¡Descarga Facial Activity & Autonomic Differences in Emotions: Study on ANS Activity y más Apuntes en PDF de Psicología solo en Docsity!
PSY( HOPHYSKIKXIY Vol. 27. No. 4 Copyright ^0 *)90 by The Society for Psychophysiological Research. Inc. Printed in U.S.A.
Voluntary Facial Action Generates Emotion-Specific
Autonomic Nervous System Activity
ROBERT W. LEVENSON, University of California. Berkeley
PAUL EKMAN, AND WALLACE V. FRIESEN University of California. San Francisco
ABSTRACT Four experiments were conducted to determine whether voluntarily produced emotional facial confiKurations are associated with differentiated patterns of autonomic activity, and if so. how this might he mediated. Suhjects received muscle-hy-muscle instructions and coaching to produce facial configurations for anger, disgust, fear, happine.ss, sadness, and surprise while heart rate, skin conductance, finger temperature, and somatic activity were monitored. Results indicated that voluntary facial activity produced significant levels of suhjective experience of the associated emotion, and that autonomic distinctions among emotions: (a) were found hoth hetween negative and positive emotions and among negative emotions, (h) were consistent between group and individual suhjects' data, (c) were found in both male and female subjects, (d) were found in both specialized (actors, scientists) and nonspecialized populations, (e) were stronger when the voluntary facial configurations most closely resembled actual emotional expressions, and (0 were stronger when experience of the associated emotion was reported. The capacity of voluntary facial activity to generate emotion-specitic autonomic activity: (a) did not require subjects to see facial expressions (either in a mirror or on an experimenter's face), and (h) could not he explained by differences in the difficulty of making the expressions or by differences in concomitant somatic activity. DESCRIPTORS: Facial action. Emotion. Autonomic activity during emotion. Emotion-specific autonomic activity.
The experiments in this report arc relevant to waves: (a) the theoretical disputes between .lames two major theoretical issues. The lirst, whether (1884) and Cannon (1927) and their followers, (b) there are dillerent patterns of autonomic nervous the empirical psychophysiological studies begin- system (ANS) activity for different emotions, is one ning with Ax (1953) and continuing for over a dec- of psychophysiology's oldest. The second, under- ade(e.g.. Averill. 1969; Funkenstein, King, & Drol- standing how voluntary facial activity can generate lette, 1954; Schachter. 1957; Sternbach, 1962), and emotion-specific autonomic activity, has a much (c) the recent revival of interest in this issue (e.g., shorter history, but has potentially important im- Ekman, Levenson, & Friesen, 1983; Roberts & plications for emotion theory. Weerts, 1982; Schwartz, Weinberger. & Singer, , „. „.. „ , 1981). Although considerable evidence for ANS Autonom,c Spec.ficty ,n kmotion specificity was reported >n the second wave of stud- Psychophysiological research on the question of ies, their long-term impact may have been blunted autonomic specificity in emotion has come in three by a number of factors (see Levenson, 1988 for a detailed discussion) including: (a) methodological M J. o l '"""•'*" ' ' ' " supported by NIMH grant ^^lems (e.g., failure to verifv subjects' emotional MH39895 and N A grant AG()7476 to the first author ,indMiMu w.->„/,>. J 1, 1. I. » J (^) State bv self-report or observauon of behavior, ob-,. , , j. * _ , ', r ^^ u u i^lMH grant MH.^8691 and Research Scientist .\ward ' , , , ,. r MH06092 to the second author *^'"'"8 physiological measurements long before or Address requests for reprints to: Robert W. Levenson. '""S ^fter the emotions were likely to have been department of Psychology. University of California. <elt by the subjects), (b) theoretical shortcomings Berkeley. C A 94720, or Paul Ekman, Department of Psy- (c g-- lack of consideration of how long an emotion chiatry. University of California, San Francisco, CA lasts and which emotions might be primary), (c) ^4 nonincreniental research (e.g., idiosyncratic meth- 363
364 Levenson, Ekman, and Friesen Vol. 27. No. 4
ods, lack of replication), and (d) the ascendance of models of emotion that emphasized cognition (e.g., Mandler, 1975; Schachter & Singer, 1962) and cul- tural relativism (e.g., Birdwhistell, 1970). The third wave of studies was stimulated by the new support for biological/evolutionary models of emotion pro- vided by evidence of cross-cultural universality in emotional facial expression (e.g., Ekman, 1989; Ek- man, Sorenson, & Friesen, 1969; Izard, 1971) as well as studies of emotion in human infants, ani- mals, and blind adults (reviewed by Ekman & Os- ter, 1979; Fridlund, Ekman, & Oster, 1987). The hypothesis that voluntary production of emotional facial configurations would produce emotion-specific patterns of autonomic activity grew out of experiences Ekman and Friesen had while developing their technique for measuring fa- cial movement, the Facial Action Coding System (FACS; Ekman & Friesen, 1976, 1978). While con- tracting various facial muscles to learn how they were related to changes in facial appearance, both researchers found themselves experiencing strong physical sensations when they contracted muscles that produced facial configurations that resembled the universal facial expressions for certain emo- tions. The first report of our work (Ekman ct al., 1983) presented findings from a study using two experi- mental tasks that represented the extremes on a physiological-cognitive dimension. In the physio- logical task (directed facial action task), subjects fol- lowed muscle-by-muscle instructions to contract voluntarily sets of facial muscles, which together would produce a facial configuration that resembled an expression that universally signals one of six emotions: anger, fear, disgust, sadness, happiness, and surprise. In the cognitive task (relived emotions task), subjects were asked to relive a past emotional experience for each of these emotions. Results from this study indicated that both the directed facial action task and the relived emotions task produced autonomic changes that enabled dis- tinctions to be made among emotions. There were differences between negative and pwjsitive emotions that were common to both tasks: (a) anger and fear produced larger heart rate increases than did hap- piness, and (b) anger produced a larger finger tem- perature increase than did happiness. There were also differences among negative emotions only in the directed facial action task: (a) the anger, fear, and sadness configurations produced larger heart rate increases than did the disgust configuration (which actually produced a heart rate decrease); and (b) the anger configuration produced a larger tem- perature increase than did the fear configuration (which actually produced a temperature decrease).
A difference among negative emotions only in the relived emotions task was that sadness produced a larger skin conductance increase than did anger, fear, or disgust. Beyond this evidence for autonomic specificity in emotion, perhaps the most intriguing finding was that voluntarily contracting facial muscles into emotional configurations produced patterned auto- nomic activation that bore some similarity to that obtained using the more conventional relived emo- tions task.
Fundamental Questions Questions raised by these initial findings caused us to launch a series of additional studies that fo- cused on voluntary facial action and emotion-spe- cific ANS activity. In this paper we will report the results of new analyses from our previously pub- lished experiment and the results from three new experiments, which address 11 fundamental ques- tions in three categories: (a) emotional report and autonomic specificity, (b) generalizability across sit- uations and populations, and (c) alternative expla- nations. Emotional report and autonomic specificity. The first set of questions concerns the extent and nature of subjective emotional experience produced by the directed facial action task and whether this task reliably produces emotion-specific autonomic dif- ferentiation: (1) Does voluntary production of emo- tional facial configurations result in subjective re- port of the associated emotion? (2) Are there reli- able autonomic differences among the negative emotions of anger, disgust, fear, and sadness, the positive emotion of happiness, and surprise? (3) Are autonomic differences among emotions that are found in group data also found in the data from individual subjects? (4) Are autonomic differences among emotions more pronounced and self-reports of the associated emotion more prevalent when subjects produce facial configurations that most closely resemble emotional expressions? (5) Arc the distinctions among negative emotions more pro- nounced when subjects report feeling the emotion associated with the facial configuration? Generalizahility across situations and popula- tions. This second set of questions concerns the ex- tent to which these findings are limited to certain experimental situations or to certain populations: (6) Are these findings limited to either male or fe- male subjects? (7) Arc these findings limilcd to spe- cialized populations whose work focuses on the face (actors or scientists who study the face)? (8) Arc these findings limited to situations in which sub- jects can see an emotional facial configuration (either in a mirror or on the face of an experimen- ter)?
(^366) Levenson, Ekman, and Friesen (^) Vol. 27. No. 4
the mouth. The coach provided feedback and sugges- tions as needed to help the subject comply with the instructions (e.g.. "that's right," "don't raise your eye- brows, lower them," "try to raise your eyelid higher"). The final facial configuration was held for 10 s. The subject was then asked if any feelings, memories, or sensations had occurred while holding the facial con- figuration. Following a 2-min rest period, the next trial began. This procedure was repeated for the six emotional configurations (anger, disgust, fear, happiness, sadness, and surprise) in one of three counterbalanced orders. Each configuration represented a universal emotional facial expression, based on evidence from cross-cul- tural studies of bolh the recognition and expression of emotion (Ekman et al., 1969; Izard, 1971; Ekman, 1989). Subjects were provided with only one oppor- tunity to make each configuration^. Although this task is somewhat novel, it does re- semble a posing task (e.g.. asking someone to "look" sad). However, in the directed facial action task no emotion was mentioned by name, and subjects were not asked to feel or think anything; subjects were asked only to contract facial muscles. Quite apart from our interest in the phenomenon of such voluntary facial action generating emotion-specific ANS activity, the directed facial action task has several methodological advantages over other techniques for sampling emo- tions: (a) examination of video recordings can verify that all of the requested muscle contractions did occur without any extraneous contractions, and (b) the mo- ment when the facial configuration was fully formed and the moment when the configuration left the face can be located and the physiological responses that occurred during this time can be extracted for analysis.
EXPERIMENT 2 Subjects In contrast to the professional actors from San Francisco used in Experiment I, Experiment 2 used Indiana University college students as subjects. One hundred and three undergraduates were recruited from the introductory psychology classes. Their participa- tion fulfilled part of a eourse requirement. These sub- jects were screened individually to determine their ability to control voluntarily their facial muscles using a procedure that required moving certain facial mus- cles singly and in combination (but not making full emotional configurations). Sixteen subjects (9 males, 7 females) who demonstrated good voluntary control participated in Experiment 2 at a later date and were paid SIS.
Apparatus Physiological. A system consisting of a Grass Mod- el 7 polygraph and a PDP 11/10 minicomputer was
Mn Experiments I, 2, and 3 there were a total of three instances in which a second attempt was allowed due to procedural problems. In these instances, an additional trial was added at the end of the task.
used for acquisition and on-line analysis of physio- logical data. Second-by-second averages were obtained for; (a) Heart rate; same as in Experiment I. (b) Skin conductance; a constant voltage device passed a small voltage between Beckman regular electrodes (same sites as Experiment I) using an electrolyte of sodium chloride in Unibasc, (c) Finger temperature; a ther- mistor was taped to the palmar surface of the distal phalanx of the second finger of the dominant hand, and (d) General somatic activity; an electromechanical transducer attached to a platform under the subject's chair generated an electrical signal proportional to the amount of movement in any direction. Four addi- tional physiological measures were obtained but will not be used for this report because they were not ob- tained in Experiment I; (a) pulse transmission time to the finger, (b) finger pulse amplitude, (c) pulse trans- mission time to the ear, and (d) respiration rate. Video. Similar apparatus to that of Experiment I was used except that a FOR.A video time code gen- erator, which superimposed the elapsed time in hun- dredths of a second on the video recording, was used instead of the frame counter. The video camera was mounted on the wall in front of the subject behind a wooden partition with a small hole in its center. Sub- jects were informed about the camera and video re- cording. Procedure In this experiment, subjects participated in only the directed facial action task. Three changes were made from the procedures used in Experiment I; (a) no mir- ror was used; (b) the coach (P.E.) was not in the same room as the subject, but instead viewed the subject's face on a video monitor and made comments over an intercom system; and (c) following each trial subjects were asked whether they experienced any emotions (the word "feelings" had been used in Experiment I), memories, or sensations. For any reported emotion, the subject was asked to rate its intensity on a 0- scale (0=no feeling of the emotion; 8 = the most in- tense feeling of the emotion ever experienced).
EXPERIMENT 3 Subjects Subjects were again nonactors. but this time were recruited from the San Francisco area using adver- tisements in local newspapers. One hundred and nine- teen respondents were screened for ability to control voluntarily their facial muscles as in Experiment 2. and were paid $8. Thirty subjects (9 males. 21 females) who demonstrated good voluntary control participat- ed in Experiment 3 at a later date and were paid an additional $17.
Apparatus Physiological. A system consisting of an LSI 11/ microcomputer and two Lafayette Instruments 6- channel polygraphs was used to obtain the same phys- iological measures obtained in Experiment 2. Video. The same equipment was used as in Exper- iment 2. The camera, however, was located in another
July, two Facial Action and Emotion-Speeific ANS Activity 367
room behind a glass partition. Subjects were again in- formed of the camera and the video recording. Procedure Subjects participated in the directed facial action task as in Experiment 2 with two changes: (a) as in Experiment 1. a mirror was provided to help subjects make the facial configurations: however, as in Exper- iment 2. the coach (P.E. or W.F.) remained in a sep- arate room and viewed the subject's face on a video monitor; and (b) after each trial the subject was asked to rate the intensity of any reported feelings on a 0- seale and the difficulty of making the configuration on a 1-5 scale (I = extremely easy. 5 = extremely difficult).
EXPKRtMKNT 4 Experiment 4 was performed to determine whether subjects could identify the target emotion only by read- ing the instructions used to construct eaeh facial con- figuration in the directed facial action task. Subjects Thirty-nine undergraduates (6 males. 33 females) at the University of San Franeisco participated in this experiment in fulfillment of a course requirement. Procedure Subjects were given a questionnaire with the fol- lowing instructions:
"We arc interested in learning what emotions you 'hink someone would fed when he or she shows a par- ticular facial expression. On the following pages a number ot facial expressions are described. After each expression IS described, you are to indicate how you think someone 1 showed that expression would feel. It is very inipor- that you don't try to make the expression on your own face. Just make your judgments from reading the deseription."
The expressions were described with the same lan- guage used in the directed facial action task. For each expression, subjects used 0-8 scales to rate the amount of anger, disgust, fear, happiness, sadness, and sutprise ihcy thought someone making that expression would feel.
Results'
Data Reduction: Experiments I, 2, and 3 Physioloijical. The videotape recording for each subject was cxatnitied to locate the standard control face and the target emotional face on each trial. The physiological data for those seconds during which each face was being held (usually 10 s) were then extracted and averaged, and a change score (target face average minus standard eontrol face average)
was calculated for the four physiological measures (heart rate, finger temperature, skin conductance, and muscle activity). Facial. The Facial Action Coding System (FACS: Ekman & Friesen. 1978) was used to de- tertnine which facial muscles were actually con- tracted on each trial. FACS is an anatomically based system which enables decomposing any facial expression into its visually distinguishable mus- cular actions through repeated slow-motion view- ing of the videotape recording. Working with the nunierieal FACS codes and silent video tape rec- ords, a rater (who was blind to both the experi- mental design and the associated target emotion) assigned a performance score (on a 0-4 scale) to each facial configuration indicating the extent to which: (a) the configuration included all of the mus- cle contractions specified in the instructions and no others, and (b) the contractions were held steadily throughout the 10-s holding period''. Thirty-six faces sampled from the three experiments and the six target aflcets were scored for reliability by a sec- ond coder (who was blind to the target emotion). The inter-coder correlation was 0.89. Sclf-ti'port. The open-ended self-reports ob- lained after each trial in Experiments I. 2. and 3 were transcribed from the video recordings. Ati as- sistant (who was blind to both the experimental design and the associated target emotion) assigned codes to each statement indicating whether the sub- ject had experienced an emotion and/or recalled an emotional memory. If a statement fell within one of these broad categories, it was further classified in terms of the specific emotion involved (22 sp)e- eific emotion subcodes were used). _._ sample of 330 self-reports was randomly selected and rated by a second coder (who was blind to the target emotion). Inter-coder agreement as to the specific emotion was 81%. Using these codes, we later determined on each trial whether the emotion associated with the facial configuration was i-eported most strongly. If a sub- ject reported only the target emotion, it was con- sidered to be the strongest reported emotion. If a subject reported more than one emotion, the target emotion had to be rated as being experienced more intensely than any other eniotion. If the target emo- tion and another emotion tied for the highest rating, this somewhat rare event (1.9% of the trials) was counted as a trial in which the target eniotion was tiot reported most strongly. Hereafter, whenever we
The ,05 level was adopted for all tests of significance unless otherwise stated.
*Beeausc this method for scoring faeial quality differed slightly from the method used in our earlier report (Ek- man. Levenson. & Friesen. l'JS?). the group data results involving high quality expressions will differ somewhat from those presented in the earlier paper.,
July. 1990 (^) Facial Action and Emolion-Specific ANS Activity 369
pcctcd hy chance (16.7%). The percentage of trials in which the target emotion was reported was sig-
- Experlmant 1 S Expsrimenl 2 D Experiment 3
" ercentoge ol trials
Target emotion Figure 1. Percentage of trials in which subjects re- ported the largol cniotion in Ihc tlircclcd facial action task. Experiments 1 and 3 used open-ended reports of "feel- ings'" and a mirror was present. I-xix-rinient 2 used ojien- cnded reports of "emotions" and there was no mirror present. AN-Anger trials. D l ^ Disgust trials. FE-'Fear trials. HA-Happiness trials. SA = Sadness trials. SU = Surprise trials. ALL = A11 trials.
nificantly greater in Experiment 2 than in Experi- ment 1. r=.^.84. / x. O O I. and Experiment 3. r = 4.30,/x.OOI. 2, There were reliable autonomic differences among the negative emotions of anger, disgust, fear, and sadness, the positive emotion of happiness, and surprise, .\n overall 3 X 2 X 6 (Experiment X Sex X Emotional Configuration) MANOVA with emo- tional configuration as a within-suhjects factor was carried out on ehanges from baseline (i.e.. target emotional faee tnmus standard control face) for all four physiological measures. _._ significant main ef- fect for emotional configuration. /•1(20/29) = 4.56. /X.OOI. indicated that there were autonomic dif- ferences among the six emotional configurations. Within the significant emotional configuration main etiect. univariate .ANOVAs revealed that the physiological variables that difterentiated the six emotional facial configurations were heart rate. /•(5/24O)=ll.l3. / x. O O I. and skin conductance. /•(5/240) = 3.16. adjusted (Huynh-Feldt) p=.O2. The effect for skin temperature approached signif- icance. /•
