Facial Reactions to Face Representations in Art: An Electromyography Study

Abstract

Facial mimicry is a reaction to facial expressions. It plays a role in social interaction. Indeed, scholars associated facial mimicry with emotional contagion and understanding others' mental states such as intentions. This is the case for facial mimicry toward human facial expressions, but we know that facial expressions are widely depicted in art through face representations (visual creations that depict facial expressions). However, despite face representation involvement in social interactions, facial reactions toward face representations in art are still unknown. The reason could be that interaction with art objects is usually analyzed within anthropology and art theories, such as conveying social agencies (a desire of action, intentions). Here, we show that facial mimicry is also observed toward face representations. This could be a means that might facilitate social interaction including emotions. Using the electromyography technique, we could show that participants mimic involuntarily face representations when these depict mouth expressions. Participant's zygomaticus and depressor were significantly activated when the pictures depict an expression including zygomaticus or depressor representation respectively. This result led us to infer that when it comes to mouth expressions, face representations in art might trigger spontaneously emotional contagion (of the expressed emotion). It might also convey information about the expressed mental states, which might help to indicate social agencies. Mimicry could participate to explain partly the social agencies of art, that might be no more just abstract concepts, but could find a real correlate in cognitive processes.

Full text

Facial Reactions to Face Representations in Art: An Electromyography
Study
Amel Achour-Benallegue
1
, David Amarantini
2
, Pierre-Vincent Paubel
3
, Jérôme Pelletier
1
,
and Gwenaël Kaminski
3, 4
1
Institut Jean Nicod, Department of Cognitive Studies, French National Center for Scientific Research - École Normale Supérieure -
School for Advanced Studies in the Social Sciences, Paris Sciences et Lettres University
2
Toulouse NeuroImaging Center, Inserm, Paul Sabatier University, Toulouse University
3
Cognition, Language, Ergonomy, Jean-Jaurés University, Toulouse University
4
Academic Institute of France, Paris, France
Facial mimicry is a reaction to facial expressions. It plays a role in social interaction. Indeed, scholars asso-
ciated facial mimicry with emotional contagion and understanding others' mental states such as intentions.
This is the case for facial mimicry toward human facial expressions, but we know that facial expressions
are widely depicted in art through face representations (visual creations that depict facial expressions).
However, despite face representation involvement in social interactions, facial reactions toward face repre-
sentations in art are still unknown. The reason could be that interaction with art objects is usually analyzed
within anthropology and art theories, such as conveying social agencies (a desire of action, intentions).
Here, we show that facial mimicry is also observed toward face representations. This could be a means
that might facilitate social interaction including emotions. Using the electromyography technique, we could
show that participants mimic involuntarily face representations when these depict mouth expressions.
Participant's zygomaticus and depressor were significantly activated when the pictures depict an expression
including zygomaticus or depressor representation respectively. This result led us to infer that when it
comes to mouth expressions, face representations in art might trigger spontaneously emotional contagion
(of the expressed emotion). It might also convey information about the expressed mental states, which
might help to indicate social agencies. Mimicry could participate to explain partly the social agencies of
art, that might be no more just abstract concepts, but could find a real correlate in cognitive processes.
Keywords: facial mimicry, faces in art, electromyography, emotional contagion, art agency
Facial expressions are configurations of facial features through
the motion of facial muscles. Facial expressions are relevant con-
figurations for nonverbal communication and displaying emotion
(Awasthi & Mandal, 2015;Reisenzein et al., 2013). They could
induce a motor reaction in the observer’s face, known as facial
mimicry. Facial mimicry is a tendency to imitate the facial expres-
sions of other people (Hess & Fischer, 2014). Previous research
describes the process of facial mimicry as similar muscle patterns
in both observer’s face and observed facial expression. These mus-
cle patterns could be measured by using the electromyography
(EMG) technique, where muscle activity is recorded through facial
electrodes. Facial mimicry occurs automatically and uncon-
sciously (Bailey & Henry, 2009;Dimberg et al., 2000;Korb et al.,
2010), but could also be modulated by social contexts, such as
empathy, cooperation, and competition, or social membership
(Hess & Fischer, 2014;Seibt et al., 2015). For example, mimicry
of smiles and sad expressions are more observed among mem-
bers of the same group compared with members of different
groups, which is interpreted as affiliative signs (Seibtetal.,
2015).
Amel Achour-Benallegue https://orcid.org/0000-0003-4183-0839
David Amarantini https://orcid.org/0000-0002-9292-2429
Jérôme Pelletier https://orcid.org/0000-0002-2411-3005
Gwenaël Kaminski https://orcid.org/0000-0001-5300-5655
This work was supported by the French National Research Agency
[ANR-13-JSH2-0006]; the Institut Universitaire de France; and FrontCog
ANR-17-EURE-0017. The funders had no role in study design, data
collection and analysis, decision to publish, or preparation of the
manuscript. We have no conflicts of interest to disclose.
We want to thank Mehdi Benallegue for his insightful comments during the
process of the study. Namely, he contributed with methodology, software and
writing–review & editing. Also, we want to thank Anubhuti Chauchan for her
relevant remarks about writing–review & editing. The OSF repository URL:
https://osf.io/yk7hz/?view_only=972e5179b5904fcbaf5b1df70416b934
Correspondence concerning this article should be addressed to Amel
Achour-Benallegue, Institut Jean Nicod, Département d'études cognitives,
Centre national de la recherche scientifique - École normale supérieure - École
des hautes études en sciences sociales, Université Paris Sciences et Lettres, 29
rue d'Ulm, Paris, France. Email: a.achour.benallegue@gmail.com
1
Psychology of Aesthetics, Creativity, and the Arts
©2021 American Psychological Association
ISSN: 1931-3896 https://doi.org/10.1037/aca0000423
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
793
This article was published Online First October 7, 2021.
2023, Vol. 17, No. 6, 793–805

The matching pattern with an observed expression is part of the
embodied emotion processes (Barrett & Lindquist, 2008;Nieden-
thal, 2007;). In embodied emotion processes, perceiving an emo-
tional stimulus and experiencing an emotion involve overlapping
mental processes (Gallese, 2019;Niedenthal et al., 2013;Nieden-
thal & Maringer, 2009). This means that the matching pattern indi-
cates also the presence of the emotion process, which can be
noticed in the following two ways: First, muscle patterns in
observers are congruent with the emotions displayed in stimuli
(Dimberg et al., 2011;Künecke et al., 2014). Studies showed, for
example, that expressions of happiness and anger enhance, in
observers, the activity of the muscles responsible for happiness and
anger expressions, respectively (Rymarczyk et al., 2011). We know
that specific muscles are more implicated in emotional expressions,
such as zygomaticus for happiness, corrugator for anger, frontalis
for fear or surprise, and depressor anguli oris for sadness (Soussi-
gnan et al., 2013). Second, observers do experience emotions that
are similar to those displayed by the observed facial expressions
(Niedenthal, 2007;Niedenthal & Brauer, 2012;Winkielman et al.,
2015). Experiencing an observed emotion means that this emotion
happens mentally and physiologically in the observer. This is
known as “emotional contagion,”which is feeling in oneself the
mimicked emotion (Dimberg & Thunberg, 2012;Hess & Fischer,
2014;Sato et al., 2013). Indeed, experiments had shown that facial
mimicry is accompanied by a self-report of the corresponding emo-
tion (McIntosh, 2006). Furthermore, when mimicry is prevented,
emotional experience and emotional information processing might be
disturbed (Niedenthal et al., 2017). Experiments had shown that
when preventing mimicry of corrugator muscle (that expresses sad
and hopeless feelings), depression may be lifted (Finzi & Rosenthal,
2014;Wollmer et al., 2012).
The above studies (Finzi & Rosenthal, 2014;McIntosh, 2006;
Wollmer et al., 2012) showed that facial mimicry might help (or at
least be associated with) emotional contagion between individuals
when interacting. Consequently, the facial mimicry process is not
a mere motor reaction, but also contributes to emotional interac-
tions (Beall et al., 2008;Hess & Fischer, 2013;Prochazkova &
Kret, 2017). Moreover, the mimicry process could be associated
with understanding others’mental states (Niedenthal et al., 2010;
Schilbach, 2016). Indeed, among the neural correlates of facial
mimicry, some studies observed the activation of the mentalizing
network (Baetens et al., 2014;Schilbach, 2016;). This network
(also called the theory of mind network) is involved in the infer-
ence of mental states such as intentions (Blakemore & Decety,
2001;Gendron et al., 2014). At this point, facial mimicry could be
associated with the metaphorical “‘direct’access to other people’s
minds”claimed by Schilbach (2016, p. 85). Emotions, as well as
intentions, might be important clues in social interaction. In social
interaction, expressing emotions is convenient to communicate to
others clues about our own mental state and to orientate their inter-
actions with us (Damasio, 1994). As put by Schilbach (2016), visi-
ble behavioral responses that are produced by facial mimicry help
to sustain the process of interaction.
Studies on facial mimicry focus on facial expressions of human
faces (known as “actual faces”). However, we know that facial
expressions are largely depicted in art. In this article, we call these
facial expressions which are depicted in art: face representations.
Face representations are the category of crafted objects such as
drawings, paintings, sculptures, and other visual creations that
depict facial expressions. Face representations include faces of dif-
ferent realism degrees due to the variety of styles, cultures, and
periods. Realistic face representations involve pictures that resem-
ble actual human faces such as classical portraits. Nonrealistic ones
are images that differ from actual human faces such as ethnographic
masks or modern portraits. Some questions arise then: what are fa-
cial reactions to face representations? Do they fall into facial mim-
icry as reactions to actual facial expressions? This issue draws
further attention since art has always represented faces with various
expressions, all over the world and throughout history. For instance,
those represented faces could be noticed in masks, portraits, anima-
tion, comics, and even in robots (Achour-Benallegue et al., 2016).
We notice that interacting with face representations has often
played an important role in social interaction. These could be
observed in idol worship, ritual ceremonies, aesthetic experience of
classical portraits, and human-robot interaction (Chen et al., 2018;
Guédron, 2011;Tateyama, 2016). As with any art object, an interac-
tion with face representations or its aesthetic understanding is a com-
plex relationship that involves social norms and cultural background,
but also calls on the cognitive processes of the observers. For exam-
ple, interpreting some will or intention (Gell, 1998) or experiencing
an emotion (Schaeffer, 2015) at the view of a face representation
could reflect those cognitive processes. We assume that facial reac-
tions to face representations could also play a role in those interac-
tions relying on embodied emotion processes. From the aesthetic
experience point of view, Gallese (2019) and Freedberg (2009)
insisted on the role of body engagement in the interaction with artis-
tic images. Freedberg (2009) reported that, beyond context, aesthetic
understanding of objects requires a sense of “the neural substrate of
human engagement with movement and embodiment, and the innate
potential for recognition of the emotions that may ensue from them”
Freedberg, 2009,p.6).
Observer’s cognitive processes and their facial reactions, in art
interaction context, have already been investigated (Gerger et al.,
2011,2014;Gernot et al., 2018;Leder et al., 2014). However,
beholder’s cognitive processes and their facial reactions when
interacting with face representations in art has not been the interest
of research yet. Having a further understanding of the reaction to-
ward face representations helps understanding, at some point, the
social interaction with face representations and contributes to
shedding light on a new facet of art and social analysis.
The review of literature about the facial process as a reaction to
facial expressions leads us to expect that facial mimicry could be a
potential process in reaction to face representations. Studies on face
perception of some face representations offer support to this postu-
late. Indeed, some studies showed that face representations are rec-
ognized as face categories even if they depict only schematic faces,
emoticons, or items with mere hints of a face (Churches et al., 2014;
Hadjikhani et al., 2009;Liu et al., 2016;Wheatley et al., 2011). This
is attested by the fact that an increase in brain wave associated with
face perception was observed in EEG studies (Eimer, 2011;Sagiv &
Bentin, 2001). If facial mimicry takes place when people observe
face representations in art like facial mimicry of actual faces, then
we may assume that perceiving facial representations might also
rely on emotional contagion and probably understanding others’
mental state processes. Facial mimicry is one of the many ways in
which viewers precognitively grasp emotions that are shown or sug-
gested in works of art (Freedberg & Gallese, 2007), and might be
the most significant one to grasp emotions in face representations.
2ACHOUR-BENALLEGUE ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
794

As put by Freedberg (2009), part of the force of masks belonging to
faraway cultures resides in our bodily responses, even if we may not
be able to pinpoint the exact emotions involved.
Some anthropological theory has already put forward the hy-
pothesis that beholders attribute a social agency to artistic objects
(Gell, 1998). The social agency is the capacity of action of an
agent in its environment. It could be the capacity of causing events
by acts of will or intention. This anthropological hypothesis (Gell,
1998) claims that beholders infer mind qualities (will, intention) in
the art objects. The focus of this theory extends beyond face repre-
sentations, but we can suggest that toward art objects representing
faces, mimicry could stand among the cognitive causes of art
agency. This would give cognitive support to art agency theory. In
the present study, we will investigate whether the mimicry process
underlies facial reactions to face representations.
In the present work, we propose to examine facial mimicry of re-
alistic and nonrealistic face representations. We hypothesize that
mimicry would be observed in face reactions to face representa-
tions. To test this hypothesis, we performed electromyography
(EMG) experiment that tests mimicry to static facial expressions in
face representations. In this experiment, we focused EMG measures
on the muscles of facial regions that are more strongly activated in
the emotions depicted by stimuli. Here, the emotions are not identi-
fied, as such, but only the expression of the muscles related to
them. The study focuses more on mimicry according to the muscu-
lar pattern than to specificemotions.Zygomaticus major,corruga-
tor supercilii,lateral frontalis,anddepressor anguli oris are the
four recorded muscle-regions. Our hypothesis states that mimicry
occurs in reaction to face representations whatever the stimuli are
realistic or nonrealistic. We expect that observers might produce a
facial reaction pattern that is similar to the pattern of facial expres-
sions depicted in the observed face representations. In other words,
we expect that EMG activity would be stronger when stimuli depict
high muscle activity than when they depict low muscle activity.
This expresses a similar muscle pattern between the observer and
the stimulus. Following the analysis of the previous hypothesis, we
also suggest more exploratory analyses to investigate realism
degree and emotional dimension effect on mimicry and emotional
contagion.
Materials and Method
Participants
Participants were 30 women and 17 men, belonging to western
culture, between the ages of 18 and 62 (M= 29.6, SD = 9.5). All
participants had normal or corrected to normal visual acuity.
Twelve participants, among 47, were eliminated because of techni-
cal or continuous behavioral artifacts (e.g., sleeping or tic move-
ments). Most of these discarded participants showed sleepiness
and were not watching the screen. All participants have received
explicit information about the experiment design and gave signed
informed consent according to the Declaration of Helsinki. They
also were paid at the end of the experiment. The study protocol
followed the local ethic guidelines from Jean Jaurès University in
Toulouse, France.
Physiological Data
After suitable skin preparation (Hermens et al., 2000) to reduce
the electrode-site impedance, EMG from zygomaticus major
(involved in smiling), corrugator supercilii (involved in frowning),
lateral frontalis (involved in eyebrow raising), and depressor anguli
oris (involved in lip lowering) of each participant was recorded at
1024 Hz using ActiveTwo system (BioSemi instrumentation, Am-
sterdam, the Netherlands). Electrode placement for these muscles
followed the recommendations of Van Boxtel (2010) for meas-
uring facial EMG activity (see the figure of electrode positioning
in the OSF repository URL: https://osf.io/yk7hz/?view_only=
972e5179b5904fcbaf5b1df70416b934). Eight BioSemi FLAT active
electrodes (11 mm width, 17 mm length, 4.5 mm height) were
placed on the skin in bipolar configuration with 10-mm interelec-
trode distance using electrode placement recommended by Dimberg
and Petterson (2000). Two additional electrodes, the common mode
sense active electrode, and the driven right leg passive electrode,
were used as reference and ground electrodes, respectively. Partici-
pants were told that their skin conductance will be recorded when
viewing the stimuli. They have been informed about the actual mea-
surement at the end of the experiment.
Procedure
EMG Task
The experiment was divided into four blocks, each one included
48 trials (Figure 1a). A 3- to 5-min break was taken between
blocks. Each stimulus was displayed for four seconds preceded by
1-s fixation cross. A blank screen was displayed as an intertrial
interval with a random duration from 9 s to 13 s to relax the
muscles. The display was carried out on a 17-in. computer screen
at a viewing distance of 60 cm. We asked participants to view dis-
played stimuli after focusing on the fixation cross. To ensure that
participants have understood the task, a pretest was performed at
the beginning of each block. Thus, each stimulus was repeated
four times in total, one per block. Stimuli were displayed ran-
domly without repetition within blocks. To avoid boredom and
loss of concentration for participants, we added cognitive tasks in
the second, third, and fourth blocks: yes/no question about famili-
arity (did participants see the stimulus before), yes/no question
about attractiveness (is the stimulus attractive), and memory task
at the end of the block respectively.
After completing the four blocks, maximum voluntary contrac-
tions (MVC) were measured (Figure 1b). This step corresponds to
a physiological data standardization measure. Participants were
asked to imitate the stimuli by contracting voluntary and strongly
some muscles. Four actual human facial expressions were dis-
played as stimuli (smile, frown, eyebrow raising, and lip lower-
ing). Each stimulus was preceded by 500 ms fixation cross and
displayed three times during 2 s to 5 s. Stimuli were displayed ran-
domly. Intertrial gray screen with “Relax”inscription was dis-
played during a random interval of 5 s to 10 s. To avoid muscle
fatigue during the MVC task, the task was split into two parts.
Each one includes a set of six-trial presentation. A 2-min break
was taken between the two parts. During this break, participants
filled up a personal data questionnaire.
FACIAL REACTIONS TO FACE REPRESENTATIONS IN ART 3
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
795

SAM Task
After the MVC task, participants assessed the 48 stimuli using
the Self-Assessment Manikin scale (SAM), which assesses emo-
tion in three independent affective spaces (pleasure, arousal, and
dominance). Participants reported their pleasure, arousal, and
dominance feelings when viewing stimuli. Each of the SAM emo-
tion dimensions (pleasure, arousal, and dominance) was displayed
on a nine-value graphical scale beneath the stimulus display (one
dimension per screen).
The two tasks of the experiment were implemented on Matlab
R2009b using the Psychophysics Toolbox extensions (see the
implementation script in the OSF repository URL: https://osf.io/
yk7hz/?view_only=972e5179b5904fcbaf5b1df70416b934).
Stimuli
Stimuli consisted of 48 cross-cultural pictures of facial expres-
sion representations (25 from western art and 23 from nonwestern
art). All belong to pictures of artworks and ethnographic objects
crafted with different techniques (25 images in two-dimensions
[2D; painting, drawing] and 23 images in three-dimensions [3D;
sculpture]). They appear in different colors and textures due to
their different styles and origins. Among these stimuli, 28 have
open eyes (observer-toward gaze or a slightly averted gaze) and 20
have closed/occluded eyes due to their expressions. To preserve
the cross-cultural diversity balance of the stimuli, we kept them as
they are despite their differences in style, technique, and eye
appearance. In all stimuli, pictures were cropped on faces and pre-
sented full-face.
The stimuli displayed different expressions from neutrals to
extreme smiling, frowning, lip lowering, or eye-opening/eyebrow-
raising (24 images with intense expression and 24 images with
neutral expression). These two controlled categories of expression
intensity are relevant for testing emotional contagion in terms of
arousal. The stimuli include two other controlled categories: 24 re-
alistic images and 24 nonrealistic images. These two categories
are relevant for testing the realism degree effect on mimicry.
Besides these categories, we fulfilled a classification according to
the expressed muscular activity. This classification provides the
muscular pattern of the represented facial expression (see the sort-
ing figure in the OSF repository URL: https://osf.io/yk7hz/?view
_only=972e5179b5904fcbaf5b1df70416b934), which is necessary
for mimicry examination (comparison of classes). The classifica-
tion ranges, for each muscle, the 48 images into three classes. The
three classes are (a) images expressing the weakest muscle activity
(weak), including the absence of activity (N= 16 for zygomaticus
and corrugator,N= 17 for frontalis, and depressor); (b) images
expressing medium muscle activity (medium; N= 16 for zygoma-
ticus,N= 15 for corrugator,frontalis, and depressor); and (c)
images expressing the strongest muscle activity (strong; N=17
for corrugator,N= 16 for zygomaticus,frontalis, and depressor).
In sum, we constructed two independent variables from stimuli to
test mimicry: realism degree including two modalities (realistic vs.
nonrealistic) and image classes counting three modalities (weak,
medium, and strong). We also constructed a supplementary inde-
pendent variable from stimuli to test emotional contagion: expres-
sion intensity including two modalities (intense vs. neutral).
Classification Procedure
Realism and Expression Intensity Categories
The 48 stimuli were selected from a database of 211 face repre-
sentations previously online assessed (Achour-Benallegue et al.,
2016). In this previous study, the realism degree and the intensity
of expression of each image were assessed by other participants
than the EMG study ones. The 211 face representations were
ranged on two axes according to the assessment results: one axis
Figure 1
Experimental Design of Facial Mimicry to Face Representations
Note. The picture depicts stimuli sequence and display time for (a) one block of mimicry task and (b)
Maximum Voluntary Contraction (MVC) task. The individual shown gave explicit written consent for the pub-
lication of his face images. See the online article for the color version of this figure.
4ACHOUR-BENALLEGUE ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
796

for realism degree and one axis for expression intensity. Then four
clusters from the intersection of the two axes extremes were high-
lighted. Forty-eight images were randomly chosen from the four
clusters: 12 from the extreme realistic degree cluster, 12 from the
extreme nonrealistic degree cluster, 12 from the extreme intensity
cluster, and 12 from the extreme neutral cluster. Thus, among the
24 realistic images, 12 represent an intense expression and 12 rep-
resent a neutral expression. Similarly, among the 24 nonrealistic
images, 12 represent an intense expression and 12 represent a neu-
tral expression.
Image Classes
The classification according to muscle activity required a sup-
plementary task. Stimuli classification was operated using an
EMG based-criterion through the following protocol: (a) a volun-
tary imitation of the stimuli facial expressions, (b) a record of this
imitation in terms of muscle activity (EMG activity), and (c) a
classification of the stimuli according to the different values of the
EMG activity. The supplementary task consisted of performing a
voluntary mimicry task on 15 new participants (equipped with an
EMG recording device) respecting the same conditions as in the
present experiment (the same 48 stimuli, 1-s fixation cross, 4-s
stimulus display, 9- to 13-s intertrial interval, four blocks, MVC
task). More precisely, we asked the new participants to reproduce
exactly on their faces the facial expression of each stimulus.
The classification was done by using three quantiles of EMG ac-
tivity means (which were computed for each stimulus). These
quantiles divide the data into three categories. Each category
includes the same number of images. First, the images to which fa-
cial reaction (EMG activity expressed in DRMS relative to the
baseline) is greater than the second quantile constitute the class of
images expressing the strongest muscle activity (strong). Then, the
images to which facial reaction is between the first and the second
quantile constitute the class of images expressing the medium
muscle activity (medium). Finally, the images to which facial reac-
tion is below the first quantile constitute the class of images
expressing the weakest muscle activity (weak). For example,
among the realistic stimuli, the strong classes of frontalis,corru-
gator,depressor, and zygomaticus activation include, respectively:
L’ébahi (painted by Boilly), the face of Boy Bitten by a Lizard
(Caravaggio), the Humored Man (sculpted by Messerschmidt),
and the Moche portrait vessel (Moche culture). Among the non-
realistic stimuli, the strong classes of frontalis,corrugator,depres-
sor, and zygomaticus activation respectively include for example:
the Kavat mask (Baining ethnicity), Reversible Head with a Fruit
Basket (painted by Arcimboldo), ‘aumakua hulu manu (Hawaiian
feathered god), and Tupilak figure (Ammassalik island).
Data Preparation
To check behavioral artifacts, participants were videotaped all
along the data acquisition using a digital video camera (Canon iVIS
HF M31). As videotape analysis showed several kinds of motion
artifacts (e.g., yawning or turning eyes away), artifacts-contami-
nated trials were removed from the analysis. Only 4% of the whole
trials were removed, 96% among them were included in clean data.
EMG Data
Raw EMG was first denoised by band-pass filtering at 20–400
Hz, then full-wave rectified by taking the absolute value, and
finally smoothed at 9 Hz to obtain the linear envelopes (Lajante et
al., 2017). For each muscle, facial EMG activity was then obtained
by computing the root-mean-square (RMS) from the EMG enve-
lope, which provides a relevant measure of the electrical activity
of the muscle during contraction (Bouisset & Maton, 1995). As
was done in numerous studies investigating facial EMG responses,
in different fields, (Kwon & Christou, 2018;Lajante et al., 2017;
Sato et al., 2008;Urgesi et al., 2016), the muscle activity during
the last second before each stimulus onset was defined as the base-
line to subsequently quantify EMG response associated to this
stimulus presentation appropriately. Facial reactions were
expressed as a change in the root mean square (DRMS relative to
the baseline) between the muscle activity during the 4 s picture
presentation and the associated baseline. To obtain physiologically
standardized data (Van Boxtel, 2010), this DRMS value was nor-
malized to the corresponding muscle DRMS (relative to the base-
line) measured during MVC test. That is, for each muscle, the
measured DRMS during the mimicry task was divided by the D
RMS of the same muscle measured during the MVC test. This
enables facial EMG analysis to compare data between different
participants. Signal processing was performed on Matlab R2009b
(see the EMG analysis script in the OSF repository URL: https://osf
.io/yk7hz/?view_only=972e5179b5904fcbaf5b1df70416b934). Raw
signals were extracted using the eeglab toolbox (Delorme & Makeig,
2004).
SAM Data
From SAM data, we reconstructed three variables: valence
(negative, average, positive), arousal (low, average, positive), and
dominance (low, average, high). The negative/low modality
includes data that is below the score 3, the average modality
includes data that is between the score 4 and score 6, and the posi-
tive/high modality includes data that is higher than 6.
Statistical Analysis
To investigate the influence of facial expression representations
on mimicry, we used a mixed-effect model, with facial reactions
(DRMS relative to the baseline) being the dependent (continuous)
variable, and image classes (weak, medium, strong), image realism
degree (realistic vs. nonrealistic) being the independent variables.
Four mixed effect models were performed, one model per muscle
region: zygomaticus,corrugator,frontalis, and depressor. All var-
iables were considered fixed-effect variables, except for partici-
pants (N= 35) and images (N= 48), which were considered
random-effect variables. Inspection of the results made us refrain
from doing any analysis involving the intermediate stimuli, and
will follow the analysis method used in previous studies that used
stimuli with high intensity of expressions (Dimberg & Thunberg,
2012;Likowski et al., 2012). Thus, we focused our comparison
only on the two extreme classes (weak vs. strong). Planned con-
trast analysis was performed to test facial mimicry. This contrast
tests whether the facial reaction (DRMS relative to the baseline)
was greater in “strong”compared with “weak”classes.
FACIAL REACTIONS TO FACE REPRESENTATIONS IN ART 5
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
797

Other post hoc contrasts were conducted to investigate: (a) the
effect of the stimuli properties, which are noncontrolled variables,
on EMG activity (technique: 2D, 3D; style: western, nonwestern;
eyes: opened, closed/occluded), as well as the effect of the expres-
sion intensity of stimuli on EMG activity; and (b) the effect of
image realism degree and the effect of valence, arousal, and domi-
nance on mimicry. The latter contrasts, first, compare the DRMS
(relative to the baseline) of the two extreme classes (“weak”vs.
“strong”) for realistic images on one side, then for nonrealistic
images on the other side. Second, these contrasts test whether the
facial reaction (DRMS relative to the baseline) was different for
the extreme modalities of valence (positive vs. negative), arousal
(high vs. low), and dominance (high vs. low). The contrasts were
included in the mixed-model procedure.
To investigate the emotional contagion of face representations,
we used a Kruskal-Wallis test, with arousal responses (scored on
nine-value scale) being the dependent (continuous) variable, and
expression intensity of images (intense vs. neutral) being the inde-
pendent variable. This ANOVA tests the effect of the expression
intensity on the affective response of arousal. Not to be confused
with the independent variable arousal (low, high), the arousal
responses scored on 9 value-scale is designated as arousal_affect.
The threshold of significance (alpha) is .05. The models were
adjusted using Bonferroni corrections pairwise comparisons. Sta-
tistical analyses were conducted using SAS Version 9.2. All
scripts (procedure, data preparation, and statistical analysis) are
available in supplementary data (see the statistics analysis script
in the OSF repository URL: https://osf.io/yk7hz/?view_only=
972e5179b5904fcbaf5b1df70416b934).
Results
Mimicry
As depicted in Figure 2,findings on mimicry show different fa-
cial reactions to the different muscles. For three muscles (zygoma-
ticus,depressor and frontalis), the facial reaction was significantly
influenced by image classes (weak vs. strong; t(1,6395) = 2.57,
p= .03 for zygomaticus,t(1,6395) = 2.89, p= .01 for depressor,
and t(1,6395) = 2.70, p= .02 for frontalis). Zygomaticus and de-
pressor results did respect our predictions, that is, the DRMS rela-
tive to baseline for zygomaticus and depressor activation in the
weakest activity class was lower than the DRMS relative to base-
line for these muscles activation in the strongest activity class (.82
(6.93) ,3.15 (6.92); and .39 (6.93) ,3.09 (6.88), respec-
tively). However, frontalis results were opposite to our predic-
tions: the DRMS relative to baseline for frontalis activation in the
weakest activity class was higher than the DRMS relative to base-
line for frontalis activation in the strongest activity class (.42
(61.07) .1.61 (61.06)). The image classes (weak vs. strong)
seem to not influence the DRMS relative to baseline for the corru-
gator muscle (t(1,3695)) = 1.09, p= .83).
Exploratory Analyses
In an exploratory way, we performed several other statistical
analyses that shed light on the category of stimuli and bring some
explanations of the previous results. These exploratory analyses
also give an insight to emotional contagion.
Category of Stimuli
None of the noncontrolled properties (technique, style, and
eyes) significantly influenced the EMG activity of each muscle.
Similarly, the EMG activity of each muscle was not affected by
the expression intensity nor the realism degree. No differences
were observed in the DRMS for zygomaticus,corrugator, fronta-
lis, and depressor activation between the following pairwise-tests:
2D versus 3D images, western-belonged versus non-western-
belonged images, opened-eye versus closed/occluded-eye images,
intense versus neutral images.
Realism Degree Effect on Mimicry
As the stimuli include as well realistic as nonrealistic face repre-
sentations, we looked closer into the data to investigate whether
the realism degree affects mimicry. Does mimicry occur in the cat-
egory of realistic stimuli and the category of nonrealistic stimuli
separately? If so, for each category of realism degree (realistic and
nonrealistic) EMG activity would still be stronger for stimuli
depicting high muscle activity than for stimuli depicting low mus-
cle activity. In realistic stimuli, no differences were observed in
the DRMS relative to baseline for zygomaticus,corrugator, and
depressor activation between image classes (see Table 1). However,
the effect of realistic image classes on the DRMS relative to baseline
for frontalis activation was significant, t(1,6395) = 3.18, p= .02.
Conversely to our predictions, the DRMS for frontalis activation
in the weakest activity class was higher than the DRMS for fron-
talis activation in the strongest activity class (.79 (61.06) .2.41
(61.26)). In nonrealistic stimuli, for all muscles, no effect of
image classes on the DRMS relative to baseline was observed.
Valence Effect on Mimicry
To investigate previous results for corrugator and frontalis,
we tested the effect of emotional dimensions (valence, arousal,
dominance) on facial reaction (DRMS relative to baseline).
Findings show a significant effect of valence on corrugator and
frontalis reactions,aswellasonzygomaticus (Table 2). Arousal
and dominance did not have any significant effect on the corru-
gator and frontalis. Besides, findings show a significant effect of
valence on corrugator only in the strong corrugator activity class
of stimuli (t(1,6385) =3.59, p= 0.0003). Negative valence was
associated with a significant corrugator activity response (DRMS
relative to baseline = 2.50), whereas, positive valence was associated
with a significant decrease of corrugator activity response (DRMS
relative to baseline = 1.81; Figure 3).
For zygomaticus,findings show a significant influence of valence
on zygomaticus activation (t(1,6385) = 3.01, p= 0.003). Positive va-
lence enhanced stronger zygomaticus activity (DRMS relative to
baseline = 3.85) compared with negative valence (DRMS relative to
baseline = 0.88). Moreover, the valence that has been attributed to
stimuli associated with strong corrugator activity was not signifi-
cantly associated with changes in zygomaticus activity (t(1,6385) =
0.98, p=0.33).
On frontalis,asignificant effect of valence was observed only in
the weak frontalis activity class of stimuli (t(1, 6385) = 2.29, p=
0.02). The negative valence was associated with a significant fronta-
lis activity response (DRMS relative to baseline = 1.71) whereas, the
positive valence was associated with a significant decrease of fronta-
lis activity response (DRMS relative to baseline = 1.21; Figure 3).
6ACHOUR-BENALLEGUE ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
798

Emotional Contagion
Our exploratory analyses show a significant effect of expression
intensity of stimuli on arousal_affect (Khi2 = 19,48; p,0,0001).
The means of arousal_affect toward intense images was greater that
the means of arousal_affect toward neutral images (5.8 .4.6).
Discussion
This study investigates whether facial expressions in face repre-
sentations are mimicked. We tested the hypothesis stating that
observers would mimic face representations regardless of realism
categories. A reaction could be labeled as mimicry if it met the fol-
lowing two conditions (Fujimura et al., 2010;Lajante et al., 2017):
(a) image classes have a significant effect on facial reactions, and
(b) the strongest activity class has a greater DRMS relative to base-
line for muscle activation compared with the weakest activity class.
Mimicry of Face Representations
To test our assumptions, we conducted analyses for the four
muscles independently (see Table 3).
Figure 2
EMG Activity of the Four Measured Muscles in the Two Extreme Image Classes (Weak vs. Strong)
Note. EMG = electromyography; RMS = root-mean-square. The EMG activity of zygomaticus (A), corrugator (B), frontalis (C), and de-
pressor (D) is represented by the estimation of DRMS relative to baseline. This represents EMG facial reactions as a response to stimulus
presentation. The figure depicts, for each muscle, the measured estimation of DRMS relative to baseline for the two extreme classes of
images: the weakest activity class ’weak’(blue bars), and the strongest activity class ’strong’(orange bars). The estimation values were multi-
pliedby1000(%). Error bars indicate the standard deviation of the estimation. See the online article for the color version of this figure.
FACIAL REACTIONS TO FACE REPRESENTATIONS IN ART 7
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
799

Zygomaticus and Depressor
The two conditions for mimicry were satisfied in two muscles:
zygomaticus and depressor. That means that participants’faces
reacted similarly to image expressions in terms of zygomaticus
and depressor muscle patterns. Consequently, face representations
depicting mouth expressions might be mimicked by observers.
When it comes to mouth expressions, observers’facial reactions to
face representations might be quite similar to the reactions toward
actual (human) faces. This brings experimental support to Freed-
berg’s suggestion on Inuit masks from Point Hope (Copenhagen;
Freedberg, 2009). Freedberg (2009) reported that we can be sure
that the force of these masks resides perhaps in our felt buccal
responses to them. Mimicry in zygomaticus and depressor is con-
sistent with previous studies on human faces stimuli showing
mimicry in zygomaticus (Seibt et al., 2015) and depressor (Soussi-
gnan et al., 2013;Philip et al., 2018). Philip et al.’s (2018) study
tested also facial reactions toward virtual faces (not human faces),
which are very realistic according to our definition of realism.
This study reported that sad virtual faces also enhanced facial reac-
tions that are congruent with the expression they display, but only
when displayed in the dynamic condition and not in the static one.
In our study, even though all stimuli were static and regardless of
their realism category, depressor activity in observers was congru-
ent with the facial expressions of these stimuli. This means that
face representations depict strongly facial expressions that might
convey salient social events.
None of the mimicry studies have been conducted on cross-cul-
tural face representations (realistic and nonrealistic), consequently,
our results are the first to show this effect in observer’s reactions
to face representations. And compared with previous studies on
human or virtual faces, our stimuli include nonrealistic images in
which the representation of zygomaticus and depressor muscle ac-
tivity could refer to more complex emotions than just happy or sad
emotions. Thus, our result on zygomaticus and depressor muscle
mimicry reactions to face representations might widen the possi-
bility of mimicry to more complex emotions (not only basic emo-
tions; Ekman & Cordaro, 2011) compared with the literature on
human faces (such as happy and angry faces).
The fact that mimicry was observed only in zygomaticus and de-
pressor could be due to a bias caused by cultural background. Par-
ticipants belonging to Western culture may have a bias that makes
them focus on mouth representations that involve zygomaticus and
depressor. Indeed, it has been shown that in recognizing the facial
expressions of emotions, Westerners focus on the position of the
mouth more than the eyes, compared to Easterners (Caldara, 2017;
Yuki et al., 2007). A cross-culturally bigger sample of participants
may bring more answers on this issue.
Reactions of zygomaticus and depressor muscles to realistic and
nonrealistic images separately did not satisfy the mimicry condi-
tions. This contrasts with the result of the global mimicry hypothesis
showing mimicry in zygomaticus and depressor. This led us to think
that our experiment did not include enough data in each of the two
categories (realistic and nonrealistic) to draw conclusions on the
issue. A larger sample of stimuli in both categories should bring
more conclusions on the effect of realism degree on facial reactions
to face representations.
Frontalis
In the frontalis muscle, the first mimicry condition was satisfied,
but not the second one. Contrary to expectation, the DRMS rela-
tive to baseline for frontalis activation in the “weak”class was
higher than in the “strong”class. This means that the muscle pat-
tern of facial reactions does not fit the displayed expressions in
stimuli, but it is inverted compared to the displayed expressions.
that is, Frontalis activity pattern is produced in reaction to images
without frontalis activity representation, and conversely, images
with frontalis activity representation do not elicit frontalis activity
pattern. To explain this result, we may suppose that this frontalis
reaction could be the result of an emotion stemming from self-rel-
evant appraisal processing (Grèzes et al., 2013;Soussignan et al.,
2013), such as surprise. In other words, participants could have
displayed frontalis reactions as a function of social meaning of
some image properties, here the images belonging to the weak
frontalis-activity class. This result is to be compared with studies,
in which participants displayed congruent facial reactions to per-
ceived facial expressions as a function of the social meaning of
perceived gaze direction (Soussignan et al., 2013). In that case,
gaze direction in stimuli affected facial reactions (higher zygomati-
cus and corrugator activity to happy and angry faces, respectively,
with direct than averted gaze, higher frontalis activity to fear faces
with averted than direct gaze). Gaze direction was interpreted as a
self-relevant clue to account for responses to salient events (Sous-
signan et al., 2013). In our study, we may suppose that a common
property among those “weak”images could have enhanced fronta-
lis-related emotion, similarly to the gaze direction. When looking
deeply in the two extreme frontalis classes we note, a posteriori,
that most of “weak”class stimuli are realistic and most of “strong”
class stimuli are nonrealistic. This realism distribution over stimuli
might have an influence on facial reaction and mimicry. Partici-
pants might have activated more their frontalis when the stimuli
were realistic than when stimuli were nonrealistic despite the
strength of the expressions of frontalis activation. Nevertheless, this
does not seem appropriate for all stimuli belonging to all classes.
Indeed, in our statistical model the effect of realism variable alone
on frontalis reaction was not significant as reported in the explora-
tory analyses. Then, the realism degree over all stimuli (whatever
the expression is) does not seem to influence frontalis reactions.
However, our results showed an influence on frontalis reactions by
realism degree in the subset of stimuli belong to the extreme fronta-
lis activity classes (expressions related to frontalis only).
Table 1
Effect of the Interaction of Image Classes With Realism Degree
on Mimicry
Realistic images Nonrealistic images
Muscles tp t p
Zygomaticus 1.76 1.00 1.87 .92
Corrugator .76 1.00 2.14 .48
Frontalis 3.18 .02 0.76 1.00
Depressor 2.82 .07 1.35 1.00
Note. The effect is represented in ttest and Bonferroni corrected proba-
bilities. The first column displays the effects of contrast between the
extreme classes (weak vs. strong) in realistic images, and the second col-
umn in nonrealistic images.
8ACHOUR-BENALLEGUE ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
800

Reactions of frontalis to realistic and nonrealistic images sepa-
rately were as follow: (a) for nonrealistic images none of the mim-
icry conditions were satisfied, and (b) for realistic images, the first
condition of mimicry was satisfied (with inverted pattern) but not
the second condition. These results show that only realistic images
lead to an inverted muscular pattern in the frontalis. Thus, this self-
relevant appraisal processing might be specific to realistic images
(and not to nonrealistic ones). Realism might be interpreted as a
self-relevant clue that contributes to frontalis reaction. Further
Figure 3
Stimuli Valence Effect on Zygomaticus, Corrugator, Frontalis, and Depressor
Reactions
Note. EMG = electromyography; RMS = root-mean-square. The EMG activity of zygomati-
cus (A), corrugator (B), frontalis (C), and depressor (D) is represented by the estimation of
DRMS relative to baseline. The figure depicts, for each muscle, the measured DRMS rela-
tive to baseline for the two valence conditions (negative vs. positive) in each stimuli class
(weak vs. strong). The EMG activity in the negative valence condition is depicted by purple
bars, and the EMG activity in the positive valence condition is depicted by yellow bars.
The estimation values were multiplied by 1000 (%). Error bars indicate the standard devia-
tion of the estimation. See the online article for the color version of this figure.
Table 3
Mimicry’s Conditions per Muscle
Muscles Condition 1 Condition 2 Mimicry
Zygomaticus Satisfied Satisfied Satisfied
Corrugator Not satisfied Not satisfied Not satisfied
Frontalis Satisfied Not satisfied Not satisfied
Depressor Satisfied Satisfied Satisfied
Note. RMS = root-mean-square; EMG = electromyography. DRMS (þ)
refers to the DRMS, relative to baseline, of EMG reaction in the positive
valence condition, and DRMS () refers to the DRMS, relative to base-
line, of EMG reaction in the negative valence condition.
Table 2
Valence Effect on Corrugator, Frontalis, and Zygomaticus
Reactions
Muscle Corrugator Frontalis Zygomaticus
Valence P,0.0001 P= 0.001 P= 0.003
RMS (þ),RMS () RMS (þ),RMS () RMS (þ).RMS ()
Note. RMS = root-mean-square. Condition 1 refers to a significant effect
of image classes on facial reactions. Condition 2 refers to a greater RMS
for muscle activation in the strongest activity class compared with the
weakest activity class.
FACIAL REACTIONS TO FACE REPRESENTATIONS IN ART 9
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
801

studies should be conducted with more balanced realism degree in
each class (weak and strong) to shed light on this result. This alter-
native hypothesis supposes that mimicry does not occur when it
comes to frontalis expressions.
Another result on frontalis activation depending on stimuli property
(valence) brings more explanation to frontalis reactions. The test of va-
lence effect on frontalis reaction showed a significant effect of stimuli
valence on frontalis activity that was stronger in negative compared to
positive valence. This suggests that participants’frontalis reacted when
they felt negative valence emotion toward face representations, and not
specifically when face representations were expressing strong frontalis
activity. Frontalis reaction might be the result of a negative emotion
stemming from self-relevant appraisal processing that is related to the
realism of face representations in art. A more detailed observation of
the valence effect on frontalis reaction indicated a significant effect of
valence on frontalis reactions only in the weak activity class. This
reflects the inverted pattern of frontalis mimicry. In the weak activity
class, the negative valence was associated with a significant frontalis
activity response (activated frontalis), whereas, positive valence was
associated with a significant decrease of frontalis activity response
(relaxed frontalis). This result suggests that, beyond stimuli realism,
the property related to the self-relevant appraisal hypothesis is related
to the negative affect conveyed by the stimuli. Further research is
needed to understand better how negative valence and realism play a
role in frontalis reactions toward face representations.
We may also consider other alternative hypotheses: (a) frontalis
reaction could have been disturbed by a lack of information related
to the full actual expression which highlights frontalis activation; (b)
frontalis reaction could have been disturbed by a combination of con-
tradictory activations which is depicted in the stimuli of the “strong”
class. This could be tested, in further studies, by comparing frontalis
reaction to the current stimuli with its reaction to face representations
which respects the combination of actual activations.
Corrugator
For corrugator, none of the mimicry conditions were satisfied,
neither in general nor in realism categories separately. One expla-
nation might be the influence of pleasure felt by participants due
to their interaction with artworks. Corrugator activity has been
shown as a good indicator of negative valence (Gerger et al.,
2011). The hedonistic experience of art (experience of pleasure)
might have attenuated the corrugator activation by enhancing posi-
tive valence in participants. This hedonistic influence could be a con-
sequence of social context influence on mimicry (Seibt et al., 2015).
Our explanation is endorsed by the results on the valence effect
on corrugator reaction. Corrugator reaction was significantly
affected by stimuli valence. The negative valence was associated
with a significant corrugator activity response (activated corruga-
tor), whereas, positive valence was associated with a significant
decrease of corrugator activity response (relaxed corrugator).
This result endorses previous studies on corrugator reaction to
negative valence (Gerger et al., 2011).
Given the mimicry hypothesis, the corrugator strong-activity
class of stimuli is supposed to enhance corrugator activity (and this
activity should be higher than the one toward stimuli from the weak
activity class). However, when we tested the effect of valence in
this strong activity class, we found a significant influence of valence
on corrugator reaction (enhanced activation in the negative
condition and decreased activation in the positive condition). This
effect could be explained by the influence of the hedonistic context
on corrugator activity in terms of the felt pleasure when viewing
highly expressive art representations.
As zygomaticus and corrugator activities are negatively corre-
lated when observing positive and negative stimuli (Fujimura et
al., 2010;Gerger et al., 2014), a better understanding of the corru-
gator reaction should take into account the zygomaticus one.
Indeed, it has been shown that positive stimuli enhance zygomaticus
activity and decrease corrugator activity, and negative stimuli
enhance corrugator activity and decrease zygomaticus activity
(Fujimura et al., 2010;Gerger et al., 2014). We observed a signifi-
cant effect of valence on zygomaticus activity (stronger zygomaticus
activity in the positive compared to the negative valence condition).
This result supports previous studies on zygomaticus reaction to
positive valence (Gerger et al., 2011,2014). However, the felt va-
lence that has been attributed to stimuli associated with strong cor-
rugator activity was not significantly associated with changes in
zygomaticus activity. This means that the positive valence attributed
to face representations expressing strong corrugator activity did not
show a significant effect on zygomaticus. This result suggests that
stimuli, representing a strong activation of the corrugator, might
have prevented zygomaticus activation despite the positive valence.
This prevention of zygomaticus activation toward positive-assessed
stimuli, associated with strong corrugator activity, might reflect a
trace of a tendency of mimicry reaction. Taken together, our find-
ings lead us to consider the hypothesis of hedonistic context influ-
ence on corrugator activity as the most relevant one.
An Interdisciplinary Study
Mimicry might contribute to a new approach for studying face
representations and their relationship with observers. Indeed, experi-
mental psychology in general and, facial expressions and mimicry,
in particular, are not that much appealed in anthropology and art
studies to understand the human behavior toward face representa-
tions. Yet a lot of knowledge has been collected on facial expressions
in experimental psychology. Our study offers an interdisciplinary
view that associates art issues (social interaction with face represen-
tations) with a specific cognitive process (facial mimicry). Conse-
quently, our hypothesis and result interpretations might extend the
anthropological and artistic discussions to findings on mimicry and
facial expressions.
Moreover, our findings on the expression intensity on arousal
responses (arousal_affect) lead us to assume that face representa-
tions might trigger an emotional contagion regardless of upper-
face or lower-face muscles. Indeed, the expression intensity of
face representations is related to the intensity of the expressed
emotion. Thus, a higher arousal_affect response to intense images
compared to neutral ones reflect an emotional contagion process in
terms of arousal affect. Given the mimicry result and by relying on
the emotional contagion one, we can suggest, as mentioned in the
literature, that the observed mimicry might be related to emotional
contagion (Dimberg & Thunberg, 2012;Hess & Fischer, 2014),
and understanding others’mental states such as intentions (Blake-
more & Decety, 2001;Niedenthal et al., 2010;Schilbach, 2016).
Thus, it might be possible to convey spontaneously emotions
(such as happiness, satisfaction, sadness, or contempt) or inten-
tions (such as caring or benevolence) through face representations
10 ACHOUR-BENALLEGUE ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
802

expressing zygomaticus or depressor activation. This does not
mean that the felt emotion in front of face representations is lim-
ited to this kind of spontaneous emotional contagion or is forced
to stay at this level. The felt emotion could also stem from more
complex processes which include regulatory strategies that depend
on participants mood (Stamatopoulou, 2018), or mental represen-
tations related to the context of the interaction (Niedenthal, 2007)
or which include memory (Adolphs, 2002). Nevertheless, the early
emotional contagion that we suggest to be induced by face repre-
sentations might apply as a priming bias for further more complex
emotional experience (Niedenthal, 1990).
Facial mimicry and its correlated processes (emotional contagion
and understanding expressed intentions) might contribute to the
social interaction with face representations. They might be an index
of face representation agencies. This could reflect Gell’santhropo-
logical theory of art (Gell, 1998), where art objects are means that
convey social agencies. In other words, art objects might transmit
the emotions and the intentions of those who created them (artiste)
or of the spiritual entities they depict (human in portraits, gods in
idols, spirits in masks). This is not to say that mimicry, emotional
contagion and understanding expressed intentions alone are a way
to impute a particular agency of the face representation, but they
would be a good way to experimentally verify the abduction (infer-
ence) of agency in these images. Mimicry should, above all, indi-
cate a potential for agency in the face representation. If we mimic
and share emotions with face representations, this means that we at-
tribute a certain mental state to them, and this mental state partici-
pate, among other things, to their agency. This does not exclude the
fact that other processes, such as imaginative thoughts, could foster
the abduction of agency in face representation.
Our study provides a new track to the issue of social interaction
with face representations. It is the first to link mimicry and its cor-
related processes (emotional contagion and understanding other’s
mental states) to the theory of art agency. It brings some support
to the art agency in the case of face representations. Social agen-
cies, such as intentions, might be no more just abstract concepts,
but could find a real correlate in cognitive processes. Conse-
quently, our findings open the door to reveal abstract concepts by
processes that are familiar to experimental psychology such as
embodied emotion processes.
Conclusion
In summary, our study showed that facial expressions in actual
human faces are not the only ones that cause facial mimicry. Face
representations in art might also stimulate facial mimicry, at least
in zygomaticus and depressor muscles (mouth surrounding muscles).
The observation of mimicry only in the lower part of the face might
be related to the participants’culture. A cross-culturally bigger sam-
ple of participants would make it possible to examine the effect of
culture on mimicry. Besides, we also have shown that facial expres-
sions might trigger emotional contagion, at least as arousal affect.
Basedonthisfinding and on the literature, we have postulated that
emotional contagion and understanding others’mental states proc-
esses might be induced by mouth expression representations through
mimicry. As put by Stamatopoulou (2018) about any expressive
dynamic stimulus, face representations in art might function as an
affective primer that could elicit embodied sense-impressions in the
perceiver. This hypothesis might be, among others such as
imaginative thoughts, a support to Gell’s anthropological theory of
art (Gell, 1998), and this, from a cognitive psychology perspective.
Further studies are needed to test these hypotheses on emotional con-
tagion and understanding others’mental states. Our analyses did not
allow us to bring to light mimicry in corrugator or frontalis.How-
ever, our results reflected an attenuated tendency to mimic stimuli
depicting expressions associated with a strong corrugator activity,
due to a hedonistic property of highly expressive face representations.
Our results also reflected a possible appraisal-based reaction in fron-
talis that could be explained by the realism of stimuli and the valence
they convey. Our analyses did not highlight mimicry toward realistic
and nonrealistic images separately, in particular in zygomaticus and
depressor. A larger sample of stimuli in both categories of realism
would allow us to further examine the mimicry of these muscles in
the two categories separately. In conclusion, mimicry, in the lower
part of the face, may play a role in the interaction with a wide range
of face representations in art. Thus, this study could serve to enrich
the facial expression and mimicry research field such as contributing
to the brain-machine interface project using emotional agents (Pillette
et al., 2017). More specifically, as an interdisciplinary study, this
work could also contribute to understanding art interactions with face
representations usually explained by anthropological and art theories.
References
Achour-Benallegue, A., Pelletier, J., & Kaminski, G. (2016). Aesthetic
impact of anthropomorphic figures in art: The case of facial expressions.
In Z. Kapoula & M. Vernet (Eds.), Aesthetics and neuroscience (pp.
55–80). Springer. https://doi.org/10.1007/978-3-319-46233-2_5
Adolphs, R. (2002). Recognizing emotion from facial expressions: Psychologi-
cal and neurological mechanisms. Behavioral and Cognitive Neuroscience
Reviews,1(1), 21–62. https://doi.org/10.1177/1534582302001001003
Awasthi, A., & Mandal, M. K. (2015). Facial expressions of emotions:
Research perspectives. In M. K. Mandal & A. Awasthi (Eds.), Under-
standing facial expressions in communication (pp. 1–18). Springer.
Baetens, K., Ma, N., Steen, J., & Van Overwalle, F. (2014). Involvement
of the mentalizing network in social and non-social high construal.
Social Cognitive and Affective Neuroscience,9(6), 817–824. https://doi
.org/10.1093/scan/nst048
Bailey, P. E., & Henry, J. D. (2009). Subconscious facial expression mim-
icry is preserved in older adulthood. Psychology and Aging,24(4),
995–1000. https://doi.org/10.1037/a0015789
Barrett, L. F., & Lindquist, K. A. (2008). The embodiment of emotion. In
G. R. Semin & E. R. Smith (Eds.), Embodied grounding: social, cogni-
tive, affective, and neuroscientific approaches (pp. 237–262). Cambridge
University Press. https://doi.org/10.1017/CBO9780511805837.011
Beall, P. M., Moody, E. J., McIntosh, D. N., Hepburn, S. L., & Reed, C. L.
(2008). Rapid facial reactions to emotional facial expressions in typi-
cally developing children and children with autism spectrum disorder.
Journal of Experimental Child Psychology,101(3), 206–223. https://doi
.org/10.1016/j.jecp.2008.04.004
Blakemore, S.-J., & Decety, J. (2001). From the perception of action to the
understanding of intention. Nature Reviews Neuroscience,2(8),
561–567. https://doi.org/10.1038/35086023
Bouisset, S., & Maton, B. (1995). Muscles, posture et mouvement. Bases et
applications de la méthode Électromyographique [Muscles, posture and
movement. Basics and applications of the electromyographic method]
(Vol. 26). Hermann.
Caldara, R. (2017). Culture reveals a flexible system for face processing.
Current Directions in Psychological Science,26(3), 249–255. https://doi
.org/10.1177/0963721417710036
FACIAL REACTIONS TO FACE REPRESENTATIONS IN ART 11
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
803

Chen, C., Garrod, O. G., Zhan, J., Beskow, J., Schyns, P. G., & Jack, R. E.
(2018). Reverse engineering psychologically valid facial expressions of
emotion into social robots. In 2018 13th IEEE international conference on
automatic face & gesture recognition (FG 2018) (pp. 448–452). https://
doi.org/10.1109/FG.2018.00072
Churches, O., Nicholls, M., Thiessen, M., Kohler, M., & Keage, H. (2014).
Emoticons in mind: An event-related potential study. Social Neuro-
science,9(2), 196–202. https://doi.org/10.1080/17470919.2013.873737
Damasio, A. R. (1994). Descartes’error: Emotion, reason and the human
brain. Grosset/Putnam.
Delorme, A., & Makeig, S. (2004). EEGLAB: An open source toolbox for
analysis of single-trial EEG dynamics including independent component
analysis. Journal of Neuroscience Methods,134(1), 9–21. https://doi
.org/10.1016/j.jneumeth.2003.10.009
Dimberg, U., Andréasson, P., & Thunberg, M. (2011). Emotional empathy
and facial reactions to facial expressions. Journal of Psychophysiology,
25(1), 26–31. https://doi.org/10.1027/0269-8803/a000029
Dimberg, U., & Petterson, M. (2000). Facial reactions to happy and angry
facial expressions: Evidence for right hemisphere dominance. Psycho-
physiology,37(5), 693–696. https://doi.org/10.1111/1469-8986.3750693
Dimberg, U., & Thunberg, M. (2012). Empathy, emotional contagion, and
rapid facial reactions to angry and happy facial expressions. PsyCh
Journal,1(2), 118–127. https://doi.org/10.1002/pchj.4
Dimberg, U., Thunberg, M., & Elmehed, K. (2000). Unconscious facial
reactions to emotional facial expressions. Psychological Science,11(1),
86–89. https://doi.org/10.1111/1467-9280.00221
Eimer, M. (2011). The face-sensitive N170 component of the event-related
brain potential. In A. Calder, G. Rhodes, M. Johnson, & J. Haxby
(Eds.), The Oxford handbook of face perception (Vol. 28, pp. 329–344).
Oxford University Press. https://doi.org/10.1093/oxfordhb/97801995
59053.013.0017
Ekman, P., & Cordaro, D. (2011). What is meant by calling emotions ba-
sic. Emotion Review,3(4), 364–370. https://doi.org/10.1177/175407391
1410740
Finzi, E., & Rosenthal, N. E. (2014). Treatment of depression with onabo-
tulinumtoxinA: A randomized, double-blind, placebo controlled trial.
Journal of Psychiatric Research,52,1–6. https://doi.org/10.1016/j
.jpsychires.2013.11.006
Freedberg, D. (2009). Movement, embodiment, emotion. Les Actes de Col-
loques du Musée du Quai Branly Jacques Chirac,1,1–11. https://doi
.org/10.4000/actesbranly.330
Freedberg, D., & Gallese, V. (2007). Motion, emotion and empathy in es-
thetic experience. Trends in Cognitive Sciences,11(5), 197–203. https://
doi.org/10.1016/j.tics.2007.02.003
Fujimura, T., Sato, W., & Suzuki, N. (2010). Facial expression arousal
level modulates facial mimicry. International Journal of Psychophysiol-
ogy,76(2), 88–92. https://doi.org/10.1016/j.ijpsycho.2010.02.008
Gallese, V. (2019). Embodied simulation. Its bearing on aesthetic experi-
ence and the dialogue between neuroscience and the humanities. Gestalt
Theory,41(2), 113–127. https://doi.org/10.2478/gth-2019-0013
Gell, A. (1998). Art and agency: An anthropological theory. Oxford Uni-
versity Press.
Gendron, M., Roberson, D., van der Vyver, J. M., & Barrett, L. F. (2014).
Perceptions of emotion from facial expressions are not culturally univer-
sal: Evidence from a remote culture. Emotion,14(2), 251–262. https://
doi.org/10.1037/a0036052
Gerger, G., Leder, H., & Kremer, A. (2014). Context effects on emotional
and aesthetic evaluations of artworks and IAPS pictures. Acta Psycho-
logica,151, 174–183. https://doi.org/10.1016/j.actpsy.2014.06.008
Gerger, G., Leder, H., Tinio, P. P., & Schacht, A. (2011). Faces versus pat-
terns: Exploring aesthetic reactions using facial EMG. Psychology of
Aesthetics, Creativity, and the Arts,5(3), 241–250. https://doi.org/10
.1037/a0024154
Gernot, G., Pelowski, M., & Leder, H. (2018). Empathy, Einfühlung, and
aesthetic experience: The effect of emotion contagion on appreciation of
representational and abstract art using fEMG and SCR. Cognitive Proc-
essing,19(2), 147–165. https://doi.org/10.1007/s10339-017-0800-2
Grèzes, J., Philip, L., Chadwick, M., Dezecache, G., Soussignan, R., &
Conty, L. (2013). Self-relevance appraisal influences facial reactions to
emotional body expressions. PLoS ONE,8(2), e55885. https://doi.org/10
.1371/journal.pone.0055885
Guédron, M. (2011). L’art de la grimace: Cinq siècles d’excès de visage
[The art of the grimace: Five centuries of facial excess]. Hazan.
Hadjikhani, N., Kveraga, K., Naik, P., & Ahlfors, S. P. (2009). Early
(M170) activation of face-specific cortex by face-like objects. Neurore-
port,20(4), 403–407. https://doi.org/10.1097/WNR.0b013e328325a8e1
Hermens, H. J., Freriks, B., Disselhorst-Klug, C., & Rau, G. (2000). Devel-
opment of recommendations for SEMG sensors and sensor placement
procedures. Journal of Electromyography and Kinesiology,10(5),
361–374. https://doi.org/10.1016/S1050-6411(00)00027-4
Hess, U., & Fischer, A. (2013). Emotional mimicry as social regulation.
Personality and Social Psychology Review,17(2), 142–157. https://doi
.org/10.1177/1088868312472607
Hess, U., & Fischer, A. (2014). Emotional mimicry: Why and when we
mimic emotions. Social and Personality Psychology Compass,8(2),
45–57. https://doi.org/10.1111/spc3.12083
Korb, S., Grandjean, D., & Scherer, K. R. (2010). Timing and voluntary
suppression of facial mimicry to smiling faces in a Go/NoGo task—An
EMG study. Biological Psychology,85(2), 347–349. https://doi.org/10
.1016/j.biopsycho.2010.07.012
Künecke, J., Hildebrandt, A., Recio, G., Sommer, W., & Wilhelm, O.
(2014). Facial EMG responses to emotional expressions are related to
emotion perception ability. PLoS ONE,9(1), e84053. https://doi.org/10
.1371/journal.pone.0084053
Kwon, M., & Christou, E. A. (2018). Visual information processing in
older adults: Reaction time and motor unit pool modulation. Journal of
Neurophysiology,120(5), 2630–2639. https://doi.org/10.1152/jn.00161
.2018
Lajante, M. M., Droulers, O., & Amarantini, D. (2017). How reliable are
“state-of-the-art”facial EMG processing methods?: Guidelines for
improving the assessment of emotional valence in advertising research.
Journal of Advertising Research,57(1), 28–37. https://doi.org/10.2501/
JAR-2017-011
Leder, H., Gerger, G., Brieber, D., & Schwarz, N. (2014). What makes an
art expert? Emotion and evaluation in art appreciation. Cognition and
Emotion,28(6), 1137–1147. https://doi.org/10.1080/02699931.2013
.870132
Likowski, K. U., Mühlberger, A., Gerdes, A. B. M., Wieser, M. J., Pauli,
P., & Weyers, P. (2012). Facial mimicry and the mirror neuron system:
Simultaneous acquisition of facial electromyography and functional
magnetic resonance imaging. Frontiers in Human Neuroscience,6, 214.
https://doi.org/10.3389/fnhum.2012.00214
Liu, T., Mu, S., He, H., Zhang, L., Fan, C., Ren, J., Zhang, M., He, W., &
Luo, W. (2016). The N170 component is sensitive to face-like stimuli:
A study of Chinese Peking opera makeup. Cognitive Neurodynamics,
10(6), 535–541. https://doi.org/10.1007/s11571-016-9399-8
McIntosh, D. N. (2006). Spontaneous facial mimicry, liking and emotional
contagion. Polish Psychological Bulletin,37(1), 31–42.
Niedenthal, P. M. (1990). Implicit perception of affective information.
Journal of Experimental Social Psychology,26(6), 505–527. https://doi
.org/10.1016/0022-1031(90)90053-O
Niedenthal, P. M. (2007). Embodying emotion. Science,316(5827),
1002–1005. https://doi.org/10.1126/science.1136930
Niedenthal, P. M., & Brauer, M. (2012). Social functionality of human
emotion. Annual Review of Psychology,63, 259–285. https://doi.org/10
.1146/annurev.psych.121208.131605
12 ACHOUR-BENALLEGUE ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
804

Niedenthal, P. M., & Maringer, M. (2009). Embodied emotion considered.
Emotion Review,1(2), 122–128. https://doi.org/10.1177/1754073908
100437
Niedenthal, P. M., Rychlowska, M., & Szarota, P. (2013). Embodied simu-
lation and the human smile: Processing similarities to cultural differen-
ces. In D. Cervone, M. Fajkowska, M. W. Eysenck, & T. Maruszewski
(Eds.), Personality dynamics: Meaning construction, the social world,
and the embodied mind (pp. 141–158). Eliot Werner Publications.
Niedenthal, P. M., Mermillod, M., Maringer, M., & Hess, U. (2010). The
Simulation of Smiles (SIMS) model: Embodied simulation and the
meaning of facial expression. Behavioral and Brain Sciences,33(6),
417–433. https://doi.org/10.1017/S0140525X10000865
Niedenthal, P., Wood, A., Rychlowska, M., & Korb, S. (2017). Embodied
simulation in decoding facial expression. In J. M. Fernández-Dols &
J. A. Russell (Eds.), The science of facial expression (pp. 397–414).
Oxford University Press.
Philip, L., Martin, J.-C., & Clavel, C. (2018). Rapid facial reactions in
response to facial expressions of emotion displayed by real versus virtual
faces. I-Perception,9(4), 1–18. https://doi.org/10.1177/2041669518786527
Pillette, L., Jeunet, C., Mansencal, B., N’Kambou, R., N’Kaoua, B., & Lotte,
F. (2017, September). Peanut: Personalised emotional agent for neuro-
technology user-training [Paper presentation]. The 7th International BCI
Conference, Graz, Austria.
Prochazkova, E., & Kret, M. E. (2017). Connecting minds and sharing
emotions through mimicry: A neurocognitive model of emotional conta-
gion. Neuroscience and Biobehavioral Reviews,80,99–114. https://doi
.org/10.1016/j.neubiorev.2017.05.013
Reisenzein, R., Studtmann, M., & Horstmann, G. (2013). Coherence
between emotion and facial expression: Evidence from laboratory
experiments. Emotion Review,5(1), 16–23. https://doi.org/10.1177/
1754073912457228
Rymarczyk, K., Biele, C., Grabowska, A., & Majczynski, H. (2011). EMG
activity in response to static and dynamic facial expressions. Interna-
tional Journal of Psychophysiology,79(2), 330–333. https://doi.org/10
.1016/j.ijpsycho.2010.11.001
Sagiv, N., & Bentin, S. (2001). Structural encoding of human and sche-
matic faces: Holistic and part-based processes. Journal of Cognitive
Neuroscience,13(7), 937–951. https://doi.org/10.1162/0898929017
53165854
Sato, W., Fujimura, T., Kochiyama, T., & Suzuki, N. (2013). Relationships
among facial mimicry, emotional experience, and emotion recognition.
PLoS ONE,8(3), e57889. https://doi.org/10.1371/journal.pone.0057889
Sato, W., Fujimura, T., & Suzuki, N. (2008). Enhanced facial EMG activ-
ity in response to dynamic facial expressions. International Journal of
Psychophysiology,70(1), 70–74.
Schaeffer, J.-M. (2015). L’expérience esthétique [The aesthetic experi-
ence]. Editions Gallimard.
Schilbach, L. (2016). The neuroscience of mimicry during social interac-
tions. In U. Hess & A. H. Fischer (Eds.), Emotional mimicry in social
context (Vol. 72, pp. 72–89). Cambridge University Press; https://doi
.org/10.1017/CBO9781107587595.005
Seibt, B., Mühlberger, A., Likowski, K. U., & Weyers, P. (2015). Facial
mimicry in its social setting. Frontiers in Psychology,6, 1122. https://
doi.org/10.3389/fpsyg.2015.01122
Soussignan, R., Chadwick, M., Philip, L., Conty, L., Dezecache, G., &
Grèzes, J. (2013). Self-relevance appraisal of gaze direction and dynamic
facial expressions: Effects on facial electromyographic and autonomic
reactions. Emotion,13(2), 330–337. https://doi.org/10.1037/a0029892
Stamatopoulou, D. (2018). Empathy and the aesthetic: Why does art still
move us? Cognitive Processing,19(2), 169–186. https://doi.org/10
.1037/a0029892
Tateyama, H. (2016). Ritual of superiority: Tolai Tubuan performance at
the national mask festival in Papua New Guinea. Journal de la Société
Des Océanistes, 142–143,21–36. https://doi.org/10.4000/jso.7550
Urgesi, C., Mattiassi, A. D., Buiatti, T., & Marini, A. (2016). Tell it to a
child! A brain stimulation study of the role of left inferior frontal gyrus
in emotion regulation during storytelling. NeuroImage,136,26–36.
https://doi.org/10.1016/j.neuroimage.2016.05.039
Van Boxtel, A. (2010). Facial EMG as a tool for inferring affective states.
Proceedings of measuring behavior (Vol. 7, pp. 104–108). Noldus Infor-
mation Technology.
Wheatley, T., Weinberg, A., Looser, C., Moran, T., & Hajcak, G. (2011).
Mind perception: Real but not artificial faces sustain neural activity
beyond the N170/VPP. PLoS ONE,6(3), e17960. https://doi.org/10
.1371/journal.pone.0017960
Winkielman, P., Niedenthal, P., Wielgosz, J., Eelen, J., & Kavanagh, L. C.
(2015). Embodiment of cognition and emotion. In M. E. Mikulincer,
P. R. Shaver, E. E. Borgida, & J. A. Bargh (Eds.), APA handbook of per-
sonality and social psychology (Vol. 1, pp. 151–175). APA.
Wollmer, M. A., de Boer, C., Kalak, N., Beck, J., Götz, T., Schmidt, T.,
Hodzic, M., Bayer, U., Kollmann, T., Kollewe, K., Sönmez, D.,
Duntsch, K., Haug, M. D., Schedlowski, M., Hatzinger, M., Dressler,
D., Brand, S., Holsboer-Trachsler, E., & Kruger, T. H. C. (2012). Facing
depression with botulinum toxin: A randomized controlled trial. Journal
of Psychiatric Research,46(5), 574–581. https://doi.org/10.1016/j
.jpsychires.2012.01.027
Yuki, M., Maddux, W. W., & Masuda, T. (2007). Are the windows to the
soul the same in the East and West? Cultural differences in using the
eyes and mouth as cues to recognize emotions in Japan and the United
States. Journal of Experimental Social Psychology,43(2), 303–311.
https://doi.org/10.1016/j.jesp.2006.02.004[Mismatch
Received August 17, 2020
Revision received June 30, 2021
Accepted June 30, 2021 n
FACIAL REACTIONS TO FACE REPRESENTATIONS IN ART 13
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
805