JOURNAL OF EXPERIMENTAL SOCIAL PSYCHOLOGY 13, 155--165 (1977)
Affective Responses to the Emotions of a Protagonist DOLF ZILLMANN
Indiana University AND
JOANNE R. CANTOR University of Wisconsin Received April 14, 1976 This study investigated the merits of various theoretical rationales for predicting viewers' affective responses to the expressed emotions of a protagonist. A film depicting a child undergoing an emotion-inducing experience was produced in six versions and shown to elementary school children. The versions effected a factorial variation in (a) the type of behavior exhibited by the protagonist in the initial sequences (malevolent, neutral, benevolent) and (b) the emotion expressed by the protagonist in the final sequence (euphoria, dysphoria). When the protagonist behaved benevolently or neutrally, the affective responses of viewers were concordant with those of the protagonist, but when he behaved malevolently, viewers' affective responses were discordant with his. The latter finding was seen to be in conflict with predictions based strictly on empathy. The rationale that was considered to account best for the findings was based on the assumption that the observer's affective disposition mediates the tendency to respond concordantly or discordantly to another's emotions.
It is commonly believed that viewers' emotional responses play an important part in their reactions to entertaining communications. The most widely held notion of the nature of these responses seems to be that they are based on an empathetic process--that the viewer, as the result of This investigation was supported in part by Grant GSOC-7205471 from the National Science Foundation to Dolf Zillmann. Requests for reprints should be sent to Dolf Zillmann, Institute for Communication Research, Radio-TV Center, Indiana University, Bloomington, IN 47401. The authors are indebted to Kenneth Day for organizing and managing the smooth running of the experimental sessions. His help in carrying out the statistical analysis of all the data is also greatly appreciated. Thanks are also due to Linda Gordon and Jan Petty for serving as experimenters. The authors would like to express their gratitude to Barry Sapolsky, who directed the shooting of the experimental film, and to Kenneth Day, Robert Cooley, Bruce Herrick, and Benjamin Strout, who collaborated with him on the production. Finally, they would like to thank David Rowland, Principal of the University Elementary School, Bloomington, Indiana, for permitting the study to be conducted at his school. 155 Copydght 9 t977by AcademicPress, Inc. All rightsof reproductionin any formreserved.
ISSN 0022-1031
156
ZILLMANN AND CANTOR
witnessing the emotional state of another person, comes to " s h a r e " that emotional state. Although research has shown that emotional reactions can indeed be induced by witnessing other people undergoing emotionarousing experiences (e.g., Berger, 1962; Stotland, 1969), it has been argued that the observer's emotional reaction may also be brought about by some aspect of the situation other than the witnessed emotional response per se, and that the observer's response may or may not be similar to that of the performer (cf. Berger, 1962). In light of this reasoning, it appears premature to conclude that affective responses to the emotions of a protagonist are always and by necessity based upon "experiential sharing." At the present time, little is known about the mechanisms governing emotional responses to entertaining communications, the specific emotional states induced by exposure to such communications, and the factors that determine the similarity or difference between a protagonist's emotional response and that of the viewer. The present study was designed to investigate the predictive adequacy of various theoretical rationales that seek to account for a viewer's responses to the expressed emotions of a protagonist. An experiment was conducted in which the emotion (euphoria, dysphoria) expressed by a protagonist was factorially varied with the type of behavior (malevolent, neutral, benevolent) he displayed prior to his expression of emotions. The notion of empathy yields obvious predictions of viewers' affective responses. According to this notion, a person takes on the emotional state he observes in another person. McDougall (1923), who termed this process "primitive passive sympathy," argued that it occurs automatically. Bandura (1969)proposed that empathy is a learned phenomenon which may develop through classical conditioning. He reasoned that because the affective expressions of other people are frequently followed by similar emotional consequences for the observer, other people's emotional responses should become conditioned stimuli for similar emotional reactions. By whatever means empathy is held to develop, the notion of empathy leads to the prediction that the viewer should experience affective responses similar to those expressed by the protagonist, independently of the type of behavior the protagonist performs initially. Predictions can also be derived from rationales which are distinct from the notion of empathy. Observers may respond primarily to the depicted emotion-inducing events rather than solely to the protagonist's display of emotions. Perhaps viewers develop predispositions to sanction or to oppose particular outcomes, dependent upon the events that precede them (cf. Zillmann & Bryant, 1975; Zillmann & Cantor, 1976). Observers may, for example, develop notions of "deservingness" based on a protagonist's behavior. Thus, a viewer may be expected to respond positively when a protagonist receives the treatment he is seen to deserve, but to respond
RESPONSES TO THE EMOTIONS OF A PROTAGONIST
157
negatively when the outcome seems unjustified and unfair (cf. Heider, 1958). Applied to the manipulations in the present design, this reasoning leads to the expectation that viewers should respond positively to a benevolent protagonist's euphoria and a malevolent protagonist's dysphoria and respond negatively to the euphoria of a protagonist who has behaved badly and the dysphoria of one who has behaved well. Since a protagonist who has behaved neutrally should not be seen as deserving either a euphoric or a dysphoric outcome, affective responses to both outcomes of a neutral protagonist should tend to be moderately negative. Finally, it may be argued that independently of quasi-formal considerations of justice, people respond to communications on the basis of their affective dispositions toward the protagonists. As James (1890) observed, people seem to enjoy seeing those people about whom they care prosper and to dislike seeing them suffer. Similarly, people seem to respond positively to their enemies' suffering and negatively to their enjoyment (cf. Bramel, Taub, & Blum, 1968; Zillmann & Cantor, 1976). Affective dispositions toward protagonists in communications may develop from witnessing the type of behavior the protagonist performs, but they need not correspond precisely with objective assessments of the " g o o d n e s s " or " b a d n e s s " of the protagonist's behavior. As Aronson (1969) has proposed, for example, a person whose behavior is always above reproach and who appears to have no " h u m a n " weaknesses may lose in attractiveness by making others feel inadequate. It is therefore conceivable that the three-level variation in initial behavior effected in the present design will not produce a corresponding variation in affective dispositions toward the protagonist. Whatever affective dispositions are observed as resulting from the manipulations of initial behavior, the disposition reasoning leads to the expectation that viewers will experience concordant emotions with the protagonist when he is liked and discordant emotions when he is disliked or resented. If neutral affect is held toward a protagonist, no appreciable affective responses to his emotions are to be expected on the basis of this reasoning.
METHOD
Subjects Forty-eight second- and third-grade students from a public elementary school in Bloomington, Indiana, served as subjects. The proportion of males to females (1:1) and second-graders to third-graders (5:3) was constant in all experimental conditions. Within these constraints, subjects were randomly assigned to conditions.
Design Each subject was exposed to one of six versions of a film. The film versions effected a factorial variation in (a) the type of behavior exhibited by the protagonist in the initial
158
ZILLMANN AND CANTOR
sequences (malevolent, neutral, benevolent) and (b) the emotion expressed by the protagonist in the final sequence (euphoria, dysphoria). Responses to the film were assessed during the film in subjects' facial expressions, and after the film in an interview.
Materials Professional production facilities were utilized to create the experimental film. The film versions were transferred to videocassette for use in the experiment. The film depicted a boy, approximately the same age as the subjects, who interacted with other characters and then underwent a concluding, emotion-inducing experience. The variation in the type of behavior enacted by the protagonist was effected in the first 3.5 minutes of the film. The variation in the emotion exhibited by the protagonist was achieved in the final minute of the film. Male volent protagonist. The film opens as the protagonist is walking down the street with a group of boys his age. He behaves aggressively, pushing and shoving one of the boys for no apparent reason. Upon arriving home, he hits and shoves his dog. I n a later scene, be is shown fixing himself a peanut butter sandwich. When his younger brother asks for some, he refuses to share his sandwich and gleefully taunts his brother with the fact that there is no more bread. Later, when his brother asks him to mend his broken airplane, he purposely damages the toy further. Neutral protagonist. The film opens with the same initial scene, but the protagonist is shown to behave neutrally toward his peers, and he does not interact with his dog. Later, when his little brother asks him for a sandwich, he gives him the remaining bread and peanut butter and allows him to make one for himself. When his brother asks him to mend his broken airplane, he tries for a moment, but then reports that it cannot be fixed. Benevolentprotagonist. The film opens with the same scene, but here, the boy is shown in a cheerful mood, interacting with his peers in a friendly manner. Arriving home, he greets his dog warmly, petting and hugging him. Later, when his little brother asks him for something to eat, he gives him half of his sandwich without hesitating. When his brother asks him to mend his toy airplane, he does so immediately and good-naturedly. Protagonist's euphoria. Following the airplane incident, the protagonist is called by his parents to come outside, and they present him with a beautiful new bicycle. He shows delight as he examines his gift, and is last seen happily riding the bike down the street. Protagonist's dysphoria. Following the airplane incident, the protagonist goes outside to ride his bike. While attempting to climb a steep curb, he loses his balance and falls offthe bike. He is last seen crying and grimacing in pain.
Procedure Each subject was tested individually. A female experimenter escorted each subject into the experimental room and seated him or her in front of a television monitor. The subject was told that a movie would be shown on the television set and then questions would be asked about the movie. The experimenter then answered any questions the subject had about the forthcoming procedure. Next, the subject was exposed to one of the six versions of the film. Without the subject's knowledge, the subject's facial expressions during the concluding segment of the film were recorded on videotape, using a hidden camera. After the film, the experimenter interviewed the subject following a fixed schedule of questions. The interview was recorded unobtrusively on audiotape. After the interview, the subject was thanked and escorted back to class.
Apparatus The film was shown on a 22" Setchell-Carlson monitor using a Sony VO- 1600 videocassette recorder. The subject's facial expressions were tape-recorded using a Sony AV-5000
RESPONSES TO THE EMOTIONS OF A PROTAGONIST
159
videotape recorder and a Shibaden camera. The interview was recorded on a Sony TC-60 audiocassette recorder.
Dependent Measures Interview. The experimenter asked the subjects a series of questions about their reactions to the film. To assess appreciation of the entire film, she asked, (1) " H o w did you like the movie?" After the subject gave a spontaneous response to this question, the experimenter asked, (2)" Would you say the movie was yukky, OK, pretty good, or neat?" These responses were given scale values of 1 through 4, respectively.1 Next, perception of emotions was assessed to determine whether the emotions depicted in the film were perceived as manipulated. The experimenter asked, choosing the phrase appropriate to the version the subject had seen, (3) "When the boy (got/fell off) the bike, how do you think he felt?" Then, if the subject had not used the word "happy" or " s a d " in response to this question, the experimenter asked, "Do you think he was happy or sad?" (The order in which these two adjectives were mentioned was systematically rotated from subject to subject.) To obtain a rating-scale measure of the intensity of the perceived emotion, the experimenter then asked, using the emotion named by the subject, (4) "Was he kinda (happy/sad), very (happy/sad), or very very (happy/sad)?" Scale values were coded as - 3 for very very sad through +3 for very very happy. 2 When a subject could not decide between happy and sad, a score of zero was assigned. To determine the subject's affective response to the concluding incident, the experimenter asked, (5) "How did you feel when the boy (got/fell off) the bike?" Again, if the subject did not use the word "happy" or " s a d " in responding, the experimenter asked, "Were you happy or sad?" (with the order of adjectives again rotated). Once an adjective was chosen, the subject was asked, (6) "Were you kinda (happy/sad), very (happy/sad), or very very (happy/sad)?" Scale values were coded as above. To determine the subject's affective disposition toward the protagonist, the subject was asked, (7) " H o w did you like the boy?" This question was followed up with a forced-choice question assessing the subject's perception of the protagonist's initial behavior: "Was he a nice boy or a bad boy?" (with the order of adjectives rotated). Then, depending on the adjective chosen, the subject was asked, (8) "Was he kinda (nice/bad), very (nice/bad), or very very (nice/bad)?" Scale values ranged from - 3 , for very very bad through +3 for very very nice, with a score of zero being assigned when a subject failed to choose between nice and bad. Appreciation of the concluding incident was assessed in responses to the question (9) " H o w did you like the way the story ended?" Scale values for all responses which were not in the form of a forced choice (Questions 1,3, 5, 7, and 9) were assigned independently by two coders who were blind to the experimental conditions of the subjects. Each response was assigned a value ranging from - 3 through + 3. For Questions 1,7, and 9, - 3 corresponded to "dislike," and +3 corresponded to "like." For Questions 3 and 5, - 3 represented "very bad" and +3, "very good." Videotaped facial expressions. The videotapes of the facial expressions of subjects as they were watching the concluding segment of the film were shown to two judges who were blind to the experimental conditions of the subjects. The judges were instructed to note any indication of euphoric or dysphoric feelings in the facial expressions and to rate each subject's emotional expressions on a scale ranging from -100, corresponding to extremely dysphoric, to + 100, These adjectives have been observed to be frequently used by children of the subjects' age. In connection with an earlier study (Zillmann & Bryant, 1975), five male and five female seven- and eight-year-olds ranked the terms from " w o r s t " to " b e s t . " The interindividual agreement was very high: Kendall's W = 0.964, p < .001. 2 Prior to the experiment, the modifiers "kinda," " v e r y , " and "very very" had been rank-ordered according to strength by five male and five female seven- and eight-year-old coders. The coders were unanimous in ranking them in ascending order from*'kinda" through "very very."
160
ZILLMANN AND CANTOR
corresponding to extremely euphoric. The mean of the ratings of the two judges constituted the datum for analysis for each subject.
RESULTS
Evidence of the Effectiveness of the Manipulations Perception of protagonist's emotions. The analyses of responses to Questions 3 and 4 ( " W h e n the boy . . . . how do you think he felt?") revealed that the protagonist's emotions were clearly recognized by the subjects. The open-ended responses to Question 3 yielded a highly significant main effect of the manipulation of emotions (F(1,42) = 597.58, p < .001). The mean score for the protagonist's emotion was 2.04 in the euphoria condition and - 2 . 0 4 in the dysphoria condition. All other effects in this analysis were trivial. The responses to Question 4 yielded redundant results. The correlation between these two measures was .94. Perception of protagonist's initial behavior. The analysis of the responses to Question 8 ( " W a s he kinda . . . . very . . . . or very very nice/bad?") revealed that the protagonist's initial behavior significantly affected subjects' impressions of him (F(2,42) = 82.17, p < .001). In comparison to when he behaved neutrally (M = 0.81), the protagonist received significantly lower ratings when he behaved malevolently (M = -2.31) and significantly higher ratings when he behaved benevolently (M = 2.25). All other effects in this analysis were negligible. Affective disposition toward protagonist. Although subjects were thus sensitive to the differentiations in the protagonist's initial behavior, their liking of the boy did not correspond precisely with their differentiated impressions of his " n i c e n e s s " or " b a d n e s s . " The correlation between the measure of the protagonist's perceived " n i c e n e s s " and liking for him was .68. In the responses to Question 7 ( " H o w did you like the b o y ? " ) , there was a significant effect of the protagonist's initial behavior (F(2,42) = 26.04, p < .001). The protagonist who had behaved malevolently was disliked (M = -1.62), and both the neutral and benevolent protagonists were liked (M = 0.88 and 1.19, respectively). The malevolent protagonist was liked significantly less (p < .01) than the other two protagonists, but liking for the benevolent and neutral protagonists did not differ significantly (p > . 10). The effects of the protagonist's emotions and the interaction were negligible. Affective Responses to the Concluding Incident Verbal responses. Responses to Questions 5 and 6 ( " H o w did you feel when the boy got/fell off the bike?"), which were highly correlated (r = .81), were analyzed in a multivariate analysis of variance. This analysis yielded a significant main effect of the protagonist's emotion
RESPONSES TO THE EMOTIONS OF
A
PROTAGONIST
161
TABLE 1 AFFECTIVE RESPONSES TO THE CONCLUDING INCIDENT AS EXPRESSED IN THE INTERVIEW
Initial behavior of protagonist Later emotion of protagonist
Malevolent
Neutral
Benevolent
Euphoria Dysphoria
-0.756 0.88b
2.00b - 1.81a
1.62b - 1.56a
Note. Scores could range from -3 (corresponding to felt -very bad") through +3 (corresponding to felt "very good"). Means having different superscripts differ significantly at p < .05 by Newman-Keuls' test. Means having the same superscripts do not differ at p = . 10. (F(2,41) = 20.37, p < .001), a negligible main effect of initial behavior, and a significant interaction b e t w e e n the two independent variables (F(4,82) = 10.68, p < .001). T h e subsequent discriminant analysis on the centroids revealed a significant primary discriminant function (Wilks' h = .26; X2(10) = 59.13,p < .001; percent of trace = 90.1). R e s p o n s e s to Question 5 were mainly responsible for this discrimination. The relative efficacy of this m e a s u r e was 1.24, c o m p a r e d to .49 for Question 6. Higher order discriminant functions p r o v e d insignificant. On the basis of the strong contribution of Question 5 to the obtained differentiation, a univariate analysis of variance was p e r f o r m e d on these responses. The means associated with this analysis are shown in Table 1. The major finding was a significant interaction b e t w e e n the effect of the p r o t a g o n i s t ' s initial behavior and that of his later emotion (F(2,42) = 26.83, p < .001). When the protagonist had b e h a v e d neutrally or benevolently, his euphoria produced positive emotional responses in subjects, and his dysphoria produced negative ones. H o w e v e r , when he had b e h a v e d malevolently, his euphoria p r o d u c e d d y s p h o r i a and his dysphoria p r o d u c e d euphoria. The analysis p e r f o r m e d on this m e a s u r e also revealed a significant main effect of the protagonist's emotion (F(1,42) = 29.18,p < .001). The mean of the three euphoria conditions was 0.96, and that of the d y s p h o r i a conditions - 0 . 8 3 . The main effect of initial b e h a v i o r was trivial. Viewing behavior. The analysis o f the data on the facial responses of subjects as they viewed the final segment of the film yielded no significant effects of the major variables under investigation. Generally speaking, facial expressions of euphoric or dysphoric feelings were very infrequent a m o n g subjects in response to the film. Viewing b e h a v i o r correlated poorly with the other m e a s u r e s , the highest correlation being with the scaled affective r e s p o n s e to the concluding incident (.32). The overall interjudge reliability for the judged facial expressions was c o m p a r a t i v e l y low (r = .75, in contrast to r - .94 for the verbal responses).
162
ZILLMANN AND CANTOR
Appreciation The three verbal measures of appreciation, Questions 1 and 2 ("How did you like the movie?") and Question 9 (' 'How did you like the way the story ended?"), were analyzed in a multivariate analysis of variance. This analysis revealed significant main effects of both independent variables and the interaction (protagonist's emotion: F(3,40) = 12.20, p < .001; initial behavior: F(6,80) = 2.83, p < .05; interaction: F(6,80) = 4.10, p < .01). The multiple discriminant analysis performed on these data yielded a significant primary discriminant function (Wilks' h = .28; X2(15) = 54.88, p < .001; percent of trace -- 86.6). Question 9 was mainly responsible for this discrimination. The relative efficacy of this measure was 1.55, as compared to .05 for both other measures. Higher order functions proved entirely negligible. Because of the strong contribution that Question 9 made to the differentiation, this measure was subjected to a univariate analysis of variance. A significant interaction between initial behavior and later emotion was obtained (F(2,42) = 12.31, p < .001). As can be seen from the interaction pattern shown in Table 2, the dysphoric ending was disliked when the protagonist had behaved neutrally or benevolently, but when the boy had behaved malevolently, the dysphoric ending was appreciated at a level comparable to the euphoric ending. The main effects in this analysis were also significant. The euphoric ending was appreciated more than the dysphoric one (F(1,42)= 38.31, p < .001; euphoria: M = 1.29; dysphoria: M = -0.73). The main effect of initial behavior was associated with F(2,42) = 5.40 (p < .01). The concluding incidents were appreciated significantly more when the protagonist had behaved malevolently (M = 0.94) than when he had behaved benevolently (M = -0.34), with the neutral condition falling in between (M = 0.19) but not differing significantly from either of the other conditions.
DISCUSSION The results of the present investigation show that there are circumstances under which the notion of empathy does not account for viewers' affective responses to the emotions of a protagonist. Although the subjects expressed affective responses similar to those of the protagonist when he had behaved either neutrally or benevolently, they failed to do so when he had behaved malevolently. In an attempt to explain the apparent failure to respond empathetically to the emotions of a malevolent protagonist, it might of course be argued that observers are capable of inhibiting to some extent their empathetic responses and that they are motivated to inhibit such responses when witnessing the emotions of malevolent people. Even if the notion of empathy were so interpreted, however, it would not predict the finding that hedonically opposite
RESPONSES TO THE EMOTIONS OF A PROTAGONIST
163
TABLE 2 APPRECIATION OF THE CONCLUDING INCIDENT AS EXPRESSED IN THE INTERVIEW
Initial behavior of protagonist Later emotion of protagonist
Malevolent
Neutral
Benevolent
Euphoria Dysphoria
0.81b 1.06n
1.62b - 1.25~
1.31b -2.00 ~
Scores could range from -3 ("dislike") through +3 ("like"). Means having different superscripts differ significantly at p < .01 by Newman-Keuls' test. Means having the same superscript do not differ at p = .10. Note.
affective responses were exhibited by observers of the malevolent protagonist. The rationale based on objective considerations of deservingness also does not account for all of the data. Assuming that the perceived " n i c e n e s s " or " b a d n e s s " of a person determines the degree to which he or she is seen to d e s e r v e beneficial or a d v e r s e c o n s e q u e n c e s , the affective responses to the " n e u t r a l " protagonist's emotions remain unexplained. Since neither reward nor punishment would seem to be d e s e r v e d by s o m e o n e perceived as having done nothing b e n e v o l e n t or malevolent, the o u t c o m e s to the neutral protagonist should h a v e produced mildly negative reactions rather than reactions that were c o n c o r d a n t with the protagonist' s emotions. The rationale based on affective dispositions toward the protagonist was the only one that accurately predicted the findings in all experimental conditions. As anticipated, affect toward the protagonist did not c o r r e s p o n d precisely to the b e n e v o l e n c e or m a l e v o l e n c e of his behavior. Either b e c a u s e his b e h a v i o r was perceived as " t o o g o o d " (cf. A r o n s o n , 1969) or for s o m e other reason, such as a decrease in subjects' perceptions of their own similarity to the protagonist (cf. Byrne, 1971), the protagonist was liked equally well when he b e h a v e d neutrally as w h e n he b e h a v e d benevolently. As predicted f r o m the affective-disposition rationale, the observers in these conditions experienced emotions that were similar to those of the protagonist, and the intensity of affect was c o m p a r a b l e in the two conditions. Also in accord with this rationale, the emotions of the malevolent protagonist, who was disliked by the subjects, induced opposite affective responses. The evidence presented in this e x p e r i m e n t supports the proposal that an o b s e r v e r who likes, or has positive sentiment toward a protagonist experiences c o n c o r d a n t emotional states with that protagonist, and an o b s e r v e r who dislikes, or has negative sentiment toward a protagonist
164
ZILLMANNAND CANTOR
experiences discordant emotional states. This is not to say that empathy is never a factor in responses to the emotions of protagonists. As indicated by the absolute values of the means of viewers' affective responses, the intensity of concordant emotions was greater than that of discordant ones. It is conceivable that a basic predisposition to empathize with the emotions of others does exist. If such a predisposition exists, however, it apparently can be overridden or even reversed when a negative disposition is held. The results clearly show that empathetic processes should not be taken for granted and that dispositional factors must be taken into consideration in order to understand affective reactions to the expression of emotions in others. It should be noted that in the present study some predictions were based on responses to the protagonist's emotions per se, and others were based on responses to the circumstances inducing the protagonist's emotions. Nonetheless, these two elements were concomitantly varied in the communication employed. Whereas the differential emphasis upon emotions or emotion-inducing events is meaningful in making predictions, the joint occurrence of the two aspects of the situation must be considered to be highly typical. Normatively positive outcomes generally produce positive emotional responses, and negative outcomes typically evoke negative responses. An attempt could, of course, be made to manipulate emotions and outcomes independently. Experimental stimuli could be produced in which a protagonist's outcomes are shown but his or her emotional reactions are obscured. Similarly, a protagonist's emotions could be depicted, with the cause of the response kept unknown. Presumably, viewers would come to infer the missing element when responding to such stimuli. Attempts could also be made to vary events and emotions factorially, resulting in some cases in inconsistent combinations of emotions and events. It seems likely that viewers would respond with considerable confusion to such stimuli, reinterpreting the hedonic quality of the observed event according to the emotion it had induced, or concluding that the protagonist was emotionally deviant. Future research, of course, can be used to bear out or disconfirm these speculations. Although the present investigation dealt specifically with a viewer's affective responses to the emotions of protagonists, the findings are considered applicable outside the realm of entertaining communications as well. The rationale most strongly supported by the findings and, indeed, all of the rationales advanced, are directly relevant to individuals' reactions to the emotions of people encountered in their daily lives. It seems, then, that affective dispositions should influence the degree to which people co-experience the emotions of others or experience opposite emotions, whether these others are protagonists in communications, public figures known only via the mass media, or friends and acquaintances encountered face-to-face in everyday situations.
RESPONSES TO THE EMOTIONS OF A PROTAGONIST
165
REFERENCES Aronson, E. Some antecedents of interpersonal attraction. In W. J. Arnold & D. Levine (Eds.), Nebraska symposium on motivation. Lincoln, Nebraska: University of Nebraska Press, 1969. Bandura, A. Principles of behavior modification. New York: Holt, Rinehart, & Winston, 1969. Berger, S. M. Conditioning through vicarious instigation. Psychological Review, 1962. 69, 450-466. Bramel, D., Taub, B., & Blum, B. An observer's reaction to the suffering of his enemy. Journal of Personality and Social Psychology, 1968, 8, 384-392. Byrne, D. The attraction paradigm. New York: Academic Press, 1971. Heider, F. The psychology of interpersonal relations. New York: Wiley, 1958. James, W. The principles of psychology. Vol. 1. New York: Henry Holt, 1890. McDougall, W. An introduction to social psychology. Boston: John W. Luce, 1923. Stotland, E. Exploratory investigations of empathy. In L, Berkowitz (Ed.), Advances in experimental social psychology. Vol. 4. New York: Academic Press, 1969. Zillmann, D., & Bryant, J. Viewer's moral sanction of retribution in the appreciation of dramatic presentations. Journal of Experimental Social Psychology, 1975, 11, 572-582. Zillmann, D., & Cantor, J. R. A disposition theory of humor and mirth. In T. Chapman & H. Foot (Eds.), Humor and laughter: Theory, research, and applications. London: Wiley, 1976.