Biological Psychology North-Holland

137

31 (1990) 1377147

FACIAL ELECTROMYOGRAPHIC REACTIONS ACTIVITY TO AUDITORY STIMULI * Ulf DIMBERG Department

AND AUTONOMIC

**

of Psycholom,

Uppsala Unioerslty, S-751 48 Uppsala. Sweden

This study explored whether high- and low-intensity auditory stimuli evoke different facial electromyographic reactions and autonomic responses. Subjects were repeatedly exposed to 95dB and 7%dB tones (1000 Hz, 40 ms rise and fall times) while their facial electromyograms from the corrugator and zygomatic muscle regions, heart rate, skin conductance responses, skin conductance half recovery time and ratings were measured. The 95dB tone evoked a “negative” reaction with increased corrugator activity and an autonomic response pattern that carried aspects of a defense reaction, that is, slowly habituating skin conductance responses with retarded recovery rate and an initial tendency to heart rate acceleration. Furthermore, the 95dB tone was rated as unpleasant. The 75-dB tone elicited an orienting response indicated by a distinct heart rate deceleration and fast habituating skin conductance responses with a relatively short recovery time. Thus, the present study demonstrated that the facial electromyographic technique is sensitive to simple environmental stimuli such as auditory stimuli and that the facial response is consistent with the autonomic response patterns and the experience of the stimuli. Keywords:

Facial electromyography, autonomic reactions, heart rate, skin conductance habituation, orienting response, defense reaction, startle.

recovery,

1. Introduction Earlier research has demonstrated that different facial electromyographic (EMG) reactions are related to different emotional activities (for a review see Fridlund & Izard, 1983). For instance, specific facial EMG activities correspond to different self-generated emotional imagery (e.g., Schwartz, Fair, Salt, Mandel, & Klerman, 1976) whereas various visual emotional stimuli spontaneously evoke different facial EMG response patterns (e.g., Dimberg, 1982, 1986). More specifically, slides of positive stimuli (such as happy faces and flowers) evoke increased zygomatic activity whereas slides of negative emotional stimuli (angry faces and snakes) spontaneously elicit increased corrugator muscle activity (Dimberg, 1982, 1986, 1988; Dimberg & Thell, 1988). The * This study was supported by grants from the Swedish Council for Research in the Humanities and Social Sciences and from Magn. Bergvalls Stiftelse, Sweden, * * Address correspondence to Ulf Dimberg, Department of Psychology, Uppsala University, Box 1854, S-751 48 Uppsala, Sweden. 0301-0511/90/$3.50

0 1990 - Elsevier Science Publishers

B.V. (North-Holland)

zygomatic muscle eievates the cheeks to a smile whereas the corrugator muscle is used when frowning (Hjortsjii, 1970). Furthermore, the evoked facial EMG response patterns are consistent with how the subjects rate the stimuli (Dimberg & Thell, 1988) and the generated facial reactions also correspond to how the subjects experience their own specific emotions (Dimberg, 1987, 1988). Moreover, the facial EMG technique differentiates not only the valence but also the intensity of affective reactions (Cacioppo, Petty, Losch, & Kim, 1986). In summary, this empirical evidence demonstrates that facial EMG is sensitive to emotional stimuli and supports the proposition that the facial muscles constitute an emotional read-output system. This is consistent with theories claiming that human emotions are biologically programmed and that specific facial expressions are linked to different emotions (Darwin, 1872; Ekman, 1972; Izard, 1977; Tomkins, 1962). One interesting question is whether facial EMG reactions are also sensitive to simple environmental stimuli, such as auditory stimuli. In previous research it has been found that EMG activity from different bodily muscle sites increased when subjects were exposed to loud auditory stimuli (Davis, Malmo, & Shagass, 1954; Fridlund, Hatfield, Cottam, & Fowler, 1986). For instance, Davis et al. (1954) exposed psychiatric patients and control subjects to bursts of white noise while EMG was detected from the masseter, the sternomastoid and the forearm extensor muscles. The burst of white noise elicited fast startle reactions which differed between patients and controls. These studies did not focus particularly on emotional reactions and did not involve measurement of different facial muscles specifically related to emotional expressions (e.g., Darwin, 1872; Ekman, 1973; Izard, 1977). One interesting question is whether auditory stimuli evoke facial muscle responses that could be related to emotional activity and the hedonic tone of the reaction. The present study focuses on this question. As reported above the facial EMG technique is sensitive in detecting differential muscle activity in the corrugator and the zygomatic muscles when subjects are exposed to different emotional stimulation. Thus, if subjects are exposed to auditory stimuli, one could expect that tone stimuli with a high aversive intensity should elicit a “negative” facial reaction indicated by increased corrugator activity whereas low intensity tones should not. Traditionally, autonomic measures have been used to detect different physiological response patterns to different stimulus intensities (for a review see Turpin, 1986); Stimuli with high intensity have been proposed to evoke a defensive reaction (DR) indicated by heart rate (HR) acceleration and slowly habituating skin conductance responses (SCRs) whereas low-intensity stimuli typically evoke an orienting response (OR) indicated by HR deceleration and habituating SCRs (Graham, 1973; Sokolov, 1963). Thus, the present study was performed to investigate whether subjects react with different facial EMG reactions and different autonomic responses when exposed to simple tone stimuli with high and low intensity within a traditional

U. Dwnherg / Facial reactions to audItoT stimulr

139

OR/DR habituation paradigm. Facial EMG was measured from the corrugator and zygomatic muscle regions whereas autonomic reactions were measured as HR and SCRs. Since intensive/aversive stimulation can be expected to be accompanied by retarded SCR recovery rates (Dimberg, 1987; Edelberg, 1973) SCR half recovery time was also measured. Finally, subjects were required to rate their experience of the stimuli on a number of scales.

2. Method 2. I. Subjects Thirty-eight subjects studying psychology or law at Uppsala University participated in the experiment. There were 19 males and 19 females who were paid about 8$ US to participate. 2.2. Appurutus

and data scoring

The subjects were tested individually in a comfortable chair situated in a sound-attenuated experimental chamber. All subjects were exposed to repeated presentations of a high (95 dB) and a low (75 dB) lOOO-Hz tone administered through headphones. The stimulus duration was 4 s and was controlled by an electronic timer. The intertrial intervals varied between 25 and 45 s. To avoid confounding effects of startle reactions the stimulus rise and fall times were set at 40 ms. Bipolar EMG was measured from the corrugator and zygomatic muscle regions. Beckman miniature Ag/AgCl electrodes were used, filled with Beckman electrode paste. The electrodes over the corrugator region were attached directly above the eyebrow on an imaginary vertical line through the inner border of the iris respectively slightly medial to the border of the eyebrow head. To detect zygomatic activity the two electrodes were placed midway along an imaginary line through the lower edge of the cheekbone to the corner of the mouth. The distance between the electrodes was about 1.5 cm. The subjects were connected to a Hewlett-Packard bioelectric amplifier with highpass and lowpass filters set at 50 Hz and 1000 Hz, respectively. The highpass filter was set at 50 Hz to reduce low-frequency intermuscle site crosstalk (Fridlund & Cacioppo, 1986). Before the electrodes were attached, the skin was cleaned with ethanol and slightly rubbed with electrode paste. The raw EMG signals, which were visually inspected using a Tektronix oscilloscope, were analyzed by a two-channel contour-following integrator which was a variant of that described by Fridlund (1979). The time constant was set at 1.0 s. The integrated DC signal was further recorded on paper on a Hewlett-

140

U. Lkmberg

/ Facial reactions to auditory stirnull

Packard 7700 series recorder and was the basis for the EMG scores. The signal was manually scored second by second and the responses were expressed as change in microvolts from the pm-stimulus level. The pre-stimulus level was defined as the instantaneous activity at stimulus onset. This scoring technique allows the evaluation of both increasing and decreasing phasic responses. Heart rate was measured with two Beckman miniature Ag/AgCl electrodes filled with Beckman electrode paste. The electrodes were attached to the subject’s chest and connected to a Hewlett-Packard bioelectric amplifier. The eIectrocardiogram was recorded on a Tandberg FM tape recorder. The R-R intervals were scored and converted to HR in beats per minute using a PDP 1 l/40 computer. HR was scored second by second during the stimulus interval, with each beat weighted by the proportion of the second it occupied. The HR data also included a pre-stimulus value which was based on the mean of the five pre-stimulus seconds. Palmar skin conductance was measured with 3eckman Ag/AgCl electrodes with a diameter of 8 mm, connected to a Hagfors type constant-voltage circuit (Venables & Christie, 1973). The electrodes were filled with 0.05 M NaCl Unibase electrode paste and were attached to the second phalanx of the first and second left-hand fingers. The signal was recorded on paper on a HewlettPackard 7700 series recorder. The SCR magnitudes were scored in microsiemens (+S) as the largest phasic response which began in the interval 1-4 s after stimulus onset. The minimum response criterion was 0.05 pS. SCR half recovery time was measured as the time for the response to recover to half amplitude. The criterion for habituation of SCRs was the number of trials to two consecutive trials with no response. After the stimulus series the subjects were required to rate how pleasant and unpleasant the stimuli were. Ratings were performed on 5-point scales, with “not at all” at one end and “very much” at the other end. They also rated the intensity of stimuli, with “ very weak” at one end and “ very high” at the other end of the scale.

2.3. Procedure

Subjects were instructed that the aim was to study physiological reactions to different auditory stimuli. To mask the purpose of facial EMG recordings they were told that heart rate and sweat gland activity in the hand and the face was measured. The order of intensity was balanced between subjects so that half of the subjects first received 6 trials of the 95-dB tone followed by 6 trials of the 75dB tone, whereas the other half had the reversed order. Immediately after exposure to the respective intensity the subjects rated their experience of the tone on the different scales.

U. Dimberg / Fucial reactions to auditmy stirnull

2.4. Design and statistical

141

analysis

Before analysis, data were collapsed in trial blocks with two trials per block. Consequently, the basic design was a 2 x 3 randomized block factorial design (Kirk, 1968) with intensity (95 dB vs. 75 dB) and trial block (3) as within-subject factors. The data were evaluated by ANOVAs and a priori t-tests. Since repeated measures are likely to result in positively biased F tests (Jennings, 1987; Kirk, 1968) the Geisser-Greenhouse conservative F test was employed throughout by reducing the degrees of freedom.

3. Results 3. I. Fucial EMG The facial EMG data were analyzed by an Intensity (95 dB vs. 75 dB) x Muscle (corrugator vs. zygomatic) X Trial block (3) X Second (4) ANOVA, with all factors as repeated measures. As predicted, the 95-dB tone evoked more overall corrugator activity than zygomatic activity, t(37) = 2.47, p < 0.01, as well as more corrugator activity as compared to the 75-dB tone, t(37) = 3.07, p < 0.005. There was no significant difference in zygomatic activity between the 95-dB and the 75-dB tone. As indicated by the Intensity X Muscle X Trial block interaction, F(l, 37) = 5.53, p < 0.05, and as can be seen in fig. 1, the effect was especially pronounced for trial block 1. Accordingly, a separate analysis on trial block 1 revealed that the different intensities evoked different muscle activity, F(1, 37) = 5.37, MS, = 1.294, for the Intensity x Muscle interaction. The 95-dB stimulus evoked a larger corrugator than zygomatic response, t(37) = 3.02, p < 0.005, and more corrugator activity as compared to the 75-dB tone, t(37) = 4.09, p < 0.005. As for the overall data, there was no

10

Trial

75dE

block

1

95dB

T.,d

75dB

block

2.

95dB

Trial

75dB

block

3.

95dB

Fig. 1. The mean EMG response (pV) for corrugator (C) and zygomatic and 95-dB tones plotted as a function of trial block.

(Z) muscle regions

to 75-

142

U. Dvnherg

/ Fucral reactions to auditory strmuli

significant difference in zygomatic activity between the 95-dB and the 75-dB tone. Furthermore, the overall ANOVA also revealed that 95 dB elicited more overall muscle activity than did 75 dB, F(1, 37) = 10.46, MS, = 2.311, and that responding was higher overall during trial block 1 and decreased over trials, F(1, 37) = 10.31, MS, = 0.741. As can be seen in fig. 1 and as indicated by the Intensity X Trial block interaction, F(1, 37) = 10.13, MS, = 0.639, this higher response tendency was most pronounced for the 95-dB tone during trial block I. Finally, the significant second factor, Ffl, 37) = 9.97. MS, = 0.220, was due to an overall increase in activity as a function of seconds. The Intensity x Second interaction, however, revealed that it was only responses to the 95-dB tone that monotonically increased over seconds, F(1, 37) = 6.06, MS, = 0.193 (m = 0.15, 0.24, 0.35 and 0.42 for the respective seconds). 3.2. Heart rare

The HR data were evaluated by an Intensity (95 dB vs. 75 dB) x Trial block (3) X Second (1 pre-onset and 4 post-onset values) repeated-measures ANOVA. FIR data were also evaluated by trend analyses. Because of technical flaws HR data are based on 34 subjects. There was an overall effect of second, F(t, 33) = 4.32, M.S, = 6.549, which was due to an overall quadratic response tendency, F(1, 33) = 8.13, MS, = 12.680. As can be seen in fig. 2 and as indicated by the Intensity X Trial block interaction, the quadratic response tendency was most pronounced for trial block 1 and then declined as a function of trials. (Note that the ANOVA was performed on HR scores in beats per minute whereas the data in the figure are plotted as change scores from the pre-stimulus level). In a separate analysis on trial block 1 the Intensity X Second interaction revealed that the different

Fig. 2. The mean

HR response to 75- and 95-dB tones expressed as change in beats per minute (bpm) from the pre-stimulus level. plotted as a function of trial block.

143

U. Dimberg / Facial reactions to auditory stmuli 1.5

Trial Fig. 3.

intensities Trend trends,

block

The mean SCR magnitude($?) to 75 and YS-dB tones plotted as a function of trial block. evoked

analysis

different

showed

response

that

patterns,

this difference

F(1, 37) = 5.70,

MS, = 4.018.

was due to different

quadratic

F(1,

33) = 18.50, MS, = 4.432. As illustrated in fig. 2 and supported by the trend analysis, the response to the 75-dB tone was a distinct deceleration which was positive in quadratic trend whereas the response to the 95-dB tone tended to be accelerative with an inversed quadratic trend. No other effects were significant. 3.3. SCR Skin conductance

response

SCRs were evaluated by an Intensity (95 dB vs. 75 dB) X Trial block (3) repeated-measures ANOVA. As illustrated in fig. 3 and by the ANOVA, 95-dB evoked overall larger responses than did 75-dB tones, F(1, 37) = 31.56, MS, = 0.399. The ANOVA also showed that there was an overall decrease in responding over trial blocks, F(1, 37) = 45.19, MS, = 0.180, which differed between intensities, F(1, 37) = 21.12, MS, = 0.151. When habituation was evaluated as two trials to criterion, that is, two consecutive trials with no response, the analysis demonstrated that habituation was slower for the 95-dB as compared to the 75-dB stimulus, F(1, 37) = 34.28, MS, = 2.518. Finally, scoring SCR recovery is typically accompanied by loss of data and recovery data are based therefore on 23 subjects. The SCR half recovery time was reliably longer to 95 dB as compared to 75 dB, F(1, 22) = 6.71, MS, = 31.944 (m = 3.54 s and 2.67 s for 95 dB and 75 dB, respectively). 3.4. Rating data The rating data were 95-dB tone was obviously

evaluated by a repeated-measures experienced as more intense than

ANOVA. The the 75-dB tone,

144

U. Dtmherg / Facial reactions to auditory stir&t

(m = 4.2 and 2.9, respectively), F(1, 37) = 104.99, MS, = 0.289. The 95dB tone was also experienced as more unpleasant than the 75-dB tone (m = 4.2 and 2.8, respectively), F(1, 37) = 62.73, MS, = 0.611, whereas the 75-dB tone was experienced as more pleasant (m = I.71 for 75 dB and 1.1 for 95 dB), F-( 1, 37) = 17.08, MS, = 0.444.

4. Discussion The present data showed that auditory stimuli of high and low intensity evoke different facial EMG responses. As predicted, a high-intensity aversive tone elicits more corrugator than zygomatic activity and more corrugator activity than does a low intensity tone. Thus, the present results demonstrate that the facial EMG technique is sensitive to simple environmental stimuli such as tone stimuli of various intensities. These responses were accompanied by different autonomic response patterns. The 95-dB tone initially evoked a HR response that tended to be accelerative whereas the 75-dB tone elicited a distinct HR deceleration. This was paralleled by larger and more slowly habituating SCRs to the 95-dB tone. They also exhibited longer half recovery times. The 95-dB tone was also experienced as more intense and more negative as shown by the unpleasant and pleasant ratings. As expected, the high-intensity/aversive tone evoked increased corrugator activity which is interpretable as a negative affective response. This interpretation was further supported by the rating data showing that the subjects experienced the 95-db tone as unpleasant. These data parallel the facial response patterns elicited by negative affective visual stimuli, such as slides of angry faces and snakes (Dimberg, 1982,1986,1988; Dimberg and Thell, 1988). Thus, the present data are consistent with the theory that the facial muscles constitute an output system for affective reactions. It is important to note, however, that an increased corrugator response does not necessarily reflect a negative emotional response. The corrugator muscle is active not only in emotional expressions such as fear, anger and sadness but even in non-emotional gestures (Ekman, 1979). To evaluate this question more specifically, ratings of specific emotions and recordings from several muscle regions should be included. It should be noted that although the 95-dB tone evoked a distinct facial EMG response, the 7%dB tone evoked little or no responses. The absence of a specific response pattern to 75 dB is consistent with the inte~retation that facial EMG reflects affective reactions in the present paradigm. That is, the 95-dB tone was perceived as highly unpleasant and was accompanied by increased corrugator activity. The 75dB tone, on the other hand, was neither unpleasant nor pleasant. Thus, the 75-dB tone was experienced as neutral and should consequently not evoke a facial response indicating an affective

U. Dimberg

/ Facial reactions to auditory stimuli

145

reaction. A further extension of the present study might be to expose subjects to auditory stimuli that are perceived as pleasant. This stimulation would then be expected to elicit increased zygomatic activity, reflecting a positive response. The interpretation that the high and the low intensities evoked response patterns associated with different hedonic tones was further supported by the autonomic data. The response to the 75-dB tone displayed characteristics typical of an orienting response with a distinct initial HR deceleration and fast habituating SCRs with a relatively short half recovery time. The response to the 95-dB tone, on the other hand, displayed aspects of a defense reaction indicated by a tendency to HR-acceleration and larger, slowly habituating SCRs with retarded recovery rate. One may question, however, whether the HR response to 95 dB indicates a startle reaction (SR) rather than a DR (Graham, 1979). The relatively weak HR acceleration and short peak latency (2 s) as well as the fast habituation support the SR interpretation for the HR response (e.g., Turpin, 1986). On the other hand, the relatively long stimulus rise and fall times (40 ms) should have been large enough to prevent the evocation of a SR (Graham, 1979). Importantly, the response shape of the facial EMG reaction did not support a simple SR interpretation. That is, the facial element of the startle reaction is typically associated with a short latency and a brief duration of about half a second (Ekman, 1984; Ekman, Friesen, & Simons, 1985). The facial response in the present experiment, on the other hand, increased as a function of stimulus duration and reached its peak during the last quarter of the exposure interval. Even if the relatively long time constant of the contour-following integrator may moderate the detection of a fast SR as well as obscure the response shape, the high muscle activity during the last stimulus second demonstrates that the corrugator response consists of components other than a fast SR. Thus, facial EMG data do not support a simple SR interpretation but rather indicate that the activity may reflect an emotional response. Furthermore, it may be that 95 dB was insufficient to elicit a distinct DR. Thus, to elucidate further the relation between facial EMG and autonomic response patterns within the OR/DR paradigm future studies should include higher stimulus intensities (e.g., Turpin & Siddle, 1983). The purpose of the present study was to measure EMG responses from facial muscles that normally are active in emotional expressions and that in earlier studies have proved to reflect negative and positive reactions to different stimuli. The results are interpreted to indicate that 95-dB tones evoke a negative affective response whereas the 75-dB tone evokes an orienting response with no specific emotional concomitants as indicated by facial EMG or ratings. However, when inspecting the facial EMG data in fig. 1, one could argue that, besides evoking a strong corrugator response, the 95-db tone also tended to evoke increased zygomatic activity. In fact it is possible to argue that

negative stimuli should evoke decreased zygomatic activity. Earlier studies (e.g.. Dimberg, 1986) have found that the increased corrugator response to a negative stimulus was accompanied by a tendency to decreased zygomatic activity. Thus, it is possible to interpret the present results as indicating that aversive tone stimuli evoke increased muscle tension in general, with the zygomatic muscle being less sensitive to this manipulation, and that the response has no specific affective concomitants. To further explore this hypothesis future studies should include recordings from several other muscle regions. IIowever, it is important to note that the statistical analysis did not support this interpretation. That is, specific comparisons between zygomatic activity to 95 dB and to 75 dB did not reveal a reliable difference. The corrugator response to 95 dB, on the other hand, was significantly larger than the response to 75dB tones as well as larger than the zygomatic activity to 95dB tones. Furthermore, these comparisons may indicate a second problem. It is possible to argue that because the corrugator and zygomatic muscles are of different sizes, the detected activity from the different muscles are not comparable. It is important to note, however, that the registration procedure and experimental procedure in the present experiment were identical to the conditions in previous studies conducted in our laboratory. In these studies (which are referred to above) it was possible to detect both corrugator and zygomatic activity which were comparable in magnitude. Thus, it is likely that the detected activities from the different muscle sites in the present study are also comparable with each other. Further, it is important to point out that important results and interpretations in the present study are based on comparisons between tone intensities within each muscle site. In summary, the present results demonstrated that auditory stimuli high in intensity evoke a response pattern that is interpretable as a negative affective reaction as indicated by both facial EMG, autonomic responses and ratings of the stimuli. Thus, the present study showed that the facial EMG technique is sensitive to simple environmental stimuli such as auditory stimuli.

References Cacioppo. J.T., Petty, RX, Losch, M.E., & Kim, KS. (1986). El~trom~ographic activity over facial muscle regions can differentiate the valence and intensity of affective reactions. fournof of PcrsonaIir), and Social P.gwholo~): 50, 260-268. Darwin, C. (I 872). The eqwessron of emorion rn man and rmimals. London: Murray. Davis. J.F., Malmo, R.B., & Shagass, C. (1954). Electromyographic reaction to strong auditory stimulation in psychiatric patients. Cunudian Jourr~ul of’Psycho/ow, 8, 177-186. Dimberg, U. (1982). Facial reactmns to facial expressions. Psychophysiolqq, 19, 643-647. Dimberg, U. (1986). Facial reactions to fear-relevant and fear-irrelevant stimuli. Biological P.wcho/o~v, 23. 153-161. Dimberg, U. (1987). Facial reactions, autonomic activity and experienced emotion: A three component model of emotional conditioning. Biolog~cuI Psych&m, 24, 105-122.

L! Drmherg / Facial reacfions fo auditori: sfimuli

147

Dimberg, U. (1988). Facial electromyography and the experience of emotion Journal of Psychophysiology, 2, 217-282. Dimberg, U., & Thell, S. (1988). Facial electromyography, fear relevance and the experience of stimuli. Journal of I’sychophysiology, 2, 213-219. Edelberg, R. (1973). M~hanisms of electrodermal adaptations for locomotion, manipulation or defense. In E.S. Stellar & J.M. Sprague (Eds.). Progress in phy~iolugicaI psychologv (Vol.5. pp. 155-209). New York: Academic Press. Ekman, P. (1973). Cross-cultural studies of facial expressions. In P. Ekman (Ed.), &ruJin and facial expression (pp. 1699222). New York: Academic Press. Ekman, P. (1979). About brows: Emotional and conversational signals. In M. van Cranach, K. Foppa, W. Lepenies, 62 D. Ploog (Eds.), Numun efhology (pp. 169-202). Cambridge. UK: Cambridge University Press. Ekman. P. (1984). Expression and the nature of emotion. In K. Scherer gL P. Ekman (Eds.), Approaches fo emotion (pp. 319-343). Hillsdale, NJ: Erlbaum. Ekman, P., Friesen, W.V., & Simons, R.C. (1985). Is the startle reaction an emotion’? Journal of Personality and Social Psychology, 49. 1416-1426. Fridlund, A.J. (1979). Contour-following integrator for dynamic tracking of electromyographic data. Psychophysiology, 16, 491-493. Fridlund. A.J., & Cacioppo, J.T. (1986). Guidelines for human electromyographic research. P