The Effects of Fear and Anger Facial Expressions on Approach- and Avoidance-Related Behaviors

Emotion
Copyright 2005 by the American Psychological Association
2005, Vol. 5, No. 1, 119 –124
1528-3542/05/$12.00
DOI: 10.1037/1528-3542.5.1.119
The Effects of Fear and Anger Facial Expressions on Approach- and
Avoidance-Related Behaviors
Abigail A. Marsh
Nalini Ambady
Harvard University
Tufts University
Robert E. Kleck
Dartmouth College
The facial expressions of fear and anger are universal social signals in humans. Both expressions have
been frequently presumed to signify threat to perceivers and therefore are often used in studies
investigating responses to threatening stimuli. Here the authors show that the anger expression facilitates
avoidance-related behavior in participants, which supports the notion of this expression being a threat-
ening stimulus. The fear expression, on the other hand, facilitates approach behaviors in perceivers. This
contradicts the notion of the fear expression as predominantly threatening or aversive and suggests it may
represent an affiliative stimulus. Although the fear expression may signal that a threat is present in the
environment, the effect of the expression on conspecifics may be in part to elicit approach.
Explanations about the evolution and potential functions of
ening or aversive. Instead, the fear expression appears to facilitate
facial expressions date back at least to Darwin (1872/1965). For
predominantly approach from perceivers.
instance, both fear and anger facial expressions have been hypoth-
Approach and avoidance are the basic responses associated,
esized to be aversive stimuli and for this reason are often used as
respectively, with aversive and appetitive motivations (Cacioppo
experimental stimuli to assess responses to threat. The anger
& Berntson, 1994; Lang, Bradley, & Cuthbert, 1997). The roots of
expression ostensibly is taken to mean that the expresser is a direct
the word aversive are defined by Webster’s Unabridged Dictio-
threat to the perceiver. The fear expression, in contrast, is thought
nary to mean “to turn away from,” whereas the roots of appetitive
to signal the presence of some threat other than the expresser,
signify “to go to, head for, or strive after.” Thus, aversive means
perhaps alerting the perceiver to danger in the environment
simply that which elicits avoidance, and appetitive does not nec-
(Adolphs, Russell, & Tranel, 1999; Breiter et al., 1996; Morris et
essarily mean “appealing” but only that which elicits approach.
al., 1996; Whalen et al., 2001). Though the notion that these two
Appetitive and aversive motivations are thought by some to be the
expressions are aversive stimuli has intuitive appeal, to our knowl-
primary motive systems that exist in the brain and to underlie more
edge this notion has not been directly tested. Nonetheless, neuro-
complex emotional responding (e.g., Lang et al., 1997). From this
logical and physiological responses to these expressions are gen-
perspective, the perception of threat is fundamentally linked to the
erally interpreted in terms of the expressions’ presumed aversive
aversive motivational system. Although not all aversive stimuli are
significance. In the study reported here, we present data that
threatening (some may be, e.g., disgusting or sad instead), threat-
suggest the anger expression elicits avoidance, supporting the idea
ening stimuli are, by definition, aversive (O
¨ hman & Mineka,
that this expression is threatening. The data, however, contradict
2001). Therefore, any stimulus signaling potential threat is ex-
the notion that the fear expression is primarily perceived as threat-
pected to activate avoidant mechanisms such as withdrawal (Ca-
cioppo & Berntson, 1994; Lang et al., 1997).
The notion that both fear and anger facial expressions are
primarily perceived as threatening would entail that both be per-
Abigail A. Marsh, Department of Psychology, Harvard University;
ceived as more aversive than appetitive, and thus that both would
Nalini Ambady, Department of Psychology, Tufts University; Robert E.
be met with primarily avoidance-related responding. However,
Kleck, Department of Psychological and Brain Sciences, Dartmouth
animal models provide a mechanism by which a fear expression
College.
This study was supported in part by a National Science Foundation
might instead constitute an appetitive stimulus. Social species, like
graduate research fellowship awarded to Abigail A. Marsh, a National
many canines and primates, use stereotyped nonverbal displays of
Science Foundation Presidential Early Career Award for Scientists and
subordination or fear to keep them from becoming targets of more
Engineers awarded to Nalini Ambady, and a Rockefeller Reiss Family
dominant conspecifics’ aggression (Blair, 1995; Preuschoft, 1999;
senior faculty grant awarded to Robert E. Kleck. We are very grateful to
Schenkel, 1967; Smith & Price, 1973). Subordination displays, in
Blair Jarvis for technical assistance and to Geoffrey Stevens for assistance
contrast to dominance displays, appear nonthreatening and appeas-
in data collection.
ing and may make the organism appear smaller, weaker, more
Correspondence concerning this article should be addressed to Abigail
juvenile, or more affiliative (Schenkel, 1967). To appease is de-
A. Marsh, who is now at the Mood and Anxiety Disorders Program,
National Institute of Mental Health, 15K North Drive, MSC 2670, Be-
fined by Webster’s Unabridged Dictionary as “to bring to a state
thesda, MD 20892. E-mail: amarsh@post.harvard.edu
of ease or content; to conciliate or satisfy.” Thus, if fear expres-
119

---

120
BRIEF REPORTS
sions are perceived as affiliative and appeasing, these expressions
cd/m2, and the mean contrast, computed as the standard deviation
may elicit approach rather than avoidance.
of the luminosity, was 58.53.
On the basis of this line of reasoning, we tested the hypothesis
Each expression appeared for 2 s, during which time participants
that anger expressions elicit avoidance from perceivers, whereas
were asked to push or pull the lever, depending on the expression.
fear expressions elicit approach. To accomplish this, we used a
The task was thus described to participants as a simple categori-
simple motor task. It has been previously demonstrated that the
zation task in which we would be measuring how quickly they
perception of aversive stimuli is associated with rapid muscle
could distinguish faces in various categories. The instructions read
extension, whereas the perception of appetitive stimuli is associ-
as follows:
ated with muscle flexion (Cacioppo, Priester, & Berntson, 1993).
Perceivers thus push a lever (extension) in response to aversive
This study is a test of reaction times. At the desk at which you’re
sitting, you’ll see a video game joystick. We are going to measure
stimuli faster than they pull (flexion), but they pull faster than they
how fast you can move the joystick to categorize pictures you see on
push to appetitive stimuli (Chen & Bargh, 1999; Da Gloria, Pahla-
the screen. You will be categorizing people’s faces. You will either
van, Duda, & Bonnet, 1994; Foerster, Higgins, & Idson, 1998;
push the joystick away from you or pull it toward you to indicate
Solarz, 1960). If both anger and fear expressions are aversive
which category each face is in.
stimuli, behavioral avoidance and muscle extension should be
facilitated by both kinds of expressions. If the fear expression is
In one condition, participants were asked to pull the lever in
affiliative or appeasing, however, this expression should predom-
response to fear expressions and to push in response to anger; in
inantly facilitate approach and muscle flexion.
the other condition, participants were asked to push the lever in
response to fear and pull in response to anger. Thus, the rate at
Method
which each participant pushed the lever to categorize, for example,
anger, could be compared with the rate at which the lever was
Participants
pulled when making the identical categorization in the other con-
dition. Each expression in a series appeared on the screen imme-
Forty-eight participants (27 women and 21 men) took part in the
diately after the expression prior to it disappeared from the screen.
study. Participants who were recruited via posters placed in the
Two additional sets of photos were used as comparisons to
psychology building were paid for their time, whereas others
confirm the validity of the lever task. These sets contained photos
participated in exchange for course credit through the Psychology
also selected to be appetitive or aversive and were presented in an
Department.
identical manner to the facial expressions described above. One set
of comparison stimuli included the faces of 16 men confirmed by
Materials
pretesting to be recognized and admired (e.g., Mahatma Gandhi,
We used a lever (the Microsoft SideWinder) affixed to a desktop
Bill Cosby) or hated (e.g., Osama bin Laden, Adolf Hitler). A
PC and the program DirectRT (Jarvis, 2004) to measure motoric
second set contained faces of 16 women who were either highly
approach and avoidance reaction times (RTs) to anger and fear
attractive (beauty pageant contestants) or disfigured (individuals
expressions. During the experiment, the lever was positioned on
with craniofacial deformities). Each of these photos was also
the desktop directly in front of the participants. Participants were
presented at approximately 4
6 in size. The average luminosity
instructed not to reposition the base of the lever or hold it in their
of the hated and admired faces was 118.66 cd/m2, and the average
hands. The base of the lever rested on a rubber pad during the
contrast was 64.55. For the disfigured and attractive faces, these
study, preventing it from sliding or moving. The lever can be
values were 116.32 cd/m2 and 57.44. Again, participants saw each
moved 360°, but participants were instructed only to move it
set of faces twice; during one presentation, participants were asked
directly forward (away from themselves) or backward (toward
to pull in response to one kind of stimulus and push to the other,
themselves) during the experiment.1 Participants were instructed to
and in the other condition the response pairings were reversed. The
move the lever as far forward or backward as they were able to
order in which each of the six series of stimulus photos (two
every time they moved it, and to move it as quickly as possible.
experimental and four comparison) was presented to participants
The time after each stimulus presentation at which the lever
was randomized across participants, as was the order in which the
reached its maximal point from the baseline position was recorded
individual stimulus photos within each series appeared.
by the DirectRT software. DirectRT records response latencies
with a precision of 10 ms.
Results
Design and Procedure
Anger and Fear Expressions
Participants twice saw a series of the same eight anger and eight
Incorrect responses to the anger–fear task constituted 11.7% of
fear expressions shown by four men and four women, all showing
the total number of responses and were analyzed separately, as
direct eye gaze. These photographs were drawn from a validated
reported below. This error rate is typical for a recognition task
set of photographs used in prior studies (Marsh, Adams, & Kleck,
2005; Matsumoto & Ekman, 1988). All stimulus individuals were
1 To ascertain that lateral motion was minimal, we examined data from
adults, although precise age information was not available. The
all fear and anger expression trials and confirmed that in over 99% of all
expressions were presented in grayscale in the center of the com-
recorded epochs (N
304,552/307,200) no lateral deviation occurred. In
puter screen. Each photograph was presented at a size of approx-
the remaining 0.09%, lateral deviations averaged only 17% of the distance
imately 6
4 . The mean luminosity of the faces was 119.86
from baseline.

---

BRIEF REPORTS
121
using fear and anger facial expressions (see Fridlund, 1994, p.
anger, t(26)
2.91, p
.01, r
.50, but male participants did not,
204). In addition, missed responses (failures to respond) consti-
t(20)
1.07, p
.30, r
.23. (See Table 2 for the means and
tuted 3% of all responses. Participants appeared to have closely
standard deviations of all responses by gender of participant and
followed the experimenter’s instructions to move the joystick as
stimulus.)
far as possible. The data showed that only 1% of the correct lever
No main effect for lever direction was found, F(1, 47)
0.46,
movements did not reach the joystick’s apex. We performed a 2
ns. As can be seen in Table 1, the mean response times across all
(participant gender)
2 (stimulus gender)
2 (expression)
2
stimuli for pushing and pulling were nearly identical. This suggests
(lever direction) analysis of variance (ANOVA) with repeated
that neither movement was intrinsically easier than the other. No
measures on the averages of participants’ log-transformed RTs for
other main effects or interactions were found to be significant. For
correct responses to each type of stimulus face.
example, no other interactions with lever direction (all ps
.20) or
The results of this ANOVA yielded a main effect for emotional
with participant gender (all ps
.15) approached significance.
expression, F(1, 45)
9.21, p
.01,
.41, such that partici-
Additionally, an analysis of the error rates, as measured by a 2
pants responded more quickly to fear than to anger expressions in
(lever direction)
2 (expression) chi-square analysis, suggests
both conditions. More important, however, the predicted interac-
that there was a trend for participants to make more errors when
tion was found for lever direction and facial expression, F(1, 45)
asked to pull in response to anger expressions (n
60) than push
7.86, p
.01,
.39, such that participants pushed the lever
(n
54), but to make more errors when asked to push in response
more quickly than they pulled it in response to anger expressions,
to fear expressions (n
40) than pull (n
27), 2(1, N
181)
t(47)
2.30, p
.05, r
.32, but pulled more quickly than they
2.57, p
.11,
.12. This pattern is again consistent with the
pushed in response to fear expressions, t(47)
2.81, p
.01, r
notion that perception of the anger expression facilitates avoid-
.38 (see Table 1).
ance, whereas perception of the fear expression facilitates
There was also a main effect for stimulus gender, F(1, 45)
approach.
5.93, p
.05,
.34, such that participants responded faster to
female than to male faces. An interaction was found between
stimulus gender and emotion, F(1, 45)
7.60, p
.01,
.38,
Comparison Photographs
such that participants responded (both pushing and pulling) to
women expressing fear faster than to any other gender– emotion
Correct responses to photographs of hated and liked men and
combination (female anger: t(47)
3.71, p
.01; male fear:
disfigured and attractive women were also analyzed using 2 (par-
t(47)
3.20, p
.01; male anger: t(47)
3.21, p
.01), but
ticipant gender)
2 (stimulus type)
2 (lever direction)
response times among these three latter types of stimuli did not
ANOVAs. For photos of hated and liked men, a significant effect
significantly differ.
of participant gender was found, such that men responded more
No main effect was found for the gender of the participant, F(1,
quickly than women across all stimuli, F(1, 46)
5.38, p
.05,
45)
0.02, ns; for the interaction between participant gender and
.32. There were no significant interactions with participant
lever direction, F(1, 45)
0.13, ns; or for the interaction among
gender, however (all ps
.60). The interaction effects between
participant gender, emotion, and lever direction, F(1, 45)
0.01,
lever direction and stimulus valence for these photos also followed
ns. A marginally significant effect was found between emotion and
the predicted pattern, F(1, 46)
3.32, p
.075,
.26. For
the gender of the participant, F(1, 45)
3.06, p
.09,
.25,
photos of disfigured and attractive women, no significant effects of
such that female participants responded more quickly to fear than
participant gender were found, and so this variable was dropped.
The results of the subsequent 2
2 ANOVA also showed a weak
resemblance to the patterns seen for the other stimuli, F(1, 45)
Table 1
1.84, p
.18,
.20 (see Table 1). The degrees of freedom were
Mean Response Times Pushing and Pulling a Lever in Response
reduced from 47 to 45 for this test because 2 participants did not
to Three Categories of Face Stimuli
provide any correct responses during these trials.
Push
Pull
Analyses of the error rates for hated and liked and disfigured
Mean
and attractive faces did not reveal significant patterns either for the
Stimulus type
M
SD
M
SD
difference
admired and hated faces
2(1, N
80)
0.62, ns, or for the
attractive and disfigured faces
2(1, N
36)
0.21, ns.2 The
Expression
Anger
1,129
204
1,194
221
65
results of a manipulation check that participants completed after all
Fear
1,149
198
1,065
185
84
of the lever tasks confirmed that the men preselected as hated were
Character
perceived to be hated (M
1.88, SD
0.71) more than were the
Hated
1,154
199
1,236
209
82
men preselected as admired (M
6.36, SD
0.31), and that the
Admired
1,132
180
1,118
183
14
Attractiveness
women preselected as attractive were rated to be more attractive
Disfigured
874
269
893
186
19
(M
1.76, SD
0.19) than the women preselected as disfigured
Attractive
865
180
820
263
45
(M
6.30, SD
0.44).
Mean all
1,051
1,054
3
stimuli
2 For the analysis of error rates for attractive and disfigured faces, the
data for the 2 participants who provided no correct responses in one or
Note.
All values are given in milliseconds. Negative values indicate the
dominant response to have been pushing (avoidance); positive values
more of these series (see ANOVA description) were removed prior to
indicate the dominant response to have been pulling (approach).
analysis.

---

122
BRIEF REPORTS
Table 2
tions. First, the perception of anger expressions appears to activate
Mean Response Times and Standard Deviations for Fear and
neural circuits involved in behavioral suppression and inhibition
Anger Expression as a Function of Sex of Perceiver and Sex of
(Blair, Morris, Frith, Perrett, & Dolan, 1999). Perhaps because of
Stimulus
this, participants’ responses to anger expressions were slowed
relative to fear expressions. In terms of effects for stimulus gender,
Push
Pull
the women’s facial expressions may have been easier to identify—
Mean
and thus respond to—than were men’s expressions (Wagner,
Variable
M
SD
M
SD
difference
Buck, & Winterbotham, 1993). The data indicated that participants
Anger expressions
responded more quickly to women’s fear expressions than to any
Male perceivers
other expressions. If women’s fear expressions were the easiest to
Male faces
1,154
238
1,205
249
51
identify, this could potentially account for the main effect for
Female faces
1,153
282
1,191
210
38
response times to both fear expressions and female faces.
Female perceivers
Male faces
1,164
211
1,183
251
19
Alternative explanations for the main findings in the present
Female faces
1,164
226
1,273
272
109
study must also be considered. In showing weakness or submis-
Fear expressions
sion, fear expressions could actually serve the maladaptive purpose
Male perceivers
of inviting attack, which is also a form of approach. In general,
Male faces
1,235
189
1,127
246
108
Female faces
1,161
280
1,096
249
65
however, the evidence would seem to contradict this possibility.
Female perceivers
As discussed above, appeasing or submissive gestures serve the
Male faces
1,205
253
1,137
182
68
symbolic purpose of “raising the white flag” in most social species
Female faces
1,112
245
1,030
202
82
and typically reduce the chances of attack from conspecifics.
Stereotyped gestures that can prevent aggressive encounters serve
Note.
All values are given in milliseconds. Negative values indicate the
dominant response to have been pushing (avoidance); positive values
members of a social group by reducing the likelihood of injuries.
indicate the dominant response to have been pulling (approach).
Moreover, evidence from cross-cultural, cross-species, and infant
research suggests that facial expressions are innate and evolved.
An expression that invited others to attack the expresser would
seem an unlikely candidate for preservation by natural selection.
Discussion
Another potential interpretation of these data is that they provide
The perception of emotional facial expressions affected motoric
support for the possibility that pushing and pulling responses to
responses associated with approach and avoidance. More specifi-
aversive and appetitive stimuli are not automatic. Specifically, a
cally, anger and fear expressions facilitated opposite movement
paradox might be seen in the fact that, although fear-related words
behaviors. Anger expressions facilitated avoidance-related behav-
such as death and hell elicit avoidance in perceivers, fear expres-
ior, supporting the notion that this expression appears aversive and
sions appear to elicit approach. The meaning of emotional expres-
perhaps threatening. Fear expressions, however, facilitated
sions appears to be processed automatically (Whalen et al., 1998).
approach-related behavior, contradicting the idea that this expres-
Moreover, approach and avoidance responses can be emitted by
sion represents a primarily aversive stimulus.
even very simple organisms and, in humans, can be reflected by
That the anger expression might generate avoidance responses is
even involuntary activities such as a potentiated startle. This sug-
unsurprising. Why, however, might the fear expression elicit pre-
gests that approach or avoidance motivations can be elicited au-
dominantly approach-related responding? Rather than serving as a
tomatically (Cacioppo & Berntson, 1994). However, it need not
threat signal, the fear expression may serve as an appeasement cue,
follow that the meaning of expressions would automatically affect
intended to ameliorate conflict or elicit conciliatory or affiliative
approach- and avoidance-related behaviors like pushing and pull-
behavior, much like the submission cues of other species (e.g.,
ing under all circumstances. Perhaps whether fear-related stimuli
Blair, 1995; Schenkel, 1967). Fear expressions have been rated to
elicit either approach- or avoidance-related behavior depends on
appear highly submissive and also highly affiliative—approxi-
the context or attentional set of the perceiver. For example, Rot-
mately equally as affiliative as happy expressions (Hess, Blairy, &
teveel and Phaf (2004) have shown that when participants are
Kleck, 2000). In addition, perceivers view fear expressions as
asked to make button-pressing movements similar to pushing and
appearing rounder, kinder, warmer, more submissive, and more
pulling a lever to categorize faces by gender, the effects of the
babyish than anger expressions (Marsh et al., 2005). Such percep-
target faces’ emotional expressions disappear. This would suggest
tions would be difficult to explain if fear were predominantly
that the behavioral responses to affectively valenced stimuli can be
threat inducing. But if fear serves to elicit approach, then perceiv-
mediated by conscious processes.
ers might perhaps construe it as affiliative in that it encourages the
The present article, however, was not aimed at testing the
formation of social bonds. Indeed, evidence exists to suggest that
automaticity of the approach and avoidance response. Instead, the
fear and other distress-related emotions like sadness and anxiety
aim was to assess the extent to which the fear expression could be
elicit not only the desire for affiliation but also caregiving from
construed as appetitive and approach-related. Essentially without
members of the social group (e.g., Batson, Duncan, Ackerman,
exception, muscle flexion is a behavior associated with appetitive
Buckley, & Birch, 1981; Radke-Yarrow, Zahn-Waxler, Richard-
stimuli and approach (see Rotteveel & Phaf, 2004). Although
son, & Susman, 1994; Schachter, 1959).
appetitive stimuli need not always elicit flexion, when flexion does
It is unclear why participants responded more quickly to fear
occur in response to a stimulus, it implies that the stimulus is
expressions than to anger expressions and to female faces more
appetitive. Thus, although directing perceivers’ attention away
quickly than to male faces. There are several possible explana-
from the emotional aspect of a fear expression might result in a

---

BRIEF REPORTS
123
reduction or an elimination of the approach response, that the
That fear and smiling expressions may share common features
approach response would be the dominant response to the percep-
should not come as a surprise, as both expressions are thought to
tion of a fear expression remains noteworthy.
share similar signal values, namely, indicating appeasement or
The specific nature of the task that participants were asked to
affiliation. It would also not come as a surprise, then, were both
perform needs to be kept in mind, however. In this paradigm,
expressions to elicit approach from those who perceive them.
participants explicitly categorized the anger and fear expressions
Future research should also address the mechanisms linking the
as they manipulated the lever. It is plausible that the need to
perception of emotional expressions to approach and avoidance. In
explicitly evaluate the stimuli changed the nature of these stimuli,
general, the neural mechanisms underlying the perception of facial
although the results of a study by Chen and Bargh (1999) contra-
expressions of emotion, and underlying the approach and avoid-
dict this possibility. Using a lever paradigm and categorization task
ance mechanisms, are still not well understood. However the two
similar to the ones used here, Chen and Bargh found that explicit
processes likely would be linked by the generation of emotion in
evaluation did not appear to affect approach and avoidance re-
the perceiver. Emotional expressions are thought to elicit emo-
sponses. In the first of their two studies, participants evaluated
tional responses in perceivers. Sometimes these responses are
words as good or bad, and pushed or pulled a lever depending on
analogous to the emotion perceived. Sometimes, however, the
each word’s perceived valence. In a second study, participants
perceiver’s emotional responses are instead complementary to the
pushed or pulled in response to all words, regardless of perceived
emotion perceived. For example, when people see an anger ex-
valence. The effects were the same across the two studies—
pression, they may have a desire to escape rather than feeling
participants pushed the lever faster to negative words and pulled
angry (Blairy, Herrera, & Hess, 1999). Such reactions to the
faster to positive words. Thus, the results of the lever task appear
perception of anger could motivate withdrawal. Perhaps emotions
to be independent of the occurrence of explicit evaluation.
associated with concern or sympathy might be elicited by the
Other elements of the experimental context should still be taken
perception of a fear expression and lead to approach. Measuring
into account when interpreting the results of this study, however.
emotional responses to the perception of emotional expressions—
Here, the fear expression was presented in a relatively neutral
either via self-report or via psychophysiological techniques—and
context. It is possible that the meaning of and responses to the fear
assessing the relationship between these emotional responses and
expression would change were the perceiver to encounter it in a
approach and avoidance might shed light on the mechanisms
potentially dangerous environment. If the study contained a po-
underlying these responses.
tential threat, perhaps an impending shock or aversive noise, a fear
expression might be perceived as a threat signal and thus an
Conclusions
aversive stimulus rather than an appetitive one. This does not
exclude the possibility that, even though fear appears to elicit
We have presented data that suggest that the fear expression
approach in a relatively neutral environment, this expression may
facilitates approach-related behavior in perceivers, indicating that
still represent a negative stimulus. Examples can be conceived in
this expression could be a social cue that serves the function of
which negative stimuli would elicit behaviors associated with
making the expresser appear approachable to perceivers. Anger
approach. For example, paramedics would be expected to respond
expressions instead may make the expresser more aversive. Po-
to the sight of an injury or blood with approach-related behavior,
tentially adaptive social functions such as these may help to
although the sight would probably still be unpleasant to them
explain why facial expressions have been selected by evolution as
rather than (or in addition to) being strictly appetitive. In this
a part of the behavioral repertoire of humans. In future investiga-
sense, one might see a distinction between the valence of a
tions of responses to emotional expressions, researchers may wish
stimulus and the behavior it elicits.
to consider the potential social implications of these expressions.
Some questions as to how to interpret the behaviors measured in
this study will be difficult to fully address before behavioral
References
responses to other types of facial stimuli, for example, smiles, are
assessed. It may be that although the predominant response to a
Adolphs, R., Russell, J. A., & Tranel, D. (1999). A role for the human
amygdala in recognizing emotional arousal from unpleasant stimuli.
fear expression in some contexts is to approach, this tendency
Psychological Science, 10, 167.
would be small relative to the tendency to approach a smiling
Batson, C. D., Duncan, B. D., Ackerman, P., Buckley, T., & Birch, K.
expression. Along these lines, it would also be beneficial to look at
(1981). Is empathic emotion a source of altruistic motivation? Journal of
responses to smiling because some see a relationship between the
Personality and Social Psychology, 40, 290 –302.
expressions of happiness and fear. Preuschoft (1995) suggests that
Blair, R. J. R. (1995). A cognitive developmental approach to morality:
the origins of the human smile may lie in the fear grimace, or
Investigating the psychopath. Cognition, 57, 1–29.
“silent bared-teeth display,” seen in some primates. The silent
Blair, R. J. R., Morris, J. S., Frith, C. D., Perrett, D. I., & Dolan, R. J.
bared-teeth display is thought to signal submission and appease-
(1999). Dissociable neural responses to facial expressions of sadness and
ment. Perhaps approach responses to fear expressions in the
anger. Brain, 122, 883– 893.
present study occurred in part because features of fear expressions
Blairy, S., Herrera, P., & Hess, U. (1999). Mimicry and the judgment of
emotional facial expressions. Journal of Nonverbal Behavior, 23, 5– 41.
(e.g., bared teeth and raised eyebrows) are perceptually similar to
Breiter, H. C., Etcoff, N. L., Whalen, P. J., Kennedy, W. A., Rauch, S. L.,
features of smiling expressions. For example, the expressers’ teeth
Buckner, R. L., et al. (1996). Response and habituation of the human
were visible in seven of the eight fear expressions used in this
amygdala during visual processing of facial expression. Neuron, 17,
study but in only one of the eight anger expressions. It is worth
875– 887.
noting, however, that in facial expression judgment tasks, perceiv-
Cacioppo, J. T., & Berntson, G. G. (1994). Relationship between attitudes
ers seldom explicitly confuse fear and happiness (Fridlund, 1994).
and evaluative space: A critical review, with emphasis on the separa-

---

124
BRIEF REPORTS
bility of positive and negative substrates. Psychological Bulletin, 115,
Nature, 383, 812– 815.
401– 423.
O
¨ hman, A., & Mineka, S. (2001). Fears, phobias, and preparedness:
Cacioppo, J. T., Priester, J. R., & Berntson, G. G. (1993). Rudimentary
Toward an evolved module of fear and fear learning. Psychological
determinants of attitudes: II. Arm flexion and extension have different
Review, 108, 483–522.
effects on attitudes. Journal of Personality and Social Psychology, 65,
Preuschoft, S. (1995). “Laughter” and “smiling” in macaques: An evolu-
5–17.
tionary perspective. De Meern, the Netherlands: Tessel Offset, B.V.
Chen, M., & Bargh, J. A. (1999). Consequences of automatic evaluation:
Preuschoft, S. (1999). Are primates behaviorists? Formal dominance, cog-
Immediate behavioral dispositions to approach or avoid the stimulus.
nition, and free-floating rationales. Journal of Comparative Psychology,
Personality and Social Psychology Bulletin, 25, 215–224.
113, 91–95.
Da Gloria, J., Pahlavan, F., Duda, D., & Bonnet, P. (1994). Evidence for a
Radke-Yarrow, M., Zahn-Waxler, C., Richardson, D. T., & Susman, A.
motor mechanism of pain-induced aggression instigation in humans.
(1994). Caring behavior in children of clinically depressed and well
Aggressive Behavior, 20, 1–7.
mothers. Child Development, 65, 1405–1414.
Darwin, C. (1965). The expression of the emotions in man and animals.
Rotteveel, M., & Phaf, R. H. (2004). Automatic affective evaluation does
Chicago: University of Chicago Press. (Original work published 1872)
not automatically predispose for arm flexion and extension. Emotion, 4,
Foerster, J., Higgins, E. T., & Idson, L. C. (1998). Approach and avoidance
156 –172.
strength during goal attainment: Regulatory focus and the “goal looms
Schachter, S. (1959). The psychology of affiliation. Stanford, CA: Stanford
larger” effect. Journal of Personality and Social Psychology, 75, 1115–
University Press.
1131.
Schenkel, R. (1967). Submission: Its features in the wolf and dog. Amer-
Fridlund, A. J. (1994). Human facial expression: An evolutionary view. San
ican Zoologist, 7, 319 –329.
Diego, CA: Academic Press.
Smith, J. M., & Price, G. R. (1973, November). The logic of animal
Hess, U., Blairy, S., & Kleck, R. E. (2000). The influence of facial emotion
conflict. Nature, 246, 15–18.
displays, gender, and ethnicity on judgments of dominance and affilia-
Solarz, A. J. (1960). Latency of instrumental responses as a function of
tion. Journal of Nonverbal Behavior, 24, 265–283.
compatibility with the meaning of eliciting verbal signs. Journal of
Jarvis, B. G. (2004). DirectRT research software (Version 2004) [Com-
Experimental Psychology, 59, 239 –245.
puter software]. New York: Empirisoft.
Wagner, H. L., Buck, R., & Winterbotham, M. (1993). Communication of
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1997). Motivated attention:
specific emotions: Gender differences in sending accuracy and commu-
Affect, activation, and action. In P. J. Lang & R. F. Simons (Eds.),
nication measures. Journal of Nonverbal Behavior, 17, 29 –53.
Attention and orienting: Sensory and motivational processes (pp. 97–
Whalen, P. J., Rauch, S. L., Etcoff, N. L., McInerney, S. C., Lee, M. B., &
135). Mahwah, NJ: Erlbaum.
Jenike, M. A. (1998). Masked presentations of emotional facial expres-
Marsh, A. A., Adams, R. B., Jr., & Kleck, R. E. (2005). Why do fear and
sions modulate amygdala activity without explicit knowledge. Journal
anger look the way they do? Form and social function in facial expres-
of Neuroscience, 18, 411– 418.
sions. Personality and Social Psychology Bulletin, 31, 73– 86.
Whalen, P. J., Shin, L. M., McInerney, S. C., Fischer, H., Wright, C. I., &
Matsumoto, D., & Ekman, P. (1988). Japanese and Caucasian facial
Rauch, S. L. (2001). A functional MRI study of human amygdala
expressions of emotion (JACFEE) [Slides]. San Francisco: Intercultural
responses to facial expressions of fear versus anger. Emotion, 1, 70 – 83.
and Emotion Research Laboratory, Department of Psychology, San
Francisco State University.
Morris, J. S., Frith, C. D., Perrett, D. I., Rowland, D., Young, A. W.,
Received June 20, 2003
Calder, A. J., & Dolan, R. J. (1996, October 31). A differential neural
Revision received May 27, 2004
response in the human amygdala to fearful and happy facial expressions.
Accepted June 1, 2004

---