Journal oj Personality and Social Psychology 1975, Vol. 32, No. S, 850-856

Empathy and Attribution: Turning Observers Into Actors Dennis T. Regan and Judith Totten Cornell University A laboratory experiment was conducted lo test Jones and Nisbelt's information-processing explanation of the often-observed tendency for individuals (actors) lo provide relatively more situalional and less dispositional causal attributions for their behavior than those provided by observers of the same behavior. According lo this explanation, aspects of the situation are phenomcnologically more salient for actors, whereas characteristics of the actor and his behavior are more salient for observers. To test this explanation, the phenomenological perspective of observers was altered without making available any additional information. Subjects watched a videotape of a getacquainted conversation after instructions either to observe a target conversant or to empathize with her. As predicted, taking the perspective of the target through empathy resulted in attributions that were relatively more situalional and less disposilional than attributions provided by standard observers. The results support Jones and Nisbctl's information-processing explanalion of actor-observer atlribulional differences, and shed additional light on the process of empathy.

Many recent experimental studies have investigated the rule-bound process by which people develop causal explanations of their own and others' behavior. Virtually all of this work emerges from an information-processing view of the attribution process, according to which differences in the causal explanations that people provide for behavior can be tied to differences in the information available for arriving at these inferences. In this context, Kclley's (1967) model of the attribution process has been particularly influential. Likening the attribution process to an informal version of scientific method, Kelley suggested that individuals systematically process information about temporal consistency, distinctivcness across objects, and consensus across individuals in arriving at a causal attribution for an observed effect. More particularly, the greater the consistency, consensus, and distinctiveness associated with an effect, the more likely the individual is to attribute that effect to the situation or entity under investigation. On the other hand, effects associated with high consistency but low consensus and distinctiveness Requests for reprints should be sent to Dennis T. Regan, Department of Psychology, Uris Hall, Cornell University, Ithaca, New York 1.48SO.

are likely to be attributed to the person or individual who experiences them. McArthur (1972) has presented considerable evidence that individuals do respond discriminately to informational variations along these dimensions in making attributions. One of the most consistently observed attributional phenomena is the tendency for individuals (actors) to arrive at relatively more situational and less dispositional or personal attributions for their own behavior, compared with the attributions for the same behavior provided by observers (e.g., Jones, Rock, Shaver, Goethals, & Ward, 1968; McArthur, 1972; Nisbett, Caputo, Legant, & Maracek, 1973). For example, Nisbett et al. (1973) found that actors explaining their choice of a girl friend or college major provided attributions that were relatively more situational (referring to qualities of the girl or the major) and less dispositional (referring to their own personal qualities) than observers explaining the same behavior. In addition, actors were less likely to assume that they would behave in the future in ways similar to their past behavior, and were also less likely to ascribe personality traits to themselves than were observers. ]n discussing these "divergent perceptions of the causes of behavior" held by actors 850

EMPATHY AND ATTRIBUTION and observers, Jones and Nisbett (1971) suggest two types of explanation. In the first place, they point to differences in the information available to actors and observers for arriving at causal explanations for behavior. Relative to the actor, the observer suffers an informational deficit. Most importantly, the observer frequently lacks historical data about the actor's behavior in similar situations. Whereas the observer can often assume that a particular behavior is typical of the actor, and thus make a dispositional attribution, the actor often knows that he has behaved differently in response to other past situations. This knowledge by the actor that his behavior is highly distinctive will often preclude his making the dispositional attributon made by the observer. This analysis is perfectly consistent with the information-processing approach to attribution outlined above. However, Jones and Nisbett suggest an additional, more interesting difference between actors and observers that may underlie the attributional difference typically found. Not only do actors and observers often differ in the information available to them; they also may typically differ in the way they process the very same available information. Because of their divergent perspectives on behavior, visual information about the surrounding situation is especially salient for the actor, whereas information about the actor and the action itself is particularly salient for the observer. The actor's visual receptors are ideally located for perceiving change and nuance in his environment, but are poorly located for monitoring his own behavior. Moreover, because of his concern with responding in a sensitive and adaptive way to changing environmental contingencies, his attention is likely to be directed outward, away from himself. For the observer, the situation is reversed. From a Gestalt perspective, the action itself, involving movement and change, is phenomenologically salient as figure against the relatively stable ground of the surrounding environment. This difference in information processing also biases the actor toward situational and away from dispositional attributions, while having the opposite effect for the observer.

851

Although highly intriguing, this suggested difference in informational salience and information processing is difficult to test. The only direct test reported in the literature was conducted by Storms (1973). Fn his ingenious experiment, Storms attempted to reverse the usual actor-observer attributional difference by artificially reversing the visual perspective adopted by each. Pairs of naive subjects (actors) engaged in a 5-minute get-acquainted conversation while being watched by matched observers. After the conversation, actors (and their matched observers) provided causal attributions for the actors' behavior along four dimensions. Compared with observers, the attributions of these standard actors were significantly more situational and less dispositional, confirming the typical finding in this area. For actors and observers assigned to the experimental condition, on the other hand, the typical or normal visual focus was reversed through the use of videotape. After the conversation, these experimental actors were shown a videotape of the conversation from the usual observer's perspective—that is, a videotape of the actor's own part of the conversation—whereas experimental observers, on the other hand, were shown a videotape from the normal perspective of their matched actor—that is, a videotape of the other actor's part of the conversation. Following this reversal of perspective, Storms found that the usual actor-observer attributional difference was also reversed: Compared with their matched observers, actors now provided causal attributions for their own behavior that were relatively less situational and more dispositional. This experiment clearly demonstrated that visual perspective can affect the causal attributions provided by actors and observers. The results are consistent with the information-processing mechanism suggested by Jones and Nisbett (1971), but, as Storms himself admits, they may also be attributable to differences in the available information created by the videotape manipulation. "When actors or observers saw a videotape of an event from a different point of view, they may have received some totally new information" (Storms, 1973, p. 172). Differences in available information and informational salience

852

DENNIS T. REGAN AND JUDITH TOTTEN

are doubtless often confounded, and it is difficult to separate them experimentally. The present experiment is an attempt to provide additional support for an information-processing or "perspective" mechanism in mediating actor-observer attributional differences, by using a method that alters the perspective or orientation of observers with no change whatever in the visual information presented to them. Tn the present experiment, the perspective of some observers of a get-acquainted conversation was altered by giving them instructions to empathize with one of the participants in the conversation. We hypothesized that the adoption of an empathic set by observers would increase the likelihood that they would provide relatively more situational and less dispositional attributions for the actor's behavior. This should occur because in empathizing with their target actor, observers should be relatively more likely to "take the role of the other" and adopt his phenomenological perspective. Empathizing observers should therefore find aspects of the situation relatively more salient and aspects of the actor's dispositional qualities less salient in attempting to provide causal attributions. Despite being provided with precisely the same visual information as control observers, therefore, empathizing observers should provide attributions for behavior that are more similar to the attributions typically given by actors themselves. The hypothesis was tested in a 2 X 2 "observer-only" factorial design. All subjects watched a videotape of a get-aquainted conversation while seated in individual cubicles. Half the subjects had been instructed to empathize with one of the two female conversants; the others were simply instructed to observe her. Cross-cutting the manipulation of observational instructions, half the subjects watched a standard tape of the conversation in which both participants were visible and audible. The other half watched a tape of the same conversation in which only the target person was visible, although again both participants were audible. This videotape manipulation was included to buttress the argument that any effect of empathic orientation on attributions is at-

tributable to differences in perspective or informational salience (information processing), rather than due to differences in the available information. The distinction between the two is admittedly not perfectly clear-cut, but it could perhaps be argued that empathic instructions simply cause subjects in the two-person videotape condition to look away from the target person and focus visual attention on the other participant. This change in visual focus might produce additional information about the other participant, which would help to produce a "situational" attribution for the target person's behavior. If this were the mechanism whereby the empathy instructions affect attributions, however, we should notice a reduced effect of instructions if the videotape showed only the target actor. With nothing but the target on the screen, empathic observers could not turn their visual focus away from the target and toward the other actor. Accordingly, a significant effect of empathic instructions in the one-person videotape condition would provide further support for an information-processing mechanism underlying actor-observer differences. METHOD Procedure When a subject arrived at the laboratory, she was greeted by the experimenter and ushered to an individual cubicle containing a television monitor. When seated, she was given a page that introduced the experiment as a study of person perception investigating "how people perceive the behavior of others." The subject read thai she would be watching a videotape of a conversation between two students who were meeting for the first time and who had simply been to'.d to get acquainted. The conversation would last about 5 minutes, and after it the subject would be asked some questions about one of the participants. The second paragraph of this introductory page contained the manipulation of observational set. Subjects in the empathic set condition were given directions modeled after Stotland's (1969) "imagine him" instructions, which were designed to make them empathize with the target. The instructions read: While you arc watching this "get acquainted" conversation, please try to empathize with Margaret, the girl on the left side of the screen. Imagine how Margaret feels as she engages in the conversation. While you arc watching the tape,

EMPATHY AND ATTRIBUTION picture to yourself just how she feels in the situation. You are to concentrate on the way she feels while conversing. Think about her reaction to the information she is receiving from the conversation, In your mind's eye, you are to visualize how it feels to Margaret to be in this conversation. After the tape is over, you will be asked about Margaret's behavior. Subjects in the observation control condition were given instructions to observe the target person. Although these instructions were not designed to promote empathy with the target, neither were they designed (like Stotland's [19691 "watch him" instructions) to completely eliminate it. Subjects in the observation control condition were told to pay close attention to all aspects of the target's behavior, to watch what she did and listen to her conversation, carefully observing all aspects of her behavior. The instructions were designed to be similar to those given to typical observers in attribution studies such as Nisbctt et al. (1973) and Storms (1973). After reading the instructions, subjects watched the videotape on a 12-inch (30-cm) television monitor. The tape depicted an actual get-acquainted conversation, in which two female students chatted for 7 minutes about their home towns, living arrangements, intellectual interests, and travel. Two simultaneous videotapes were made of this conversation, and subjects were randomly assigned to one of the tape forms. The two-person tape showed the target person full face and the other conversant in profile; both participants were seated at a table, and were shown from head to foot. The one-person tape showed the target person only, although the other person could still be heard. The target person again appeared full face, and was shown from the waist up. Thus, her image was somewhat larger on the one-person tape. After watching the conversation, subjects were given the dependent variables to fill out. The major dependent variable was exactly that used by Storms (1973). Subjects were given a set of four pages of questions, arranged in random order. On each page there were three questions. The first question asked the subject to describe the target's behavior along one of the four dimensions of friendliness, talkativeness, nervousness, and dominance, by checking the appropriate point on a 9-point scale anchored at the extremes. The next two questions asked how much influence the subject thought the following two factors had in causing that behavior: (A) Personal characteristics of the target: How important were personal characteristics (personality, traits, character, personal style, attitudes, mood) in causing Margaret to behave that way? (B) Characteristics of the situation: How important were characteristics of the situation (factors such as being in an experiment, the "get acquainted" conversation, conversation topics, the way the other student behaved) in causing Margaret to behave that way?

853

Each of these questions was also answered on a 9-point scale, labeled at the extremes extremely important (9) and extremely unimportant (1). Since no a priori distinction was made among the four types of behavior, and since the hypothesis was concerned with the relative strength of situational versus dispositional attributions, the appropriate measure was the difference between situational and personal attributions, summed over all four behaviors. Since we predicted an increase in the tendency to make relatively situational attributions in the empathic set conditions, we subtracted the importance attributed to dispositional factors from the importance assigned to situational factors. Thus a higher value on this index indicates that a subject's attributions were relatively more situational and less dispositional. After completing the attributional items, subjects in the two-person videotape conditions indicated the percentage of time that they estimated they had spent watching the target and the percentage of time they had spent watching the other person in the conversation, with the constraint that the total should add up to 100%. Subjects were then thanked for their participation and carefully debriefed. No subject indicated undue suspicion or any awareness of the hypothesis.

Subjects The subjects were 40 female undergraduate volunteers enrolled at Cornell University. They were run in individual cubicles in groups ranging from one to three subjects. They were randomly assigned to the four experimental conditions, with the exception that when more than one subject was run at a particular time, all subjects at that time saw the same videotape. Since there was no opportunity for interaction among subjects, and assignment to instruction condition was random even when two or three subjects were run at the same time, each subject was considered independent in the data analysis.

RESULTS The major hypothesis of this experiment was that an empathic orientation would make observers relatively more likely to provide situational attributions and relatively less likely to provide dispositional attributions for an actor's behavior. Table 1 presents the mean situational, dispositional, and situational minus dispositional scores summed over all four behaviors. An analysis of variance performed on the difference scores indicated a significant effect of observational set on causal attributions. Subjects who witnessed the conversation after receiving empathy instructions provided relatively more situational and less dispositional (M = +1.8) at-

DENNIS T. REGAN AND JUDITH TOTTEN

854

TABLK 1 SlTTJATIONAl,, DlSl'OSITlONAL, AND SlTUATlONAL

MINUS DISPOSITJONAI. ATTRIHUTION SCORES TOTALED OVER i\LL FOUR BEHAVIORS Instructions Videotape condition

Attribution Situational

Di.spositional Siti aLional minus d ^positional 2-person

SiU ational Dis jositioiial Siti ational minus d ^positional

Kmpatliic Observer set set 27.8 2S.O

25.2 29.8

2.8 28.7 28.0

27..! 28.4

• It

tributions for the target person's behavior than did subjects who received standard observer instructions (M — —2.8): /*'(!, 36) = 5.79, p < .025.1 Neither the effect of videotape condition, F(l, 36) = 2.14, nor the interaction, /*"(!, 36) < t, was significant. Table 1 also indicates that the empathy and observer cells in the one-person videotape condition differ significantly from each other at p < .05. Considering situational and dispositional attributions separately, in both cases the means were ordered in the predicted direction: Compared with observer controls, subjects given empathy instructions provided relatively more situational and less dispositional attributions. The effect of instructions on dispositional attributions approached significance, -F(l,36) =3.59, p < .10; the effect of instructions on situational attributions was nonsignificant, F(l,36) -~ 2.12. Regarding the four target behaviors, the effect of instructions on the situational minus dispositional index was highly significant for talkativeness, 7?(1,36) = 6.28, p < .02, marginally significant for dominance, F ( l , 3 6 ) = 3.78, p < .10, and nonsignificant for friendliness and nervousness. The data clearly support the hypothesis: "Taking the role of the other" via instructions to empathize results in observers' causal attributions for another's behavior becoming relatively more situational and less 1

All probability values reported in this paper are two-tailed.

dispositional. In this respect, empathic observers provide attributions more like those typically offered by actors themselves. The fact that the effect of empathic instructions was significant even when subjects watched the one-person videotape is consistent with the notion that empathy affects the attribution process at least in part by altering the way in which the available information is processed. There is, of course, the possibility that the empathy instructions caused subjects to pay more attention to the situation surrounding the target actor. Any such tendency might result in more information being gathered about the situation, which in turn would increase the likelihood of situational attributions. But if this were the only mechanism in operation, we would expect a stronger effect of empathy instructions in the twoperson videotape condition, where the depiction of the other actor on the tape should make it easier to focus visual attention on the situation. However, the results indicated a significant effect of instructions in the oneperson videotape condition, where only the target actor was shown; this effect was not at all weaker than that in the two-person videotape condition. It will be recalled that subjects in the twoperson videotape condition were asked to indicate the percentage of time they spent looking at Margaret, the target actor. These data should be interpreted with great caution, both because of the small sample size involved and because they are based on retrospective judgments by subjects that doubtless contain a large amount of error. Subjects in the empathy condition reported a lower percentage of time looking at the target (M = 57.5%) than observer control subjects (M = 75.0%): ^(18) =2.99, p < .01. Correlations were computed to investigate the relationship between the percentage of time looking at the target and the situational minus dispositional attribution index. The overall correlation for all 20 subjects in the two-person videotape condition was +.06 ( n s ) , and the correlation in the observer control condition was —.08 ( n s ) . More interestingly, the correlation for subjects who received the empathy instructions was +.61

EMPATHY AND ATTRIBUTION

855

(.05

Empathy and attribution: turning observers into actors.

A laboratory experiment was conducted to test Jones and Nisbett's information-processing explanation of the often-observed tendency for individuals (a...
575KB Sizes 0 Downloads 0 Views