index finger) or ‘‘happy’’ (right middle finger) for the affect ex-
pressed on the face. Behavioral data were analyzed in SPSS and
consisted of reaction times (excluding error and post-error trials)
and accuracy rate. For the skin conductance experiment, which
took place outside of the scanner, this paradigm was adapted in or-
der to distinguish individual skin conductance responses by using
an ISI of 14000–16000 ms (mean ISI = 15000 ms) and 111 presenta-
tions of happy or fearful expression faces. Stimuli were displayed on
a 16 inch monitor, with the subjects sitting at a distance of w20 in-
ches and responding by pushing standard keyboard buttons.
fMRI Data Acquisition
Functional data were acquired on a 1.5 Tesla GE Signa MRI scanner,
using a gradient-echo, T2*-weighted echoplanar imaging (EPI) with
blood oxygen level-dependent (BOLD) contrast pulse sequence.
Twenty-four contiguous axial slices were acquired along the AC-
PC plane, with a 64 3 64 matrix and 19 cm field of view (voxel size
3 3 3 3 4.5 mm). A total of 397 volumes were acquired (TR = 2000,
TE = 40, flip angle = 60). Structural data were acquired using a 3D
T1-weighted spoiled gradient recalled (SPGR) pulse sequence
with isomorphic voxels (1.5 3 1.5 3 1.5 mm) in a 24 cm field of
view (256 3 256 matrix, 124 slices, TR 34 ms, TE 3 ms).
SCR Data Acquisition
Skin conductance responses were acquired via silver electrodes
(0.5 VDC excitation)attached to the palmar surfaces of the left index
and middle finger. Signal was amplified and low-pass filtered (10 Hz
cut-off) via a SA Bioamp amplifier (James Long Company, Caroga
Lake, NY). Filtered analog SCR data were then digitized and stored
at a sampling rate of 100 Hz, using Powerlab Chart 5 software (ver-
sion 5.02, ADInstruments, Colorado Springs, CO).
fMRI Data Analysis
All images were analyzed using SPM2 (Wellcome Department of Im-
aging Neuroscience, London, UK; see http://www.fil.ion.ucl.ac.uk/
spm/spm2.html) implemented in Matlab 6.5 (Mathworks, Inc., Na-
tick, MA). The first five volumes were excluded from the analysis
to allow for signal stability following onset transients. Data were cor-
rected for differences in slice timing. Images were motion corrected,
coregistered with subjects’ SPGR scans, normalized to the MNI
template space, resampled at 2 3 2 3 2 mm, and smoothed with
a Gaussian kernel of 8 mm3FWHM (Friston et al., 1995a). A 128 s
temporal high-pass filter was applied to the data to remove low-fre-
quency noise. Regressors for the stimulus events (convolved with
a canonical HRF) were created for congruent-congruent, congru-
ent-incongruent, incongruent-congruent, and incongruent-incon-
gruent trial types, with error and post-error trials modeled sepa-
rately. We also included a regressor-of-no-interest reflecting the
mean whole-brain activity on an acquisition-by-acquisition basis.
This modelwas appliedto normalized data across subjects in agen-
eralized linear model approach (Friston et al., 1995b) and submitted
to random-effects analyses using one-sample t tests.
Our amygdala search volume was defined by a bilateral amygdala
mask using the AAL parameters available in the WFU PickAtlas
(Maldjian et al., 2003), with a 3D dilation factor of 1 to ensure that
we captured the entire amygdala. Group-level contrasts were
thresholded within the amygdala at p = 0.005 (uncorrected) and
a ten voxel spatial extent, which yielded an equivalent correction
for multiple comparisons and enriched for larger activation clusters
(Forman et al., 1995). Our frontal cortical search volume was defined
by a bilateral medial and lateral frontal gray matter mask using the
WFU PickAtlas. We searched for significant voxels that met both
the p = 0.005 threshold and a more stringent p = 0.001 threshold,
also with a ten voxel spatial extent. As has been the custom in pre-
vious research (Phelps et al., 2004), we employed a more stringent
threshold in the cortex than in the amygdalae, as cortical activations
are known to be relatively easier to detect than amygdalar activa-
tions. Reported voxels correspond to standardized Montreal Neuro-
logical Institute (MNI) coordinate space. For the displayed section,
the right side of the subject is the right side of the image.
For the psychophysiologic interaction (PPI) analyses (Friston
et al., 1997), we extracted the deconvolved time course from a
5 mm radius sphere around the group peak activation voxel for the
rostral cingulate [210, 48, 0]. Dorsomedial prefrontal activity was
similarly extracted from a 5 mm radius sphere around its group
peak voxel [22, 38, 38]. Activity within the amygdala mask was
then regressed on a voxel-wise basis against the product of this
time course and the vector of the psychological variable of interest,
with the physiological and the psychological variables serving as
regressors of no interest. The results were then taken to
Neural Circuit for Emotional Conflict Resolution
using amygdala activity extracted from the cluster showing both
negative coupling with the rostral cingulate and greater conflict-
related activity (i.e., yellow cluster in Figure 6A) and was performed
using an ordinary least-squares algorithm implemented in Matlab.
Effective connectivity analyses were implemented using the dy-
namic causal modeling tool in SPM2 (Friston et al., 2003). Predic-
tions about the observed data consist of a model with combined in-
trinsic connectivity in the absence of experimental manipulation and
bilinear modulation, which reflects the effects of experimental vari-
ables. These effects were modeled by the following simplified equa-
tion (Friston et al., 2003):
in which dz1/dt is the state vector per unit time for the target region
and z2corresponds to time series data from the source region. uiin-
dicates the direct input to the model (i.e., all correct trials), while um
indicates input from the modulatory variable onto intrinsic pathways
specified by the model. Activity in the target region is therefore de-
termined by an additive effect of the intrinsic connectivity with the
source region (Az2), the bilinear variable (umBz2, corresponding to
the modulatory experimental manipulation), and the effects of direct
input into the model (Cui).
Time series data were extracted from a 6 mm diameter sphere
around each individual’s peak voxel within the rostral cingulate
resolution). Search volumes for these contrasts were the rostral cin-
gulate cluster defined in the group contrast and the AAL-defined
and two-sample t tests at the group level to determine significance.
SCR Data Analysis
Analysis was done in Matlab. The data were downsampled to 10 Hz,
a derivative of the time course was calculated, and local maxima
and minima were identified. We analyzed the magnitude of the
SCR (max-min) for all events if they began within 4 s of the stimulus
onset, were not associated with error and post-error trials, and
were >0.02 mS. SCR magnitudes were then subjected to ANOVA
and regression analyses in SPSS.
We would like to thank Chris Kelly for writing the SCR analysis script
and Chris Summerfield, Emily Stern, and Ethan Kross for comments
on the manuscript. This work was funded (in part) by the Howard
Hughes Medical Institute, the Kavli Institute for Brain Sciences, the
Neurobiology and Behavior Research Training Program (A.E.,
NICHD HD 07430), a Doris Duke fellowship (D.M.P.), and by Johnson
& Johnson (J.H.).
Received: February 8, 2006
Revised: June 12, 2006
Accepted: July 28, 2006
Published: September 6, 2006
Algom, D., Chajut, E., and Lev, S. (2004). A rational look at the emo-
tional stroop phenomenon: a generic slowdown, not a stroop effect.
J. Exp. Psychol. Gen. 133, 323–338.
Bechara, A., Tranel, D., Damasio, H., Adolphs, R., Rockland, C., and
Damasio, A.R. (1995). Double dissociation of conditioning and de-
clarative knowledge relative to the amygdala and hippocampus in
humans. Science 269, 1115–1118.
tal cortical function and anxiety: controlling attention to threat-
related stimuli. Nat. Neurosci. 7, 184–188.
(1999). Conflict monitoring versus selection-for-action in anterior
cingulate cortex. Nature 402, 179–181.
Botvinick, M.M., Braver, T.S., Barch, D.M., Carter, C.S., and Cohen,
J.D. (2001). Conflict monitoring and cognitive control. Psychol. Rev.
Botvinick, M.M., Cohen, J.D., and Carter, C.S. (2004). Conflict mon-
itoring and anterior cingulate cortex: an update. Trends Cogn.Sci.8,
Breiter, H.C., Etcoff, N.L., Whalen, P.J., Kennedy, W.A., Rauch, S.L.,
Buckner, R.L., Strauss, M.M., Hyman, S.E., and Rosen, B.R. (1996).
Response and habituation of the human amygdala during visual
processing of facial expression. Neuron 17, 875–887.
Broadbent, D.E. (1958). Perception and Communication (London:
Bush, G., Luu, P., and Posner, M.I. (2000). Cognitive and emotional
Carmichael, S.T., and Price, J.L. (1995). Limbic connections of
the orbital and medial prefrontal cortex in macaque monkeys.
J. Comp. Neurol. 363, 615–641.
Carroll, N.C., and Young, A.W. (2005). Priming of emotion recogni-
tion. Q. J. Exp. Psychol. A 58, 1173–1197.
Carter, C.S., Botvinick, M.M., and Cohen, J.D. (1999). The contribu-
tion of the anterior cingulate cortex to executive processes in cogni-
tion. Rev. Neurosci. 10, 49–57.
Compton,R.J.,Banich, M.T.,Mohanty,A., Milham, M.P., Herrington,
J.,Miller, G.A.,Scalf,P.E.,Webb, A.,andHeller,W. (2003).Payingat-
tention to emotion: an fMRI investigation of cognitive and emotional
stroop tasks. Cogn. Affect. Behav. Neurosci. 3, 81–96.
Critchley, H.D., Mathias, C.J., Josephs, O., O’Doherty, J., Zanini, S.,
Dewar, B.K., Cipolotti, L., Shallice, T., and Dolan, R.J. (2003). Human
cingulate cortex and autonomic control: converging neuroimaging
and clinical evidence. Brain 126, 2139–2152.
Davidson, R.J., Pizzagalli, D., Nitschke, J.B., and Putnam, K. (2002).
Depression: perspectives from affective neuroscience. Annu. Rev.
Psychol. 53, 545–574.
Delgado,M.R.,Olsson, A.,andPhelps,E.A. (2006).Extending animal
models of fear conditioning to humans. Biol. Psychol. 73, 39–48.
Derrfuss, J., Brass, M., Neumann, J., and von Cramon, D.Y. (2005).
Involvement of the inferior frontal junction in cognitive control:
meta-analyses of switching and Stroop studies. Hum. Brain Mapp.
Devinsky, O., Morrell, M.J., and Vogt, B.A. (1995). Contributions of
anterior cingulate cortex to behaviour. Brain 118, 279–306.
Drevets, W.C., and Raichle, M.E. (1998). Reciprocal suppression of
regional cerebral bloodflow during emotional versus higher cogni-
tive processes: implications for interactions between emotion and
cognition. Cogn. Emot. 12, 353–385.
Duncan, J., and Owen, A.M. (2000). Common regions of the human
frontal lobe recruited by diverse cognitive demands. Trends Neuro-
sci. 23, 475–483.
Egner, T., and Hirsch, J. (2005a). Cognitive control mechanisms re-
tion. Nat. Neurosci. 8, 1784–1790.
Egner, T., and Hirsch, J. (2005b). The neural correlates and func-
tional integration of cognitive control in a Stroop task. Neuroimage
Ekman, P., and Friesen, W.V. (1976). Pictures of Facial Affect (Palo
Alto, CA: Consulting Psychologists).
Etkin, A., Klemenhagen, K.C., Dudman, J.T., Rogan, M.T., Hen, R.,
Kandel,E.R., and Hirsch, J. (2004). Individual differences in trait anx-
iety predict the response of the basolateral amygdala to uncon-
sciously processed fearful faces. Neuron 44, 1043–1055.
Etkin, A., Pittenger, C., Polan, H.J., and Kandel, E.R. (2005). Toward
tions. J. Neuropsychiatry Clin. Neurosci. 17, 145–158.
K.L. (2006). Beyond threat: amygdala reactivity across multiple ex-
pressions of facial affect. Neuroimage 30, 1441–1448.
Forman, S.D., Cohen, J.D., Fitzgerald, M., Eddy, W.F., Mintun, M.A.,
and Noll, D.C. (1995). Improved assessment of significant activation
in functional magnetic resonance imaging (fMRI): use of a cluster-
size threshold. Magn. Reson. Med. 33, 636–647.
Friston, K.J., Ashburner, J., Poline, J.B., Frith, C.D., Heather, J.D.,
and Frackowiak, R.S. (1995a). Spatial registration and normalization
of images. Hum. Brain Mapp. 2, 165–189.
Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.B., Frith, C.D.,
and Frackowiak, R.S. (1995b). Statistical parametric maps in func-
tional imaging: a general linear approach. Hum. Brain Mapp. 2,
Friston, K.J., Buechel, C., Fink, G.R., Morris, J., Rolls, E., and Dolan,
R.J. (1997). Psychophysiological and modulatory interactions in
neuroimaging. Neuroimage 6, 218–229.
Friston, K.J., Harrison, L., and Penny, W. (2003). Dynamic causal
modelling. Neuroimage 19, 1273–1302.
Garcia, R., Vouimba, R.M., Baudry, M., and Thompson, R.F. (1999).
The amygdala modulates prefrontal cortex activity relative to condi-
tioned fear. Nature 402, 294–296.
Gratton, G., Coles, M.G., and Donchin, E. (1992). Optimizing the use
of information: strategic control of activation of responses. J. Exp.
Psychol. Gen. 121, 480–506.
Hariri, A.R., Mattay, V.S., Tessitore, A., Fera, F., and Weinberger,
D.R. (2003). Neocortical modulation of the amygdala response to
fearful stimuli. Biol. Psychiatry 53, 494–501.
Hsu, M., Bhatt, M., Adolphs, R., Tranel, D., and Camerer, C.F. (2005).
Neural systems responding to degrees of uncertainty in human
decision-making. Science 310, 1680–1683.
Hull, A.M. (2002). Neuroimaging findings in post-traumatic stress
disorder. Systematic review. Br. J. Psychiatry 181, 102–110.
Isenberg, N., Silbersweig, D., Engelien, A., Emmerich, S., Malavade,
K., Beattie, B., Leon, A.C., and Stern, E. (1999). Linguistic threat
activates the human amygdala. Proc. Natl. Acad. Sci. USA 96,
James, W. (1890). The Principles of Psychology (Cambridge, MA:
Harvard University Press). Reprinted in 1981.
Kawasaki, H., Kaufman, O., Damasio, H., Damasio, A.R., Granner,
M., Bakken, H., Hori, T., Howard, M.A., IIIrd, and Adolphs, R.
(2001). Single-neuron responses to emotional visual stimuli re-
corded in human ventral prefrontal cortex. Nat. Neurosci. 4, 15–16.
Kawasaki, H., Adolphs, R., Oya, H., Kovach, C., Damasio, H., Kauf-
man, O., and Howard, M., IIIrd. (2005). Analysis of single-unit re-
sponses to emotional scenes in human ventromedial prefrontal cor-
tex. J. Cogn. Neurosci. 17, 1509–1518.
Kerns, J.G., Cohen, J.D., MacDonald, A.W., IIIrd, Cho, R.Y., Stenger,
V.A., and Carter, C.S. (2004). Anterior cingulate conflict monitoring
and adjustments in control. Science 303, 1023–1026.
Kim, H., Somerville, L.H., Johnstone, T., Alexander, A.L., and
Whalen, P.J. (2003). Inverse amygdala and medial prefrontal cortex
responses to surprised faces. Neuroreport 14, 2317–2322.
Kim, H., Somerville, L.H., Johnstone, T., Polis, S., Alexander, A.L.,
Shin, L.M., and Whalen, P.J. (2004). Contextual modulation of amyg-
dala responsivity to surprised faces. J. Cogn. Neurosci. 16, 1730–
guiere, F. (2004). Early amygdala reaction to fear spreading in occip-
ital, temporal, and frontal cortex: a depth electrode ERP study in
human. Neuron 42, 665–676.
Kumari, V., Mitterschiffthaler, M.T., Teasdale, J.D., Malhi, G.S.,
Brown, R.G., Giampietro, V., Brammer, M.J., Poon, L., Simmons,
A., Williams, S.C., et al. (2003). Neural abnormalities during cognitive
generation of affect in treatment-resistant depression. Biol. Psychi-
atry 54, 777–791.
LeDoux, J.E. (2000). Emotion circuits in the brain. Annu. Rev. Neuro-
sci. 23, 155–184.
MacDonald, A.W., IIIrd, Cohen, J.D., Stenger, V.A., and Carter, C.S.
(2000). Dissociating the role of the dorsolateral prefrontal and ante-
rior cingulate cortex in cognitive control. Science 288, 1835–1838.
MacLeod, C.M. (1991). Half a century of research on the Stroop ef-
fect: an integrative review. Psychol. Bull. 109, 163–203.
Maldjian, J.A., Laurienti, P.J., Kraft, R.A., and Burdette, J.H. (2003).
An automated method for neuroanatomic and cytoarchitectonic
atlas-based interrogation of fMRI data sets. Neuroimage 19, 1233–
Mathews, A. (1990). Why worry? The cognitive function of anxiety.
Behav. Res. Ther. 28, 455–468.
Mathews, A., and MacLeod,C. (1985). Selective processing of threat
cues in anxiety states. Behav. Res. Ther. 23, 563–569.
Mayr, U., Awh, E., and Laurey, P. (2003). Conflict adaptation effects
in the absence of executive control. Nat. Neurosci. 6, 450–452.
McKenna, F.P. (1986). Effects of unattended emotional stimuli on
color-naming performance. Curr. Psychol. Res. Rev. 5, 3–9.
Musil, S.Y., and Olson, C.R. (1988a). Organization of cortical and
subcortical projections to medial prefrontal cortex in the cat.
J. Comp. Neurol. 272, 219–241.
Musil, S.Y., and Olson, C.R. (1988b). Organization of cortical and
subcortical projections to the anterior cingulate cortex in the cat.
J. Comp. Neurol. 272, 203–218.
Nachev, P., Rees, G., Parton, A., Kennard, C., and Husain, M. (2005).
Volition and conflict in human medial frontal cortex. Curr. Biol. 15,
Ochsner, K.N., and Gross, J.J. (2005). The cognitive control of emo-
tion. Trends Cogn. Sci. 9, 242–249.
Oya, H., Kawasaki, H., Howard, M.A., IIIrd, and Adolphs, R. (2002).
emotion categories of complex visual stimuli. J. Neurosci. 22, 9502–
Paxinos, G. (1990). Human Nervous System (San Diego: Academic
Penny, W.D., Stephan, K.E., Mechelli, A., and Friston, K.J. (2004).
Modelling functional integration: a comparison of structural equa-
tion and dynamic causal models. Neuroimage 23 (Suppl 1), S264–
Pessoa, L., Kastner, S., and Ungerleider, L.G. (2003). Neuroimaging
studies of attention: from modulation of sensory processing to top-
down control. J. Neurosci. 23, 3990–3998.
Petrovic, P., Dietrich, T., Fransson, P., Andersson, J., Carlsson, K.,
and Ingvar, M. (2005). Placebo in emotional processing–induced
expectations of anxiety relief activate a generalized modulatory net-
work. Neuron 46, 957–969.
Phelps, E.A., Delgado, M.R., Nearing, K.I., and LeDoux, J.E. (2004).
Extinction learning in humans: role of the amygdala and vmPFC.
Neuron 43, 897–905.
Posner, M.I., and DiGirolamo, G.J. (1998). Executive Attention: Con-
flict, Target Detection and Cognitive Control. In The Attentive Brain,
R. Parasuraman, ed. (Cambridge, MA: MIT Press), pp. 401–423.
Posner, M.I., Petersen, S.E., Fox, P.T., and Raichle, M.E. (1988). Lo-
calization of cognitive operations in the human brain. Science 240,
of medial prefrontal cortex decreases the responsiveness of central
amygdala output neurons. J. Neurosci. 23, 8800–8807.
Raichle, M.E., MacLeod, A.M., Snyder, A.Z., Powers, W.J., Gusnard,
D.A., and Shulman, G.L. (2001). A default mode of brain function.
Proc. Natl. Acad. Sci. USA 98, 676–682.
Lasko, N.B., Orr, S.P., and Pitman, R.K. (2000). Exaggerated amyg-
dalaresponsetomasked facial stimuli inposttraumaticstressdisor-
der: a functional MRI study. Biol. Psychiatry 47, 769–776.
Ridderinkhof, K.R., Ullsperger, M., Crone, E.A., and Nieuwenhuis, S.
(2004). The role of the medial frontal cortex in cognitive control. Sci-
ence 306, 443–447.
Rosenkranz, J.A., and Grace, A.A. (2002). Cellular mechanisms of in-
fralimbic and prelimbic prefrontal cortical inhibition and dopaminer-
gic modulation of basolateral amygdalaneurons in vivo. J. Neurosci.
Rushworth, M.F., Walton, M.E., Kennerley, S.W., and Bannerman,
D.M. (2004). Action sets and decisions in the medial frontal cortex.
Trends Cogn. Sci. 8, 410–417.
Neural Circuit for Emotional Conflict Resolution
Sheline, Y.I., Barch, D.M., Donnelly, J.M., Ollinger, J.M., Snyder, Download full-text
A.Z., and Mintun, M.A. (2001). Increased amygdala response to
pressant treatment: an fMRI study. Biol. Psychiatry 50, 651–658.
Martis, B., Macklin, M.L., Lasko, N.B., Cavanagh, S.R.,Krangel, T.S.,
et al. (2005). A functional magnetic resonance imaging study of
amygdala and medial prefrontal cortex responses to overtly pre-
sented fearful faces in posttraumatic stress disorder. Arch. Gen.
Psychiatry 62, 273–281.
Stroop, J.R. (1935). Studies of interference in serial verbal reactions.
J. Exp. Psychol. 18, 643–662.
Tipples, J., and Sharma, D. (2000). Orienting to exogenous cues and
attentional bias to affective pictures reflect separate processes.
Br. J. Psychol. 91, 87–97.
van Hoesen, G.W., Morecraft, R.J., and Vogt, B.A. (1993). Connec-
tions of the Monkey Cingulate Cortex. In Neurobiology of Cingulate
Cortex and Limbic Thalamus: A Comprehensive Handbook, B.A.
Vogt and M. Gabriel, eds. (Boston, MA: Birkhauser), pp. 249–284.
Vogt, B.A., Finch, D.M., and Olson, C.R. (1992). Functional heteroge-
neity in cingulate cortex: the anterior executive and posterior evalu-
ative regions. Cereb. Cortex 2, 435–443.
Vuilleumier, P., Armony, J.L., Driver, J., and Dolan, R.J. (2001). Ef-
fects of attention and emotion on face processing in the human
brain: an event-related fMRI study. Neuron 30, 829–841.
ies of shifting attention: a meta-analysis. Neuroimage 22, 1679–
Weissman, D.H., Warner, L.M., andWoldorff, M.G.(2004). Theneural
mechanisms for minimizing cross-modal distraction. J. Neurosci.
Whalen, P.J., Bush, G., McNally, R.J., Wilhelm, S., McInerney, S.C.,
Jenike, M.A., and Rauch, S.L. (1998). The emotional counting Stroop
paradigm: a functional magnetic resonance imaging probe of the
anterior cingulate affective division. Biol. Psychiatry 44, 1219–1228.
Williams, J.M., Mathews, A., and MacLeod, C. (1996). The emotional
Stroop task and psychopathology. Psychol. Bull. 120, 3–24.
Yang, T.T., Menon, V., Eliez, S., Blasey, C., White, C.D., Reid, A.J.,
Gotlib, I.H., and Reiss, A.L. (2002). Amygdalar activation associated
with positive and negative facial expressions. Neuroreport 13,