واکنش ها به حالات چهره: اثرات بافت اجتماعی و اضطراب سخنرانی در پاسخ به حالات خنثی، خشم و شادی

Abstract Male and female participants (n=19) high or low in speech fear viewed pictures of faces posed in anger, neutral, and joyful expressions for 8 s each. Zygomaticus major and corrugator supercilii EMG, skin conductance, and heart rate were measured during picture viewing, and subjective ratings were made after each picture. Compared to anger expressions, joy expressions were responded to with greater zygomatic EMG, less corrugator EMG, and greater heart rate and skin conductance. Physiological response to neutral expressions was similar to response to anger expressions . Expressions posed by women were responded to physiologically more negatively than expressions posed by men. More fearful participants exhibited more negative and less positive facial expressions, and skin conductance responses suggesting greater attention when viewing negative expressions. Results suggest that reactions to facial expressions are influenced by social context, and are not simple mimicry.

Introduction When people are exposed to pictures of positive and negative facial expressions, they produce facial expressions that mimic the affective displays they are viewing (Dimberg, 1982). Specifically, viewed anger expressions produce increased corrugator supercilii (“frown”) muscle activity, and joyful expressions produce increased zygomaticus major (“smile”) muscle activity. These expressions are made spontaneously, without instruction to respond, and occur even when participants are told not to respond expressively or are given instructions to, for example, frown at the joyful face (Dimberg et al., 2002). The expressions are also quite rapid, appearing within about a half second following exposure to the stimuli (Dimberg et al., 1998 and Dimberg et al., 2002). These findings and others have been taken as evidence that mimicry of facial expression is automatically controlled and thus independent of other environmental modifiers (Chartrand and Bargh, 1999), though it is open to debate whether the processes controlling these expressions are emotional, social, or some combination (Fridlund, 1991). Though automatically elicited, mimicked facial expressions are influenced by social variables; for example, when one is cooperating with a confederate mimicry occurs, but a competitive situation can evoke counter-mimicry (Lanzetta and Englis, 1989). Similarly, mimicking of a politician’s emotional expressions is greater when one agrees with that person’s political positions (Bourgeois and Hess, 1999 and McHugo et al., 1991), and negative expressions can be elicited in response to positive displays by politicians with whom one disagrees (McHugo et al., 1991). It has recently been argued that mimicry is not only affected by, but is also a cause of, interpersonal empathy (Chartrand and Bargh, 1999). In addition, emotional expressiveness is greatly influenced by the social context in which the expression occurs (Brody and Hall, 1993, Vrana and Rollock, 1998 and Vrana and Rollock, 2002). An expressive display is influenced by the gender, ethnicity, and power status of both the sender and receiver of the message (Kirouac and Hess, 1999). Such contextual effects may be controlled via different neural pathways (Dimberg et al., 2002) and emerge over a different time course (Vrana and Rollock, 1998) than the initial automatic expression. Despite the importance of these contextual variables, they have not been examined systematically in research on responses to facial emotional displays. Indeed, in the majority of these studies the gender and ethnicity of the participants or the persons posing the facial expressions are not reported. Some studies have employed only female participants in the belief that they are more emotionally expressive (e.g., Dimberg, 1997), though gender differences in expressivity are dependent on the particular emotion, task, and context (Brody and Hall, 1993). One study (Dimberg and Lundquist, 1990) that did examine male and female participants’ expressions in response to pictures of men and women posed in happy and angry expressions found that female participants evidenced greater reactivity, but that expressions did not differ depending on the gender of the person posing the angry or happy face. This occurred even though the female stimuli were rated as more intense (e.g., rated as more angry in the anger expressions and more friendly in the happy expressions) than the male stimuli. The current study further examines social contextual variables within Dimberg’s well-validated and productive paradigm for investigating facial reactions to facial expressions by varying and analyzing for gender differences in both participants and the stimulus persons displaying the facial expressions. Social context affects responses to expressive displays in part because context affects one’s interpretation of the expression and its meaning (Hess et al., 1999). In this regard, another area that has not been fully investigated is that of facial reactions to a neutral facial expression. Studies have generally employed anger and joy faces for negative and positive expressions, respectively, and when neutral control conditions have been included these have generally been pictures of flowers, mushrooms, or simple geometric figures rather than neutral faces (e.g., Dimberg and Christmanson, 1991 and Merckelbach et al., 1989). A simple mimicry account of facial actions to facial expressions predicts that no reaction would be elicited by a neutral expression. However, if context and interpretation are involved, one might expect a negative response to a neutral expression given its ambiguous meaning. Alternatively, neutral expressions might be interpreted differently depending on the context. One recent study (Dimberg et al., 2000) examined unconscious reactions to happy, neutral, and angry expressions within 1 s following masked presentation, and found corrugator and zygomaticus responding to neutral expressions to be roughly midway between that found for happy and angry expressions. These results appear to favor a mimicry account; however, masking and the timing of response measurement may not have allowed contextual factors to operate (e.g., Vrana and Rollock, 1998). The current study will examine the effect of neutral expressions by having participants view faces with angry, joyful, and neutral expressions for enough time to allow unconscious perception of and response to the stimuli. The facial expression viewing paradigm has also been employed to study social anxiety. One might predict that, if social (facial) stimuli elicit fear in socially anxious people, they will respond as people do to other fear-relevant stimuli (e.g., Dimberg, 1986) and exhibit a more negative expression to social stimuli than will non-fearful people. On the other hand, one might predict that socially fearful people will be less attentive to external cues because of being self-focused on their own presentation and social performance (Hope et al., 1989), and thus overall will be less appropriately facially responsive to the expressions of others. The results have been mixed thus far. In one study, women reporting public speaking fear and non-fearful women viewed slides of angry and happy facial expressions (Dimberg, 1997). The high fear participants exhibited greater corrugator EMG to the angry expressions than did the low fear participants, and the low fear participants exhibited greater zygomaticus EMG to the happy expressions than did the high fear group. However, another study (Dimberg and Christmanson, 1991) found that, whereas the low fear group showed the expected pattern of facial expressions (e.g., increased zygomaticus EMG to happy faces and increased corrugator EMG to angry faces), the positive and negative facial stimuli did not affect the high fear group’s expressions. The former study defined high and low fear groups based on a median split of unselected participants on a speech fear questionnaire, whereas the latter study chose participants based on extreme scores on the same questionnaire. This seems to give weight to the self-focus hypothesis in that the most fearful participants did not discriminate among facial expressions. However, other methodological differences between the studies preclude direct comparison. The current study will choose extreme groups of men and women based on the same public speaking fear questionnaire as these earlier studies, and examine expressions to men and women posed in angry, neutral, and joyful facial expressions. Autonomic nervous system measures may assist in teasing apart an attentional (e.g., self-focus) versus a fear or aversion explanation: Greater heart rate deceleration indicates greater attention devoted to an interesting, meaningful, or novel visual stimulus, whereas increased heart rate indicates a defensive response to phobic or aversive stimulation (Klorman et al., 1977). Generally, aversive pictures elicit greater heart rate deceleration than positive pictures (Lang et al., 1993), except when the material is truly phobic to the viewer. Thus, greater heart rate deceleration would be expected from social phobics if they were attending more to the social stimuli. Initial data (Dimberg et al., 1986) show that slides of neutral faces elicit greater heart rate deceleration than nonsocial stimuli (mushrooms), however high and low speech fear participants did not differentially respond. The high fear group exhibited greater skin conductance responses (SCRs) to the social than the nonsocial stimuli, whereas the low fear group did not, an indication of greater attention to the social stimuli. Another study (Merckelbach et al., 1989) also found greater SCRs to slides of faces than to nonsocial stimuli, and in addition found increased SCRs to faces with angry compared to happy expressions. Inhibited spontaneous eye blinking, an index of attention, was seen to the angry expressions compared to happy expressions or nonsocial stimuli. However, they did not find SCR or eyeblink differences between social phobics and non-phobics. Thus, there is evidence that social stimuli engage attention more than nonsocial stimulation, but little indication that socially anxious people differ from non-anxious people in engagement with social stimuli.

4. Results 4.1. Facial EMG 4.1.1. Zygomaticus major EMG Zygomaticus EMG differed depending on the type of facial expression being viewed, F(2,25)=14.95, P<0.0007, ε=0.6415. As expected, EMG was greater when viewing joy compared to angry facial expressions; viewing joy expressions also produced greater zygomaticus EMG than viewing neutral expressions, and neutral did not differ from anger. The gender of the person making the facial expression also had a significant effect on the reaction. Overall, participants smiled more at faces of men than women, F(1,12)=29.15, P<0.0002. However, the top panel of Fig. 1 shows that this difference was significant only for joyful and anger expressions, and in the opposite direction for neutral expressions, expression×picture gender F(2,25)=6.69, P<0.05, ε=0.6506. There were no significant effects involving participant gender. Zygomaticus (top panel) and corrugator (bottom panel) EMG response mean (with ... Fig. 1. Zygomaticus (top panel) and corrugator (bottom panel) EMG response mean (with standard error bars) while viewing anger, neutral, and joy expressions posed by male and female models. Figure options Low fear participants smiled more than high fear participants, F(1,13)=4.91, P<0.05. The two fear groups also differed in their zygomaticus response to the emotional expressions, expression×fear group F(2,25)=4.42, P<0.05, ε=0.6835. As can be seen in the top panel of Fig. 2, the low fear group smiled more than the high fear group especially when viewing the joyful and neutral expressions. There was also an expression×picture gender×fear group interaction, F(2,25)=6.98, P<0.05, ε=0.6506. Data for male versus female pictures were compared for each expression separately for each fear group. After the Bonferroni correction the only significant finding was that, for the low fear group only, zygomaticus response was greater to male than female joyful expressions. Zygomaticus (top panel) and corrugator (bottom panel) EMG response mean (with ... Fig. 2. Zygomaticus (top panel) and corrugator (bottom panel) EMG response mean (with standard error bars) while subjects low and high in speech fear view anger, neutral, and joy expressions. Figure options 4.1.2. Corrugator supercilii EMG Corrugator EMG differed depending on the type of facial expression being viewed, F(2,27)=69.6, P<0.0001, ε=0.5943. As expected, EMG was greater when participants were viewing the anger expressions compared to the joy expressions. Unexpectedly, the corrugator response was significantly greater while participants viewed the neutral expressions compared to either the anger or joy expressions. Overall, there was greater corrugator EMG in response to slides of female faces than of male faces, F(1,13)=97.30, P<0.0001. This was significant for all three expressions, but the difference was greatest for anger expressions, expression×picture gender F(2,27)=5.14, P<0.05, ε=0.8305. 2 These data are shown in the bottom panel of Fig. 1. No effects of participant gender were found. The interaction between fear group and facial expression was not significant, F(2,27)=1.89, P=0.19, ε=0.5943, although the interaction between fear group and the linear component of facial expression (anger, neutral, joy) was marginal, F(1,13)=4.01, P<0.07. Post hoc exploration found that the high fear group exhibited significantly greater corrugator EMG in response to the anger expressions than did the low fear group (see Fig. 2). 4.2. Skin conductance response (SC) Participants responded with reductions from baseline while viewing the neutral and anger expressions; in contrast there was a slight increase in SC while viewing joyful expressions that was significantly different from responses during the other two expressions, F(2,29)=21.36, P<0.0001, ε=0.8082. Skin conductance was greater (less decrease) in response to faces of men compared to women, F(1,14)=10.36, P<0.0001. An expression×picture gender interaction, F(2,29)=8.72, P<0.01, ε=0.5399, and follow-up analyses found that the SC response was significantly greater (less decrease) to faces of men compared to women only for neutral expressions. The top panel of Fig. 3 displays these results. There were no effects involving participant gender. Skin conductance (top panel) and heart rate (bottom panel) response mean (with ... Fig. 3. Skin conductance (top panel) and heart rate (bottom panel) response mean (with standard error bars) while viewing anger, neutral, and joy expressions posed by male and female models. Figure options There was less decrease from baseline when viewing faces overall for the low fear group compared to the high fear group, F(1,14)=47.63, P<0.0001. The two groups also responded differently to the different facial expressions, expression×fear group F(2,29)=36.79, P<0.0001, ε=0.8082. The top panel of Fig. 4 shows that SC was significantly greater to joy expressions than anger or neutral expressions for the low fear group, but there was significantly more SC decrease to joy expressions than anger or neutral expressions for the high fear group. An expression×fear group×picture gender interaction, F(2,29)=18.88, P<0.0004, ε=0.5399, found that, for the low fear group, the pattern of SC responses (greater for joy compared to neutral or anger expressions) was the same regardless of whether the expressions were posed by men or women. However, for high fear participants SC was greater to neutral than joy or anger expressions when viewing men, and greater to anger compared to joy expressions, and then joy compared to neutral expressions, when viewing women. Skin conductance (top panel) and heart rate (bottom panel) response mean (with ... Fig. 4. Skin conductance (top panel) and heart rate (bottom panel) response mean (with standard error bars) while subjects low and high in speech fear view anger, neutral, and joy expressions. Figure options 4.3. Heart rate Heart rate (HR) responded marginally differently to the three types of emotional expressions, F(2,25)=4.13, P<0.06, ε=0.5746: although all main effect mean responses represented decreases from baseline, there was greater deceleration to neutral or anger expressions than during joyful expressions. HR decelerated more during pictures of women than men, F(1,12)=36.81, P<0.0001, but this was significant only when viewing joyful expressions, F(2,25)=9.17, P<0.005, ε=0.6656. The bottom panel of Fig. 3 shows these data. There were no effects involving participant gender. Overall, HR decelerated more for participants in the high fear group than the low fear group when viewing the facial expressions, F(1,12)=7.32, P<0.02. However, a fear group × expression interaction, F(2,25)=9.54, P<0.01, ε=0.5746 and follow-ups found that the groups differed only during joyful expressions. These data can be seen in Fig. 4. 4.4. Ratings As expected, the joyful slides, compared to the neutral slides, were rated as more positively valent, eliciting greater feelings of control, and being more preferred as a member of an audience to which one was going to give a speech; in turn the neutral slides were rated more highly on these three dimensions compared to the anger slides (valence: F(2,30)=64.99, ε=0.5457; dominance: F(2,30)=24.20, ε=0.6099; speech: F(2,30)=69.40, ε=0.6269; all P<0.0001). All of the comparisons between categories were significant. Anger slides were rated as the most arousing, with neutral slides rated as least arousing, F(2,30)=6.17, P<0.01, ε=0.8446; only the comparison between neutral and anger was significant. These data are presented in Table 1. Table 1. Rating means (and standard deviations) Rating Slide category Anger Neutral Joy Valence 7.2 (2.3) 10.6 (0.9) 15.2 (2.0) Arousal 11.9 (3.2) 8.1 (1.7) 10.2 (4.2) Control 7.9 (3.6) 11.6 (2.6) 14.1 (3.1) Speech preference 5.2 (2.7) 10.8 (2.0) 14.6 (2.6) Data were collected on a 0–20 scale, with higher ratings reflecting more positive valence, higher arousal, and greater speech preference, respectively. Table options There were two participant gender×picture gender interactions, each showing a preference by male participants for pictures of women. Men reported more positive valence overall to slides of women (M=11.6, S.D.=1.4) than of men (M=10.4, S.D.=1.0) regardless of the facial expression, whereas female participants reported the same mean (M=11.0) for pictures of men and women, interaction F(1,15)=4.84, P<0.05. This resulted as well in a marginal main effect indicating pictures of women were rated more positively than pictures of men, F(1,15)=3.55, P<0.08. A similar pattern was found for the question “How much would you like to give a speech with this person in the audience?” (participant gender×picture gender F(1,15)=8.97, P<0.01; picture gender F(1,15)=5.25, P<0.05). No main effects or interactions were found for fear group, except for a marginal main effect on the “How much would you like to give a speech with this person in the audience?” question, F(1,15)=4.24, P<0.06, indicating that the high speech fear group rated themselves as liking to give a speech less to anyone, regardless of facial expression (high fear: M=9.5, S.D.=1.0; low fear: M=10.9, S.D.=1.6).