toward a theory of behavioral contagion1

0 downloads 0 Views 1MB Size Report
Experi- ments dealing with contagion are reviewed and theoretical statements derived from this review. The basic theoretical argument is that the lowering of the ...
Psychological Review 1966, Vol. 73, No. 2, 179-192

TOWARD A THEORY OF BEHAVIORAL CONTAGION1 LADD WHEELER 2 Naval Medical Research Institute, Bethesda, Maryland Behavioral contagion is defined operationally and is contrasted with conformity, imitation, social pressures, and social facilitation. Experiments dealing with contagion are reviewed and theoretical statements derived from this review. The basic theoretical argument is that the lowering of the avoidance gradient in an approach-avoidance conflict is essential to the occurrence of contagion.

Redl (1949), the social scientist most closely associated with the concept of contagion, lists four points as necessary for its occurrence:

equally open display of an entire lack of fear or guilt. Combining these four points does not provide a definition of contagion, as the points more nearly constitute a theory than a definition. When Redl and his associates began systematic research on contagion, they defined the concept as follows:

1. Existence of an acute conflict area within the imitator; strong impulse urge toward fulfillment of a vehement need on one hand, sufficient pressure from ego or superego forces to keep it down on the other. An incident of behavioral contagion, for the 2. High degree of lability of the imi- purposes of the present investigation betator's "personality balance" in the area comes, then, an event in which a recipient's has changed to become "more like" concerned; impulses strong enough to behavior that of the actor or initiator. This change press for release, controls only just has occurred in a social interaction in which strong enough to prevent that release. the actor has not communicated intent to 3. Existence of a similar type of evoke such a change [Polansky, Lippitt, & strong urge toward impulse expres- Redl, 1950, p. 322]. sion along the same line in the initiUnfortunately, the above definition ator ; his urge must lie in the same di- does not adequately distinguish conrection as that of his imitator. tagion from other types of social influ4. Open acting out in favor of im- ence. For example, the Asch situapulse satisfaction by the initiator, with tion fits the definition. 1 Behavioral contagion3 will be deFrom Bureau of Medicine and Surgery, Navy Department, Research Task MR005.12- fined in this paper through the use of 2005.01, Subtask 1. The opinions and state- reduction sentences (Carnap, 1936,

ments contained herein are the private ones of the writer and are not to be construed as official or reflecting the views of the Navy Department or the Naval Service at large. 2 I thank Irving Altaian and Albert Bandura for their comments on earlier versions of this paper. An extensive correspondence with J. Barnard Gilmore and a careful study of his (1965) paper have been most valuable. His paper may be obtained from him at the University of Waterloo.

8 The term "behavioral contagion" is used advisedly. Many psychologists, because of certain aspects in the development of the concept of contagion, associate it with the spread of an emotional climate. While emotions do engender like emotions under certain circumstances, this paper will be concerned with overt behavior which may or may not follow or be accompanied by the spread of emotions.

179

180

LADD WHEELER

1937). One advantage in using reduction sentences to form a definition is that the definition is then "open." Other reduction sentences may be added to the definition as knowledge of the phenomenon increases. One reduction sentence will be used at this time to define contagion. It is labeled below as S (1). S (1). If the set of test conditions Tt exists, then contagion has occurred if and only if Person X performs Behavior N (B N ) where T\ is specified as follows: (a) A set of operations has been performed on Person X which is known to produce instigation toward BN in members of the class to which X belongs; (b) BN exists in the response repertoire of X, and there are no physical restraints or barriers to prevent the performance of B N ; (c) X is not performing B N ; (d) X observes the performance of BN by Person Y.4 S (1) defines contagion completely in terms of observable events and conditions. It is necessary to the next section of this paper to draw certain inferences about the processes underlying S (1). If Person X is instigated to BN, is not physically restrained from performing BN) and is not performing BN, the implication is that X possesses internal restraints against the performance of BN. Thus, there is an intrapersonal conflict between the instigation to perform BN and the internal restraints against performance of BN. In other words, X is experiencing an approach-avoidance conflict. S (1) also implies that X's observation of the performance of BN by Person Y 4 B» is not to be taken as a completely specific behavior, but as a behavior class. For example, aggression is a class of behavior, while the verbal statement, "You idiot!," is simply one member of that class. Thus, if the model says, "You idiot!," and the observer says, "I'll poke you in the nose!," both statements are to be considered as BN,

changes the relative strengths of the approach tendency and the avoidance tendency. The implication is supplied by the fact that X changes his behavior after observing the behavior of Person Y. The discussion of what changes take place in the approach and avoidance tendencies will be deferred until the theoretical section of the paper. An example of the general situation described in S (1) might be appropriate at this point. Every Sunday an individual, X, is frustrated in his quest for sleep by the sound of the power mower of the neighbor across the street, Z. Because of the strong restraints against directing aggression toward a neighbor in our culture, S has confined himself to politely requesting every Sunday that Z wait until afternoon to mow his lawn. One Sunday, as X is approaching Z to make this request, another neighbor, Y, also approaches Z. Y curses Z and threatens him with physical destruction. Without even waiting for Z's reaction to this attack, X also becomes quite aggressive toward Z. In the remainder of this paper, Person X of both the definition S(l) and the example above will be referred to as the observer. Person Y of S (1) and the above example will be referred to as the model. RELATIONSHIP BETWEEN CONTAGION AND OTHER TYPES OF SOCIAL INFLUENCE One may ask why the concept of behavioral contagion is necessary when psychologists already have conformity, imitation, social facilitation, and social pressures (not to mention half a dozen other related terms). Let us briefly examine some of these terms. The term "conformity" has been used so indiscriminately that one must be somewhat arbitrary in defining it. The mass of recent research on con-

BEHAVIORAL CONTAGION formity has been concerned with the Asch effect or modifications thereof. Consequently, the definition of conformity presented here will be of Aschtype conformity. Such a definition might be written as follows: If Line ab is sufficiently longer than Line cd so that Person X has in the past normally given the correct judgment of relative line length when making this judgment in the absence of other individuals, and if Person X is asked to make this judgment immediately after and in the presence of three or more other individuals who have audibly stated that Line cd is longer than Line ab, then Person X is conforming if and only if he states that Line cd is longer than Line ab. This definition could, of course, be written in somewhat more general terms. Asch-type conformity is not restricted to the judgment of line length and has, of course, been extended into opinions. But regardless of what is being judged, it is apparent from the above definition that conflict is produced in Person X by the judgments of other individuals. The conflict is between making the subjectively veridical but socially rejected judgment and making the subjectively counterfactual but socially accepted judgment. According to the definition, Person X must resolve this conflict by himself. That is, the actions of the other individuals create the conflict and do not aid in resolution of the conflict. On the other hand, the reduction sentence, S (1), used to define contagion implies that conflict in Person X exists prior to the appearance of other individuals and that the presence of other individuals contributes to conflict resolution. Thus, there is a definite and major difference between conformity and contagion. An interesting example of the difference between conformity and contagion

181

may be seen in the effect of a nonunanimous majority in the Asch situation. Prior to the appearance of a partner who gives the veridical judgment, the subject (S) is instigated to give the veridical judgment, but experiences internal restraints against doing so because the unanimous majority has given a different judgment. S resolves this conflict created by the majority by giving the nonveridical judgment; the socially created restraints against giving the veridical judgment are too great. A partner who then begins to give the veridical judgment reduces S'& restraints, which is to say that contagion occurs, and 51 ceases to conform to majority opinion. Thus, we have a well-documented situation in which contagion is not only directly opposed to conformity, but appears to be the more powerful of the two phenomena. At this point in the paper, it has been argued that while contagion and conformity both involve conflict, they differ sharply in the role other individuals play in the conflict. A third term which has been of great importance in research on social influence is "pressures toward uniformity" (Festinger, 19S4). What is the relation of this term to contagion and conformity ? In essentials, there is no difference between conformity and pressures toward uniformity. In both cases, the individual possesses an opinion or belief which he then finds to be in conflict with the opinions or beliefs of others. In both cases, the individual must resolve this conflict. Any appearance of differences between the two types of social influence is due to the greater elegance of the pressures toward uniformity formulation and to different research techniques. That is, experiments conducted to investigate pressures toward uniformity have typically involved face-to-face discussions of opinions in which pressures toward

182

LADD WHEELER

uniformity may be evidenced by the individual changing his own opinion (the usual evidence for conformity), by the individual changing the opinions of others (impossible in an Asch-type situation), or by the individual rejecting the others as comparison persons (possibly and likely in an Asch-type situation, but the relevant data are never collected). Another type of social influence is "social facilitation." According to Thorpe (1956, p. 120), social facilitation is said to occur when "the performance of a more or less instinctive pattern of behavior by one member of a species will tend to act as a releaser for the same behavior in another or in others, and so initiate the same lines of action in the whole group." Thorpe is an ethnologist and would probably forgive students of human behavior for insisting that the behavior need not be instinctive. Bandura and Walters (1963, p. 79) provide the following illustration of social facilitation (or in their terms, of the eliciting effect of a model): An obvious eliciting effect may be observed in cases in which an adult, who has lost the idioms and pronunciations of the local dialect of the district in which he was raised, returns for a visit to his home. The original speech and pronunciation patterns, which would take a stranger years to acquire, may be quickly reinstated.

One may easily demonstrate social facilitation by reaching for a cigarette during a conversation. Other smokers will also light cigarettes. Social facilitation was nicely demonstrated by Starch (1911) by having 5"s copy in longhand literary passages written in longhand. Although 5"s were not instructed to do so, they imitated the longhand style in which the passages were written. Starch called this phenomenon "unintentional or unconscious imitation."

In such situations there is no apparent conflict. There would appear to be no definite restraints and no particularly noticeable instigation. By what process then does social facilitation occur? There are several candidate answers which are probably all accurate to some extent. In the first place, it is probably true that merely conceiving of a behavior increases its probability of occurrence and is usually the first step in performing that behavior. If there are no restraints present, observing and thus conceiving of a behavior is likely to lead step by step to performance. We may call this cognitive-behavioral chaining. In the second place, there is probably no behavior in one's repertoire which is entirely without restraints. If there are no other restraints, mere inertia constitutes a restraint. Observing someone else perform a behavior reduces these minor restraints; after all, it did not seem to hurt the other fellow. Third, there are many behaviors which are conditioned responses to the conditioned stimulus of another person performing the behavior. Thus, we frequently yawn in the presence of others who are yawning because all members in the group have approximately the same need for oxygen. After this has happened on a number of occasions, one's personal yawn may be elicited solely by the sight of another person yawning. In short, social facilitation is mediated by several factors, separately or in combination. It is distinguished from conformity, contagion, and social pressures by lack of any marked conflict. What of the by now almost generic term "imitation" ? The work of Miller and Bollard (1941) is, of course, the classic in the field of imitation. The primary argument in their book is that

183

BEHAVIORAL CONTAGION imitation is learned via reward and punishment and that imitation then generalizes across drives, models, and situations. These authors describe two ways of learning imitation: matched-dependent behavior and copying. In both cases, the responses are connected to a cue from an independent subject or model. In both, likewise, the punishment of non-matched responses and the reward of matched responses eventually result in conformity of behavior between the leader and the follower. The essential difference between the two processes is that in matcheddependent behavior the imitator responds only to the cue from the leader, while in copying he responds also to cues of sameness and difference produced by stimulation from his own and the model's responses [p. 159].

After learning to imitate, it presumably makes no difference whether learning took place via matched-dependent behavior or copying. Following established principles of stimulus and response generalization, learned imitation will have more or less of an effect in different situations. Stated somewhat differently, to the extent that Situation A2 is similar to the learning situation, Al (i.e., the observer has the same drive; the discrimi-

native and background stimuli are the same; the model is the same; the effective responses are the same), any imitation learned in Situation Al will manifest itself in Situation A2. It is apparent that if one wishes to define imitation as I have attempted to define other types of social influence, one could not better the definition of English and English (1958), "action that copies the action of another more or less exactly, with or without intent to copy [p. 253]." Thus, imitation may be considered a generic term subsuming contagion, conformity and social pressures, and social facilitation. In order to make the relationships among the several uniformity processes clearer to the reader, Table 1 describes social pressures and conformity, contagion, and social facilitation in terms of initial conditions for the observer, the model's behavior, the hypothetical processes occurring in the observer, and finally the observer's behavior. Table 1 performs the function of distinguishing between the various social influence processes. Another necessary scientific function is to synthesize phenomena, and this function is attempted in Figure 1.

TABLE 1 DYNAMICS OF SELECTED INFLUENCE PROCESSES Stages in influence process

Behavioral contagion

Social pressures and conformity

Social facilitation

Observer's initial conditions

Instigated to BN. In- Instigated to Bp. No No restraints against restraints. BN or Bp. No instigaternal restraints tion to BN or Bp. against BN.

Model's behavior

Model performs BN.

Hypothetical processes

Reduction of model's Creation of restraints Cognitive-behavioral restraints against BN. against Bp. Conflict chaining, CS elicits between BN and Bp. CR, inertia overcome. Fear reduction.

Observer's behavior

Observer performs BN.

Model performs BN.

Model performs BN.

Observer performs BN Observer performs BN. (or rejects model or induces model to perform Bp).

184

LADD WHEELER

along the ordinate from 3; to the zero point be the y intercept. If the two lines so constructed intersect in the contagion area, the x intercept will be greater than the y intercept. If the two lines intersect in the conformity area, the x intercept will be less than the y intercept. If the two constructed lines intersect in the social facilitation area, the x intercept may be greater than, less than, or the same as the y intercept. This is consistent with the earlier representation of social facilitation as being mediated by some combination of reFIG. 1. Three-dimensional model of conformity, contagion, and social facilitation. straint reduction, elicitation of a con(The model performs BN ; Bp is any and all ditioned response, and cognitive-bebehavior or behaviors other than BN. The havioral chaining. Restraint reduction strength of forces in the conflict is the sum would, of course, reduce the combined of the strength of forces in the observer toward BN and toward Bp. Combined strength of forces, while elicitation of strength of forces ranges from 0.0 to C, a conditioned response and cognitivewhere C is the maximum possible in any behavioral chaining would act as addiconflict. The ratio of B N to BP forces in tional forces and would thus increase the observer is directly related to the proba- the combined strength of forces. Furbility that BN will be performed. The case in which the ratio of B N to Bp forces is thermore, social facilitation is shown to greater than 1.0 is not relevant to the involve small forces both before and discussion.) after the behavior of the model. Social facilitation overlaps with both conformThe front of the block in Figure 1 is ity and contagion, thus describing the concerned with the combined strength twilight zone in which these latter two of forces in the observer's conflict both phenomena may be indistinguishable. prior to and after the performance by The third dimension of the block in the model of Behavior N (B N ). Note Figure 1 represents the ratio of BN that for any initial level of forces in the forces (B N is the behavior performed conflict, the forces in contagion must by the model) to BP forces (B P is be to some degree less after the be- any other behavior) prior to the havior of the model, and in conformity model's performance of BN. It is indithey must be to some degree greater cated that for contagion the BN forces after the behavior of the model. This are slightly less than the BP forces, may be verified by drawing anywhere and we know from the face of the block on the face of Figure 1 a line parallel that the sum of the BN and Bp forces to the ordinate. Let the intersection is relatively large prior to the behavior of this line with the abcissa be x, and of the model. On the other hand, the let the distance along the abcissa from forces involved in social facilitation x to the zero point be the x intercept. are approximately equal, and it is Now draw anywhere on the face of shown on the face of the block that the Figure 1 a line parallel to the abcissa. sum of these forces is small; that is, Let the intersection of this line with there is high lability, but no marked the ordinate be y, and let the distance conflict. In the case of conformity

BEHAVIORAL CONTAGION the Bp forces are greater than the BN forces, but the ratio varies with the type of situation. In the easy discrimination kind of situation originally used by Asch (1956), BP forces are far greater than BN forces. When opinions are at issue (cf. Festinger, 1954), the ratio of BN to BP forces is somewhat greater, and the ratio is even greater in the autokinetic situation (Sherif, 1936). This distinction of various conformity situations is, of course, consistent with the well-known fact that conformity varies directly with the ambiguity of the stimuli to be judged. It should be noted that Figure 1 does not contain the situation in which the ratio of BN to BP forces is greater than 1.0 prior to the model's performance of BN. This is because such a ratio would lead the observer to perform BN independent of the model's performance of BN. There would be no social influence. CONTAGION : EMPIRICAL REVIEW AND THEORETICAL IMPLICATIONS Little research relevant to behavioral contagion has been conducted. What research that has been done will be organized according to the major question (s) to which the research contributed. After research relevant to any one question has been described, theoretical implications of the research will be discussed. The central theoretical statement to the following discussion is that behavioral contagion as defined in S (1) is mediated by the lowering of the observer's avoidance gradient in an approach-avoidance conflict. Specificity of Response Matching In an experiment by Grosser, Polansky, and Lippitt (1951), 5s were told that certain unspecified toys in a collection of toys were forbidden. In this

185

way restraints against playing with any of the toys were created. 5s played with the toys more when they observed an experimental confederate playing with the toys than when they did not. In addition, 5s tended to play with toys they had initially rated as most attractive, regardless of which toys the confederate played with. The investigators attributed the successful contagion to a change in the overall perceived safety of the situation. Similar results with older 5s were obtained by Wheeler, Smith, and Murphy (1964). Army recruits were placed in a waiting room containing magazines, darts, and shuffleboard. They were told that they could read magazines while they waited. Thus, mild restraints against playing with the games were created. 5s played with the games more frequently when they observed an experimental confederate playing with one of the games than when the confederate read or simply walked about the room or when the confederate was not in the room. This was true whether or not the quantity of game equipment allowed 5 to play the same game as the confederate. Thus, contagion was demonstrated under conditions that made specific response matching impossible. These experiments suggest that depending upon the initial strength of the approach and avoidance gradients, the observer's behavior may or may not be an exact imitation of the model's behavior. In a situation involving several potential actions directed toward the same end or expressing the same need pattern, observation of the model's performance of Behavior A will reduce the observer's fear of performing A1 as well as A. Depending then upon the relative strengths of the approach gradients for A and A1, the observer may perform A1 as a consequence of witnessing the performance of A.

186

LADD WHEELER

Balancing oj Approach and Avoidance Tendencies Kimbrell and Blake (1958) created two degrees of thirst in 5s. With an experimental confederate, 5s were placed in the vicinity of a water fountain with a sign which read: "Do not use this fountain." The measure of contagion in each of the two thirst conditions was the difference in the percentage of 5s who drank if the confederate did not drink and the percentage who drank if the confederate did drink. Contagion was greater in the Weak Thirst condition because most 5s in the Strong Thirst condition drank regardless of what the confederate did. Freed, Chandler, Blake, and Mouton (1955) obtained somewhat similar results. 5s approaching the main door of a university building encountered a sign urging use of an inconvenient side door. In three separate conditions, the sign implied three different degrees of prohibition. An experimental confederate either did or did not conform to the injunction in each of the conditions. The measure of contagion in each condition was the difference between the percentage of 5s obeying the sign when the confederate conformed and the percentage obeying the sign when the confederate deviated. This difference was greatest when the sign implied the least degree of prohibition. When the sign implied strong prohibition, 5s obeyed the sign whether the confederate did or not. An implication of these experiments is that the probability oj the occurrence of behavioral contagion is greatest when the avoidance gradient is just slightly higher than the approach gradient. If the restraints are barely sufficient to prevent overt expression of the approach tendency, a small reduction in fear will be sufficient to produce the approach behavior. If the re-

straints are far stronger than the approach tendency, massive fear reduction is necessary to produce a change in behavior. If the avoidance gradient is below the approach gradient, the behavior will be performed whether the model is present or not. Consequences to the Model oj his Behavior Bandura, Ross, and Ross (1963b) exposed children to a television program depicting an aggressive model. In one condition, the model was rewarded for his aggression; that is, the model obtained good things for being aggressive. In a second condition, the model was thrashed for this aggression. After viewing the program, children were placed in a room containing numerous toys that could be used to reproduce the model's aggressive response patterns. Children who had observed the rewarded model were more aggressive than children who had observed the punished model. Walters, Leat, and Mezei (1963) placed children in the vicinity of an attractive assortment of toys with which they were forbidden to play. The children were then shown a film of another child playing with the toys. Some of the children saw the child rebuked by his mother for playing with the toys. Others saw the child rewarded through his mother's nurturant interaction. After viewing the film, the children were left alone in the room containing the toys. Children who had seen the model rewarded played with the toys more frequently than children who had seen the model punished. It is apparently not necessary for the observer to know the consequences of the model's behavior, however. Bandura, Ross, and Ross (1963a), Walters and Thomas (1963), and Walters, Thomas, and Acker (1962) found 5s to be more aggressive after watching

BEHAVIORAL CONTAGION an aggressive model who had not been obviously rewarded or punished for aggression. It is probably the case that if the observer expects the model to be punished, the lack of any overt punishment is perceived as rewarding. Walters and Parke (1964) and Walters, Parke, and Cane (1965) found that children violated prohibitions against playing with toys less after having observed a punished model than after having observed a rewarded model or a model who was neither rewarded nor punished. An implication of these studies is that the observer vicariously performs the behavior performed by the model, and the observer vicariously experiences the consequences of the behavior •which are overtly experienced by the model. To the extent then that the model is rewarded or not punished, the observer's avoidance gradient is lowered.

187

the initiator was a well-dressed, highstatus individual than when he was poorly dressed. The results with regard to the status of the model suggests that general characteristics of the model may have the same effect as immediate consequences of the model's behavior. If the observer perceives the model engaging in behavior which has no immediate visible consequences, the observer may estimate the likelihood of the model's subsequent punishment or reward. One basis for such an estimate is the previous success of the model in obtaining rewards and avoiding punishments. Thus, to the extent that the model appears to be or is known to be generally rewarded or not punished for whatever behavior he emits, the observer's avoidance gradient is lowered for whatever behavior the model performs. Characteristics of the Observer

Characteristics of Effective

Models

Polansky et al. (1950) and Lippitt, Polansky, and Rosen (1952) found that children high in prestige were the most effective models for contagion. However, many of the incidents of contagion tabulated and analyzed did not involve restraints, and thus the conclusions from the two studies are not as relevant to the current discussion as they might be. A field experiment on the contagion of aggressive behavior, done within the larger study by Polansky et al. (1950), found effective models to be highly impulsive. This may mean simply that impulsive boys acted sooner than others and thus had more opportunity to become models. Lefkowitz, Blake, and Mouton (1955) investigated the contagion of pedestrians crossing against a red light. Contagion was greater when

Polansky et al. (1950) and Lippitt et al. (1952) found that observers most susceptible to influence through contagion had high prestige. The field experiment on the contagion of aggression, also reported by Polansky et al. (1950), indicated that impulsive children were most likely to be influenced toward aggression through the observation of aggressive models. These findings duplicated exactly those concerning effective models. Smith, Murphy, and Wheeler (1964) found no relationship between conforming in an Asch-type situation and being influenced by a model to violate a prohibition against playing with games available in a waiting room. Moreover, while the usual relationship between conformity and authoritarianism was obtained, there was no relationship between contagion and authoritarianism.

188

LADD WHEELER

There is evidence that susceptibility to contagion is increased by unusual or extreme physiological states. Novakovsky (1924) linked starvation, fatigue, illness, and extreme cold to susceptibility to modeling influences. A number of the incidents mentioned involved behavior that possessed potentially negative consequences. Schachter and Singer (1962) found that an adrenalin injection increased the contagion of anger, particularly if S could not explain his unusual physiological state as a product of the injection. Bandura et al. (1963b) investigated sex differences in observers in an experiment on the contagion of aggression. 5"s watched either a televised nonaggressive model, a model punished for his aggression, or a model rewarded for his aggression. After observing the adult male models, the children were placed in a playroom, and aggressive behavior was observed. Compared to a control condition in which no model was observed, total aggression expressed by the females was lowered by the nonaggressive model and not affected by the other models. In contrast, the total amount of aggression produced by the males was decreased in the model-punished condition and increased in both the model-rewarded condition and the condition in which a nonaggressive model was observed. These results were explained in a reasonable although post hoc manner, and it is not necessary to repeat the explanation here.5 The evidence relevant to characteristics of the observer having been reviewed, no general statement tying the 5 Other studies have shown the sex of the observer to influence contagion. Such results are important to the present discussion in pointing out that socialization creates restraints and dispositions that affect susceptibility to contagion in different ways depending upon what behavior is involved in the contagion.

experiments together is readily apparent. We will let the review stand on its own merits. The only point that might be stressed is that the kind of person most susceptible to influence through contagion does not appear to be anxious, authoritarian, or of low self-esteem. Locus of Restraints It has been argued in this paper that contagion is mediated by restraint reduction or fear reduction. Thus, it may be useful in making predictions concerning contagion to recognize that the locus of restraints may be different from one situation to another. As the locus of restraints changes, what is feared should change. In turn, situational determinants of contagion may change. While there has been no research directed toward this possibility, certain predictions present themselves. If the restraints are group derive*d, the observer will fear rejection by the group for his lack of impulse control. The most effective model should be a solid supporter of the group norms, an individual who is quite unlikely to be rejected. If this individual performs the forbidden action, he is in essence declaring that the group no longer forbids the action, thus reducing the observer's fear. However, if there is too much of a status differential between the model and the observer, the observer may feel that the model is using "idiosyncrasy credits" (Hollander, 1958) which he himself does not possess. In such a situation, fear would not be reduced, and contagion would not occur. To the extent that the observer is attracted to the group and does not have access to other desirable groups, fear of rejection should be strong, and contagion should be less likely to occur. To the extent that other group members express disapproval of the model's action, contagion will not occur. This

BEHAVIORAL CONTAGION effect will be most striking when disapproval is expressed by high-status group members. In general, when the restraints are group derived and a group member performs the forbidden act, the behavior should spread quickly throughout the group. Every member that performs the act reduces the restraints further. In a different situation, the locus of restraints is the observer's superego. The observer fears guilt and regret for engaging in the behavior. Thus, the model's action must somehow reduce the anticipation of guilt by the observer which is possible by making the action seem more "right" and less "wrong." For example, Bandura and Walters (1959) have suggested that adolescents are more likely to engage in sexual intercourse during double dating than at other times. When there are superego restraints, the apparent morality of the model should be of great importance. An adolescent girl is not likely to have her restraints against sexual behavior lowered if the model is either a prostitute or a nun. The prostitute model could not reduce the observer's fear of guilt because of the model's apparent lack of sexual morality. The nun model would be rejected as a false nun. The most effective model should have slightly higher apparent sexual morality than the observer. Another prediction is that to the extent that the model performs the behavior without overt guilt or fear responses, the observer's fear should be reduced. In a still different situation, the locus of restraints is an authority figure. The optimal conditions for eliciting contagion in such a situation are complex. As pointed out earlier, the most effective model is often one who generally appears to escape punishment for his behavior. But the model who fits this description perfectly is often not an effective model because he is not sub-

189

ject to the authoritarian restrictions in the first place. Thus, a speeding police car is not likely to reduce restraints against breaking the speed limit. The most effective model for reducing authoritarian restraints would be one who performed the behavior in the presence of the authority without punishment by the authority. In addition, the model must originally appear to be subject to the same authoritarian restriction as the observer. An experiment by Milgram (1965) provides an example of contagion when the locus of restraints is an authority figure. JTs were instigated to cease shocking another individual, but they were ordered by an authority to continue the shocking. Two models who were apparently under the same orders from the authority refused to continue the experiment and were not punished by the authority for their disobedience. Virtually all 5"s ceased to obey the authority. On the other hand, models who obeyed the authority had no effect at all, as one would predict from the theory of contagion. Unlike the situation in which the locus of restraints is the superego, similarity between the observer and model is probably not relevant. This is because the observer fears punishment from an external source, and any model who performs the behavior without negative consequences will reduce restraints, no matter how dissimilar he may be to the observer. There is no intention to imply that the discussion of group-derived restraints, superego restraints, and authoritarian restraints is exhaustive; there are undoubtedly other common loci of restraints. Nor is there any intention to imply that the usual situation has only one locus of restraints. The intention is merely to suggest the potential predictive value of specifying the locus of restraints.

190

LADD WHEELER

Interpersonal Consequences of Contagion Festinger, Pepitone, and Newcomb (1952) hypothesized that when members of a group are deindividuated, internal restraints are reduced, and previously suppressed behaviors are performed. This in turn leads to satisfaction with the group. Essentially the same hypothesis was suggested by Hoffer (1951). Some support for this hypothesis was provided by Festinger et al. (1952) who found (a) a negative correlation between the frequency of negative statements about parents and the ability to remember which person in the group said what, and (b) a positive correlation between the frequency of negative statements about parents and the desire to have further discussions with the other group members. The first correlation was interpreted as a positive relationship between restraint reduction and deindividuation, while the second correlation was interpreted as a positive relation between restraint reduction and attraction to the group. Singer, Brush, and Lublin (1964) found that in groups of girls discussing pornographic literature, members were more attracted to groups in which obscene words were spoken. This was interpreted as a positive relation between restraint reduction and attraction to the group. Deindividuation is a feeling of anonymity, and one would expect it to exist to the extent that the individual cannot be singled out as the only person performing a behavior. Thus, observation of a model performing a behavior should produce in the observer some feeling of anonymity with regard to his own performance of that behavior. If the model appears to be "getting away with it," the overall result should be a lowering of the ob-

server's restraints against performing the behavior, the restraints being lowered both by the feeling of anonymity and by the knowledge that the behavior will not be punished. Now it is probably true that whenever restraints are reduced, by whatever means, to the point of behavior occurring and going unpunished, there will be satisfaction with the situation. One type of satisfaction with the situation would be "liking" for the model and a knowledge that he and the model have something in common. It has been shown in a series of studies that when an individual perceives some attribute in common between himself and a model, he will tend to perceive himself as having other of the model's attributes (Burnstein, Stotland, & Zander, 1961; Stotland & Dunn, 1962, 1963; Stotland & Hillmer, 1962; Stotland & Patchen, 1961; Stotland, Zander, & Natsoulas, 1961). From these considerations, we may predict that after the occurrence of contagion in a situation in which the observer is not punished for his behavior, the observer will like the model more than previously and will feel more similar to him on a variety of dimensions, that is, he will be less individuated from the model, Note that while in the Festinger et al. (1952) formulation deindividuation leads to restrain reduction, the contagion formulation suggests that most of the deindividuation occurs after the restraint reduction. CONCLUDING REMARKS This paper has focused upon examples of contagion which involve socially undesirable behavior, such as aggression, adolescent sexual intercourse, and violation of formal prohibitions. But it is not necessary that contagion be confined to socially undesirable behavior. Experiments have shown that

BEHAVIORAL CONTAGION volunteering behavior is increased by the observation of others volunteering (Rosenbaum & Blake, 1955 ; Schachter & Hall, 1952). The latter authors specifically attributed the effect of volunteering models to the reduction of group-derived restraints. The incident in the Kew Gardens section of New York in which a number of people observed a woman attacked three times and eventually murdered could probably have been prevented by one individual. Had any one individual made a visible attempt to aid the victim, the fear would probably have been reduced in other of the observers, who would have then acted. This paper has dealt with socially undesirable behavior because when the behavior elicited through contagion is socially desirable, it is sometimes difficult to rule out other explanations. For example, there are probably strong pressures upon an individual to volunteer when those around him are volunteering. However, it should be emphasized that theoretically contagion is not limited to socially undesirable behavior. The theory is concerned with all behavior which the observer perceives as possibly bringing negative consequences to the performer. REFERENCES ASCH, S. E. Studies of independence and conformity: A minority of one against a unanimous majority. Psychological Monographs, 1956, 70(9, Whole No. 416). BANDURA, A., Ross, DOROTHEA, & Ross, SHIELA A. Imitation of film-mediated aggressive models. Journal of Abnormal and Social Psychology, 1963, 66,3-11. (a) BANDURA, A., Ross, DOROTHEA, & Ross, SHIELA A. Vicarious reinforcement and imitative learning. Journal of Abnormal and Social Psychology, 1963, 66, 601-607. (b) BANDURA, A., & WALTERS, R. H. Adolescent aggression. New York: Ronald, 1959. BANDURA, A., & WALTERS, R. H. Social learning and personality development. New York: Holt, Rinehart & Winston, 1963.

191

BURNSTEIN, E., STOTLAND, E., & ZANDER, A. Similarity to a model and self-evaluation. Journal of Abnormal and Social Psychology, 1961, 62, 257-264. CARNAP, R. Testability and meaning: I. Philosophy of Science, 1936, 3, 420-471. CARNAP, R. Testability and meaning: II. Philosophy of Science, 1937, 4, 1-40. ENGLISH, H. B., & ENGLISH, AVA C. A comprehensive dictionary of psychological and psychoanalytical terms. New York: Longmans, Green, 1958. FESTINGER, L. A theory of social comparison processes. Human Relations, 1954, 7, 117-140. FESTINGER, L., PEPITONE, A., & NEWCOMB, T. Some consequences of deindividuation in a group. Journal of Abnormal and Social Psychology, 1952, 47, 382-389. FREED, A., CHANDLER, P. J., BLAKE, R. R., & MOUTON, JANE S. Stimulus and background factors in sign violation. Journal of Personality, 1955, 23, 499. GILMORE, J. B. When should we predict imitative behavior; when shall we understand it? Paper read at Society for Research in Child Development, Minneapolis, 1965. GROSSER, D., POLANSKY, N., & LIPPITT, R. A laboratory study of behavioral contagion. Human Relations, 1951, 4, 115-142. HOFFER, E. The true believer. New York: Harper, 1951. HOLLANDER, E. P. Conformity, status, and idiosyncrasy credit. Psychological Review, 1958, 65, 117-127. KIMBRELL, D., & BLAKE, R. R. Motivation factors in the violation of a prohibition. Journal of Abnormal and Social Psychology, 1958, 56, 132-133. LEFKOWITZ, M. M., BLAKE, R. R., & MOUTON, JANE S. Status factors in pedestrian violation of traffic signals. Journal of Abnormal and Social Psychology, 1955, 51, 704-706. LIPPITT, R., POLANSKY, N., & ROSEN, S. The dynamics of power. Human Relations, 1952, 5, 37-64. MILGRAM, S. Liberating effects of group pressure. Journal of Personality and Social Psychology, 1965, 2, 127-134. MILLER, N. E., & DOLLARD, J. Social learning and imitation. New Haven: Yale Univer. Press, 1941. NOVAKOVSKY, S. Arctic or Siberian hysteria. Ecology, 1924, 2(5). POLANSKY, N., LIPPITT, R., & REDL, F. An investigation of behavioral contagion in groups. Human Relations, 1950, 3, 319348.

192

LADD WHEELER

REDL, F. The phenomenon of contagion and "shock effect" in group therapy. In K. R. Eissler (Ed.), Searchlights on delinquency. New York: International Universities Press, 1949. Pp. 315-328. ROSENBAUM, M., & BLAKE, R. R. Volunteering as a function of field structure. Journal of Abnormal and Social Psychology, 1955, 50, 193-196. SCHACHTER, S., & HALL, R. Group-derived restraints and audience persuasion. Human Relations, 1952, S, 397-406. SCHACHTER, S., & SINGER, J. E. Cognitive, social, and physiological determinants of emotional state. Psychological Review, 1962, 69, 379-399. SHERIF, M. The psychology of social norms. New York: Harper, 1936. SINGER, J. E., BRUSH, CLAUDIA A., & LUBLIN, SHIRLEY C. Conformity and deindividuation: Studies in conflict and restraint. Research Bulletin No. 44, June 1964, Pennsylvania State University. SMITH, S., MURPHY, D. B., & WHEELER, L. Relation of intelligence and authoritarianism to behavioral contagion and conformity. Psychological Reports, 1964, 14, 248. STARCH, D. Unconscious imitation in handwriting. Psychological Review, 1911, 18, 223-228. STOTLAND, E., & DUNN, R. E. Identification, "oppositeness," authoritarianism, self-esteem, and birth order. Psychological Monographs, 1962, 76(9, Whole No. 528). STOTLAND, E., & DUNN, R. E. Empathy, self-esteem, and birth-order. Journal of Abnormal and Social Psychology, 1963, 66, 532-540. STOTLAND, E., & HILLMER, M. L., JR. Identification, authoritarian defensiveness, and

self-esteem. Journal of Abnormal and Social Psychology, 1962, 64, 334-342. STOTLAND, E., & PATCHEN, M. Identification and changes in prejudice and in authoritarianism. Journal of Abnormal and Social Psychology, 1961, 62, 265-274. STOTLAND, E., ZANDER, A., & NATSOULAS, T. Generalization of interpersonal similarity. Journal of Abnormal and Social Psychology, 1961, 62, 250-256. THORPE, W. H. Learning and instinct in animals. London: Mathuen, 1956. WALTERS, R. H., LEAT, MARION, & MEZEI, L. Response inhibition and disinhibition through empathetic learning. Canadian Journal of Psychology, 1963, 16, 235-243. WALTERS, R. H., & PARKE, R. D. Influence of response consequences to a social model on resistance to deviation. Journal of Experimental Child Psychology, 1964, 1, 269280. WALTERS, R. H., PARKE, R. D., & CANE, VALERIE A. Timing of punishment and the observation of consequences to others as determinants of response inhibition. Journal of Experimental Child Psychology, 1965, 2, 10-30. WALTERS, R. H., & THOMAS, E. L. Enhancement of punitiveness by visual and audiovisual displays. Canadian Journal of Psychology, 1963, 17, 244-255. WALTERS, R. H., THOMAS, E. L., & ACKER, C. W. Enhancement of punitive behavior by audiovisual displays. Science, 1962, 136, 872-873. WHEELER, L., SMITH, S., & MURPHY, D. B. Behavioral contagion. Psychological Reports, 1964, 16, 159-173. (Received December 1, 1964)