Socially Induced Affect
Socially induced affect refers to an emotional experience in one person that is induced by someone else's affect, that person's observable emotions or feelings. This definition implies two partiesa person directly showing affect (the model) and a person observing the model and experiencing emotion as a consequence of the affect of the model. For example, a soldier's distress due to the loss of a loved one induces feelings of distress in his or her team unit members.1 In this case, the soldier is the model and the team members are the observers.
The transfer of feelings from model to observer is incidental in the sense that it is caused not by an intended action of a person, but only by the presence of the other.2Identified originally by researchers working on problems of social facilitation (e.g., Zajonc, 1965), this transmission of affect from one person to another does not depend on the relationship that may exist between them; it occurs between strangers as well as between friends. Results from a large number of experiments document the phenomenon of socially induced affect.3
Socially induced affect has long been an important topic in psychology. (For historical accounts see Gladstein, 1984; Deutsch and Madle, 1975; Wispé, 1986; for extended discussions see Gladstein, 1983; Goldstein and Michaels, 1985; Hatfield et al., 1992; Hoffman, 1977; Stotland 1969.) Interest continues today in clinical, developmental, and social psychology (Demos, 1984; Gladstein, 1984; Hatfield et al., 1992). It has been used to explain processes in social learning (Bandura, 1971), helping behavior (Batson and Coke, 1983), the avoidance of people in distress (Berger, 1962), the patient-therapist relation (Freud, 1921/1957), and crowd behavior (Le Bon,
1920/1982). Since affective experience plays an important role in human behavior and performance, it is not surprising that socially induced affect is suggested in such a wide range of social phenomena.
In considering the effects of socially induced affect on performance, however, several points should be kept in mind. First, if an affective state alters performance, than the effects of socially induced affect should be similar to those of direct affect (based on one's own experiences and feelings) with the same characteristics, such as intensity or length of feeling. In these instances, socially induced affect is simply one of several methods that could be used to induce the desired affective state. To the extent that socially induced affect is different than directly induced affectfor example, if it is easier to use in some circumstances or its effects are more subtleit may provide a unique way to influence performance.4Second, in any given situation, there are ways to enhance performance other than by inducing affect. The method used should be the one that produces change in the most efficient, cost-effective, and reliable way. Only by comparing the effects of alternative methods can this judgment be made. Third, when considering the influence of socially induced affect on performance, it should be noted that the meaning of "performance" depends on context: What is the goal of the activity? If the goal is to train technicians, the performance of interest may be average improvement in the speed and accuracy with which each person performs his or her tasks after training. If the goal of an Army unit is to overcome a series of obstacles, performance would be judged at the team level: How well did the team do? Performance is enhanced when the chances of reaching the intended goal are improved. We discuss issues of performance following a review of what is known to date about socially induced affect.
In this chapter, we first summarize what is known about socially induced affect from research. We then discuss alternative mechanisms that may explain the phenomenon, including the role of cognition, classical conditioning, and mimicry. We also consider several basic issues unresolved by the research completed to date, such as the role of culture and the difference between socially induced affect and direct affect. The major section on performance presents examples of the possible role played by socially induced affect in seven applied settings. These examples are intended to suggest fruitful areas for further research as well as offer implications for performance.
CONCORDANT AND DISCORDANT AFFECT
A typical experiment on socially induced affect takes the following form. On entering a laboratory, the subject meets another person who has also agreed to participate in the experiment. The experimenter tells both
people that she is studying physiological changes that occur when a person experiences various types of physical stimuli. The two people draw lots to determine who will experience the stimuli and who will serve as a "control" subject. The control subject (observer) is told that his physiological changes will be assessed while the other person experiences the stimuli. Often, when the other subject plunges his hand into cold water and winces with pain, the observer twinges with pain, and when the other subject smiles with pleasure in response to an unknown stimulus, the observer expresses a more positive mood. When these reactions occur, the subject has experienced socially induced affectthat is, feelings caused by the observable feelings of another person.
Experiments of this type have demonstrated both concordant and discordant affect (Heider, 1958). Concordant induction is the transmission of affect in the same directioneither positive or negativeas in the experiment described above. Discordant induction is the transmission of the opposite affect: for example, when a performer's achievement produces envy in an observer or when a disliked person's failure produces happiness in the observer.5 Four possibilities are then defined by the direction of the model's affect (positive or negative) and the observer's affect (positive or negative).6Our reviews of the research on socially induced affect is organized in terms of the framework of these four possibilities (after Heider, 1958).
A representative sampling of the many studies that have addressed the question of whether affect can be socially induced is shown in Table 10-1. Each study appears in the cell for which it provides evidence.7For example, the study of Haviland and Lelwica (1987) appears in the modelpositive, observer-positive cell since it provides evidence for positive concordant induction. The study by Berger (1962) appears in both the model-negative, observer-positive and model negative, observer-negative cells since it provides evidence for both discordant and concordant induction.
For experiments that use physiological arousal (electrodermal response) as the dependent measure, the direction of observers' responses cannot be established unless another measure (e.g., self-report) is used at the same time; thus, studies that use only arousal are included as evidence for both discordant and concordant induction. Because of this approach, the number of studies in the discordant cells is higher than it would be if concordant induction were considered the default or usual response and a more specific burden of proof were required to demonstrate discordant induction. If the studies based on physiological arousal are removed from the two discordant cells in Table 10-1, the empirical support for the existence of discordant induction diminishes substantially, although it does not disappear. Discordant induction appears to exist, but much more research is needed to determine under what conditions. Although there is evidence for all four types of induction, by far the largest amount of evidence is in the negative-negative cell. It is clear that observing a
TABLE 10-1 Studies on Occurrence of Socially Induced Affect
person's negative affect generates arousal and negative affect in an observer, but there is evidence for each type of induction.
Much of the work establishing that affect in one individual can arouse emotion in another was done in the area of social learning (Bandura, 1971). In social learning theory, a central concern is how the affective response of a model to an object influences the conditioning of a response to that object of a person observing the model. Thirty years ago, Berger (1962) provided evidence that observers physiologically respond when watching models experience pain in a learning situation. He labeled the observer's affective reaction to the model's emotional response vicarious instigation. Berger's article fostered an abundance of research (see Green and Osborne, 1985, for a review). Berger (1962) noted that in investigations of such vicarious conditioning or reinforcement, researchers took it for granted that the unconditioned response of an observer is the same as the emotion experienced by the model who is engaged in a learning task (see also Lindahl, 1977).
The basic paradigm for much of this work was established in Berger's (1962) seminal paper. He compared the electrodermal responses of participants who watched a model that they either believed was experiencing electrical shock or did not believe was experiencing electrical shock. His finding that someone who is observing another person in distress displays electrodermal responses has been replicated a number of times (e.g., Craig, 1968; Craig and Lowery, 1969; Hygge and Ohman, 1976a, 1976b). Electrodermal responses in observers are found not only when the model experiences pain (Craig and Wood, 1969; Vaughan and Lanzetta, 1980, 1981), but also failure (Craig and Weinstein, 1965), reward (Krebs, 1975), and pleasure (Stotland, 1969). Krebs (1975) also found changes in the heart rate of those watching a model experience positive and negative emotions. Although most investigators have used physiological measures of observers' responses, studies using self-reports, facial expressions, or both have also found evidence for both concordant and discordant socially induced affect (Aderman and Unterberger, 1977; Coyne, 1976; Gotlib and Robinson, 1982, Hsee et al., 1990; Hygge, 1978; Marks and Hammen, 1982). Children, too, have been found to experience emotions when watching another person experience emotion (Eisenberg et al., 1988; Zillmann and Cantor, 1977), and there is some evidence that affect can be socially induced in infants (Haviland and Lelwica, 1987).
A number of methods have been used in studies investigating aspects of socially induced affect. Even though each method has weaknesses, each has been successfully used to assess such affect. The outcomes across different dependent measures suggest that socially induced affect is not a phenomenon tied to a specific method. Supportive results are obtained even when direct effects of a model's situation and physical movement are controlled. Although it cannot be fully determined whether observers in many of these studies were experiencing concordant or discordant induction, the investigations have combined physiological and self-report measures that
provide data indicating that much of the emotional arousal is concordant with a model's emotional state.
There is only a small amount of research that specifically examined discordant induction, but a few studies using self-report measures have demonstrated it. Interestingly, there appears to be more evidence of discordant inductionwhen a model experiences positive affect and the observers experienced negative affectthan when the model experiences negative affect (and the observers experience positive affect). Are people more likely to experience or report feeling negative in response to another's positive affect than they are to experience or report feeling positive in response to another's negative affect? More studies of discordant induction would be helpful in better understanding its dynamics. In particular, studies that use facial efference as measured by electromyographs as a dependent measure would be useful: this approach might allow experimenters to obtain direct information on the valence of the observer's reaction and would thus add to the current evidence for discordant induced affect.
It is critical to note that the socially induced affect may not be all of the emotion that an observer feels. Inasmuch as an observer's situation is necessarily different from a model's, a perfect match in emotion cannot be expected. Berger (1962) described a vicariously instigated emotion as the prelude to the observer's own emotional state: the socially induced sadness experienced in response to seeing a friend weep, for example, may lead to a feeling of concern in the observer not felt by the friend. Finally, other affect-inducing circumstances may be occurring simultaneously with a model's emotional response. That there are physiological or experiential differences between individuals does not exclude the possibility that affect in an observer is, at least partially, socially induced.
To understand how socially induced affect influences human performance, one must consider how one person's emotions can induce emotion in another. Little systematic work has directly examined the mechanisms by which affect is socially induced. In the past, conditioning and cognitive explanations of empathic responses have been popular (Allport, 1924; Bandura, 1969; Heider, 1958; Stotland, 1969). More recently, mechanisms that focus on mimicry have been proposed for these phenomena (Hatfield et al., 1992; Vaughan and Lanzetta, 1980, 1981).
Several explanations have put forth the finding that people become emotionally aroused when observing another person experience emotion. In
reviewing those explanations, the critical issue is not whether cognition can cause emotional arousal, but rather the role it plays in the social induction of affect.
Appraisal of Consequences for One's Well-Being
The appraisal of a situation as having consequences for one's wellbeing is a primary source of emotional arousal (Lazarus, 1982, 1984). Thus, one way emotion can be socially induced is if a model's situation or actions indicate a potential change in the observer's well-being. For example, Allport (1924) claimed instigation of fear in another was due to direct fear of the situation. Certainly fear of a performer's situation or stimulus can induce emotional arousal in an observer; fear and anxiety may well be experienced when one comes upon another person getting mugged. Concordant induction would be explained by a model and an observer appraising the situation in the same way (McCosh, 1880); discordant induction would occur if the observer and the model appraised the situation in different ways. (For findings that show that different cognitive appraisals can lead to different emotions see Roseman, 1984; Smith and Ellsworth, 1985). Other examples of self-relevant instigation of affect in an observer include times when one person's anger induces another to be afraid of what the person will do or when one person is afraid because another's fear suggests a danger in their shared environment. In these cases, the implications for an observer's immediate well-being provide a fairly straightforward explanation for how one person's affect elicited affect in another. When another's emotion signals an actual or potential change in one's well-being, emotion is likely to be induced.
One interesting aspect of socially induced affect, however, is that it is sometimes generated in a person who does not appraise the situation as being relevant to his or her well-being. This type of induction is not explainable from the theoretical view that emotions are aroused only when the situation is appraised as being relevant for personal well-being (Lazarus, 1982, 1984; Lazarus and Smith, 1988). Examples of situations in which immediate self-interests are not at stake include the positive feelings associated with being around a happy person, a mother's distress when seeing her daughter's sadness at failing at an athletic completion, a child's pleasure in watching another child being teased to the point of tears, and a person's unhappiness at seeing a disliked coworker experiencing joy.
Several studies strongly suggest that socially induced affect cannot be solely explained by appraisals of implications for well-being. Affect can be socially induced in an observer even when the model is a stranger to the observer and when the observer knows that he or she is not going to experience the same treatment. In studies in which observers are assured that they
will not experience the negative stimulus, they still show socially induced affect (Hygge and Ohman, 1976a; Vaughan and Lanzetta, 1980, 1981; see also Craig and Lowery, 1969). An even more mundane example is the affect induced by a person's reading a novel or watching a play or movie.
Imagining Oneself in Another's Shoes
Bandura (1969) and Stotland (1969) postulated that imagining oneself in another's situation or in similar situations is the reason one responds emotionally when watching another individual experience emotion. Bandura (1969) and Stotland (1969) seemed to propose a strong version of this argumentthat imagination is the sole way affect is socially induced. However, the evidence suggests that a weak version of the argument is more probableimagination is one mechanism.
Stotland (1969) tested his assumption that empathy (concordant affect) is the result of a cognitive or symbolic process involving an observer imagining him- or herself in the model's situation or imagining how the model must be feeling. Subjects who imagined how the model was feeling, or how they would feel in the situation, responded differently to different model or performer states. His most consistent finding was that there were no differences in the physiological responses of observers among conditions during which they viewed the model ostensibly experience pleasure, neutral feelings, or pain when subjects were instructed simply to watch the model's movements. Because this ''watch-him" condition was taken as a noncognitive control, the null results in combination with the differences found while subjects were using their imagination was taken to mean that cognition is the mechanism responsible for vicarious instigation. This conclusion may not be the best one, however. The instructions given to subjects in the "watch-him" condition may have objectified the model. In the "watch-him" condition, the instructions made the model an object of scrutiny; this observational mind-set may have distanced him from the observer and lessened any response (Hoffman, 1977). Furthermore, the model had his back to the observers, so that any noncognitive mechanism that relies on facial expression could not have operated. A lack of response to the model's affect under this condition does not rule out a noncognitive mechanism.
If Stotland's view is taken to mean that certain activities or perceptual attitudes can eliminate socially induced affect, his findings are supportive. His results are then limited to the finding that imagining oneself in an affective situation leads to physiological changes and self-reported emotion congruent with the situation. This suggests only that imagining oneself in the situation of a model may be sufficient to generate the changes found in vicarious-instigation research, not that imagining oneself in the situation is necessary for affect to be socially induced.
Using ideas in Gestalt psychology, the notion that inconsistency or imbalance among cognition is aversive has played an important role in social psychology for decades (Abelson, 1983; Festinger, 1957; Harary, 1983; Heider, 1946, 1958; Zajonc, 1968, 1983). In trying to explain socially induced affect, cognitive consistency theories provide a means by which the effects of the relationship between a model and an observer can be understood as an essential variable in the occurrence of both concordant and discordant cases of the phenomenon. Heider (1958) argued that the preference for balanced cognitive states was a reason for a person who likes another to experience affect concordant with the other's affect and for a person who dislikes another to experience affect discordant with the other's affect.
How might different consistency theories claim affect is socially induced? First, a model and a model's affect may form a cognitive unit; for the situation to be balanced, the observer must have similar sentiments about each of the two elements in the unit (see Heider, 1958; Zajonc, 1968). If Paul feels positive about Maria, and Maria is experiencing positive affect, then Paul must feel positive about Maria's feelings for the situation to be balanced. Similarly, if Paul feels negative about Maria, and Maria is experiencing positive affect, then Paul must feel negative about Maria's feeling for the situation to be balanced. Whatever Paul is feeling about Maria's affect may be thought of as the part of Paul's overall emotional state that is socially induced. The mechanism involves the preference for consistency; an observer generates a positive or negative feeling about the perceived situation in order to maintain balance.
A second way in which cognitive consistency theories explain socially induced affect is in relation to "ought" (Heider, 1958) or the requirements of justice. People like to believe that what occurs is just (Lerner, 1980). Good people should be rewarded (including experiencing positive emotions), and bad people should be punished (including experiencing negative emotions). The notion of preference for "justice" has been the reasoning behind the few studies that have explicitly examined discordant induction (Aderman and Unterberger, 1977; Bramel et al., 1968; Zillmann and Cantor, 1977). If an observer believes the model to be a good person or to have committed a good act, then the model's happiness should be pleasant to the observerit is a balanced situation. Similarly, if an observer believes the model to be a bad person or to have committed a bad act, then the model's happiness will be unpleasant to the observerit is an imbalanced situation.
The "justice" explanation relies on the inherent affective consequence of states of (in)balance or (in)consistency. However, one need not postulate the justice motive or the external "ought" to consider cognitive consistency as a mechanism for socially induced affect. There is also no need for an
observer to form an affective link between him- or herself and the model's affect as is implied by the unit formation process. In consistency theories, any cognitive imbalance is aversive and motivates change; furthermore, the resolution of imbalance is pleasant (Festinger, 1957; Heider, 1946, 1958). Thus, for example, when another person about whom one is negative feels good it causes imbalance, which is aversive. Similarly, when a positive other feels bad, the imbalance generates negative affect. Since both the existence and the restoration of balance is understood to be pleasant, the mere perception of a balanced or imbalanced relation between a model and his or her affect is enough to cause either positive or negative feelings in the observer.
The advantage of this proposed mechanism is that it explains both discordant and concordant induction and indicates how the relationship between an observer and a model would influence the type of induction. Furthermore, balance is a cognitive mechanism that does not rely on appraisals of well-being for the generation of affect. The pleasantness of balance and the unpleasantness of imbalance are simply attributes of the cognitive system. For example, an observer does not need to realize that it is good for his or her enemy to fail; just the juxtapositions of the cognition of "enemy" and "enemy's feeling negative" is enough to cause at least some level of positive affect.
Although cognitive consistency has appeal as an explanatory mechanism for socially induced affect, it is by no means time to abandon all the other discussed mechanisms. First, the consistency drive is only one of many that may cause affective reactions (Abelson, 1983; Zajonc, 1968): the social induction of affect is much too pervasive and reliable a phenomenon to consider it to be dependent on what can be a weak mechanism. In addition, this mechanism can only account for positive or negative changes in an observer's emotional state; anything more complex would no doubt require more cognitive work. Nonetheless, cognitive consistency theories explain how some degree of positive or negative affect is generated in an observer and how the relationship between an observer and a model might influence what is induced.
A simple way by which the affect of others might induce a person's affect is through classical conditioning (Hoffman, 1990). If the emotional behaviors of others are frequently followed by affectively congruent emotional consequences for the observersuch as the pairing of others' smiling with gifts or kind wordsthe emotional expressions of others should become conditioned stimuli capable of eliciting congruent emotional reactions in observers (Allport, 1924; Bandura, 1965).
Englis et al. (1982) also propose that classical conditioning may be responsible for discordant responses. They give the example of a sadistic mother's smile signaling punishment for a child; repeated association of this display of pleasure with the incongruent outcome of punishment might lead to the acquisition of a discordant emotional response to the mother's smile. Their experiments found differences in the facial expressions of observers watching a model experience emotion that depended on whether the observers' rewards and shocks had been symmetrical with the model's smiles and grimaces or had been asymmetrical to the model's expression. These data provide some support for the notion that classical conditioning can alter the pattern of an observer's facial movements produced when a model is either pained or happy. Specifically, if a person's expressions have been predictive of an observer's outcome, the observer's own facial response to the other's expressions is then associated with that contingency.
Although Englis et al. (1982) limit their discussion and data to the instance when a specific individual's display, such as a mother's smile, is paired with incongruent outcomes, such as punishment, the conditioning mechanism might have broader implications. For example, people may learn that expressions of pleasure by disliked others are coupled with aversive outcomes; thus, positive expressions by any disliked person might induce negative affect. Classical conditioning provides a plausible mechanism for how discordant affect might be induced and how such induction might be related to the social relation between the model and the observer. Again, more research will surely yield new insights about this mechanism.
One obvious mechanism of social induction is mimicry of a model (Hatfield et al., 1992; Vaughan and Lanzetta, 1980, 1981). This section reviews the evidence for mimicry, with an emphasis on facial expressions.
Imitation of expressions is a phylogenetically ancient and basic form of intraspecies communication found in many vertebrate species (Brothers, 1990). McDougall (1908) held that a model's pain display is an unconditioned stimulus for mimicry. If mimicry is such a basic phenomenon, then the cognitive resources necessary for mimicry may be minimal. Of course, mimicry may occur as a result of the imaginal processes discussed above; in these cases the mechanisms may work together to create socially induced affect. It is also possible, however, that cognitive representations are not needed to match the movements of another; simply perceiving another's expression may be enough. Thus, mimicry may provide a mechanism for induction that is both noncognitive and does not rely on implications for well-being.
In the small amount of research on motor mimicry, there is support for the notion that people imitate others. Berger and Hadley (1975) examined
whether muscle activity in an observer tends to parallel muscle use in a model. They recorded activity in the arms and lips of observers (using electromyographs (EMG)) who were watching videotapes of arm wrestling and stuttering. They found greater EMG activity in the observers' muscles that corresponded to the muscles being used by the models than in the muscles that did not correspond to the muscles being used by the models.
Evidence for facial mimicry is found in Dimberg's (1982, 1988) studies of facial reactions to facial expressions. He reports that 8-second presentations of slides of posed angry and happy faces elicited facial EMG responses in subjects consistent with the posed expression. Specifically, zygomatic activity (muscle used to pull cheeks back, as in a smile) was higher when subjects viewed a happy versus an angry face, and corrugator activity (muscle used to wrinkle the brows) was elevated when subjects viewed an angry face, and it was decreased when subjects viewed a happy face.
Mimicry of nonstatic faces has been demonstrated in several studies. Vaughan and Lanzetta (1980, 1981) report that a model's facial display of pain instigates congruent facial activation in an observer. Mimicry of positive expressions has been found by Bush et al. (1989), who tested facial reactions of individuals watching a videotaped comedy routine. Subjects saw two target comedy routines, one of which had smiling faces dubbed into the presentation during soundtrack laughter. Half the subjects had been told to inhibit their facial expressions. The half whose expressions were spontaneous displayed greater zygomatic and orbicularis oculi activity (narrows eyes, as when smiling, but also evident in pain grimacesee Englis et al., 1982) during the dubbed segments than during the segments without smiling faces.
There are a number of studies suggesting that mimicry is a mechanism by which affect might be socially induced. Support for the idea that mimicry can lead to affective resonance is found in the psychoanalytic literature (for a review, see Basch, 1983). In addition, because their data suggest that the instigated facial response of a observer is time locked with the model's expression, Vaughan and Lanzetta (1980) consider mimicry a likely mechanism for vicarious instigation, since a mimetic response should occur simultaneously with a model's response. A cognitive strategy such as retrieval of congruent imagery would take longer to initiate and subside.
Research suggests that facial mimicry plays a role in the generation of affect. Although other types of mimicry may relate to socially induced affect, most investigations of mimicry have concentrated on the face, and the best evidence currently available is about the role of the face (see Hatfield et al., 1992).
How might mimicry lead to socially induced affect? The observer's physical imitation of the model may lead to changes in the emotion of the observer. James (1890/1950) suggested that facial action leads to emotion,
rather than emotion leading to facial action. Although this hypothesis is by no means universally accepted, the data tend to support the notion that facial efference can both initiate and modulate emotional experience (for a review, see Adelmann and Zajonc, 1989). Understanding of facial feedback is limited, however: more knowledge of the mechanisms, power, and moderating conditions of facial feedback is needed. Basic questions such as the frequency with which facial action alters affective state need to be addressed before one can understand the role of facial mimicry in socially induced affect.
Evidence that facial action might play a central role in socially induced affect is given by Vaughan and Lanzetta (1981). When they showed to undergraduate observers a videotape of a model displaying periodic pain expressions, they found that observers who were instructed to pose an expression of pain while the model was expressing pain experienced greater changes in skin conductance and heart rates than both observers instructed to inhibit their facial actions and those given no instructions regarding facial action. These data suggest that facial efference similar to that of a model increases the induction of emotion in an observer.
An example of spontaneous mimicry leading to socially induced affect is given by Bush et al. (1989), who evaluated amusement in relation to mimicry of smiling behavior. As described above, they showed videotapes of comedy routines, with smiling faces dubbed into one performance during soundtrack laughter and not into the other. Half the subjects had been told to inhibit their facial activity. Those in the spontaneous condition reported greater amusement during the routine with the dubs than during the nondubbed routine; furthermore, those in the spontaneous condition reported greater amusement during the dubbed routine than those told to inhibit their expressions. These findings support the perspective that mimicry of others' facial expressions can produce affective responses.
In considering its relation to performance, mimicry may combine with cognitive processes to influence choices. The issue is one of interpreting another person's expressions. The attribution of intentions on the basis of nonverbal behavior is a topic that has received considerable attention in the literature (e.g., Ekman and Friesen, 1969; Druckman and Bjork, 1991:Ch.9). As we discuss below, mimicry may be diagnostic of another's interest, attention, or liking for a model. It may also be a source of information used by observers during their "calculations" about responses or strategies.
At present, investigations of socially induced affect have focused on proving that it occurs and some of the mechanism by which it may occur. Few studies have evaluated the amount of influence that socially induced
affect has in people's overall or specific emotional experiences. Until there is better understanding of the frequency with which socially induced affect occurs and how strong it is when it does occur, it will be difficult to know the amount of influence it has in real-life situations. The efficacy with which each of the mechanisms described above induces affect needs to be evaluated and compared with each other and with methods of inducing direct affect. Key characteristics that should be evaluated include intensity of induced affect, duration of change in affect, and individuals' awareness of the changes in affect and their sources.
Also unexamined are factors that might moderate socially induced affect. Affect may be more or less likely to be socially induced in various situations or in different people. Buck (1984), for example, has found that people differ in their ability to send and receive nonverbal information. Some people's emotions may be more visible or "readable" to others; these individuals may tend to induce affect in others more often or more strongly. Those who pick up on nonverbal emotional cues may be more readily, more frequently, or more potently influenced by the emotions of others.
One potential moderating variablethe relationship between model and observerhas received both theoretical (James, 1890/1950; Heider, 1958) and empirical attention. Krebs (1975) found that subjects who believed they were similar to the model reported feeling worse and were more aroused while waiting for the model to receive a punishment than those who believed they were dissimilar. Zillmann and Cantor (1977) found that in conditions in which children reported liking the protagonist of a stimulus film, they also reported feelings similar to his (see also Miller, 1987).
Finally, cultural differences may influence socially induced affect. For example, the mimicry mechanism for socially induced affect may be more prevalent in cultures with norms for relatively more facial movement associated with emotion than in cultures with norms for relatively less movement (see Mesquita and Frijda, 1992). In "collectivist" cultures, the situation of a member of one's group may be perceived as being more directly relevantperhaps even truly self-relevantto the individual than in "individualistic" cultures (see Markus and Kitayama, 1991). There are, of course, other possible cultural influences on socially induced affect and it is important to keep in mind that all the mechanisms discussed in the preceding section above may function differently in different cultural settings. Those mechanisms can provide a basis for theoretically guided research on socially induced affect. One direction would be to design experiments that can distinguish among alternative interpretations or theories for observed effects. Another direction would be to compare effects obtained in different types of situations and cultures. The results could address issues of the extent to which interpretations of socially induced affect are situational or culture-specific.
The next section explores some potential applications or functions of socially induced affect. One should be reminded that others' emotions are only one source of affect; a sadist who happily smiles while hurting his or her victim is not likely to induce positive affect in the victim! Research needs to evaluate the relative strength of socially induced affect in various situations and to explain other routes to socially induced affect that may complement or counter the effects of mimicry.
AFFECT AND PERFORMANCE
As noted above, affect socially induced in individuals may influence their performance in the same ways that direct affect does. We first summarize general findings on the effects of affect on performance in several domains. We then explore areas in which socially induced affect may have a more unique role.
Several studies have found that mood influences performance of certain tasks. Slife and Weaver (1992) report that subjects put in a negative affective state were less skillful than other subjects in such metacognitive tasks as predicting their answers to math problems and rating their performances after attempting to answer the problems, but no differences in actual task performance were reported. Wolff and Gregory (1991) examined specific areas in which negative affect might influence performance. They found no differences in performance on verbal tests between those put in a temporary dysphoric mood and control subjects. They did find a significant decrement in performance by the dysphoric subjects in such visual-motor subtests as block design and object assembly. These losses may have been due to dysphoria's causing motor slowing or interference with perceptual organization. Radenhausen and Anker (1988) examined reasoning performance in depressed and elated subjects. They found a marginally significant difference in solving syllogisms, with those in negative moods doing worse. No differences in perceptual performance was found. Saavedra and Earley (1991) induced positive or negative affective states in subjects and examined the effect on a problem solving task. They report that subjects in the positive affect condition outperformed those in the negative affect condition. In addition, subjects in the positive affect condition were more likely to stay with their goal and task choice in a second round.
These results suggest that various types of skills are influenced by affect, a notable exception being math performance. It appears that, in general, more negative affect causes decrements in performance, and more positive affect facilitates performance. Note that for the most part, the skills examined in these studies are cognitive in nature. There is virtually no evidence on whether affect influences physical performance. It seems plausible that positive or negative affect could reduce reaction time or the
performance of complex motor behaviors, for example, by distracting a performer. Clearly, before the effects of socially induced affect on performance can be established, a better understanding of the general effects of affect on cognitive and motor performance is needed.
There are two notable differences between socially induced and direct affect. First, socially induced affect is likely to play a larger role in interpersonal or social situations. Because socially induced affect might be a natural occurrence in social situationssuch as when a small group is attempting to complete a task or when individuals are being trainedpossible influences of socially induced affect on performance in these types of situations should be considered. Second, socially induced affect may be more diffuse and less tied to the appraisal of a situation than is direct affect. Socially induced affect may be more likely than direct affect to be mistakenly attributed to an aspect of the situation other than the people in it. Affect caused by appraising some aspect of the environment as having implications for one's own well-being has a source already connected to it. Affect induced socially by one of the mechanisms other than appraisal (classical conditioning, cognitive consistency, and mimicry) may not be tied to a specific object and thus may be more easily misassociated with some salient element of the situation. In a similar vein, Schwarz et al. (1991) reported that when people know the causes of their affective state, they are less likely to use it as information in reacting to their environment (e.g., in evaluating an argument). Keeping these factors in mind, we turn to specific ways in which socially induced affect might influence performance.
Results obtained from studies on a number of topics have implications for the relationship between socially induced affect and performance. Although support for a relationship is usually indirect, the evidence identifies seven areas of performance likely to be influenced by socially induced affect: performance of flight cockpit crews; attention; teaching and influencing; negotiating; panic behavior in groups; helping behavior; and group cohesion. Further research will be needed to distinguish specific effects of socially induced affect from other elements in the situation that may contribute to the observed performance, but the review does suggest possible socially induced affect of applications.
Flight Crew Performance
A recent study by Chidester et al. (1990) compared flight crews with two different types of leaders. One type of crew was led by a captain who displayed a positive emotional style characterized by frequent smiles, an optimistic outlook, and a sensitivity to other's feelings. Members of those crews also showed positive attitudes as they went about their work. Another type of crew was led by a captain whose expressive style was nega-
tive. He or she frowned, complained a lot, was irritable, and evinced a pessimistic attitude. Members of those crews, like their captains, appeared negative in interpersonal relations with their colleagues. The two crews differed in other ways. Members of the "negative" crew made more errors in performing their mission than members of the "positive" crew: they missed discovering indications of hydraulic loss, they flew at airspeeds above or below those required for the aircraft's configuration, and they made improper landings or missed an approach.
The differences in performance between the two types of crews may have been due to the differences between the captains in the way they communicated to their crew members. Just as positive feelings can energize a performer, negative feelings can distract him or her. However, other differences between the captains, correlated with expressive styles, may also have contributed to crew performances: for example, the positiveexpressive captains may have delegated tasks more effectively, adopted different standards for evaluating workload, or had higher achievement motivation than the negative-expressive captains. But the direction of causation may have been reversed: negative captains may have been reacting to inept crews while the positive captains may have been responding to efficient crews. Common sense suggests that the effects were reciprocal, with influences from captain to crew and from crew to captain. Without data from more detailed research it is difficult to eliminate these plausible rival hypotheses for the findings.
The Chidester et al. (1990) study, while suggestive, illustrates problems of design that prevent attributing crew performance specifically to socially induced affect. Proper controls for other factors were not instituted: for example, crew members' affect may have been a result of the types of relationships formed between them and their captains. In this way, the study is similar to those summarized in the rest of this section. While clearly relevant, the studies are primarily explorative in the sense of providing bases for better designed experiments that can isolate effects produced by socially induced affect. Such experiments would control for factors correlated with a leader's expressed affect and also assess whether the expressed affect (positive or negative) is transmitted to group members (concordant or discordant) whose performance is evaluated. These kinds of experiments can be designed for each of the topics discussed below.
There may be motivational differences between direct and socially induced affect. For example, negative affect caused by a person's appraising a situation as detrimental to his or her well-being may be more likely to carry with it suggestions for specific action. This situation may be detrimental to
performance in that the tendency for the action motivated by the appraisal and affect may interfere with the performance of the person's task. But for a person experiencing socially induced negative affect, because the person may be less aware of the cause (which is not as self-generated), it may precipitate a search for a cause; the negative affect may be used as information that there is trouble in the environment. (These differences may be less important in the case of positive affect, as there is less motivation associated with positive states.) For example, a person watching a radar screen for possible enemy incursions may be somewhat distracted if he or she is angry at being selected for this duty. A second person sitting next to the first, angry, person may be more distracted, however, if she or he is experiencing negative affect socially induced from the first person. The second person may have to exert effort to both deal with the negative affect and to discover the cause of the negative affect. That person, who is experiencing socially induced affect, might suffer a greater decrement in performanceresponding as quickly as possible to signs of danger from the screenthan the first person.
Mimicry, discussed above as a mechanism for transmitted affect, may also have value as a diagnostic clue to another's interest or attention. Bush et al. (1986) found that supporters of President Reagan spontaneously mimicked his expressions more than did nonsupporters. This finding suggests that liking is indicated by mimicry. Less clear is the mechanism for this relationship. If mimicry indicates the amount of attention paid to the liked or disliked model, then it may be diagnostic of the extent to which observers are involved or interested in the model. An instructor could, for example, monitor the mimetic facial actions of his or her students: indications that they are not longer mimicking might signal that the time has come to review or clarify the material.
Teaching and Influencing
Better performance in teaching and influencing means changing knowledge or attitudes among students or an audience. There are several ways in which socially induced affect might contribute to such an outcome. In general, to the extent that the affective component of the message is integral to it, speakers more able to communicate that component are more effective. For example, an easily mimicked speaker may be better able than a less easily mimicked speaker to make the audience feel what he or she is feeling about the topic. Beyond that, however, socially induced affect in students or an audience can influence the teaching and persuasion processes in two ways: it may change how the observers respond to the message in ways that make learning or persuasion more or less likely; or it may be misattributed to or associated with various components of the situation, which influences the degree of persuasion or learning experienced.
A socially induced affective state might influence the way in which an audience responds to a persuasive message; for a review, see Schwarz et al. (1991). A speaker may induce either positive or negative affect in the audience. Positive moods have been found to increase persuasion (Janis et al. 1965). The reason for this appears to be that people in a positive mood pay less attention to the message, or do not engage in the effort to carefully consider the message; people in negative or neutral moods more systematically process the content of the message, and are thus more likely to detect flaws in it than are those in positive moods (Bless et al., 1990; Kuykendall and Keating, 1990; Mackie and Worth, 1989; Schwarz et al., 1991; Worth and Mackie, 1987). Thus, if simple persuasion is desired, inducing a positive mood in an audience would be beneficial. However, if the arguments in favor of the speaker's case are strong, it might be preferable to induce more negative moods, so that the audience carefully processes the information. Attitude change that results from careful processing is found to be more stable than attitude change in the absence of such processing (Petty and Cacioppo, 1986). In learning situations in which the goal is to increase students' thinking about the content of the message, positive moods may be detrimental.
Socially induced affect might also have an effect through misattribution to some component of a communication. One component is the speaker. For example, liking for a source might be increased if socially induced positive affect is associated with the source. A speaker who smiles a lot might create positive affect among the audience through people's mimicry of him or her. The audience might then associate this increase in positive affect with the source and thus like the speaker more.
This liking for the source could influence persuasion. People tend to be more persuaded by people they like (e.g., Roskos-Ewoldsen and Fazio, 1992). There is evidence, however, that the effect of liking for a speaker is limited. The physical attractiveness of a speaker, which increases liking for a speaker, influences acceptance of emotionally toned messages, but not of rationally toned messages (Pallak et al., 1983). Similarity of source to the targetalso a cause of likingincreases persuasion (Dembroski et al., 1978), but primarily when the topic is a matter of subjective preference rather than objective reality (Goethals and Nelson, 1973). Thus, positive affect induced in an audience may increase the degree to which people are persuaded when the arguments are based on emotion or the topic is not a matter of objective reality.
In terms of enhancing performance, this approach would seem best when nonobjective factors may increase performance. For example, there may be a limited number of objective reasons why an individual should follow a code of conduct. A speaker who is liked may be more likely to persuade individuals to follow the code than one who is disliked. This
could be valuable in an academic situation, for example, in which cheating is detrimental to the goals of the institution and group, but may not be clearly disadvantageous for an individual. A liked speaker may also be more effective under conditions of uncertainty, when facts are not readily available. When there are too few facts to support the preferability of any specific action (e.g., does the group head north or south to get around an obstacle?), a liked leader may be able to persuade his or her subordinates to take the action.
A second component with which socially induced affect might be associated is the topic of the communication. Through socially induced affect, a speaker may be able to influence the observers' attitudes toward specific topics. For example, if an instructor frowns during the discussion of a particular topic, the students may mimic the action and thus experience more negative affect. One consequence of this may be unintended ''learning." For example, if an instructor is describing a new safety procedure to some trainees, the instructor may (unknowingly) frown during the presentation because he or she does not enjoy or approve of the procedure. The induction of this negative affect in the students may cause them to dislike the procedure and may cause them to be less persuaded that it is a good procedure. Unintended teaching and learning may not be all bad, however. The obvious case is a teacher who expresses a positive attitude toward a topic perhaps originally considered negative by a pupil. Concordant affect has been assumed in these examples. The effects may be reversed if discordant affect is induced. For example, if discordant affect occurs when a teacher is talking enthusiastically about a topic, the topic may become associated with negative affect for the students. If, as predicted by the classical conditioning and cognitive consistency mechanisms, discordant affect is induced when an observer dislikes the model, then a disliked teacher may consistently induce the opposite emotions in his or her pupils. This is an example of why it is important to be aware of variables that moderate or interact with socially induced affect.
There is a large research literature on the social-psychological aspects of negotiation and related forms of social interaction (e.g., Druckman, 1977; Pruitt, 1981; Pruitt and Rubin, 1985). Many of the studies deal with the "self-aggravating aspects of negotiations that are relatively independent of the substantive questions involved and that can impede or facilitate negotiations on any issue" (Frank, 1968:192). These aspects consist primarily of the emotional reactions of negotiators to their opponents' offers and proposals. Negative feelings may be the result of the competitive situation, a dislike for the opponent, or the realization that compromise
will be needed if an agreement is to be attained. They can also be induced socially by the expressions of one's opposite number. Although we know much less about this source of affect in negotiation than other sources, there is reason to believe that it may have a strong influence on bargaining behavior, particularly when negotiation is viewed as an "expression game" (Goffman, 1969).
Goffman's expression-game metaphor is discussed by him in the context of strategic interaction. Emphasizing the nonverbal exchanges in negotiation, the expression game consists of alternating moves and countermoves, with negotiators' both managing the impressions they send and processing information they receive. Focusing on interaction processes, expression games are "assessment contests" designed to conceal or convey information; they illustrate the diagnostic value of nonverbal behavior in terms of clues to unarticulated intentions. The "clues" are the essence of the game, not merely factors that influence other forms of behavior (Druckman et al., 1982). Thus, an opponent's expressions are a source of information about his or her intentions, including the judgment that he or she is telling the truth (see the chapter on detecting deception in Druckman and Bjork, 1991). With regard to socially induced affect, inferences about another's intentions may be based on the feelings engendered in oneself by the other's expressions. When negative affect or anxious feelings are induced, a negotiator may be quick to conclude that the opponent is concealing information vital to obtaining an agreement.
A negotiator's feelings can move the process either toward or away from agreement. Research by Carnevale and Isen (1986) showed that when positive affect was induced, few contentious tactics were used, and joint benefits were improved. The positive affect served to offset the competitive feelings aroused by the bargaining task. However, although a positive opponent may be more liked than a negative opponent, the positive feelings may not translate into agreements. Johnson (1971b) found that when a scripted opponent acted in a warm manner toward the negotiating subject, the latter person liked the opponent more but did not reach a better agreement than when faced with a "cold" scripted opponent. The warm opponent served to reinforce the negotiator's feelings about the superiority of his or her positions. The induced good feelings were apparently attributed to the opponent's judgment about his or her position leading, perhaps, to a false estimate of the probability of agreement.
Findings obtained in another study by Johnson (1971a) suggest that a strategy consisting of alternating between negative (acting cold) and positive (acting warm) affect may be effective. Negotiators compromised more and evinced a larger change in attitudes when faced with a scripted opponent who alternated between showing anger and warmth than when faced with opponents who were either angry or warm throughout the interactions.
This finding suggests an advantage in "fine-tuning" one's expressions. Another way to do this is by creating expectations for toughness (showing anger) early and a willingness to compromise (conveying warmth) later in the process. A pattern in the other directiongoing from relatively soft to tough postureshas been shown to produce impasses (Druckman and Bonoma, 1976). Of particular interest here is the effect of induced feelings on bargaining behavior. Although based only on a few studies, the finding coincides with effects obtained in many studies for alternative concession strategies (e.g., Druckman and Harris, 1990). Still to be determined, however, is whether the observed effect of the other's behavior is due primarily to expressions or to bargaining moves. Both are sources of information about intentions. Both are bases for the emotional experience during negotiation. A more precise rendering of the specific role played by socially induced affect in negotiation awaits further research.
Panic Behavior in Groups
Panic is a state that is detrimental to the performance of any task, particularly a group task under potentially dangerous conditions when order is essential to performance. Panic often occurs in group situations, and socially induced affect may well play a role in its occurrence. The affective states of members of a group appear to be related to the likelihood of panic and to survival in a crisis. Panic increases under perception of personal threat (Klein, 1976). Sako and Misumi (1982) found that the perceived possibility of escape from electric shock in a bottle-neck situation was associated with actual escape; the more people thought they would escape, the more they did. Kugihara et al. (1980) found that increased aggression and competition among members of a group decrease the percentage of members of a group that escape a panic situation. Overall, then, if negative affectfearcan be prevented from being passed from one person to another, or if positive affect can be induced in the members of a group, then performancesustained organization or order and greater survivalin panic situations should be improved.
Research on mass hysteria (e.g., outbreaks of psychogenic physical symptoms in a group) suggests that relationships among people is an important determinant of who experiences the hysterical symptoms. An individual is more likely to exhibit symptoms if a friend experiences symptoms (Small et al., 1991; Stahl and Lebedun, 1974; Wong et al., 1982). Like the finding on panic in a group, this finding suggests that the emotional state of one's friends is likely to have a strong influence on one's own emotional state. Within a large group, it may be pockets of friends that panic first. One way to mitigate panic might be to strengthen social ties between leaders trained not to panic and the members of their groups.
There is a fair amount of research that indicates that empathy can cause helping (Batson and Shaw, 1991). When one feels what another in distress is feeling, one may be more likely to help that person, although this may occur simply to alleviate one's own distress. On any group task in which members may possess different levels of skills, a factor that encourages one to help another may facilitate success. If liking another person increases socially induced affect, members of groups that like each other may be more likely to help one another. Another way to increase socially induced affect in groupsand potentially mutual helpingwould be to use classical conditioning. A group member who has learned that the emotions of a fellow group member have implications for his or her own well-being may be more likely to experience socially induced affect and to help the fellow group member.
Socially induced affect can increase a group's cohesion. A particular member's positive affect may produce good feelings in other members. Those induced positive feelings may serve to increase one's liking for other group members. Such increased liking contributes to enhanced cohesion (Braaten, 1991). Moreover, empathy and group cohesion have been shown to be related (Roark and Sharah, 1989).
A number of studies have found that cohesiveness is associated with efficacy and success (Jaffe and Nebenzahl, 1990; Spink, 1990; Williams and Widmeyer, 1991), but some do not (Keyton and Springston, 1990). In a review of 40 years of research, Mudrack (1989) concluded that the effect of group cohesiveness on performance is not clear, and although Evans and Dion (1991) found in a meta-analysis of studies testing the effects of cohesiveness on group performance that the relation appears both stable and positive, they caution against generalizing these findings to real work groups. One problem is the variety of definitions of cohesiveness. To understand how socially induced affect might influence group performance, research needs to establish what components of cohesiveness are influenced by socially induced affect, and what effect these components have on group performance.
The effect of cohesion on group performance depends on the goals in the situation. Zaccaro and Lowe (1988) found that interpersonal cohesiveness, defined as satisfactory relationships with other members of the group, inhibited optimal productivity on an additive group task by causing taskinterfering interactions among group members. However, when groups performed a disjunctive taskdetermining together what items would be needed to survive in the subarctic after a plane crashsuch cohesiveness was asso-
ciated with better performance (Zaccaro and McCoy, 1988). Thus, it may be best to induce positive affect only in groups whose tasks require cooperation and communication and to not induce positive affect or, even in some instances, to induce negative affect for groups performing simple additive tasks. The lack of positive affect, or the induced negative affect, would reduce interpersonal cohesion and interfere with communication among the group's members.
One possible benefit of positive feelings for a group is suggested by work done by Horwitz (1954). He had members of a sorority work together to solve jigsaw puzzles. Some of the puzzles were interrupted by the experimenter. When interrupted, the women were asked to vote on whether to continue the task. At the end of the series of tasks, participants were asked individually to list the puzzles on which they had worked. Horwitz found greater memory for interrupted tasksknown as the Zeigarnik effectfor puzzles on which the individual had not wanted to continue, as long as the group had voted to continue that puzzle. This suggests that the individual had taken on the group's goals: motivation was conveyed from the group to the individual. This socially induced motivation may promote individual performances within a group by increasing motivation for the group's goals in the individual. Similarly, in a longitudinal study of 54 work groups in several private and public organizations, such as engineering project groups and product assembly groups, Greene (1989) found that group acceptance of goals increased productivity.
There is evidence for the transmission of affect from one person to another. The evidence is stronger for the induction of the same affect (concordant socially induced affect) than for the induction of an opposite affect (discordant socially induced affect). When positive feelings are induced, it is more likely that a message will be accepted. Negative feelings lead to less acceptance of the communication due, perhaps, to a more critical evaluation of the message. The strength of socially induced affect, whether positive or negative, may depend on similarity and liking for the source, as well as cultural factors. Less is known about possible explanations for why it occurs. Plausible mechanisms involve cognition, conditioning, and mimicry. Perhaps all these processes are involved: for example, the initial induction may be due to conditioning or mimicry while its later effect on performance is due to cognitive appraisal processes.
A number of interesting implications for performance arise from the research to date:
• Mimicry of another's facial expressions can facilitate persuasion or learning. It facilitates persuasion to the extent that it indicates liking for the
source of a message; it facilitates learning to the extent that it indicates attention to and interest in the communication.
• Negative or anxious feelings conveyed by sources of deception may lead an observer to be suspicious of his or her intentions.
• Bargainers who alternate between negative and positive affect or who express negative affect in the early phases of negotiation and positive affect in the later phases may elicit more compromise from their opponents than bargainers who express either positive or negative affect throughout the negotiation.
• Better response to panic situations may occur in a group when a leader conveys positive affect, especially in a cohesive group.
• By increasing one's liking for a group, socially induced affect may enhance member motivation to perform, in terms of contributing to its goals.
Fewer implications can be drawn from research for the effects of socially induced affect on physical performance or motor skills. Nor has the research clearly distinguished between effects produced by direct and socially induced affect.
Both the basic exploration of socially induced affect and questions about its role in performance continue to raise interesting research issues. These types of questions are related. Greater understanding of the basic processes of socially induced affect can yield insights into the way it operates in realworld settings, especially in relation to performance. Examining how socially induced affect influences performance will, in turn, suggest ideas about the mechanisms responsible for the phenomenon.
1 We use the concept of "induction" rather than "contagion" for several reasons. First, induction refers to an action caused by something else (e.g., the ipecac induced vomiting); contagion implies that something has been transferred from one individual to another (e.g., a contagious virus). Second, with regard to affect, induction indicates that a model's emotion caused emotion in an observer; contagion suggests that a model's emotions were transferred to the observer. Third, the term induction better describes the general phenomenon of interest, as when affect expressed by one person causes affect in another person. Fourth, it does not imply that an observer's response must be identical to that of the performer: thus socially induced affect includes the possibility that an observer's affect is opposite that expressed by the model (discordant affect).
2 Our definition of socially induced affect does not include situation's in which a person's extra-emotional behavior changes the emotions of another person. For example, if an officer tells a subordinate that he or she failed at a particular task, and the subordinate feels a negative emotion as a result of this evaluation, the subordinate's emotions have been influenced by an exchange of information. However, the subordinate's emotions were not influenced by the officer's emotions or their direct behavioral manifestations (e.g., facial expression). Therefore, work examining the effect of an evaluation on a person's self-esteem or motivation, for example, is not directly relevant to the present discussion. What would be relevant to socially induced affect, as defined in this chapter, is an understanding of how the emotions that accompany
the officer's message influence the subordinate's emotional reactions, and how these, in turn, might influence future performance.
3 The phenomenon could also be referred to as socially induced emotion. The literature is not consistent in its use of the terms "emotion" and "affect," which leads to some confusion in measurement. If self-reports are used, the phenomenon being measured is emotion; if facial expressions and other observable physical responses are used, affect is being measured. Physiological measures, used in many experiments, may indicate either emotion or affect. To the extent that they correlate with a person's reported feelings, they indicate emotion. To the extent that they indicate arousal and are manifest in observable expression they reflect affect. The problem consists of distinguishing between whether emotion or affect is indicated by a measure. Although we try to maintain the distinction between affect and emotion noted above, in reporting on the research we use the term the investigators use.
4 of course, this does not mean that socially induced affect would always be superior to directly induced affect. If it is simpler to induce the desired state directly, and there are not apparent advantages in inducing affect through social means, then direct, nonsocial induction would be preferred.
5 A "pure" discordant affect would occur when an entirely uninvolved detached observer experienced the opposite feeling of the performer: for example, a person watching someone on the news. Demonstrations of such effects depend on achieving proper controls for the effects of relationship (of the model and the observer) on the experience of affect. Such controls have not been completely achieved in the experiments completed to date and future experimentation should attempt to achieve insulation of an observer from a relationship with the model.
6 A more complex instance is the induction of a different emotion of the same "valence" (direction), such as when a model's anger causes fear or sadness in the observer. This instance illustrates an induction of similar valence (concordant) but a different "flavor" of that affect (discordant emotion). Although this is an interesting case for future research, it is not treated in this chapter.
7 Our definition of affect as observable emotion does not perfectly match the term as used in some of the research reviewed. For studies that use observers' self-reports as the dependent measure, the phenomenon being measured is emotion, not affect, in our terms, since it is not externally observed or measured.