Brain responses to dynamic facial expressions of pain
Daniela Simona,b,*, Kenneth D. Craigc, Wolfgang H.R. Miltnera, Pierre Rainvilleb
aDepartment of Biological and Clinical Psychology, Friedrich-Schiller-University, Jena, Germany
bDe ´partement Stomatologie, Universite ´ de Montre ´al, Montre ´al, Canada
cDepartment of Psychology, University of British Columbia, Vancouver, Canada
Received 1 June 2006; received in revised form 24 August 2006; accepted 30 August 2006
The facial expression of pain is a prominent non-verbal pain behaviour, unique and distinct from the expression of basic emo-
tions. Yet, little is known about the neurobiological basis for the communication of pain. Here, subjects performed a sex-discrim-
ination task while we investigated neural responses to implicit processing of dynamic visual stimuli of male or female faces
displaying pain or angry expressions, matched on expression intensity and compared to neutral expression. Stimuli were presented
in a mixed blocked/event-related design while blood oxygenation level dependent (BOLD) signal was acquired using whole-brain
functional magnetic resonance imaging (fMRI) at 1.5 Tesla. Comparable sustained responses to pain and angry faces were found
in the superior temporal sulcus (STS). Stronger transient activation was also observed to male expression of pain (Vs neutral
and anger) in high-order visual areas (STS and fusiform face area) and in emotion-related areas including the amygdala (highest
peak t-value = 10.8), perigenual anterior cingulate cortex (ACC), and SI. Male pain compared to anger expression also activated
the ventromedial prefrontal cortex, SII/posterior insula and anterior insula. This is consistent with the hypothesis that the implicit
processing of male pain expression triggers an emotional reaction characterized by a threat-related response. Unexpectedly, several
areas responsive to male expression, including the amygdala, perigenual ACC, and somatosensory areas, showed a decrease in
activation to female pain faces (Vs neutral). This sharp contrast in the response to male and female faces suggests potential differences
in the socio-functional role of pain expression in males and females.
? 2006 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
Keywords: Facial expression of pain; FACS; Face recognition; Amygdala; Threat
The detection and interpretation of facial expressions
is critical to our appreciation of the social environment.
It relies on a network including the fusiform gyrus
(FFA) and the superior temporal sulcus (STS), respon-
sible for the visual analysis of faces. It moreover com-
somatosensory cortices, insula and anterior cingulate
cortex (ACC), all suggested to be involved in processing
the emotional content of a face (reviewed in Adolphs,
2002). Among these areas, the amygdala is a key struc-
ture playing a pivotal role in the automatic detection of
socially relevant and threatening facial expressions of
emotion (for review see Zald, 2003).
Pain is a fundamental experience associated with an
actual threat to the body’s integrity. Beside the robustly
observed neuronal response in a network including the
primary and secondary somatosensory cortices (SI,
SII), ACC, insula and regions of the frontal cortex
(for review see Apkarian et al., 2005), experimental pain
of sufficient intensity is characterized by a motor
response including a facial expression. Although pain
and emotion-related processes clearly overlap to some
extent (Rainville, 2004), pain experiences always involve
a sensory dimension and pain expression has been
0304-3959/$32.00 ? 2006 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
*Corresponding author. Tel.: +49 30 2093 4823; fax: + 49 30 2093
E-mail address: firstname.lastname@example.org (D. Simon).
Pain 126 (2006) 309–318
described to be unique and distinct from those of basic
emotions (Craig et al., 2001).
Although the facial expression of pain is one of the
most prominent non-verbal pain behaviours, little is
known about the neurobiological basis for pain commu-
nication. Evolutionary theories propose that the facial
expression of pain following acute injury might play a
role in survival by automatically alarming onlookers
in situations of immediate threat and/or by eliciting
empathic behaviour towards the individual experiencing
pain (Williams, 2002).
Recently, Botvinick et al. (2005) found an overlap of
brain responses in the self produced by thermal painful
stimulation and pain perceived in others by viewing film
clips of patients displaying pain behaviours (including
facial expression). In line with other studies examining
explicit processing of vicarious pain, activation to pain
faces was detected within the supracallosal ACC and
the anterior insula – the affective division of the pain
matrix (also see Hutchison et al., 1999; Morrison
et al., 2004; Singer et al., 2004; Jackson et al., 2005). This
may reflect an empathic response.
Currently, there is no study that examined brain
responses to pain expression under implicit conditions;
i.e. when processing capacity for pain-related cues is
partly consumed by a competing task. Hence, the pres-
ent study assesses neural responses to pain faces under
such conditions. It also constitutes a first attempt to
determine the specificity of brain responses to pain
expression in contrast to threatening angry faces.
An automatic threat-related response in the amygdala
during implicit processing of pain faces was hypothe-
sized (e.g. Whalen et al., 2001). Additionally, taking
the literature on processing of emotional facial expres-
sions into account (e.g. Phan et al., 2002), activation
of MPFC, anterior insula, ACC, SI, SII/posterior insula,
FFA and STS was investigated. By incorporating actors’
and observers’ gender into the analyses possible influ-
ences of sex on BOLD-responses were examined.
2. Materials and methods
Seventeen healthy volunteers (8 males and 9 females, mean
age = 23.1 ± 4.2) participated in the experiment after provid-
ing written informed consent approved by the local Ethics
Committee. All subjects had no history of neurological and
psychiatric impairment, were free of medication, right-handed
and had normal vision or corrected to normal vision. Subjects
received monetary compensation for their participation (40
Participants were shown series of 1-s dynamic visual stimuli
(clips) of 8 actors (4 males, mean age: 24.4 ± 7.5) displaying
pain, angry or neutral facial expressions (see Fig. 1A) and
taken from a newly developed and validated set of stimuli
(Simon et al., under review). The 1sec-stimuli were always
cut backward from the peak of the expression in order to
Fig. 1. (A) Examples of two females and one male actor showing neutral, anger and pain faces. Five time points in the clip (1, 250, 500, 750 and
1000 ms) are displayed. (B) The mixed blocked/event-related design. (a) Randomization of the sex of the model within blocks, (b) single stimulus-
event sequence within each of three blocks with white curves approximating the expected BOLD-response modelled separately for male or female
face stimuli (black/orange, shades of red, or blue/green) (c) blocks of pain (P), anger (A) or neutral (N) faces interspersed with blocks of visual
D. Simon et al. / Pain 126 (2006) 309–318
avoid different lengths and variability of exposure to the visual
stimuli. The specificity of the pain faces compared to facial
expressions of other basic emotions was confirmed in a valida-
tion phase by two trained judges using the Facial Action Cod-
ing System (FACS) (Ekman and Friesen, 1978) and by
subjective evaluations in an independent group of 15 normal
volunteers (inter-rater reliability: Cronbach’s a = 0.97). FACS
coding was not performed as a function of time but reflected
the total amount (frequency and intensity) of facial actions
present in the 1-s clips. Referring to the FACS investigators
guide (Ekman et al., 2002, chapter 12, p.174) the stimuli corre-
sponded well with the prototypes or major variants of the
intended emotional facial expressions (anger = brow lowered,
lid tightened, lip tightened; pain = brow lowered, cheek raised,
lid tightened, upper lip raised, lips stretched and lips part).
Based on the prototypical facial expression, anger stimuli
included an average of 3.38 (SEM = 0.53) target action units
(AUs) and only 0.38 (SEM = 0.26) non-target AUs (89.6%
AUs were part of anger prototypes or their variants). Pain
stimuli included an average of 6.88 (SEM = 0.30) AUs and
only 0.63 (SEM = 0.26) non-target AUs (92.3% AUs were part
of pain prototypes or their variants).
Pain and anger stimuli were selected to match on expres-
sion-intensity ratings (Likert scale ranging from 0 = not at
all to 5 = the most intense possible; Pain Vs Anger:
P > 0.05). Discriminability with other emotions was unequivo-
cal as intensity ratings for non-target emotions (fear, disgust,
sadness, surprise and happiness) were negligible (see Table
1A; range of the means: 0.00–0.62). Although the pain and
anger clips were comparably intense, pain clips were rated as
slightly more arousing (Pain Vs Anger: P = 0.01) and unpleas-
ant than anger clips (Pain Vs Anger: P = 0.001). However, no
interaction with the sex of the actor and emotion was detected
for valence and arousal ratings (Ps > 0.05; see Table 1B).
Moreover, FACS coding confirmed that male and female
expressions were comparable in frequency and intensity of
the expression (P (cor) > 0.05).
2.3. Task design
Subjects were told that the study aimed at investigating the
ability to discriminate gender from briefly presented clips of
different facial expressions. They were instructed to respond
as quickly and accurately as possible in a response window fol-
lowing each stimulus. Our design was hybrid (Mechelli et al.,
2003; Visscher et al., 2003), containing a block structure (facial
expression) and an event-related structure (sex of the actor)
(Fig. 1B). Participants were administered 15 blocks of clips
per face category, interspersed with visual baseline (fixation
cross) during three runs (?8.5 min/run). The order of blocks
was pseudo-randomized within and across runs. Each block
lasted 16 s and included eight trials, starting with the 1-s clip
and followed by a 1-s response window during which subjects
indicated the sex of the actor by pressing one of two keys with
the index and middle finger of the left hand. For each actor one
version of a dynamic stimulus per emotion was repeatedly pre-
sented during the experiment. Clips of male and female actors
were pseudo-randomized within blocks with a resulting mean
stimulus onset asynchrony (SOA) of 4 s (range 2–8 s) with
respect to the events. Both SOA and epoch/block-duration
are comparable to Mechelli et al. (2003) who were able to dif-
ferentiate between sustained and transient BOLD-responses.
This design was implemented to examine the effect of the ‘cat-
egory of facial expression’ (across blocks), ‘gender of the actor’
(across trials), and interactions between those factors. Subjects
were trained in the gender discrimination task prior to the
scanning session using distinct stimulus material.
2.4. Behavioural analysis
Accuracy and latency of gender discrimination were
acquired during the scanning session. Immediately after scan-
ning, each block of stimuli was presented a second time and
subjects provided ratings of overall expression intensity
(0 = ‘not at all’ to 5 = ‘the most intense possible’), valence
Subject’s intensity rating by emotion category
Target expressionExpression perceived
Note. Values correspond to the mean (SEM) intensity rating of each expression category (expression perceived) for each target expression condition
averaged across actors and judges. Mean (SEM) for the target expression is reported in bold. SEM reflects variability between actors.
aMean rating of the target emotion is significantly different from the mean rating on the other scales (Fisher’s protected least significance difference
test; all Ps < 0.001).
Mean (SEM) ratings by emotional category for male and female actors
Target expression Intensity of the target expression Valence of the target expressionArousal of the target expression
Male actorsFemale actorsMale actorsFemale actorsMale actorsFemale actors
D. Simon et al. / Pain 126 (2006) 309–318
(?4 = ‘clearly unpleasant’ to +4 = ‘clearly pleasant’) and
arousal (?4 = ‘highly relaxed’ to +4 ‘high level of arousal’).
Comparisons of stimulus categories on accuracy, latency,
and post-experimental ratings (intensity, valence and arousal)
was performed using repeated measures analysis of variance
(ANOVA) with a Greenhouse–Geisser correction (P < 0.05).
For post-hoc comparisons Bonferroni correction was applied.
2.5. fMRI data acquisition and analysis
Blood oxygenation level-dependent (BOLD) signal was
acquired using functional magnetic resonance imaging (fMRI)
at 1.5 Tesla (Siemens Sonata, Erlangen, Germany). A total of
142 whole-brain volumes were acquired in each of the three
runs using thefollowing
TE = 42 ms, 40 consecutive slices, 3 · 3 · 3 mm voxel, flip
angle = 90?, FOV = 192 mm, 64 · 64 matrix. Analyses were
performed using the software Brainvoyager QX (Version 1.3;
Brain Innovation, Maastricht/Netherlands). Preprocessing
included slice-time correction, realignment, motion correction,
co-registration, spatial normalization and smoothing (10 mm
FWHM Gaussian kernel). Temporal smoothing was done with
a high-pass filter of 5 cycles per time series.
Statistical analysis was performed by multiple linear regres-
sion of the signal time course at each voxel. First, a separate
GLM was specified for each subject including parameter esti-
mates of block and event-related activity at each voxel for each
regressor (block design: neutral, anger, pain; event-related
painfemale). The expected BOLD signal change for each stimula-
response function (modified gamma function of d = 2.5, s =
1.25). Contrast images were calculated by applying appropriate
linear contrasts to the parameter estimates for the parametric
regressor of each event. In order to account for variability
among the differentsubjects andruns, the individual single-sub-
ject contrasts were entered into a second-level random-effects
analysis resulting in t-statistic (df = 16) at each voxel. Further-
parisons (male Vs female observer) were conducted.
Based on earlier studies on facial expressions of emotion
and on pain experience and ‘pain empathy’, directed searches
were conducted specifically on the amygdalae, MPFC, anterior
insula, ACC and somatosensory cortices (SI and SII/posterior
insula). Additionally, the response of the visual areas FFA and
STS was examined. Based on those limited directed-search
areas defined a priori, a statistical threshold of P < 0.005
(uncorrected) was used corresponding to a t-score >3.3. More-
over, in order to protect against type I error, only activation
foci that survived the mentioned threshold within a cluster size
equal to or greater than four contiguous voxels (extent thresh-
old P 108 mm3) were accepted.
parameters: TR = 3500 ms,
3.1. Behavioural data
3.1.1. Ratings of emotion expression
The analysis of expression intensity, valence and
arousal ratings resulted in a main effect of emotional
P < 0.0001; valence: F(2;30)= 16.80, P < 0.0001; arousal:
P < 0.0001].
revealed that subjects perceived anger, as well as pain
faces as more intense, unpleasant and arousing than
neutral faces [intensity: pain Vs neutral: P < 0.0001;
anger Vs neutral: P < 0.0001; valence: pain Vs neutral:
P < 0.0001; anger Vs neutral: P = 0.007; arousal: pain
Vs neutral: P < 0.0001; anger Vs neutral: P < 0.0001],
while pain and anger did not differ significantly from
each other (Ps > 0.05). While intensity and valence rat-
ings were not influenced by the observer’s sex, a main
effect of this factor was detected for arousal ratings
(F(2;15)= 8.14; P = 0.028). This reflected the slightly
higher levels of arousal perceived in the clips by female
The observed high overall performance in the gender
discrimination task (correct response rate: 94.2% ± 1.4)
confirms that subjects attended to the faces. As demon-
strated in Table 2, however, gender discrimination was
significantly influenced by emotional expression (Main
effect of face category: F(2;32)= 5.95; P = 0.015). Pair-
wise comparisons revealed that observers’ judgements
were slightly less accurate to anger and pain than to neu-
tral (pain Vs neutral: P = 0.014; anger Vs neutral:
P = 0.027; pain Vs anger: P = 0.666). The correct
response rate was not influenced by the actors’ or the
observers’ gender. Moreover, facial expressions or gen-
der of the actors and observer did not influence the
latency of the subjects’ responses (Table 2).
4. Neuroimaging data
4.1. Block-design analysis
The first analysis of brain imaging data revealed areas
with sustained activity during the blocks of pain or
Mean (SEM) accuracy and reaction time (RT) in the gender
Collapsed across actor sex Male actors Female actors
Accuracy (% correct)
aNote that RT is calculated from the onset of the stimulus but that
subjects were instructed to respond after the end of the 1-s stimulus.
D. Simon et al. / Pain 126 (2006) 309–318
Peak activation in a priori search areas in the event-related analyses
ROI Pain > neutral Pain > anger
Male > female actors
ROI, region of interest; AMY, amygdala; MPFC, medial prefrontal cortex; ACC, anterior cingulate cortex; aINS, anterior insula; SI, primary
somatosensory cortex; SII/pINS, secondary somatosensory cortex/posterior insula; FFA, fusiform face area; STS, superior temporal sulcus; L, left;
R, right; (x, y, z), Talairach coordinates of peak voxel [Activation threshold: P < 0.005, uncorr. (ROI), cluster P4 contiguous voxels].
aSignificant peaks that did not meet the cluster-size threshold (see Section 2).
D. Simon et al. / Pain 126 (2006) 309–318
contrasted to neutral blocks produced BOLD-signal
increases bilaterally in the STS (x, y, z = +51, ?54,
+7, t = 6.68, and ?54, ?43, +7, t = 6.40). Similar effects
were observed in response to blocks of anger stimuli
(+54, ?46, +10, t = 6.88, and ?48, ?52, +7, t = 4.33).
STS activation did not differ significantly between pain
and anger. Pain compared to neutral stimuli also pro-
duced a decrease in BOLD-response in the left somato-
sensory cortex (?51, ?19, +49, t =
that was not observed in response to anger expression
but did not reach significance in the direct contrast
between pain and anger. Finally, BOLD-signal was sig-
nificantly stronger to pain compared to anger expression
in the region of left SII/posterior insula (?45, ?31, +19,
t = 4.49). No other ROI showed sustained response in
the block-design analysis.
? 4.35), an effect
4.2. Event-related analysis
The second analysis contrasted the transient response
produced by pain, anger and neutral expression of male
and female actors within the blocks of trials. Table 3
lists the significant responses to pain faces (Vs neutral
and anger) found in the a priori search areas. As shown
in the Table, the activation pattern was dramatically
affected by the actor’s gender.
Pain-evoked activation (Vs neutral) was most robust
in response to male faces in the amygdala and strongest
in the right side as shown in Fig. 2A. Additional
responses to male expression of pain were found in SI,
FFA, and STS (Table 3). A peak activation was also
observed in the sub-genual ACC (+6, +20, ?8,
t = 3.76), but it failed to reach the cluster-size criterion
(see Section 2). In sharp contrast, male anger compared
to neutral expression produced only a significant
decrease in the left posterior ACC. The direct contrast
between pain and anger confirmed pain-specific increas-
es in the amygdala (Fig. 2B), MPFC, ACC, anterior
insula, somatosensory cortices (SI, SII/post insula),
FFA and STS (see Table 3 and Fig. 3).
The increase in BOLD-signal to pain expression
(Pain > Neutral) produced by male actors was observed
in response to female actors only in the left STS (see
Table 3; Male actors and Female actors in Pain > Neu-
tral). In contrast, responses to female pain expression
(Vs Neutral) were characterized by relative decreases
in the left amygdala, ACC, and somatosensory cortices
(right SI and bilateral SII/post insula) (Table 3; note
that a similar decrease just failed to reach the cluster-size
criteria in the rightamygdala:
t = ? 4.04). Fig. 2C clearly shows the negative effect
in the amygdala for the contrasts of pain with both neu-
tral and anger. Habituation was also noted in the amyg-
dala. However, post-hoc statistical analysis revealed no
significant interaction of habituation with the sex of
the actor, or the emotion condition (Ps > 0.05). The
habituation effect only interacted with the side of the
activation (F(1,16)= 4.97; P = 0.04), due to more rapid
habituation in the right amygdala. Similar decreases
were found in the somatosensory areas (bilateral SI
and SII/post insula) in response to female expression
of anger (Vs neutral). Additional analyses examining
BOLD-signal changes relative to the baseline visual-fix-
ation condition (not shown) revealed that those negative
changes found in the amygdala, the ACC, and the
somatosensory areas in response to pain expression
(Vs Neutral) are due, at least in part, to a stronger
increase in BOLD-signal produced by neutral female
faces. There was no specific effect of female expression
of pain Vs anger (Table 3).
The direct comparison of BOLD responses evoked by
male and female actors confirmed the robust effect on
pain-related responses. Male actors led to stronger
pain-evoked responses than female actors in the amyg-
dala, ACC, and somatosensory areas (Table 3, Male >
Female). This effect of the actor’s gender may also be
summarized by a robust response to male expression
of pain and to female neutral expression.
4.3. Effect of observers’ gender
Most of the effects reported above were detected in
male and female subjects analysed separately (not
shown), indicating that both subgroups contributed to
the response to pain and anger expression. Nevertheless,
Fig. 2. Bilateral activation of the amygdala to (A) male pain relative to
male neutral faces and (B) to male pain relative to male angry faces in
male and female observers. Statistical parametric maps (P < 0.005) are
overlaid on a T1 scan (radiological convention: left = right). (C) Bar
graphs display the corresponding b values (±SEM) for the peak
response in the right amygdala.
D. Simon et al. / Pain 126 (2006) 309–318
there were a few ROIs where the sex of the subject sig-
nificantly modulated the BOLD-response to pain or
anger. Those include male observers showing stronger
decreases to female expression of pain (Vs neutral; or
stronger increases to neutral female expression) in the
left SII/posterior insula (?39, ?10, ?5, t = ?5.2) and
right ACC (+9, +41, +4, t = ? 4.5) and (Vs anger) in
the bilateral FFA (right: +59, ?4, ?24, t = ?3.8; left:
?42, ?46, ?11, t = ? 4.6). Moreover, enhanced visual
activation was found in female observers to males
expressing pain relative to anger (right STS: +60, ?40,
?2, t = 4.6) and in male observers to angry male faces
contrasted with neutral male faces (left STS: ?57,
?37, +10, t = 4.1).
Our findings demonstrate activation of several emo-
tion-related areas, including the amygdala during the
automatic processing of facial pain expression as com-
pared to neutral faces under conditions of slight atten-
tional distraction induced by the sex-discrimination
task. Importantly, those results do not merely reflect
a non-specific response to the perception of emotional
facial expressions, as several areas showed a unique or
stronger response to pain compared to anger expres-
between these activation patterns and the sex of the
5.1. Cerebral network involved in the processing of pain
Considering previous findings that threat-related
stimuli activate the amygdala (Adolphs, 2002), its strong
activation to male pain expression supports the interpre-
tation that these stimuli may signal a potential threat
automatically detected by the observer and promoting
vicarious fear conditioning, as suggested by Botvinick
et al. (2005). Furthermore, viewing another person
expressing acute pain without knowing the source of
pain involves some ambiguity regarding this potential
threat. This is consistent with findings of stronger amyg-
dala engagement when a threat to the observer is uncer-
tain, indirect or ambiguous (Whalen et al., 2001). In
contrast, anger expression directed at the observer sig-
nals a less ambiguous danger emanating from the
Engagement of the amygdala and other target areas
in the processing of male facial expressions of pain
may be integrated into a common explanatory model
reflecting a response pattern to biologically relevant
stimuli. Visual analysis of pain, anger and neutral
expressions (Haxby et al., 2000) resulted in sustained
BOLD-increases in FFA, a region associated with the
processing of facial identity (e.g. Sergent et al., 1992;
Kanwisher et al., 1997). The STS, responsible for the
processing of changeable aspects of faces including
facial expressions (Allison et al., 2000), consistently
Fig. 3. Peak BOLD-responses in the event-related analysis contrasting pain and anger expressed by male actors (all subjects). Peak activation is
shown by the cross-lines in MPFC (A), peri-genual ACC (B), anterior insula/inf. frontal gyrus (C), SI face area (D), bilateral SII/posterior insula (E),
FFA (F), and STS (G). Statistical parametric maps (P < 0.005) are overlaid on a T1 scan (radiological convention: left = right).
D. Simon et al. / Pain 126 (2006) 309–318
showed sustained responses to both types of dynamic
stimuli compared to neutral. Pain expressed by males,
however, evoked stronger transient activation of the
extrastriate areas FFA and STS, possibly mediated
by amygdalar interconnections with these regions
(Vuilleumier et al., 2001).
Male pain faces also seemed to engage somatosensory
cortices. As suggested by Adolphs and colleagues (2000),
facial displays may be interpreted by covert simulation
of the other individual’s state with the help of these
brain regions. The amygdala, continuously updated with
somatosensory input, might have facilitated the process-
ing of male pain faces via its feedback projections as
characterized by transient responses in those areas. Con-
sidering that the amygdala may also be activated direct-
ly by thalamic inputs (LeDoux, 2000), a contribution of
this area in the sensory analysis of biologically meaning-
ful information at early stages is likely (Liddell et al.,
The anterior insula, also receiving input from the
amygdala (Amaral et al., 1992), is known to be engaged
in the integration of threat perception and bodily arous-
al states (Critchley, 2003). Hence, transient insular acti-
vation to male pain faces could reflect monitoring of the
bodily arousal changes associated with those stimuli
(Damasio, 1999). Furthermore, the amygdala sends pro-
jections to the anterior cingulate and MPFC cortex
(Amaral et al., 1992). While the MPFC was mainly
found to be engaged in cognitively bound emotional
tasks (Phan et al., 2002), the activation found to males’
pain expression is in line with its proposed role in the
top-down modulation of emotional responses during
processing of biologically significant signals (LeDoux,
2000). Engagement of the rostral–ventral ACC, known
to be associated with processing conflicts caused by
emotional distractors (e.g. Vuilleumier et al., 2001),
might reflect inhibitory mechanisms towards these sali-
ent, but task-irrelevant, pain expressions.
Surprisingly, those effects were not found to female
pain faces, which solely elicited stronger responses in
extrastriate areas. Although we must consider the fol-
lowing interpretations as preliminary, given the fact that
this finding was unexpected, we speculate that this
robust interaction with the actor’s sex may relate to
the different meanings that the pain expression may con-
vey in males and females. As males are generally found
to display less facial pain expression and experience less
pain in response to a controlled acute pain stimulus than
females (Sullivan et al., 2000; Robinson and Wise,
2003), a comparable intensity of pain expression of both
genders may signal a more intense noxious stimulus in
males. One may also speculate that male pain expres-
sions could be more strongly associated with situations
of potential threat for the observer (e.g. acute traumatic
pain produced by an external agent that may also harm
the observer). Furthermore, pain expression may lead to
a variety of behavioural responses that may vary
between males and females and/or constitute different
levels of ambiguous threat to the observer (e.g. aggres-
sion). Finally, the lack of amygdala response to female
pain expression could reflect a spontaneous top-down
inhibition of the defence response and the promotion
of helping behaviour.
Remarkably, in some of the target areas – namely the
amygdala, rostral–ventral ACC, somatosensory cortices
and FFA – enhanced activation was found to female
neutral faces. Although this effect was less pronounced
than the response to male pain expression, it also
occurred in both male and female observers. A possible
explanation for this unexpected result stems from devel-
opmental studies demonstrating that still faces are not
‘‘socially neutral’’, and produce amygdala activation in
children (Thomas et al., 2001). Considering that gender
influences the interpretation of the same emotional
expressions (e.g. Plant et al., 2004) and that females
are generally expected to display more emotions (e.g.
Guinsburg et al., 2000), one could speculate that neutral
female faces have been perceived as socially more salient
than neutral male faces. However, imaging studies on
this topic are still scant and only a few have used
dynamic stimuli in which the still faces may contrast
more clearly with expressive displays. However, some
studies reported enhanced neural responses to female
neutral faces in terms of mate-selection and attractive-
ness (Aharon et al., 2001; Fischer et al., 2004). Although
we did not include ratings of attractiveness, our results
strongly suggest a robust effect of the actor’s sex largely
independent from the observer’s sex.
5.2. The vicarious pain experience – neural mirroring in
the pain matrix?
Based on the idea of ‘mirror neurons’ (Rizzolatti
et al., 2001), researchers recently demonstrated activa-
tion of brain regions associated with the experience of
pain – namely insula and ACC – in response to visual
stimuli evoking pain experiences in others (Morrison
et al., 2004; Singer et al., 2004; Botvinick et al., 2005;
Jackson et al., 2005). However, activation of these areas
has also been associated with other emotional experienc-
es (Damasio et al., 2000), and pain-related responses
may reflect an emotional rather than a pain-specific
While the present study did not directly examine the
overlap between real and vicarious pain experience, it
included another emotional facial expression to test
the specificity of the activation patterns to pain expres-
sion. Although male pain faces compared to anger
evoked transient activation in areas of the pain matrix
(SI, SII/posterior insula, anterior insula, ACC), the fact
that these responses were accompanied by strong amy-
gdalar engagement points to a threat-related response
D. Simon et al. / Pain 126 (2006) 309–318
in addition to, or in contrast with, an internal simulation
of pain. Additionally, the observed somatosensory acti-
vation to female neutral faces runs counter to a simple
explanation based on a generalized empathic response
produced by a mirroring system.
Previous studies have used passive viewing conditions
or tasks explicitly focusing on the other person’s pain
experience, allowing for the processing of pain-related
information with full attentional capacity. Involvement
of insula and ACC has been more strongly associated
with explicit processing (Phan et al., 2002; Straube
et al., 2006) and mirroring mechanisms may be facilitat-
ed under such task conditions, while amygdalar engage-
ment might be actively suppressed (Pessoa et al., 2005).
Although the low attentional load of the gender discrim-
ination task left enough cognitive ressources to process
the pain-related information, it might have prevented
top-down modulation of amygdala responses.
Reconciling the findings of previous studies on vicar-
ious pain with these results, one may speculate that ‘mir-
roring’ and the pain-empathic response are not triggered
automatically when pain-related information (e.g. male
pain face) constitutes a potential threat and might rather
depend on the attentional focus to the pain cues. In
future research this potential moderating effect should
directly be addressed using a design including both an
implicit and explicit processing task. Moreover, in order
to further investigate the potential of facial pain dis-
plays’ to trigger rapid threat processing, one could
incorporate these stimuli into a masking paradigm
(e.g. Whalen et al., 1998).
This is the first imaging study investigating implicit
processing of dynamical prototypical facial pain expres-
sions. Our findings suggest that male pain faces consti-
tute an ambiguous threat to the observer. Supporting
the idea that the facial expression of pain is unique
and distinct from the expression of basic emotions (Wil-
liams, 2002), the results indicate some specificity of pain-
related responses as compared to anger. Unexpectedly,
threat-related responses were exclusively observed to
male pain expression. This suggests important sex differ-
ences in the meaning of pain expression and might con-
tribute to the understanding of the functional role of
We thank Leo Tenbokum and Jean-Maxime Leroux
for their help in the realisation of this study. This
research was supported by an operating grant from
the Natural Sciences and Engineering Research Council
(NSERC) of Canada (PR) and a scholarship of the
German Academic Research Exchange Service (DAAD)
awarded to D.S.
Adolphs R, Damasio H, Tranel D, Cooper G, Damasio AR. A role for
somatosensory cortices in the visual recognition of emotion as
Adolphs R. Neural systems for recognizing emotion. Curr Opin
Aharon I, Etcoff N, Ariely D, Chabris CF, O’Connor E, Breiter HC.
Beautiful faces have variable reward value: Fmri and behavioral
evidence. Neuron 2001;32:537–51.
Allison T, Puce A, McCarthy G. Social perception from visual cues:
role of the STS region. Trends Cogn Sci 2000;4:267–78.
Amaral P, Amaral DG, Price JL, Pitka ¨nen A, Carmichael ST.
Anatomical organization of the primate amygdaloid complex. In:
Aggleton JP, editor. The amygdala: Neurobiological aspects of
emotion, memory, and mental dysfunction. New York: Wiley
Apkarian AV, Bushnell MC, Treede RD, Zubieta JK. Human brain
mechanisms of pain perception and regulation in health and
disease. Eur J Pain 2005;9:463–84.
Botvinick M, Jha AP, Bylsma LM, Fabian SA, Solomon PE, Prkachin
KM. Viewing facial expressions of pain engages cortical areas
involvedin the direct experience
Craig KD, Prkachin KM, Grunau RVE, The facial expression of pain.
In: Turk DC, Melzack R, editors. Handbook of pain assessment,
vol. 2nd ed. New York: Guilford Press; 2001. p. 153–69.
Damasio AR. The feeling of what happens: Body and emotion in the
making of consciousness. Heartcourt Brace: Springer; 1999.
Damasio AR, Grabowski TJ, Bechara A, Damasio H, Ponto LL,
Parvizi J, et al. Subcortical and cortical brain activity during
the feelingof self-generated
Ekman P, Friesen WV. Manual for the Facial Action Coding
System. Palo Alto, CA: Consulting Psychologist Press; 1978.
Ekman P, Friesen WV, Hager JC. Facial Action Coding System
(FACS). Salt Lake City: A Human Face; 2002.
Fischer H, Sandblom J, Herlitz A, Fransson P, Wright CI, Backman
L. Sex-differential brain activation during exposure to female and
male faces. Neuroreport 2004;15:235–8.
Guinsburg R, de Araujo Peres C, Branco de Almeida MF, de Cassia
Xavier Balda R, Cassia Berenguel R, Tonelotto J, et al. Differences
in pain expression between male and female newborn infants. Pain
Haxby JV, Hoffman EA, Gobbini MI. The distributed human
neuralsystem forface perception.
Hutchison WD, Davis KD, Lozano AM, Tasker RR, Dostrovsky JO.
Pain-related neurons in the human cingulate cortex. Nat Neurosci
Jackson PL, Meltzoff AN, Decety J. How do we perceive the pain of
others? A window into the neural processes involved in empathy.
Kanwisher N, McDermott J, Chun MM. The fusiform face area: a
module in human extrastriate cortex specialized for face percep-
tion. J Neurosci 1997;17:4302–11.
LeDoux J. Cognitive–emotional interactions: listen to the brain. In:
Lane RD, Nadel L, editors. Cognitive neuroscience of emo-
tion. New York: Oxford University Press; 2000.
lesion mapping.J Neurosci
of pain. Neuroimage
itsdisorders.Br Med Bull
D. Simon et al. / Pain 126 (2006) 309–318
Liddell BJ, Brown KJ, Kemp AH, Barton MJ, Das P, Peduto A, et al. Download full-text
A direct brainstem–amygdala–cortical ‘alarm’ system for sublim-
inal signals of fear. Neuroimage 2005;24:235–43.
Mechelli A, Henson RN, Price CJ, Friston KJ. Comparing event-
related and epoch analysis in blocked design fMRI. Neuroimage
Morrison I, Lloyd D, di Pellegrino G, Roberts N. Vicarious responses
to pain in anterior cingulate cortex: Is empathy a multisensory
issue? Cogn Affect Behav Neurosci 2004;4:270–8.
Pessoa L, Padmala S, Morland T. Fate of unattended fearful faces in
the amygdala is determined by both attentional resources and
cognitive modulation. Neuroimage 2005;28:249–55.
Phan KL, Wager T, Taylor SF, Liberzon I. Functional neuroanatomy
of emotion: a meta-analysis of emotion activation studies in PET
and fMRI. Neuroimage 2002;16:331–48.
Plant EA, Kling KC, Smith GL. The influence of gender and social
role on the interpretation of facial expressions. Sex roles
Rainville P. Pain and emotions. In: Price DD, Bushnell MC, editors.
Psychological methods of pain control: basic science and clinical
perspectives. Progress in pain research and management 2004;vol.
29. Seattle, WA: IASP Press; 2004. p. 117–41.
Rizzolatti G, Fogassi L, Gallese V. Neurophysiological mechanisms
underlying the understanding and imitation of action. Nat Rev
Robinson ME, Wise EA. Gender bias in the observation of experi-
mental pain. Pain 2003;104:259–64.
Sergent J, Ohta S, MacDonald B. Functional neuroanatomy of face
and object processing. A positron emission tomography study.
Singer T, Seymour B, O’Doherty J, Kaube H, Dolan RJ, Frith CD.
Empathy for pain involves the affective but not sensory compo-
nents of pain. Science 2004;303:1157–62.
Straube T, Mentzel HJ, Miltner WHR. Neural mechanisms of
automatic and direct processing of phobogenic stimuli in specific
phobia. Biol Psychiatry 2006;59:162–70.
Sullivan MHL, Tripp DA, Santor D. Gender differences in pain and
pain behaviour: the role of catastrophizing. Cogn Ther Res
Thomas KM, Drevets WC, Whalen PJ, Eccard CH, Dahl RE, Ryan
ND, et al. Amygdala response to facial expressions in children and
adults. Biol Psychiatry 2001;49:309–16.
Visscher KM, Miezin FM, Kelly JE, Buckner RL, Donaldson DI,
McAvoy MP, et al. Mixed blocked/event-related designs separate
Vuilleumier P, Armony JL, Driver J, Dolan RJ. Effects of attention
and emotion on face processing in the human brain: an event-
related fMRI study. Neuron 2001;30:829–41.
Whalen PJ, Rauch SL, Etcoff NL, McInerney SC, Lee MB, Jenike
MA. Masked presentations of emotional facial expressions mod-
ulate amygdala activity without explicit knowledge. J Neurosci
Whalen PJ, Shin LM, McInerney SC, Fischer H, Wright CI. Rauch
SLA, functional MRI study of human amygdala responses to facial
expressions of fear versus anger. Emotion 2001;1:70–83.
Williams AC. Facial expression of pain: an evolutionary account.
Behav Brain Sci 2002;25:439–88.
Zald DH. The human amygdala and the emotional evaluation of
sensory stimuli. Brain Res Brain Res Rev 2003;41:88–123.
in fMRI. Neuroimage
D. Simon et al. / Pain 126 (2006) 309–318