Effect of individual differences in nonverbal expressiveness on transmission of emotion

EFFECT OF I N D I V I D U A L DIFFERENCES IN NONVERBAL EXPRESSIVENESS ON TRANSMISSION OF E M O T I O N Howard S. Friedman Ronald E. Riggio ABSTRACT! This study tested the possibility that individual differences in nonverbal expressiveness may function as a mediating factor in the transmission of emotion through social comparison. In a quasiexperimental design, small groups consisting of one expressive person and two unexpressive people were created in which the participants sat facing each other without talking for two minutes. Self-report measures of mood indicated that tile feelings of the unexpressive people were influenced by the expressive people but the expressive people were relatively unlikely to be influenced by the unexpressive people. The findings have implications for the role of nonverbal communication in the emotional side of group interaction. Certain individuals such as tile " l i f e " of tile party, tile charismatic politician, and tile terrified patient sitting next to you in the dentist's waiting room seem especially able to communicate their feelings to others. Tile presence of such an individual may, on tile one hand, arouse emotion in those nearby and may incite them to join in collective action such as a panic. On tile other hand, such an expressive person may, if calm, provide reassurance or induce relaxation. Feelings may, in part, be affected by tile moods of others through the process of social comparison (Schachter, 1959). In their pioneering work on emotion, Schachter and Singer (1962) started from Maranon's findings that people injected with This research was supported by NIMH Grant #RO3MH31453 and by an Intramural Research Grant from UC Riverside to Howard Friedman. We would like to thank Louise M. Prince and Dan Segall for their assistance and I-liot Smith, Joe Schwartz and Keith Widaman for suggestions. Requests for reprints should be addressed to: Howard S. Friedman, Department of Psychology, University of California, Riverside, CA 92521. 96 Journal oi Nonverbal Behavior 6(2), Winter, 198I 0191-5886/87/1600-0096500.95 @ "1981 Human Sciences Press 97 HOWARD S. FRIEDMAN, RONALD E. RIGGIO adrenalin, but provided with situational cues of "calm," reacted "as if" they felt emotion. Schachter and Singer'went on to argue that emotion involves physiological arousal plus a cognitive label supplied by others. Less well known is the fact that Maran~on also asserted that "emotion is not produced in this way only--that is to say, by the perception within our bodies of tile Visceral trembling--but also by simply seeing in another individual the expressive phenomena, tile gestures of emotion ~' (1950, p. 485). In other words, a given emotion may spread directly from one person to another through emotional expression. However, Maran'on (1950) also noted that this spread of emotion does not always occur but rather "only in favorable conditions of the environment...I could this moment before an audience of students and interested persons imitate the gestures of sadness and neither feel sadness nor make it felt to my hearers. But tonight a great actor may repeat from the stage these same gestures and may sadden the spectators"(p. 486). Part of the difference between tile two situations lies in the expectations of the audience, their willingness to be moved. But a portion also Lies in the "eloquence" of a "great" actor. The present paper takes up this issue, tile effect of individual differences in emotional expressiveness on the transmission of emotion. It seems likeh/that a key element of this charismatic ability to arouse emotion in others involves nonverbal communication. Since the time of Darwin, nonverbal cues such as facial expressions [lave been seen as central to tile transmission of feeling and the addition of force to spoken language (Argyle, 1975; Ekman, Friesen, & Ellsworth, 1972). In the original studies of emotional social comparison (Schachter, 1959; Wrightsman, 1960), social comparison and "homogenization of emotional state" occurred even when subjects were told they could not talk to each other. It has also been shown that social comparison can be affected by gaze (EIIsworth, Friedman, Perlick, & Hoyt, 1978). Thus, a key mediating factor in the transmission of emotion may be individual differences in nonverbal expressiveness. Especially relevant are cues about negative emotions. The original notion of emotional social comparison was constructed around anxiety resulting from uncertainty, and fear is tile primary emotion studied in subsequent social comparison research. Recent discussion makes the likelihood of a tie between unexplained arousal and negative affect quite explicit (Marshall & Zimbardo, 98 JOURNAL OF NONVERBAL BEHAVIOR 1979). Hence, we were primarily interested in the transmission of cues about anxiety, fear, and anger. Recent research has provided conceptual clarification and an operational definition of tile concept of individual differences in nonverbal expressiveness (Friedman, Prince, Riggio, & DiMatteo, 1980). Nonverbally expressive people possess a healthy dramatic flair which is easily recognized by others. They are colorful, entertaining, and more likely to be involved in positions and occupations that involve interacting with and influencing other people. Furthermore, individual differences in expressiveness can be easily and reliably measured with a short self-report measure known as tile Affective Communciation Test (ACT) (Friedman et al., 1980). The ACT has been validated in terms of friends' ratings, social characteristics, personality variables, and nonverbal skills. In our recent report on tile ACT (Friedman et al., 1980), we noted that the tone of sessions in which raters judged videotapes seemed heavily influenced by the mood of the most expressive (high on ACT) person in the group. Further, research in progress suggests that expressive people spontaneously give more intimate greetings and emit more emotions than unexpressives. The present study used tile ACT to study expressiveness and the spread of. emotion. Expressive and unexpressive individuals were paired and their mood was assessed before and after silent interaction. It was expected that, as Mara~'on predicted, expressive individuals would naturally and directly affect the emotions of unexpressives. METHOD Design A quasi-experimental design modified from those described by Schachter (1959) was employed in which mood was assessed at two times--before and after the opportunity for social comparison. Individual differences in nonverbal expressiveness were measured by a pre-test, and small groups were created on tile basis of these scores. Mood was not manipulated but instead was allowed to vary naturally among groups. It was hypothesized that the unexpressive people would be influenced by the mood of the highly expressive people but that the highly expressive people would not be influenced by the unexpressive people. Undergraduate students who had participated in an earlier experiment on nonverbal expressiveness were recruited to a study of "fluctuations in moods" on the basis of their scores on the Affective 99 HOWARD S. FRIEDMAN, RONALD E. RIGGIO Communication Test [Friedman et al., 1980). One highly expressive person participated in each experimental session. However, to control for tile possibility that unexpressive people might be more influenced by anyone (high or low on expressiveness), we included two unexpressive people in each group, three people total. Twenty-seven groups were run, including 27 highly expressive people and 54 unexpressive people. Overall, the 27 highly expressive subjects had a mean ACT score of 90.7 while the 54 unexpressive subjects had a mean ACT score of 58.8, t[79) = 12.6, p < .001 : This mean difference (31.9) represents two standard deviations as defined by tile ACT norm group. When considered within the 27 groups, the expressives were of course also found to be much more expressive than the (average of tile two) unexpressives: matched t(26) = 14.4, p < .001. Procedure Pre-test. Subjects entered tile experimental room, were seated facing away from each other, and independently filled out a mood questionnaire. The directions instructed the subjects to describe how they felt at this instant [on 11-point bipolar scales) and were told that "we want to see |low your moods change during a period of two minutes when you sit silently and let your thoughts wander." Interaction. On cue, subjects were told to turn their cilairs to face the other two people and were reminded that they could look at tile other two people but could not talk to them. Subjects were seated six feet (2 m.) apart. Tile experimenter started a stopwatch and left tile room. He returned after two minutes and gave the subjects a second questionnaire. Post-test. The post-test questionnaire told subjects that "some of your moods probably changed somewhat as a function of your thoughts," and then presented mood scales identical to those filled out initially. Hence subjects were encouraged to believe that it was not odd for their moods to change in such a silort time. However, subjects did not know that this was a study of social influence and did not know that they were selected on the basis of their expressiveness scores. The experimenter did not know which subject was the expressive one. Measures. The first dependent measure on the questionnaire asked subjects, "At this instant, how anxious do you feel?" (11-point scale from "not at all anxious" to "very anxious"). Also of interest were measures of fear and anger. Subjects were also asked how happy and bored they felt but these measures showed no consistent change and are not considered further. 100 JOURNAL OF NONVERBAL BEHAVIOR Predictions Rather than produce one key test statistic, the quasiexperimental design of this study leads to three nonindependent predictions which together bear on tile hypothesis. 1. Since the unexpressive people are presumably being influenced by tile expressive people but the expressive people )lave no other expressives to influence their moods during tile two minutes, tile unexpressive people should change their mood more than tile expressive people. 2. Since tile unexpressives are presumably being influenced by the expressives, tile unexpressive people should finish tile session with a mood more similar to the initial mood of the expressive people than they had at the start. 3. Since the unexpressive people are thought to be moving towards the mood of the expressive people but not vice versa, the discrepancy between the mood of tile unexpressive people at the end of the session and initial mood of the expressive people should be less than tile discrepancy between tile final mood of the expressive people and the initial mood of the unexpressive people. In other words, the unexpressives should come to look more like the expressives did than vice versa. Analyses Testing the three hypotheses using the three dependent variables involved three multivariate analyses of variance. In these analyses, both subjects and groups were considered to be random factors, while the comparison factor, in each instance, was fixed. Since unexpressive subjects began the session with moods that could be either higher or lower than the expressive subject, change could occur in both directions. The following descriptive statistics refer to the initial moods of tile expressive-unexpressive pairs: In 59% of the cases, the high expressive was initially lower on anxiety than was the unexpressive while in 41% of the cases, the expressives were initially higher or equal on anxiety; in 41% of the cases, the high expressive was initially lower on anger; and in 39% of the cases, the high expressive was initially lower on fear. Since unexpressives could become more like expressives by moving either up or down in mood depending on their group, all scores of changes in mood ratings used absolute values. Because groups were a random factor, and since subjects were nested in groups, the tests of the comparison factor of interest in each anova 101 HOWARD S. FRIEDMAN, RONALD E. RIGGIO design used the group by comparison factor interaction as the error term. To test wilether expressives or.unexpressives changed more over time, the absolute values of the differences in the expressives' mood scores over time were compared to the corresponding differences for the unexpressives. To test wtlether the ~nexpressives finished the session with a mood more similar to the initial mood of the expressive people than they had at the start, the absolute values of the differences between tile unexpressives at time two and the expressives at time one were compared to the absolute values of tile differences between the unexpressives at time one and the expressives at time one. To test whether the unexpressives came to look more like the expressives originally looked than the expressives came to look like the unexpressives originally looked, we compared the absolute values of the differences between the unexpressives at time two and the expressives at time one to the absolute values of the differences between the expressives at time two and the unexpressives at time one. Additional Control Condition A potential confounding would arise if ACT scores are naturally correlated with mood stability, tilat is, if unexpressive people are generally more labile. To check this possibility 28 additional people (with ACT scores ranging from 34 to 92) participated in an almost identical procedure (with two mood measures), except that ttley remained alone. The correlations between expressiveness (ACT score) and mood change were .06 for anxiety, .01 for fear, and .08 for anger. Thus, as expected, unexpressives are not generally more changing. It still may be the case that unexpressives are more subject to influence by expressives, a conclusion hinted at by previous research (Friedman et al., 1980). This dimension may be an integral part of tile personality difference between expressives and unexpressives. RESULTS As predicted, the mood scores of the unexpressives changed significantly more than the mood scores of tile expressives, Wilk's Lambda Criterion F(3,24)=3.63, p<.05. Given the short period of time in which the change could occur, it is likely due to an external factor, namely the influence (or lack thereof) of the others. 102 JOURNAL OF NONVERBAL BEHAVIOR Comparing tile initial scores of tile expressives with the initial and final scores of the unexpressives showed a tendency for tile unexpressives to finish the session with a mood more similar to tile initial mood of the expressives than they had at the start, Wilk's Lambda Criterion F(3,24)=3.85, p<.05. Comparing the similarity of tile final scores of tile unexpressives and the initial scores of the expressives to tile similarity of the final scores of the expressives and the initial scores of the unexpressives showed, as predicted, that the unexpressives did indeed come to look more like the expressives than vice versa, Wilk's Lambda Criterion F(3,24)=3.31, p<.05. In sum, there is support for the relevance of nonverbal expressiveness to tile transmission of emotion. When together in a small group, unexpressives change their mood more than expressives and come to look more like the expressives than vice versa. DISCUSSION Although the conditions under which people seek social comparison information have been well-studied, factors affecting one's ability to obtain this information have not. Little is known about tile actual communication process. The present study suggests that one important factor affecting the spread of emotion is the nonverbal emotional expressiveness of the other. In any novel situation, where pressures for social comparison are high, certain individuals may have a disproportionate influence on the group, whether they intend to or not. However, it may also be the case that the unexpressive individuals are more receptive to emotional expressivity in others, that is, unexpressives are more likely to look to others for cues in defining the emotional tone of a particular social situation. Only further research focusing on the particular mediating cues involved can clarify tile precise mechanisms operating in tile transmission of emotion. Nevertheless, we do have indirect evidence supporting tile idea that nonverbal facial expressiveness may be a key factor. In a related study, the expressive facial cues (greater number of head movements, more smiles, and faster rate of speech, together called "facial animation") of a number of students were carefully scored (Riggio, 1981). It was found that charismatic persons (i.e. those who scored high on tile Affective 103 HOWARD S. FRIEDMAN, RONALD E. RIGGIO Communication Test) were more likely to score highly on this facial animation factor than were unexpressives. Since, in the present study, subjects were seated at close range with a clear and direct view of each others' faces, facial cues may indeed have been tile mediating mechanism. Future research in a variety of related topics in social psychology should focus directly on the actual process of social interaction (Friedman, 1979). Tile notion of individual differences in nonverbal communication abilities provides a promising new perspective for viewing face-to-face interaction (Rosenthal, 1979). A focus on tile cues that are sent and received draws attention to the ongoing interchange in face-to-face interaction. Furthermore, cues of emotion such as those transmitted through facial expression can be objectively defined and measured. The how of social comparison (rather than merely tile when and why) can thus be studied. In addition, although skills rather than traits are involved, nonverbal abilities can be viewed as personality variables which are of direct relevance to social interaction. Thus, in addition to its interest as a phenomenon in itself, individual differences in nonverbal expressiveness can be employed as a control or blocking variable in studies of the transmission of emotions. Finally, it is worth noting that tile present analysis does not depend on the validity of the Schachter and Singer (1962) theory of emotion. While it may be the case that people in the present study looked to others for a label for their emotional state, a simpler explanation may be that the spread of emotion occurred through Maraffon's other mechanism--the observation of expressive phenomena. REFERENCES Argyle, M. Bodily communication. New York: International Universities Press; 1975. Ekman, P., Friesen, W.V., & EIIsworth, P. Emotion in the human face: Guidelines for research and integration of findings. New York: Pergamon Press, Inc., 1972. Ellsworth, P., Friedman, H. S., Perlick, D., & Hoyt, M. Effects of direct gaze on subjects motivated to seek or avoid social comparison. Journal of Experimental Social Psychology, 1978, 14, 69-87. Friedman, H. S. The concept of skill in nonverbal communication: Implications for understanding social interaction. In R. Rosenthal (Ed.) Skill in nonverbal communication. Cambridge, MA: Oelgeschlager, Gunn, & Hain, 1979. Friedman, H. S., Prince, L. M., Riggio, R. E., & DiMatteo, R. Understanding and assessing nonverbal expressiveness: The Affective Communication Test. Journal of Personality and Social Psychology, 1980, 39, 333-351. 104 JOURNAL OF NONVERBAL BEHAVIOR Maranon, G. The psychology of gesture. Journal of Nervous and Mental Disease, 1950, 112, 469-497. Marshall, G.D., & Zimbardo, P. G. Affective consequences of inadequately explained physiological arousal. Journal of Personality and Social Psychology, 1979, 37, 970-988. Riggio, R. E. Verbal and nonverbal cues of deception. Doctoral Dissertation. University of California, Riverside, 1981. Rosenthal, R. (Ed) Skill in nonverbal communication. Cambridge, MA: Oelgeschlager, Gunn, & Hain, 1979. Schachter, S. The psychology of affiliation: Experimental studies of the source of gregariousness. Stanford: Stanford University Press, 1959. Schachter, S., & Singer, J. Cognitive, social and physiological determinants of emotion. Psychological Review, 1962, 69, 379-399. Wrightsman, L.S. Effects of waiting with others on changes in level of felt anxiety. Journal of Abnormal and Social Psychology, 1960, 61, 216-222.