Facial Mimicry and Emotional Contagion to Dynamic Emotional Facial Expressions and Their Inffuence on Decoding Accuracy

International Journal of Psychophysiology 40Ž. 2001 129᎐141

Facial mimicry and emotional contagion to dynamic emotional facial expressions and their inﬂuence on decoding accuracy

Ursula Hessa,U, Sylvie Blairy b

aDepartment of Psychology, Uni¨ersity of Quebec at Montreal, P.O. Box 8888, Station Centre-Ville, Montreal, QC, Canada H3C 3P8 bFree Uni¨ersity of Brussels, ERASME Hospital, Brussels, Belgium

Received 27 June 2000; received in revised form 26 July 2000; accepted 1 August 2000

Abstract

The present study had the goal to assess whether individuals mimic and show emotional contagion in response to relatively weak and idiosyncratic dynamic facial expressions of emotions similar to those encountered in everyday life. Furthermore, the question of whether mimicry leads to emotional contagion and in turn facilitates emotion recognition was addressed. Forty-one female participants rated a series of short video clips of stimulus persons expressing anger, sadness, disgust, and happiness regarding the emotions expressed. An unobtrusive measure of emotional contagion was taken. Evidence for mimicry was found for all types of expressions. Furthermore, evidence for emotional contagion of happiness and sadness was found. Mediational analyses could not conﬁrm any relation between mimicry and emotional contagion nor between mimicry and emotion recognition. ᮊ 2001 Elsevier Science B.V. All rights reserved.

Keywords: Facial mimicry; Emotional contagion; Emotion recognition

1. Introduction munication of affective statesw e.g. Freud, 1921, based on a theory by Lipps, 1907; Bavelas et al., It has been suggested that mimicry ᎏ the 1986x . For example, RogersŽ. 1957 saw the imita- imitation of others’ non-verbal displays by an tion of a client’s non-verbal behavior as a means observer ᎏ plays an important role in the com- to communicate empathy and some schools of therapyŽ. see, e.g. Siegel, 1995 advocate imitation as a means of understanding the client’s internal U Corresponding author. state. E-mail address: [email protected]Ž. U. Hess . Facial mimicry in this context is usually concep-

0167-8760r01r$ - see front matter ᮊ 2001 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 8 7 6 0Ž. 0 0 00161-6 130 U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 tualized as an automatic, reflex-like processŽ see, al.Ž. 1993 . Similarly, Cappella Ž. 1993 , based on e.g. Lipps, 1907; Hoffmann, 1984; Hatfield et al., evidence in favor of the facial feedback hypothe- 1993. , with the observer’s facial expression match- sisŽ. FFH in particular, proposes that facial feed- ing the observed facial expression. Emotional back from mimicry causes contagion. contagion is a closely related concept that is However, evidence is accumulating that emo- sometimes defined in overlapping termsŽ e.g. Hat- tional contagion may not be causally related to field et al., 1993. . It is therefore useful to define mimicryŽ Gump and Kulik, 1996; Blairy et al., the specific use of the two terms in the framework 1999. . Also, in their review of the literature, Hess of the present study. Specifically, we consider as et al.Ž. 1999 could not find any consistent evi- mimicry the congruent facial reactions to the dence that mimicry facilitates emotion recogni- emotional facial displays of others. That is, tion. Together, these findings throw doubt on the mimicry is defined exclusively as an expressive notion that emotion recognition is related to a component. In contrast, we define emotional con- reflex-like mimicry process via contagion. tagion as an affective state that matches the So why do people mimic at all since this process other’s emotional display. seems to not be related to either emotional con- In a recent review, Hess et al.Ž. 1999 conclude tagion or emotion recognition accuracy? Before that evidence from studies on both adults and answering this question a second look at the infants strongly suggests that, in general, people evidence reported above is necessary. First, de- adopt facial, postural, and vocal behaviors that spite the evidence for facial mimicry reported are congruent with the displays they observe, and above, it is not clear whether individuals mimic that these displays often represent mimicryŽ see the type of expressions they are likely to en- also Dimberg, 1990. . However, some examples of counter in real life. This, because evidence for counter-mimicry effectsŽ e.g. Lanzetta and Englis, mimicry in adults is largely based on studies that 1989; Hess, 1998. have also been reported. Speci- employed very intense, prototypical facial expres- fically, Lanzetta and EnglisŽ. 1989 found mimicry sions presented as still photographs. For example, in a collaborative task situation but counter- the extensive studies on mimicry and contagion in mimicry in a competitive task situation. This, and adults by Dimberg and LundqvistŽ e.g. Dimberg, evidence that mimicry may depend on the type of 1990; Lundqvist, 1995; Lundqvist and Dimberg, task the participant is engaged inŽ Hess et al., 1995. employed stimuli selected from the ‘Pic- 1998. , suggests that mimicry may not be an auto- tures of facial affect’Ž. Ekman and Friesen, 1976 , matic, reflex-like mechanism. Furthermore, a which are a set of highly recognizable and proto- number of studies suggest that individuals tend to typic facial expressions. Such stimuli may in fact report emotional states that match the facial elicit a reflex-like response due to their extremity emotion displays to which they have been exposed that is not found for less extreme expressions. Žsee, e.g. Hatfield et al., 1993; Strayer, 1993; Laird This notion is supported by the observation that et al., 1994; Schneider et al., 1994; Lundqvist and studies finding evidence for the situation depen- Dimberg, 1995. . dence of mimicry employed somewhat weaker The two processes, mimicry and emotional con- and more natural expressionsŽ Lanzetta and En- tagion, have been suggested to be causally elated. glis, 1989; Gump and Kulik, 1996; Hess et al., This idea goes back to LippsŽ. 1907 who sug- 1998.Ž. . Also, McHugo et al. 1991 and Bourgeois gested that the imitated expression leads ᎏ via a and HessŽ. 1999 using video exert of news pro- feedback process ᎏ to emotional contagion. As grams featuring politicians found that mimicry regards the influence of emotional contagion on was modulated by the political attitude of the empathyŽ the capacity to recognize the emotional observer. That is, observers were more likely to state of others.Ž. , Lipps 1907 as well as Hoffmann mimicry a politician if they shared his political Ž.1984 imply that emotional contagion should in beliefs than when not. turn facilitate emotion recognition. Related ideas In sum, studies finding clear evidence for facial have more recently been expressed by Hatfield et mimicry and emotional contagion tend to employ U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 131 prototypical, high intensity, still photographs as Specifically, participants were asked to decode stimulus material, whereas those studies that a series of video clips of emotional expressions of found evidence of situational influences on happiness, anger, sadness, and disgust. These fa- mimicry employed more naturalistic, less proto- cial expressions were recorded during an emotio- typical or weaker, facial stimuli. Thus, the ques- nal imagery task and represent the spontaneous, tion of whether individuals mimic the type of idiosyncratic expressions of the expressor. Partici- expressions they are likely to encounter in every- pants’ own facial expressions were measured us- day life deserves further investigation. ing facial electromyography and their emotional As regards the lack of evidence for the facilita- state was assessed using an unobtrusive self- tion of emotion recognition by mimicry, it is report measure. Mediational analysesŽ Baron and possible that the use of prototypical stimulus ma- Kenny, 1986. were used to assess the influence of terial may not suffice to uncover subtle improve- mimicry on emotional contagion and on emotion ments in decoding accuracy due to mimicry. recognition. Specifically, the process described by LippsŽ. 1907 is relatively elaborate and demands both a certain empathic ability and introspection. Yet, emotion 2. Method displays can be decoded by using other sources of information. The sender’s emotion displays, the 2.1. Participants facial, vocal, postural, etc., expressions emitted by the sender can be used to draw inferences regard- Forty-one female volunteers with a mean age ing the presumed emotional state of the sender of 24.3 yearsŽ. S.D.s8.6 from the University of using a pattern-matching approachŽ e.g. Buck, Quebec at Montreal participated individually. 1984. . For example, the presence of upturned Participants were recruited on campus and via corners of the mouth and of wrinkles around the announcements in classrooms. Facial EMG data eyes can be interpreted as signaling happiness for some episodes had to be deleted for two whereas eyebrows drawn together in a frown may participants due to movement artifacts. signal anger. This process should be especially useful for the decoding of the prototypical, highly 2.2. Dependent measures recognizable expressions often used in studies on mimicry and contagion. In those circumstances, 2.2.1. Facial EMG the additional information provided by the process Facial EMG was measured on the left side of described by Lipps may not in fact add to the the face. Electrode placements were chosen ac- already high level of decoding accuracy. cording to Fridlund and CacioppoŽ. 1986 . Activity Thus, facial mimicry may be expected to have a of the Orbicularis oculi, and Zygomaticus major facilitative effect on emotion recognition mainly was employed to assess smiling, activity of the in situations where the emotion displays are rela- Corrugator supercilii Ž.eyebrow was employed to tively weak and non-prototypical and where the assess frowning, and activity of the Leätor labii participants do not know the sender and have no alesque nasii was employed to assess the ‘sneer’ previous attitude towards the sender. Finally, it found in disgust expressions. Muscle activity was may also be that contagion, but not mimicry facil- measured using bipolar placements of Med. Asso- itates emotion recognition. ciates Inc. AgrAgCl miniature surface electrodes The present study is aimed to address these with Med Associates Inc. electrolyte gelŽ. TD41 . issues. Given the preceding considerations, it was The skin was cleansed with PDI disposable elec- considered important to use expressions that were trode prep padsŽ. 70% alcohol and pumice . A not preselected according to their correspon- Contact Precision Instruments system with a 60- dence to an emotion stereotype. Furthermore, to Hz notch filter was used to amplify the raw EMG enhance ecological validity dynamic facial expres- signals, which were integrated with 200-ms time sions were chosen. constant. The smoothed EMG signal was sampled 132 U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 at 10 Hz and stored to disk. In addition, skin being. The questionnaire is an adaptation of a conductance level and interbeat interval were questionnaire developed by Philippot et al.Ž. 1994 measured. The results pertaining to these mea- describing sensations that may be experienced sures will not be reported in the framework of during such a task and contains 16 scales describ- this article as these measures were included to ing a variety of physical sensationsŽ e.g. butterflies bolster the notion that we are interested in the in the stomach, feeling cold. as well as the items participants physical reactions to the stimuliŽ see relevant to the assessment of emotional conta- below. . Furthermore, due to equipment failure, gion:´´ enerveragressifŽ. irritatedraggressive , activity of the Zygomaticus major could not be tristerdeprime´´Ž. sadrdepressed , gairen- assessed for most subjects. The implications of thousiasteŽ. cheerful , repulse´ Ž repulsed . . The this will be discussed below. labels were chosen to correspond to the emotion displays serving as stimulus material without us- 2.2.2. Ratings ing the same terms as those employed for the Following the presentation of each stimulus rating scales. The correspondence of the two sets sequence, subjects were asked to assess the emo- of emotion labels was assessed in a separate study tions portrayed as well as their intensities using where participants were asked to indicate for an emotion profile. For this, subjects indicated each of a number of emotion terms to which the degree to which the expression reflected each ‘basic’ emotion they refer. The terms used for the of the following seven emotions: joieŽ. happiness , contagion scale were recognized as referring to tristesseŽ. sadness , peur Ž.Ž. fear , colere` anger , the target emotion by at least 90% of the respon- degout´ˆŽ. disgust , surprise Ž surprise . , and mepris ´ dents with the exception of ‘enerve’´ which was Ž.contempt . Participants clicked a point along each confused with fear. The scales were continuous emotion scale indicating the intensity with which and anchored by ‘not at all’ at one extremity and the face reflected that specific emotion. The scales ‘strongly’ at the other. Presenting these scales were represented by a 200-pixel long, bounded together with the physical sensations allowed to rectangle on the screen, the pixels were graded in reduce the possibility that participants became color from light gray to dark gray, with the darker aware that we were interested in their own emo- end of the scale indicating greater intensity of the tional states. This questionnaire has been exten- emotion. Each scale contained an emotion label sively used in paradigms related to the one em- and was anchored with the verbal labels ‘not at ployed in the present study. In the past, partici- all’ and ‘very intensely.’ Furthermore, participants have not reported suspicions regarding the pants were asked to indicate how difficult the task link between the emotion terms in the scale and was. the judgment task during the debriefing sessions where they are explicitly encouraged to speculate 2.2.3. Decoding accuracy about the purpose of the ‘well-being’ scale. Par- Decoding accuracy was defined as the observers’ ticipants typically report that the scale serves to ability to correctly infer the imagined emotion. assess stress. The notion that we are mainly An expression was considered as accurately iden- concerned with the participants’ physical reac- tified when the emotion receiving the highest tions to the task is further sustained by the heart intensity rating on the emotion profile corre- rate and skin conductance measures taken. sponded to the target emotion. An accurately identified expression received a score of 1 and a 2.3. Stimuli misidentified expression received a score of 0. Stimuli were recorded during an emotional 2.2.4. Self-reported emotional state imagery task. The task consisted of either imagin- For each type of emotion expression partici- ing the events described in an emotion script or pants were asked once, following the rating task, thinking about an emotion-inducing event. Fol- to fill out a questionnaire regarding their well- lowing each imagery period, self-reports of the U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 133 participants’ emotional state during the imagery more, AU 17Ž. pushing up of the chin and AU15 period were obtained. occurred once each. For the present study, we pre-selected video The expressions were presented in one of two recordings from those imagery periods for which different random orders using an Apple Macin- the participants described having felt the target tosh Quadra 840AV with a 14-inch color monitor emotion. From this pool of sequences we selected at a rate of 30 framesrs. those that contained at least one visible expressive displayŽ as assessed by an experienced FACS 2.4. Procedure coder, see below.Ž . For each emotion happiness, anger, disgust, and sadness. one sequence from two male and two female stimulus persons was The experimenter explained to the participants selected from this pool of expressions. For the that their task would be to judge the emotionŽ. s experiment sequences were edited to a length of portrayed by a series of stimulus persons. They 15 s. The resulting set of 16 episodes was digitized were informed that during the experiment physio- and converted to black and white. As mentioned logical measures would be taken. To reduce the above, we were interested in using the idiosyn- possibility that participants were aware that we cratic expressive displays of the stimulus person. were interested in their facial expressions we Thus, expressive displays for the different emo- employed a cover story suggesting that the experi- tion conditions were expected to vary across sti- ment was concerned with changes in skin temper- muli. In order to be able to establish a criterion ature and that the transducers affixed to their for the presence of mimicry we coded the actual face were intended for this purpose. Further- expressive displays of the stimulus persons using more, participants were informed that they would the Facial Action Coding systemŽ FACS, Ekman be filmed during the experiment. To avoid that and Friesen, 1978. . The expressions were coded participants focus on the camera during the ex- by the first author who is an experienced FACS periment, the video camera was hidden. Partici- coder. pants who signed a consent form repeating this The FACS data revealed that AU4, describing information were seated in a comfortable arm- the drawing together and lowering of the brows, a chair and the electrodes were attached. In order facial movement involving the activity of Corruga- to familiarize the participants with the rating task tor supercilii, was present for all anger displays. they were asked to complete two practice trials Furthermore, three of the four anger displays during which the experimenter answered ques- contained AU23 describing the pressing together tions regarding the procedure. of the lips. Sadness displays also involved AU4. In For each trial, the participant first saw a video addition, AU1Ž describing the drawing together of sequence showing the stimulus person’s neutral the inner corner of the eyebrows. and AU15 facial expression and then a video sequence Ždescribing the pulling down of the corner of the showing the emotional facial expression. Each lips. occurred once each. sequence was approximately 15 s long. There was All four happiness expressions contained AU12 a 5-s interval between the neutral and the expres- Ždescribing the pulling up of the corner of the sive sequence. Sequences were shown at a rate of lips, an activity involving Zygomaticus major. ac- 30 framesrsŽ. standard NTSC video rate . Fol- companied by AU6Ž describing the movement lowing this, the rating scales were presented, once that creates the wrinkles around the eyes and the participant had finished the ratings the next involves Orbicularis oculi activity. . The four dis- set of sequences was shown. The stimuli were gust displays showed little similarity. However, presented in one of two different random orders. AU12 occurred as part of three of the four dis- The well-being questionnaire was presented fol- plays. AU10Ž lifting of the upper lip as in a sneer, lowing the rating task. an activity involving Leätor labii alesque nasii At the end of the experiment, participants were actiïty. occurred twice as did AU23. Further- interviewed regarding their hypotheses. They were 134 U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 then fully debriefed and any remaining questions mimicry should be indexed by significantly higher were answered by the experimenter. levels of Corrugator supercilii than Orbicularis oculi and Leätor LAN activity. However, Leätor LAN 2.5. Artifact control and data reduction activity should be significantly lower for sadness than for anger mimicry. Mimicry to happy dis- The video records for all participants were in- plays should be indexed by significantly higher spected for movements that could disrupt the levels of Orbicularis oculi than Corrugator super- psychophysiological measures. Using a visual edit- cilii and Leätor LAN activity. For disgust dis- ing computer program PHYSIO3Ž. Banse, 1995 , plays, Corrugator supercilii activity should be sig- periods corresponding to such movements were nificantly lower than Orbicularis oculi and Leätor set missing and excluded from further analyses. LAN activity. The means for participants’ facial For the purpose of the following analyses the EMG activity are shown in Fig. 1. Note that the periods during which the subjects saw the neutral data was transformed to z-scores and that the face of the stimulus person served as baselines. zero level refers thus to the mean across all data This period was chosen to exclude that reactions points and not to an absence of difference from to the stimulus persons are confounded with reac- baseline. tions to the emotional facial expressions. One way Significant main effects of muscle site, indicat- analyses of variance using a multivariate ap- ing a differentiated pattern of muscle activity proach were employed to assess whether facial across sites, emerged for participants’ facial activ- s reactions to the neutral faces differed across ity during the decoding of anger, F2,37 7.29, s s - emotions. The results were non-significant for all P 0.002, sadness, F2,37 11.85, P 0.001, and - s - three muscle sites Ž.F 1 . Difference scores were happy episodes, F2,37 36.86, P 0.001, but not therefore calculated using the neutral face base- for disgust episodes. Furthermore, the contrasts s s line. The difference scores were averaged across specified for the anger, F1,38 9.91, P 0.003, s - the four episodes for each emotion. For the sadness, F1,38 21.99, P 0.001, and happy s - ANOVAs, EMG data was transformed into episodes F1,38 45.41, P 0.001, were also sig- within-subjects z-scores prior to averaging. nificant. The test comparing Leätor LAN activity for anger and sadness episodes was marginally significant in the predicted direction, tsy1.70, 3. Results Ps 0.098. Thus, clear evidence emerged for mimicry during the decoding of anger, sadness, 3.1. Facial mimicry and happiness displays.

The presence of facial mimicry implies a pat- 3.2. Emotional contagion tern of facial activity in response to the emotional display of others. To assess facial mimicry, we To assess emotional contagion, planned con- therefore verified first whether a distinct pattern trasts were conducted on the relevant scales of of facial activity emerged in response to the sti- the well-being questionnaire. As we can not as- muli. For this, the standardized difference scores sume that the different emotion scales are scaled were analyzed using a one-way repeated measures identically, the ratings were first z-transformed. ANOVA across muscle sites. This analysis across Since use of z-scores obliterates information re- muscle sites is allowable as facial EMG data were garding the grand mean for the scales, a separate previously transformed into z-scores and the data analysis was conducted on the grand means. are thus on the same scale. We then used planned Across all conditions, participants reported to be contrasts to assess whether the facial activity con- quite cheerfulŽ. means24.46, S.D.s31.55 as well formed to the expected patterns specified as a as somewhat irritatedŽ mean s 17.13, S.D.s function of the mimicry indices described above. 17.38. . This emotional state is quite consistent Specifically, for both anger and sadness, with the experimental situation, which was calm U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 135

Fig. 1. Observers’ facial EMG as a function of the emotional facial expression of the sender. and relaxed but presented the participants with sadnessrdepression, respectively. During the de- occasional frustrations when they had difficulty coding of anger expressions the pattern of the with the decoding task. Importantly, this implies means did not conform to the planned contrast as that the emotional contagion effects reported repulsed emerged as the highest rated emotion, below modulated the participants’ affective state and participants reported feeling significantly s s rather than changing it. more repulsed than cheerful, F1,40 7.14, P The planned contrasts assessed whether partic- 0.011. Yet it is interesting to note, that the French ipants reported feeling more cheerful when de- word ‘repulse’´ connotes a desire to get away from coding expressions of happiness, more repulsed something. Thus, reporting this feeling in re- when decoding expressions of disgust, more sponse to an anger expression, while not indica- sadrdepressed when decoding expressions of sad- tive of empathy, represents an adequate emotio- Ž ness, and more irritatedraggressive when decod- nal reaction to a threat display see, e.g. Dimberg ¨ . ing expressions of anger. The means are shown in and Ohman, 1996 . No difference emerged for Fig. 2, note that the zero level of the z-scored the decoding of disgust expressions. In sum, the pattern of results is consistent with contagion of data refers to the mean across all data points and happiness and sadness. that negative scores indicate levels lower than the mean and not the use of a bimodal scale. 3.3. Decoding accuracy Significant contrasts in the expected direction emerged for self-reported affect during the de- Decoding accuracy was 45% for anger expres- s s coding of happy, F1,40 8.74, P 0.005 and sad sions, 87% for happiness expressions, 43% for s s faces, F1,40 4.45, P 0.041, for which partici- disgust expressions, and 75% for sadness expres- pants reported significantly more cheerfulness and sions. 136 U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141

Fig. 2. Observers’ self-reported emotional state as a function of the emotional facial expression of the sender.

3.4. Mediational analyses would be confirmed if facial activity predicts self- reported emotional state and decoding accuracy, The notion that mimicry leads to emotional whereas self-reported emotional state predicts contagion which in turn positively influences de- decoding accuracy after facial activity is con- coding accuracy, was assessed using mediational trolled for. Table 1 shows the beta values for the analyses as proposed by Baron and KennyŽ. 1986 . paths. For each emotion expression the self- For this, decoding accuracy was first regressed on report on the corresponding scale was employed facial activity as assessed by facial EMGŽ. path a . as well as EMG data from all muscle sites. For Second, self-reported emotional feeling state was this analysis the untransformed difference scores regressed on facial activityŽ. path b . Finally de- were employed. coding accuracy was regressed on both self- Inspection of Table 1 does not provide evi- 1 reported feeling state and facial activityŽ. path c . dence for a mediation of decoding accuracy The mediational pathway proposed by Lipps through contagion from mimicry. Considering first the two analyses for the expressions for which evidence for both mimicry and contagion emerged 1As self-reported emotional feeling state was assessed only in the previous analyses, that is, expressions of once for each emotion, the corresponding data for decoding happiness and sadness. For the decoding of hap- accuracyŽ. 0 or 1 and facial EMG for the specific expressions for which the participant’s emotional state had been assessed piness expressions, we note no evidence that ei- were entered in the equation. ther higher levels of activity of Orbicularis oculi U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 137

or lower levels of activity of Corrugator supercilii note that lower levels of Orbicularis oculi Žwhich are associated with either self-reports of cheerful- is congruent with, but certainly not indicative of a ness or decoding accuracy. Furthermore, no sad expression. are related to self-reported sad- evidence for a link between cheerfulness and ness and to decoding accuracy; however, no direct accuracy in decoding happy expressions emerges. link between emotional state and decoding accu- Regarding the decoding of sad expressions, we racy emerged once facial activity was controlled 138 U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 for. Similarly, lower levels of Leätor LAN are was at times somewhat difficult. However, partici- associated with an increase in decoding accuracy. pants also reported higher levels of cheerfulness At best, this pattern suggests that lower levels of when decoding expressions of happiness, higher facial movements incongruent with mimicry are levels of sadness when decoding expressions of related to an increase in decoding accuracy but sadness, and higher levels of anger when decod- no mediation through emotional contagion seems ing expressions of anger. to be involved. Thus, the present study presents evidence for The pattern for the decoding of anger is even mimicry and contagion in situations where rela- less supportive of Lipps’ model. Here, the pres- tively realistic, low intensity, idiosyncratic emotio- ence of movements that are incongruent with nal facial expressions served as stimulus material. mimicry, specifically, Orbicularis oculi and Leätor This finding corroborates the contention by Hess LAN activity is positively related to self-reports of et al.Ž. 1998 that people mimic facial displays in irritationraggressiveness. It is possible to specu- situations where they are interested in the emo- late, based on the lower decoding accuracy for tional state of the expressorŽ as was the case for this expressionsŽ. 43% , that anger expressions the participants of this study, who were explicitly were particularly difficult to decode and that par- asked to rate this state based on the emotional ticipants therefore showed both some grimacing expressions. . and reported more irritationraggressiveness. Yet, this finding is clearly not in support of Lipps’ 4.1. Lipps’ model model. For the decoding of disgust expressions no evidence for either mimicry or contagion emerged. No evidence for a link between mimicry and We thus can not expect mediation through these contagion was found. Furthermore, mimicry was processes. Low levels of both Currugator supercilii not systematically related, neither directly, nor and Orbicularis oculi activity were found to be indirectly via contagion, to decoding accuracy. related to higher decoding accuracy, suggesting Some evidence emerged that lower levels of facial that lower levels of facial movements that are movements that are incongruent with mimicry incongruent with mimicry are related to an in- can be related to an increase in decoding accu- crease in decoding accuracy, but again no evi- racy, but no evidence for mediation via contagion dence for mediation via contagion emerges. emerged in these cases. Evidence for a positive Furthermore, Leätor LAN activity is positively link between facial activity congruent with related to self-reported feelings of repulsedness. mimicry and self-reported emotional contagion emerged only for the decoding of disgust expressions, for which, however, no evidence for either 4. Discussion mimicry or contagion on a group level was found. A number of positive relationships between facial The present study presents evidence that par- movements not congruent with mimicry and self- ticipants mimicked dynamic facial expressions of reported contagion emerged for the decoding of happiness, anger, and sadness. Furthermore, evi- happiness, anger, and disgust expressions, sug- dence for emotional contagion was found for gesting an unspecific link between facial activity happiness, sadness, and, tentatively, for anger. It per se and emotional contagion per se. is important to note that the emotional contagion Thus, no evidence for the process proposed by effects modulate the affective experience rather Lipps was found. This finding confirms previous than shape it. Specifically, participants reported negative findings regarding the relationship generally moderate levels of both cheerfulness between mimicry and emotion recognition accu- and irritationraggression ᎏ a self-report that is racyŽ. e.g. Blairy et al., 1999 . It can be argued that congruent with the participants general positive the failure to find a relationship between mimicry attitude towards the task, which was reflected in and decoding accuracy in studies using highly debriefing interviews, and the fact that the task prototypical stimulus material was due to the fact U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 139 that participants based their judgments on pat- Yet, even though the measurement of the activity tern matchingŽ. Buck, 1984 rather than on the muscle was deemed important, we feel that its more elaborate empathic process suggested by absence does not present a serious impediment to Lipps. However, this argument is less valid in the the interpretation of the results as it has been present case, where the expressions employed shown that crow’s feet wrinkles around the eye were much closer to real life expressions than reliably accompany genuine happinessŽ e.g. those typically employed in research on mimicry Duchenne, 1862r1990. and can thus be employed in so far as they were relatively weak and non- as a marker for a happy expression and AU6Ž the prototypical. The use of video stimuli as opposed movement due to Orbicularis oculi activity that to stills further enhanced ecological validity. causes crow’s feet wrinkles around the eyes. was present in the stimulus expressions. Thus, mimicry 4.2. Limitations of the study of happy expressions should be indexed by Orbic- ularis oculi activity as well as by Zygomaticus One concern regarding the present study is the major activity. relatively low level of decoding accuracy for anger and disgust displays. It may be argued that the 4.3. Summary decoding task was too difficult to allow the detec- tion of facilitative effects of mimicry and con- The present study demonstrates mimicry and tagion. However, mimicry, and tentatively, emotional contagion effects to dynamic, idiosyn- contagion was observed for anger expressions. cratic, relatively weak emotional expressions. Furthermore, for happy expressions, which were However, these two processes were not found to very well recognized, no evidence for a facilitation be related to each other. Nor was either mimicry of decoding accuracy due to mimicry and conta- or contagion related to decoding accuracy. Thus, gion was found. Thus, decoding difficulty alone mimicry does not seem to serve to facilitate emo- should not have been an issue. tion communication nor does it seem to be a Furthermore, the value of the stimulus material symptom or ‘side-effect’ of emotional contagion. lies in its relative ecological validity. The stimulus This raises the question of why people mimic? material was chosen from a set of expressions It has been suggestedŽ Bavelas et al., 1986, obtained during a study involving emotional 1988. that mimicry serves to communicate empa- imagery and the selection criterion was that the thy and Hess et al.Ž. 1999 conclude in their review expressor reported feeling the target emotion that evidence for this effect is among the clearest during imagery and not the prototypicality of the in the literature on mimicry. That is, observers’ expression. Thus, these expressions are closer to facial mimicry informs the expressor of their ‘un- those that participants may encounter in real life. derstanding’ of the expressor’s emotional state. However, the use such idiosyncratic expressions However, the present data does not support the implies a relatively lower decoding accuracy when notion that this communication could be accurate compared to highly prototypical material or posed since mimicry was not systematically related to expressions in generalŽ see also Motley and Cam- either recognition accuracy nor emotional conta- den, 1988. . Finally, and most importantly, the gion. mode of emotion recognition described by Lipps Furthermore, it has been suggested that emo- should be most useful in situations where the tional contagion may be mediated by a social decoding task is not trivially simple. comparison process. Specifically, Gump and Kulik A second possible problem is posed by the Ž.1997 suggest, based on the extension of social technical difficulties that rendered the Zygomati- comparison theory of SchachterŽ. 1959 , that emo- cus major data unusable. Zygomaticus major is tional contagion may be one of the processes by the muscle that pulls the corners of the mouth up which emotional states of individuals in similar and the predictions for mimicry of happy expres- situations may converge. In support of this no- sions are partially based on activity of this muscle. tion, they point to studies showing stronger conta- 140 U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 gion effects when participants were in similar References rather than dissimilar situations compared to the modelŽ. Kulik and Mahler, 1987; Sullins, 1991 .2 Banse, R., 1995. Physio3Ž. PC software . Humboldt University, Both notions described above imply that the Berlin. ᎐ individual first attributes a specific internal state Baron, R.M., Kenny, D.A., 1986. The moderator mediator variable distinction in social psychological research: con- to the interaction partner and that these attribu- ceptual, strategic, and statistical considerations. J. Pers. tions then influence the internal state of the Soc. Psychol. 51, 1173᎐1182. individualŽ. indirect contagion . That is, observers Bavelas, J.B., Black, A., Lemery, C.R., Mullett, J., 1986. ‘I ‘mimic’ the emotions they decode and do not show how you feel’: motor mimicry as a communicative act. ᎐ reflexively adopt the emotion displays they see. J. Pers. Soc. Psychol. 50, 322 329. Bavelas, J.B., Black, A., Chovil, N., Lemery, C.R., 1988. Form However, the present stimulus set can not be and function in motor mimicry: topographic evidence that employed to address the question whether people the primary function is communicative. Hum. Commun. mimic what they see, that is, show reflex-like Res. 14, 275᎐299. mimicry of the facial expressions they are pre- Blairy, S., Herrera, P., Hess, U., 1999. Mimicry and the judg- sented with, or what they know, that is, show the ment of emotional facial expressions. J. Nonverbal Behav. 23, 5᎐41. expression corresponding to the decoded affect. Bourgeois, P., Hess, U., 1999. Emotional reactions to political To adequately address this question, stimuli need leaders facial displays: a replication. Psychophysiology 36, to be constructed for which the decoded affect S36. and the displayed stimulus are incongruent such Buck, R., 1984. Nonverbal receiving ability. In: Buck, R.Ž. Ed. , that different expressions would be shown when The Communication of Emotion. Guilford Press, New York, pp. 209᎐242. ‘mirroring’ than when displaying the decoded af- Cappella, J.N., 1993. The facial feedback hypothesis in human fect. interaction: review and speculation. Special issue: emotio- In sum, the present study could show clear nal communication, culture, and power. J. Lang. Soc. Psy- evidence for mimicry and contagion to non-proto- chol. 12, 13᎐29. typical dynamic emotion displays. No evidence Dimberg, U., 1990. Facial electromyography and emotional reactions. Psychophysiology 27, 481᎐494. was found that either process serves to facilitate Dimberg, U., Ohman,¨ A., 1996. Behold the wrath: psychophys- decoding. This finding adds to and extends previ- iological responses to facial stimuli. Motiv. Emotion 20, ous negative findings in this regard. The nature of 149᎐182. the stimulus material and the relatively lower Duchenne, G.B., 1862r1990. The mechanism of human facial decoding accuracy undermine the argument that expression. In: Cuthbertson, R.A.Ž. Ed. and Trans. , Cam- bridge University Press, Cambridge. the observed lack of the mediational relationship Ekman, P., Friesen, W.V., 1976. Pictures of Facial Affect. proposed by LippsŽ. 1907 is due to the ease of the Consulting Psychologists Press, Palo Alto, CA. task. These findings raise the question of why Ekman, P., &. Friesen, W.V., 1978. The Facial Action Coding people mimic? System: A Technique for the Measurement of Facial Move- ment. Consulting Psychologists Press, Palo Alto, CA. Freud, S., 1921. Group psychology and the analysis of the ego. In: Strachey, J.Ž. Ed. and Trans. , The Standard Edition of Acknowledgements the Complete Psychological Works of Sigmund Freud, vol. 18. Hograth Press, LondonŽ. Original work published 1921 . Fridlund, A.J., Cacioppo, J.T., 1986. Guidelines for human electromyographic research. Psychophysiology 23, 567᎐589. The study was supported in part by a grant Gump, B.B., Kulik, J.A., 1997. Stress, affiliation, and emotio- from the Fond FCAR to the first author. nal contagion. J. Pers. Soc. Psychol. 72, 305᎐319. Hatfield, E., Cacioppo, J.T., Rapson, R.L., 1993. Emotional Contagion. Cambridge University Press. Hess, U., 1998. Mimicry: fact and fiction. Proceedings of the 2 Interestingly, the evidence for counter mimicry reported Xth Conference of the International Society for Research by Lanzetta and colleaguesŽ Lanzetta and Englis, 1989; on Emotion, Wurzburg,¨ Germany, 4᎐8 August. McHugo et al., 1991. was also found in studies in which Hess, U., Blairy, S., Philippot, P., 1998. Facial reactions to participants were either in competition with the model or held emotional facial expressions: affect or cognition? Cognit. strong opinions contrary to those of the model. Emotion 12, 509᎐532. U. Hess, S. Blairy rInternational Journal of Psychophysiology 40() 2001 129᎐141 141

Hess, U., Blairy, S., Philippot, P., 1999. Facial mimicry. In: attitudes on emotional reactions to expressive displays of Philippot, P., Feldman, R., Coats, E.Ž. Eds. , The Social political leaders. Nonverbal Behav. 15, 19᎐41. Context of Nonverbal Behavior. Cambridge University Motley, M.T., Camden, C.T., 1988. Facial expression of emo- Press, New York, NY, pp. 213᎐241. tion: a comparison of posed expressions versus spontaneous Hoffmann, M.L., 1984. Interaction of affect and cognition on expressions in an interpersonal communications setting. empathy. In: Izard, C.E., Kagan, J., Zajonc, R.B.Ž. Eds. , West. J. Speech Commun. 52, 1᎐22. Emotion, Cognition, and Behavior. Cambridge University Philippot, P., Chapelle, G., Blairy, S., 1994. Respiratory feed- Press, Cambridge, pp. 103᎐131. back in the generation of emotion. In: Frijda, N.Ž. Ed. , Kulik, J.A., Mahler, H.I.M., 1987. Effects of preoperative Proceedings of the VIIIth conference of the International room-mate assignment on preoperative anxiety and recov- Society for Research on Emotion, Cambridge, England. ery from coronary-bypass surgery. Health Psychol. 6, Rogers, C.R., 1957. The necessary and sufﬁcient conditions of 525᎐543. therapeutic personality change. J. Consult. Psychol. 21, Laird, J.D., Alibozak, T., Davainis, D. et al., 1994. Individual 95᎐103. differences in the effects of spontaneous mimicry on emo- Schachter, S., 1959. The Psychology of Afﬁliation. Stanford tional contagion. Motiv. Emotion 18, 231᎐247. University Press, Stanford, CA. Lanzetta, J.T., Englis, B.G., 1989. Expectations of cooperation Schneider, F., Gur, R.C., Gur, R.E., Muenz, L., 1994. Stan- and competition and their effects on observers’ vicarious dardized mood induction with happy and sad facial expres- emotional responses. J. Pers. Soc. Psychol. 33, 354᎐370. sions. Psychiatry Res. 51, 19᎐31. Lipps, T., 1907. Das Wissen von fremden Ichen. In: Lipps, T. Siegel, E.V., 1995. Psychoanalytic dance therapy: the bridge Ž.Ed. , Psychologische Untersuchungen Ž Band 1 . . Engel- between psyche and soma. Am. J. Dance Therapy 17, mann, Leipzig, pp. 694᎐722. 115᎐128. Lundqvist, L.O., 1995. Facial EMG reactions to facial expres- Strayer, J., 1993. Children’s concordant emotions and cogni- sions: the case of facial emotional contagion? Skand. J. tions in response to observed emotions. Child Dev. 64, Psychol. 36, 130᎐141. 188᎐201. Lundqvist, L.O., Dimberg, U., 1995. Facial expressions are Sullins, E.S., 1991. Emotional contagion revisited: effects of contagious. Int. J. Psychophysiol. 9, 203᎐211. social comparison and expressive style on mood congru- McHugo, G.J., Lanzetta, J.T., Bush, L.K., 1991. The effect of ence. Pers. Soc. Psychol. Bull. 17, 166᎐174.