Journal ol Personality and Social Psycholoty 1975, Vol. 32, No. 6, 1134-1146
Empathy and Altruism Dennis Krebs Harvard University The psychophysiological responses of 60 subjects were measured as they observed a performer play a roulette game. Half of the subjects were led to believe that they were similar to the performer in personality and values, and half were led to believe that they were dissimilar. Half of the subjects in each condition believed that the performer won money and experienced pain as he played the game, and half believed that he performed a cognitive and motor skill task. Subjects who observed a performer who ostensibly experienced pleasure and pain exhibited greater psychophysiological reactions than subjects who did not. Subjects who believed they were similar to the performer tended to react more strongly than subjects who believed they were different from him. Similar subjects also reported identifying most with the performer and feeling the worst while he waited to receive shocks. It was concluded that the similar subjects empathized most with the performer who appeared to experience pleasure and pain. When required to make a choice between helping themselves at a cost to the performer or helping the performer at a cost to themselves, the subjects who reacted most empathically behaved most altruistically. The results were interpreted as casting some light on century-old questions about the human capacity for altruism.
The classical significance of altruism is that it appears to constitute an exception to the pervasive principle that behavior is controlled by rewards and punishments and the associated generalization that people are essentially selfish (Jones, Sontag, Beckner, & Fogelin, 1969). Recently it appears that psychologists have begun to investigate the phenomenon of altruism (Krebs, 1970; Midlarsky, 1969; Wispe, 1972). However, in general, contemporary research on "positive forms of social behavior," "aiding responses," and "helping behavior" is only peripherally relevant to the questions that have concerned scholars for centuries. Psychologists have manipulated various antecedents of helping behaviors and studied their effects, and they have measured a number of correlates of prosocial events; however, they have done little to examine the extent to which the acts that they investigated were oriented to the welfare of either the person who was helped or the helper. It is the extent of self-sacrifice, the expectation of gain, and the orientation to the needs of another that define acts as Requests for reprints should be sent to Dennis Krebs, Department of Psychology, Simon Fraser Universtiy, Burnaby 2, British Columbia, Canada.
altruistic. People help people for a variety of reasons. Some of them may be altruistic; some of them may not. The present study sought to cast some light on the phenomenon of altruism by investigating the idea that empathic reactions mediate altruistic responses. It sought to address this classic issue by quantifying the affective reactions of people who help. A number of investigators (e.g., Aronfreed, 1968; Hebb, 1971; Stotland, 1969) have suggested that people experience the joys and sorrows of their fellows vicariously in certain circumstances. If it were established that people do in fact empathize with others, it could be inferred that there are situations in which it would be in a person's best interest to sacrifice a direct gain for himself in order to help another. When the reward value of an empathic reaction exceeded the reward value of a direct reaction, it would be more rewarding to give (and experience the pleasure of another vicariously) than to receive (and experience a lesser pleasure directly). The main reason why classical debates about the human capacity for altruism have flourished is because scholars have defined altruism as helping behavior that is not motivated by expectations of reward. As Krebs
1134
EMPATHY AND ALTRUISM (1970) pointed out, the existence of altruistic behavior can never be proved when defined in this manner because it requires proving the null hypothesis (i.e., establishing the absence of expectation of reward). In the present conception, helping behavior that is governed by expectations of vicariously experienced pleasure is considered altruistic because it involves self-sacrifice and because its primary purpose is to help another, even though it in no way violates the law of effect. This behavior is other oriented in the sense that it is only when the helper identifies with, takes the role of, imaginatively steps into the shoes of, or reacts as if he were the person he helps that he experiences the consequences that govern his behavior. Two recent studies have created conditions designed to induce empathy and to measure its effect on helping behavior (Aronfreed & Paskel, cited in Aronfreed, 1968; Aderman & Berkowitz, 1970). Although both studies supported the idea that empathic reactions mediate altruistic behavior, neither study established that the subjects who behaved altruistically actually empathized with the person they helped. In fact, in the Aderman and Berkowitz study, subjects who purportedly empathized with a person in a tape recording were never given a chance to help him; rather, the measure of altruism was the extent to which they helped the experimenter. The attempt to induce and quantify empathic reactions has a history in psychology that is longer than the history of attempts to measure altruism. As pointed out by Tagiuri (1969), past studies on empathy are replete with measurement artifacts. In most cases, empathy has been defined as the cognitive ability to predict what another person thinks. This conception is inappropriate for the present purpose, which requires a measure of vicarious affective arousal. Relatively few studies have attempted to quantify the emotional concomitants of empathy. The most significant attempt was made in a pioneering series of studies by Stotland (1969) and his colleagues. Stotland created two basic conditions for empathy: He encouraged subjects to identify with a "performer" by inducing them to believe that they were similar to him,
1135
and he led them to believe that the performer experienced pleasure and pain (by convincing them that a "diathermy" machine generated heat that was pleasureful or painful to the performer's hands). Stotland quantified the empathic responses of his subjects by measuring the psychophysiological concomitants of their emotional reactions and asking them to describe their affective state. Stotland's (1969) exploratory studies supply useful guidelines for the measurement of empathic reactions; however, several aspects of the procedure render the results ambiguous: The diathermic treatment did not seem credible, the psychophysiological measures were primitive, and no attempt was made to control for sources of psychophysiological reaction other than the imagined pain and pleasure of the other. From a tradition quite different from that of Stotland (1969), a second set of studies, namely, studies on vicarious instigation, may provide a means for improving the methodological rigor necessary to establish empathy (Bandura & Rosenthal, 1966, 1968; Berger, 1962; Craig & Lowery, 1969; Craig & Wood, 1969). According to Berger, vicarious instigation occurs when "an observer responds emotionally to a performer's unconditioned emotional response (UER)" (p. 450). Empathic reactions are one type of vicariously instigated response (sadistic reactions are another). One of the most attractive features of Berger's studies is the care that was taken to control for direct sources of emotional arousal, which Berger (1962) called "pseudovicarious effects." Examples of pseudovicarious effects are, as cited by Berger: [a] "An observer may respond with fear to the firing of a gun, regardless of whether the target is a bull's-eye or another human being." [b] "A sudden loud scream by a performer . . . may elicit a fear response." [c] "A child may fear that he will be spanked when he sees that his brother is being spanked for no apparent reason." (p. 451). The significance of studies on vicarious instigation is that they supply a method of increasing the certainty of the assumption that the emotional reactions of a person who observes another experience pleasure and/or displeasure are vicariously instigated, that is, elicited by the emotional
1136
DENNIS KREBS
reactions he attributes to the other rather than by other stimuli. Although the logic of the design of Stotland's (1969) studies did not make such precision necessary, the conclusions he drew would be strengthened if they were supported by a study that controlled for pseudovicarious effects. The first purpose of the present study was to combine the aspects of Berger's design that allowed him to conclude that the responses he elicited were vicarious and the aspects of Stotland's design that allowed him to conclude that the vicarious responses he measured were empathic. It first sought to add to the sophistication of past studies on empathy and to extend their generality by casting the apparent pleasure and pain of the performer in a meaningful context through means of sophisticated techniques of meaurement, a credible similarity manipulation, measurement of more than one psychophysiological reaction, and elicitation of vicarious reactions to both pleasure and pain. The second purpose of the study was to test whether the subjects who experienced the strongest empathic reactions would treat the person with whom they empathized most altruistically. METHOD Subjects Subjects were 60 male adults who responded to an advertisement in a college newspaper soliciting subjects for an experiment on "personality and psychophysiology."1 The average age of the subjects was 20. The pay for the experiment was advertised as ranging from $2 to $7, with an average amount of $5. Subjects were advised that the experiment involved two separate testing sessions.
Induction and Quantification of Empathy Subjects met in groups for the first session. They were given three pencil-and-paper tests and were informed that they would be paired with another subject on the basis of their scores on the tests.2 1 Four subjects were discarded: One knew the technician, one admitted prior knowledge of the design, and two were discarded because of equipment failure. 2 The tests were: a modified version of the Jackson Personality Inventory (Jackson, 1967) which contained a scale from the Maudsley Personality Inventory (Eysenck, 1958), the Adjective Check List (Gough & Heilbrun, 1965), and the Allport-Vernon Scale of Values (Allport, Vernon, & Lindzey, 19S1).
After a few days, the subjects were phoned, and the second session was arranged at a time "convenient to both participants." When subjects arrived for the second session, they were taken to a two-room laboratory. Subjects were seated in a room that contained electrodes and psychophysiological supplies. They faced a blind that covered a one-way mirror. An assistant of the experimenter connected electrodes to the subjects that would measure skin conductance, vasoconstriction, and heart rate. Psychophysiological responses were recorded on a six-channel Grass Model 7 polygraph. The electrodes employed to measure skin conductance are described by O'Connell and Tursky (1960). The plethysmograph employed to measure vasoconstriction is described by Tursky and Greenblatt (1967), and a detailed description of the cardiographic equipment is supplied by Lerner (1969). Similarity manipulation. After the subjects were hooked up, they were given a set of instructions informing them that they had been randomly selected to serve as an observer in the experiment. The "performer" was actually a confederate of the experimenter. Subjects were told that they had been paired with the performer on the basis of a computer analysis of their responses to the personality tests. Half of the subjects were informed that they had been paired with the performer because they were similar to him, and half were informed that they had been paired with the performer because they were different from him. The instructions informed subjects that the first purpose of the experiment was to measure the psychophysiological reactions of the performer as he reacted directly to various stimuli, and the second was to measure the indirect responses of the observer as he watched the performer at his task. The thrust of the instructions was to lead subjects to believe that they were members of a control group and that the experimenter was interested mainly in the responses of the performer. It was assumed that the aura of science created by the demeanor of the experimenter and the psychophysiological equipment would mitigate against the inference that the purpose of the manipulations was to measure empathy. The performer was a 2 2-year-old graduate student who had recently arrived in the area. He was visible to the subjects from the chest up. Electrodes similar to those attached to the subjects were visible on the performer (although psychophysiological responses were not actually recorded from them). The experimenter told the performer that an observer would watch him through the one-way mirror but that he would not ever meet him. After checking the performer's electrodes, the experimenter asked the performer whether he had any questions. An exchange occurred that was designed to strengthen the similarity manipulation: [Similar performer]: About how long will this take? [Experimenter] Not more than half an hour. [Similar performer]: Good, I've got to pick up a paper at [a location that implied that the
EMPATHY AND ALTRUISM performer held interests similar to those of the observer]. [Experimenter]: Are you a student here? [Similar performer]: Yes, I'm in [a field consistent with the observer's most preferred value as revealed by the Allport-Vernon (1951) Scale of Values]. In the exchange between the experimenter and the dissimilar performer, the performer affiliated himself with a field consistent with the observer's least preferred value, and when appropriate, said he was not a student. The experimenter went on to say that there would be three parts to the experiment. He explained that in the first part, the performer would be required to watch a series of three lights and listen to some tones. Habituation trials. Subjects watched the performer through the one-way mirror as he reacted to a series of stimuli that were later associated with pain and pleasure. The purpose of presenting the stimuli beforehand was to habituate subjects to their physical effects. The stimuli consisted of three lights arranged in a- series approximately 8 cm apart and a number of tones. Subjects' psychophysiological responses were monitored, and elicitation of the stimuli continued until subjects were habituated to them. Reward-punishment manipulation. After subjects were habituated to the experimental stimuli, the experimenter again addressed the performer. He informed him that the second part of the experiment would involve playing a roulette game. Half of the subjects were led to believe that the performer would receive rewards and punishments, and half were led to believe that he would perform relatively innocuous motor and conceptual tasks. The subjects who believed that the performer received rewards and punishments were expected to impute strong affective responses to him; the other half were not. The high-affect and low-affect condition intersected the similarity condition to produce four groups with IS subjects in each: high affect-similar, high affect-dissimilar, low affect-similar, and low affectdissimilar. Subjects in the two high-affect groups heard the experimenter inform the performer that if the roulette ball landed on an odd number, he would receive an electric shock in his right arm and that the shock would range in intensity from very mild to very severe, depending on the precise outcome. If the ball landed on an even number, the performer would win some money. The amount he won would be indicated by the color of a poker chip that would be dispensed from a machine in front of him. After each spin of the roulette wheel, one of the three lights came on in front of the performer. The left-hand light was always followed by punishment, the right-hand light was followed by reward, and the middle light signaled that the trial would be taken over (no outcome). Subjects in the low-affect group heard the performer receive the same instructions as subjects in the high-affect groups, except that no mention was made of winning money or receiving shock. Instead,
1137
subjects in the low-affect groups were led to believe that the left-hand light would signal that the performer would perform a reaction time task, and the right-hand light would signal that the performer would perform a cognitive task. The experimenter asked the performer to flex one of his arms as soon as possible after the "shock" light went off and to attempt to guess the color of the chip that would fall at the offset of the "reward" light. The experimenter assigned the same meaning to the middle light as he did for subjects in the high-affect groups. The rationale for these tasks was given in psychophysiological terms. The experimenter ensured that they made sense to the performer before he began (and that through him their meaning would be conveyed to the subjects who were observing him). The purpose of varying the intensity of the reward and punishment outcomes in the high-affect groups was to help to standardize the uncertainty associated with the comparable task in the lowaffect groups and to make the subjects' task more interesting. Conditioning trials. When the reward-punishment manipulation was completed, subjects watched the performer receive 18 randomly arranged "conditioning" trials. The temporal format for each trial was as follows: tone sounds
p spins roulette wheel
one of 3 lights flash
M = 8 sec M = 12 sec
one of 3 outcomes occur
intertrial interval < M = 40 sec
Six trials ended in reward/chip, six ended in shock/ reflex, and six ended in no outcome. When the outcome was reward/chip, a colored chip fell into a tray. When the outcome was shock/reflex, the performer jerked his arm. Observers in the highaffect groups were led to believe that the performer won $3 and received 6 shocks. Observers in the low-affect group were not given any information about the performer's earnings but believed that he earned $3 for participating in Session 2 of the experiment. Subjects in all groups observed the same sequence of physical stimuli. When the shock/reflex light came on, the performer looked concerned. He flexed the muscles in his forearm at its outset. When the reward/chip light came on, the performer glanced toward the machine that dispensed chips. Two assistants unobtrusively observed the performer and were able to discern whether his actions signified that he was receiving rewards and punishments or reaching to the less arousing stimuli. In terms of Berger's (1962) paradigm for vicarious instigation, the arm jerk and chip were expected to supply evidence to subjects in the high-affect groups that the performer experienced an unconditioned emotional response ("the performer's UER need not actually occur," Berger, 1962, p. 450). The unconditioned emotional response imputed to the performer was expected to serve as an unconditioned stimulus to the observer and to elicit an unconditioned emo-
1138
DENNIS KKEBS
tional response from him, which was expected to become conditioned to the original conditioned stimulus (the lights that signaled shock and reward). In order to ensure that responses to the conditioned stimulus were vicarious rather than pseudovicarious, they were compared with the responses of subjects in the low-affect groups, who were exposed to the same physical stimuli (the onset and offset of the lights, the ejection of the chip, and the arm jerk). The possibility that subjects in the high-affect groups indiscriminately responded more strongly to all experiment stimuli without responding to the emotional reactions they imputed to the performer (e.g., because they were nervous about electric shock or afraid that they would later receive shocks) was checked by comparing their responses to the nooutcome light to those of subjects in the low-affect groups. Satisfying the conditions for vicarious instigation is a necessary but not sufficient condition for establishing that (vicariously conditioned) emotional responses are emphatic (Berger, 1962). The similarity manipulation was designed to increase the disposition of similar subjects to empathize with the performer. The subjects who empathized most were expected to report feeling the same way as they thought the performer felt. It was predicted that subjects in the high-similar group would evidence the strongest psychophysiological reactions and congruent emotional responses to the lights signaling that the performer would receive rewards and punishments but not to the light signaling no outcome.
Quantification of Altruism After the final conditioning trial, subjects observed the experimenter enter the performer's room and inform him that there would be one last "bonus" trial. The experimenter explained that the trial would be the same as the previous trials except that (a) the performer would stand the chance of winning from nothing at all to $2 if the roulette ball landed on an even number or a shock that ranged in intensity from barely perceptible to maximally painful if the ball landed on an odd number, and (b) the amount of reward and intensity of shock would be determined by the decision of another person rather than by chance. The odds of winning and losing remained SO: SO. The experimenter then addressed the subjects through an intercom. He asked them to open a large manila envelope that lay on a table next to them. The envelope contained a set of written instructions and $2 in dimes. The instructions explained that the nature of the outcome of the final trial (reward or punishment) would be determined by a spin of the roulette wheel, as before, but that the subjects would determine ahead of time both the amount of money the performer would win if he obtained an even number and the amount of punishment he would receive if the number he obtained was odd. If subjects had simply been required to estimate how much money or shock the performer would
obtain, they would have been expected to maximize the favorability of his outcomes. Although many past studies have employed no-cost measures of helping (Krebs, 1970), such measures supply a questionable index of altruism because they fail to test the extent to which a subject is willing to jeopardize his own welfare in order to help another. Therefore, the instructions went on to inform subjects that they would have a chance to win money or receive a shock on the bonus trial and that the more favorable they made the outcomes for the performer, the less favorable these outcomes would be for themselves. Subjects were asked to select 1 of 21 pairs of outcomes that varied from maximally altruistic to maximally selfish. For the most altruistic of the alternatives, the subject agreed to allow the performer to win $2 if the number was even and to escape all shock if the number was odd. The cost to the subject was to agree to receive 100% of a maximally painful shock if the performer lost and no money if he won. The alternatiyes decreased systematically in monetary benefit to the performer (in the event of a win) in proportion to the intensity of shock the subject was willing to accept in the event of a loss. The second most altruistic alternative entailed accepting 90% maximum shock/ no money in order to assign no shock/$1.80 to the performer, etc., until the llth alternative, when neither the subject nor the performer stood the chance of winning any money or receiving any shock. The remaining alternatives entailed a selfish choice. The selfish alternatives increased systematically until the most selfish alternative, in which the subject inflicted a 100% maximally painful shock on the performer if he lost and no money if he won, in order to assign a $2.00/no-shock set of outcomes to himself. Subjects were informed that their decision would be anonymous. They were asked to implement it by depositing whatever dimes they decided to give to the performer in the appropriate slot in an electronically programmed box. They were given to understand that the intensity of the shock would be determined electronically by the number of dimes that they deposited and that they would keep the money they assigned to themselves. The outcome of the "bonus" trial was always a win.
Interview After the bonus trial had been completed, the electrodes were disconnected, and subjects were shown to an interview room. Subjects were asked to reconstruct their experience in the experiment and to describe their cognitive and affective reactions to everything that happened. They were then given a posttest questionnaire that supplied a check on all experimental manipulations, measured the extent to which the plight of the performer was salient to the subjects, and assessed the subjects' impression of the performer. Subjects were asked to rate the extent to which they identified with the
EMPATHY AND ALTRUISM performer and to supply an indication of their affective experience while the "shock," "reward," and no-outcome lights were on. Subjects were debriefed and advised that they would receive a complete description of the experiment after about a month.
RESULTS Checks on Experimental Manipulations It is always possible in laboratory experiments of this sort for subjects to discern the purpose of the experiment and to try to respond accordingly. The posttest interview began with a persistent attempt to discover how the subjects perceived all facets of the experiment. None of the subjects reported believing that the purpose of the experiment was to measure empathy, and none reported attempting to control his psychophysiological responses voluntarily. Responses to the posttest questionnaire indicated that the reward-punishment manipulation was effective. Subjects in the two high-affect groups reported believing that the performer experienced more pleasure while the reward light was on than subjects in the two low-affect groups (M = 8.0 vs. 6.25, on a 9-point scale), t ( 5 6 ) = 2.97, p < .0025, for the one-tailed prediction).3 Subjects from the high-affect groups also reported believing that the performer experienced less pleasure while the shock light was on (M — 2.7 vs. 4.9), t = 6.946, p < .0001. There were no significant differences between the responses of subjects in the high-similar and high-dissimilar groups (M = 8.2 vs. 7.8 for reward and 2.6 vs. 2.8 for punishment) or lowsimilar and low-dissimilar groups (M = 6.3 and 6.2 for reward and both 4.9 for punishment). It should be noted that although subjects in the low-affect group believed that the performer felt quite neutral (4.9 on a 9-point scale) while waiting to make a reflex reaction, they unexpectedly attributed considerable pleasure to the performer as he waited 8 Unless otherwise reported, all t tests followed from a 2 X 2 analysis of variance. The degrees of freedom associated with the t score are the same as those associated with the within-subjects mean square error of the analysis of variance, which was 56 (Lindquist, 19S3). One-tailed tests were employed unless otherwise reported.
1139
for a chip to fall (6.2). The difference between the two reactions was statistically significant (£ = 2.84, p< .025, two-tailed). Responses to the posttest questionnaire also supported the assumption that the similarity manipulation was generally effective. Subjects in the two similar groups rated the performer as significantly more similar to them than subjects in the two dissimilar groups (M = 5.2 vs. 2.7), t = 5.234, p < .0001. There were no significant differences between perception of similarity in the high-similar versus low-similar group (both Ms =5.2). However, subjects in the high-dissimilar group unexpectedly viewed themselves as more similar to the performer than subjects in the low-dissimilar group (Ms = 3.5 and 2.0, respectively), / = 2.21, p < .05, twotailed). Measurement of Empathy In all, 14 indices of empathy were tapped: 3 subjective reports and 11 psychophysiological reactions. In all cases, subjects from the high-similar group were expected to emit the strongest response. The one-tailed tests that follow compare the mean responses of subjects in the high-similar group with the mean response of subjects in the other three groups combined. When subjects from the high-similar group did not evidence the greatest response, the prediction was disconfirmed. Subjective measures of empathy. Subjects in the high-similar group reported identifying with the performer to the greatest extent M = 6.1 vs. 4.8 for the other three groups), t = 2.15, p < .025. Subjects in the highsimilar group reported feeling the worst of all subjects while waiting for the performer to receive a punishment (M = 3.5 vs. 5.0), t = 2.54, p < .01. Self-report measures relating to affective reactions while waiting for the performer to receive a reward failed to differ among groups (F = 1.94). The mean response of subjects in the high-dissimilar group was slightly higher than the mean response of subjects in the high-similar group (6.3 vs. 5.9). The posttest interview indicated that the machine that dispensed chips was a source of great amusement and pleasure to
1140
DENNIS KREBS
subjects in all conditions, and as reported earlier, subjects from all groups believed that the performer enjoyed waiting for a chip to fall. Measurement oj psychophysiological responses. Lacey (1949) and others have shown that changes in galvanic skin response and blood pulse volume generally correlate quite highly with the level of the base from which they originate. (Changes in heart rate are relatively independent of their base). Therefore, in the present study, electrodermal activity was measured in ohms and was converted to log skin conductance. Responses were scored as a proportion of their base. The basal level of skin conductance was denned as the most stable level of skin conductance in the 6 sec prior to the onset of a stimulus. The average correlation between the basal level of skin conductance and the proportion of change to the lights that comprised the main experimental stimuli was .022. Blood pulse volume was measured as a proportion change of raw mV score with the basal level denned as the mean of the maximum and minimum blood pulse volume in the 6 sec prior to the onset of a stimulus. The average correlation between base rate and change to the lights was .066. The measure of arousal associated with a change in blood pulse volume is termed vasoconstriction. Heart rate was measured as the average increase or decrease in beats per minute from a basal level that was defined as the mean of 1.050 i— HIGH AFFECT-SIMM AR
HIGH A F F E C T - D I S S I M I L A R
LOW AFFECT-SIMILAR LOW AFFECT-DISSIMILAR
FIGURE 1. Proportion of change in log skin conductance to instructions.
1.900 1.800
i.roo u o < o
1.600 1.500
o 1.400 \° 1.300 o a-°~
HIGH AFFECT-DISSIMILAR
1.200 LOW AFFECT-SIMILAR 1.IOO 1.000
A LOW A F F E C T - D I S S I M I L A R
.900 FIGURE 2. Proportion of change in blood pulse volume to instructions.
the maximum and minimum beats per minute in the 6 sec prior to the onset of stimulus (cf. Craig, 1968). The mean correlation between raw heart rate (number of beats per minute) and heart rate change to the onset of the three lights was .152. The relationship between heart rate and vicarious arousal is more complex than the relationship between skin conductance and vasoconstriction and vicarious arousal. Heart rate acceleration is associated with heightened arousal when arousal occurs over a period of time, but heart rate deceleration is associated with attentiveness during the immediate intake of information. Thus, the appropriate measure of empathy is heart rate deceleration immediately after stimuli signaling that another will be rewarded or punished or heart rate acceleration while waiting for another to receive rewards or punishments (cf. Craig, 1968; Craig & Lowery, 1969). Psychophysiological reactions to the instructions. Figures 1-3 display the mean changes in basal level of skin conductance, vasoconstriction, and heart rate from the final habituation trial to the first conditioning trial, the time during which subjects from the high-affect groups learned that the performer could win money or receive shocks and subjects from the low-affect groups learned that the performer would perform a less
EMPATHY AND ALTRUISM
HIGH AFFECT-SIMILAR
LOW AFFECT-SIMILAR
HIGH AFFECT-DISSIMILAR LOW A F F E C T - D I S S I M I L A R
FIGURE 3. Beats-per-minute change in heart rate instructions.
arousing task. Observers in the high-affect groups were expected to input stronger emotional responses to the performer (especially fear of receiving a shock) than subjects in the low-affect groups. Subjects in the high-similar group were expected to be most prone to experience the responses vicariously and therefore to evidence the strongest psychophysiological reactions. Subjects from the high-similar group evidenced the greatest increase in skin conductance (t = 2.27, p < .02), vasoconstriction (£ = 2.80, p < .005), and
1141
heart rate acceleration (t = 1.12, p < .12) to the instructions. Subjects in the high-dissimilar group emitted the second strongest responses. Psychophysiological reactions to the reward light. Figures 4-6 display the mean psychophysiological reaction of subjects to the light that signaled a reward/chip outcome. Subjects in the high-similar group evidenced the greatest increases in skin conductance (t = 2.01, p < .025, heart rate deceleration (t = 1.65, p = .06), and heart rate acceleration (t = 1.30, p< .10). In all cases, skin conductance and vasoconstriction were denned as the maximum increase during the 12 sec in which the lights were on, heart rate deceleration was denned as the maximum deceleration during the 6 sec after the onset of lights, and heart rate acceleration was denned as the maximum acceleration during the final 6 sec (in anticipation of the outcome). There were no significant differences between groups in vasoconstriction. Subjects from the high-dissimilar group emitted the second strongest responses to the light that signaled reward. Psychophysiological reactions to the shock light. Figures 4-6 also show that subjects in the high-similar group tended to react most strongly to the light signaling that the per-
O HIGH AFFECT-SIMILAR • HIGH AFFECT-DISSIMILAR A LOW AFFECT-SIMILAR A LOW AFFECT-DISSIMILAR
REWARD/CHIP LIGHT
SHOCK/PUNISH LIGHT
NO OUTCOME LIGHT
FIGURE 4. Proportion of change in log skin conductance to lights.
DENNIS KREBS
1142
O HIGH AFFECT-SIMILAR
1.3200
• HIGH A F F E C T - D I S S I M I L A R A LOW AFFECT-SIMILAR
1.2800
A LOW AFFECT-DISSIMILAR
1,2400
5 1.2000 p 1.1600 cc 1.1200 ce 1.0800 1.0400 1.0000 REWARD/CHIP LIGHT
SHOCK/REFLEX LIGHT
NO OUTCOME LIGHT
FIGURE 5. Proportion of change in blood pulse volume to lights.
former would receive a shock. They evidenced the greatest increase in skin conductance (t — 2.35, p < .02) and heart rate deceleration (£ = 2.38, p
Empathy and Altruism Dennis Krebs Harvard University The psychophysiological responses of 60 subjects were measured as they observed a performer play a roulette game. Half of the subjects were led to believe that they were similar to the performer in personality and values, and half were led to believe that they were dissimilar. Half of the subjects in each condition believed that the performer won money and experienced pain as he played the game, and half believed that he performed a cognitive and motor skill task. Subjects who observed a performer who ostensibly experienced pleasure and pain exhibited greater psychophysiological reactions than subjects who did not. Subjects who believed they were similar to the performer tended to react more strongly than subjects who believed they were different from him. Similar subjects also reported identifying most with the performer and feeling the worst while he waited to receive shocks. It was concluded that the similar subjects empathized most with the performer who appeared to experience pleasure and pain. When required to make a choice between helping themselves at a cost to the performer or helping the performer at a cost to themselves, the subjects who reacted most empathically behaved most altruistically. The results were interpreted as casting some light on century-old questions about the human capacity for altruism.
The classical significance of altruism is that it appears to constitute an exception to the pervasive principle that behavior is controlled by rewards and punishments and the associated generalization that people are essentially selfish (Jones, Sontag, Beckner, & Fogelin, 1969). Recently it appears that psychologists have begun to investigate the phenomenon of altruism (Krebs, 1970; Midlarsky, 1969; Wispe, 1972). However, in general, contemporary research on "positive forms of social behavior," "aiding responses," and "helping behavior" is only peripherally relevant to the questions that have concerned scholars for centuries. Psychologists have manipulated various antecedents of helping behaviors and studied their effects, and they have measured a number of correlates of prosocial events; however, they have done little to examine the extent to which the acts that they investigated were oriented to the welfare of either the person who was helped or the helper. It is the extent of self-sacrifice, the expectation of gain, and the orientation to the needs of another that define acts as Requests for reprints should be sent to Dennis Krebs, Department of Psychology, Simon Fraser Universtiy, Burnaby 2, British Columbia, Canada.
altruistic. People help people for a variety of reasons. Some of them may be altruistic; some of them may not. The present study sought to cast some light on the phenomenon of altruism by investigating the idea that empathic reactions mediate altruistic responses. It sought to address this classic issue by quantifying the affective reactions of people who help. A number of investigators (e.g., Aronfreed, 1968; Hebb, 1971; Stotland, 1969) have suggested that people experience the joys and sorrows of their fellows vicariously in certain circumstances. If it were established that people do in fact empathize with others, it could be inferred that there are situations in which it would be in a person's best interest to sacrifice a direct gain for himself in order to help another. When the reward value of an empathic reaction exceeded the reward value of a direct reaction, it would be more rewarding to give (and experience the pleasure of another vicariously) than to receive (and experience a lesser pleasure directly). The main reason why classical debates about the human capacity for altruism have flourished is because scholars have defined altruism as helping behavior that is not motivated by expectations of reward. As Krebs
1134
EMPATHY AND ALTRUISM (1970) pointed out, the existence of altruistic behavior can never be proved when defined in this manner because it requires proving the null hypothesis (i.e., establishing the absence of expectation of reward). In the present conception, helping behavior that is governed by expectations of vicariously experienced pleasure is considered altruistic because it involves self-sacrifice and because its primary purpose is to help another, even though it in no way violates the law of effect. This behavior is other oriented in the sense that it is only when the helper identifies with, takes the role of, imaginatively steps into the shoes of, or reacts as if he were the person he helps that he experiences the consequences that govern his behavior. Two recent studies have created conditions designed to induce empathy and to measure its effect on helping behavior (Aronfreed & Paskel, cited in Aronfreed, 1968; Aderman & Berkowitz, 1970). Although both studies supported the idea that empathic reactions mediate altruistic behavior, neither study established that the subjects who behaved altruistically actually empathized with the person they helped. In fact, in the Aderman and Berkowitz study, subjects who purportedly empathized with a person in a tape recording were never given a chance to help him; rather, the measure of altruism was the extent to which they helped the experimenter. The attempt to induce and quantify empathic reactions has a history in psychology that is longer than the history of attempts to measure altruism. As pointed out by Tagiuri (1969), past studies on empathy are replete with measurement artifacts. In most cases, empathy has been defined as the cognitive ability to predict what another person thinks. This conception is inappropriate for the present purpose, which requires a measure of vicarious affective arousal. Relatively few studies have attempted to quantify the emotional concomitants of empathy. The most significant attempt was made in a pioneering series of studies by Stotland (1969) and his colleagues. Stotland created two basic conditions for empathy: He encouraged subjects to identify with a "performer" by inducing them to believe that they were similar to him,
1135
and he led them to believe that the performer experienced pleasure and pain (by convincing them that a "diathermy" machine generated heat that was pleasureful or painful to the performer's hands). Stotland quantified the empathic responses of his subjects by measuring the psychophysiological concomitants of their emotional reactions and asking them to describe their affective state. Stotland's (1969) exploratory studies supply useful guidelines for the measurement of empathic reactions; however, several aspects of the procedure render the results ambiguous: The diathermic treatment did not seem credible, the psychophysiological measures were primitive, and no attempt was made to control for sources of psychophysiological reaction other than the imagined pain and pleasure of the other. From a tradition quite different from that of Stotland (1969), a second set of studies, namely, studies on vicarious instigation, may provide a means for improving the methodological rigor necessary to establish empathy (Bandura & Rosenthal, 1966, 1968; Berger, 1962; Craig & Lowery, 1969; Craig & Wood, 1969). According to Berger, vicarious instigation occurs when "an observer responds emotionally to a performer's unconditioned emotional response (UER)" (p. 450). Empathic reactions are one type of vicariously instigated response (sadistic reactions are another). One of the most attractive features of Berger's studies is the care that was taken to control for direct sources of emotional arousal, which Berger (1962) called "pseudovicarious effects." Examples of pseudovicarious effects are, as cited by Berger: [a] "An observer may respond with fear to the firing of a gun, regardless of whether the target is a bull's-eye or another human being." [b] "A sudden loud scream by a performer . . . may elicit a fear response." [c] "A child may fear that he will be spanked when he sees that his brother is being spanked for no apparent reason." (p. 451). The significance of studies on vicarious instigation is that they supply a method of increasing the certainty of the assumption that the emotional reactions of a person who observes another experience pleasure and/or displeasure are vicariously instigated, that is, elicited by the emotional
1136
DENNIS KREBS
reactions he attributes to the other rather than by other stimuli. Although the logic of the design of Stotland's (1969) studies did not make such precision necessary, the conclusions he drew would be strengthened if they were supported by a study that controlled for pseudovicarious effects. The first purpose of the present study was to combine the aspects of Berger's design that allowed him to conclude that the responses he elicited were vicarious and the aspects of Stotland's design that allowed him to conclude that the vicarious responses he measured were empathic. It first sought to add to the sophistication of past studies on empathy and to extend their generality by casting the apparent pleasure and pain of the performer in a meaningful context through means of sophisticated techniques of meaurement, a credible similarity manipulation, measurement of more than one psychophysiological reaction, and elicitation of vicarious reactions to both pleasure and pain. The second purpose of the study was to test whether the subjects who experienced the strongest empathic reactions would treat the person with whom they empathized most altruistically. METHOD Subjects Subjects were 60 male adults who responded to an advertisement in a college newspaper soliciting subjects for an experiment on "personality and psychophysiology."1 The average age of the subjects was 20. The pay for the experiment was advertised as ranging from $2 to $7, with an average amount of $5. Subjects were advised that the experiment involved two separate testing sessions.
Induction and Quantification of Empathy Subjects met in groups for the first session. They were given three pencil-and-paper tests and were informed that they would be paired with another subject on the basis of their scores on the tests.2 1 Four subjects were discarded: One knew the technician, one admitted prior knowledge of the design, and two were discarded because of equipment failure. 2 The tests were: a modified version of the Jackson Personality Inventory (Jackson, 1967) which contained a scale from the Maudsley Personality Inventory (Eysenck, 1958), the Adjective Check List (Gough & Heilbrun, 1965), and the Allport-Vernon Scale of Values (Allport, Vernon, & Lindzey, 19S1).
After a few days, the subjects were phoned, and the second session was arranged at a time "convenient to both participants." When subjects arrived for the second session, they were taken to a two-room laboratory. Subjects were seated in a room that contained electrodes and psychophysiological supplies. They faced a blind that covered a one-way mirror. An assistant of the experimenter connected electrodes to the subjects that would measure skin conductance, vasoconstriction, and heart rate. Psychophysiological responses were recorded on a six-channel Grass Model 7 polygraph. The electrodes employed to measure skin conductance are described by O'Connell and Tursky (1960). The plethysmograph employed to measure vasoconstriction is described by Tursky and Greenblatt (1967), and a detailed description of the cardiographic equipment is supplied by Lerner (1969). Similarity manipulation. After the subjects were hooked up, they were given a set of instructions informing them that they had been randomly selected to serve as an observer in the experiment. The "performer" was actually a confederate of the experimenter. Subjects were told that they had been paired with the performer on the basis of a computer analysis of their responses to the personality tests. Half of the subjects were informed that they had been paired with the performer because they were similar to him, and half were informed that they had been paired with the performer because they were different from him. The instructions informed subjects that the first purpose of the experiment was to measure the psychophysiological reactions of the performer as he reacted directly to various stimuli, and the second was to measure the indirect responses of the observer as he watched the performer at his task. The thrust of the instructions was to lead subjects to believe that they were members of a control group and that the experimenter was interested mainly in the responses of the performer. It was assumed that the aura of science created by the demeanor of the experimenter and the psychophysiological equipment would mitigate against the inference that the purpose of the manipulations was to measure empathy. The performer was a 2 2-year-old graduate student who had recently arrived in the area. He was visible to the subjects from the chest up. Electrodes similar to those attached to the subjects were visible on the performer (although psychophysiological responses were not actually recorded from them). The experimenter told the performer that an observer would watch him through the one-way mirror but that he would not ever meet him. After checking the performer's electrodes, the experimenter asked the performer whether he had any questions. An exchange occurred that was designed to strengthen the similarity manipulation: [Similar performer]: About how long will this take? [Experimenter] Not more than half an hour. [Similar performer]: Good, I've got to pick up a paper at [a location that implied that the
EMPATHY AND ALTRUISM performer held interests similar to those of the observer]. [Experimenter]: Are you a student here? [Similar performer]: Yes, I'm in [a field consistent with the observer's most preferred value as revealed by the Allport-Vernon (1951) Scale of Values]. In the exchange between the experimenter and the dissimilar performer, the performer affiliated himself with a field consistent with the observer's least preferred value, and when appropriate, said he was not a student. The experimenter went on to say that there would be three parts to the experiment. He explained that in the first part, the performer would be required to watch a series of three lights and listen to some tones. Habituation trials. Subjects watched the performer through the one-way mirror as he reacted to a series of stimuli that were later associated with pain and pleasure. The purpose of presenting the stimuli beforehand was to habituate subjects to their physical effects. The stimuli consisted of three lights arranged in a- series approximately 8 cm apart and a number of tones. Subjects' psychophysiological responses were monitored, and elicitation of the stimuli continued until subjects were habituated to them. Reward-punishment manipulation. After subjects were habituated to the experimental stimuli, the experimenter again addressed the performer. He informed him that the second part of the experiment would involve playing a roulette game. Half of the subjects were led to believe that the performer would receive rewards and punishments, and half were led to believe that he would perform relatively innocuous motor and conceptual tasks. The subjects who believed that the performer received rewards and punishments were expected to impute strong affective responses to him; the other half were not. The high-affect and low-affect condition intersected the similarity condition to produce four groups with IS subjects in each: high affect-similar, high affect-dissimilar, low affect-similar, and low affectdissimilar. Subjects in the two high-affect groups heard the experimenter inform the performer that if the roulette ball landed on an odd number, he would receive an electric shock in his right arm and that the shock would range in intensity from very mild to very severe, depending on the precise outcome. If the ball landed on an even number, the performer would win some money. The amount he won would be indicated by the color of a poker chip that would be dispensed from a machine in front of him. After each spin of the roulette wheel, one of the three lights came on in front of the performer. The left-hand light was always followed by punishment, the right-hand light was followed by reward, and the middle light signaled that the trial would be taken over (no outcome). Subjects in the low-affect group heard the performer receive the same instructions as subjects in the high-affect groups, except that no mention was made of winning money or receiving shock. Instead,
1137
subjects in the low-affect groups were led to believe that the left-hand light would signal that the performer would perform a reaction time task, and the right-hand light would signal that the performer would perform a cognitive task. The experimenter asked the performer to flex one of his arms as soon as possible after the "shock" light went off and to attempt to guess the color of the chip that would fall at the offset of the "reward" light. The experimenter assigned the same meaning to the middle light as he did for subjects in the high-affect groups. The rationale for these tasks was given in psychophysiological terms. The experimenter ensured that they made sense to the performer before he began (and that through him their meaning would be conveyed to the subjects who were observing him). The purpose of varying the intensity of the reward and punishment outcomes in the high-affect groups was to help to standardize the uncertainty associated with the comparable task in the lowaffect groups and to make the subjects' task more interesting. Conditioning trials. When the reward-punishment manipulation was completed, subjects watched the performer receive 18 randomly arranged "conditioning" trials. The temporal format for each trial was as follows: tone sounds
p spins roulette wheel
one of 3 lights flash
M = 8 sec M = 12 sec
one of 3 outcomes occur
intertrial interval < M = 40 sec
Six trials ended in reward/chip, six ended in shock/ reflex, and six ended in no outcome. When the outcome was reward/chip, a colored chip fell into a tray. When the outcome was shock/reflex, the performer jerked his arm. Observers in the highaffect groups were led to believe that the performer won $3 and received 6 shocks. Observers in the low-affect group were not given any information about the performer's earnings but believed that he earned $3 for participating in Session 2 of the experiment. Subjects in all groups observed the same sequence of physical stimuli. When the shock/reflex light came on, the performer looked concerned. He flexed the muscles in his forearm at its outset. When the reward/chip light came on, the performer glanced toward the machine that dispensed chips. Two assistants unobtrusively observed the performer and were able to discern whether his actions signified that he was receiving rewards and punishments or reaching to the less arousing stimuli. In terms of Berger's (1962) paradigm for vicarious instigation, the arm jerk and chip were expected to supply evidence to subjects in the high-affect groups that the performer experienced an unconditioned emotional response ("the performer's UER need not actually occur," Berger, 1962, p. 450). The unconditioned emotional response imputed to the performer was expected to serve as an unconditioned stimulus to the observer and to elicit an unconditioned emo-
1138
DENNIS KKEBS
tional response from him, which was expected to become conditioned to the original conditioned stimulus (the lights that signaled shock and reward). In order to ensure that responses to the conditioned stimulus were vicarious rather than pseudovicarious, they were compared with the responses of subjects in the low-affect groups, who were exposed to the same physical stimuli (the onset and offset of the lights, the ejection of the chip, and the arm jerk). The possibility that subjects in the high-affect groups indiscriminately responded more strongly to all experiment stimuli without responding to the emotional reactions they imputed to the performer (e.g., because they were nervous about electric shock or afraid that they would later receive shocks) was checked by comparing their responses to the nooutcome light to those of subjects in the low-affect groups. Satisfying the conditions for vicarious instigation is a necessary but not sufficient condition for establishing that (vicariously conditioned) emotional responses are emphatic (Berger, 1962). The similarity manipulation was designed to increase the disposition of similar subjects to empathize with the performer. The subjects who empathized most were expected to report feeling the same way as they thought the performer felt. It was predicted that subjects in the high-similar group would evidence the strongest psychophysiological reactions and congruent emotional responses to the lights signaling that the performer would receive rewards and punishments but not to the light signaling no outcome.
Quantification of Altruism After the final conditioning trial, subjects observed the experimenter enter the performer's room and inform him that there would be one last "bonus" trial. The experimenter explained that the trial would be the same as the previous trials except that (a) the performer would stand the chance of winning from nothing at all to $2 if the roulette ball landed on an even number or a shock that ranged in intensity from barely perceptible to maximally painful if the ball landed on an odd number, and (b) the amount of reward and intensity of shock would be determined by the decision of another person rather than by chance. The odds of winning and losing remained SO: SO. The experimenter then addressed the subjects through an intercom. He asked them to open a large manila envelope that lay on a table next to them. The envelope contained a set of written instructions and $2 in dimes. The instructions explained that the nature of the outcome of the final trial (reward or punishment) would be determined by a spin of the roulette wheel, as before, but that the subjects would determine ahead of time both the amount of money the performer would win if he obtained an even number and the amount of punishment he would receive if the number he obtained was odd. If subjects had simply been required to estimate how much money or shock the performer would
obtain, they would have been expected to maximize the favorability of his outcomes. Although many past studies have employed no-cost measures of helping (Krebs, 1970), such measures supply a questionable index of altruism because they fail to test the extent to which a subject is willing to jeopardize his own welfare in order to help another. Therefore, the instructions went on to inform subjects that they would have a chance to win money or receive a shock on the bonus trial and that the more favorable they made the outcomes for the performer, the less favorable these outcomes would be for themselves. Subjects were asked to select 1 of 21 pairs of outcomes that varied from maximally altruistic to maximally selfish. For the most altruistic of the alternatives, the subject agreed to allow the performer to win $2 if the number was even and to escape all shock if the number was odd. The cost to the subject was to agree to receive 100% of a maximally painful shock if the performer lost and no money if he won. The alternatiyes decreased systematically in monetary benefit to the performer (in the event of a win) in proportion to the intensity of shock the subject was willing to accept in the event of a loss. The second most altruistic alternative entailed accepting 90% maximum shock/ no money in order to assign no shock/$1.80 to the performer, etc., until the llth alternative, when neither the subject nor the performer stood the chance of winning any money or receiving any shock. The remaining alternatives entailed a selfish choice. The selfish alternatives increased systematically until the most selfish alternative, in which the subject inflicted a 100% maximally painful shock on the performer if he lost and no money if he won, in order to assign a $2.00/no-shock set of outcomes to himself. Subjects were informed that their decision would be anonymous. They were asked to implement it by depositing whatever dimes they decided to give to the performer in the appropriate slot in an electronically programmed box. They were given to understand that the intensity of the shock would be determined electronically by the number of dimes that they deposited and that they would keep the money they assigned to themselves. The outcome of the "bonus" trial was always a win.
Interview After the bonus trial had been completed, the electrodes were disconnected, and subjects were shown to an interview room. Subjects were asked to reconstruct their experience in the experiment and to describe their cognitive and affective reactions to everything that happened. They were then given a posttest questionnaire that supplied a check on all experimental manipulations, measured the extent to which the plight of the performer was salient to the subjects, and assessed the subjects' impression of the performer. Subjects were asked to rate the extent to which they identified with the
EMPATHY AND ALTRUISM performer and to supply an indication of their affective experience while the "shock," "reward," and no-outcome lights were on. Subjects were debriefed and advised that they would receive a complete description of the experiment after about a month.
RESULTS Checks on Experimental Manipulations It is always possible in laboratory experiments of this sort for subjects to discern the purpose of the experiment and to try to respond accordingly. The posttest interview began with a persistent attempt to discover how the subjects perceived all facets of the experiment. None of the subjects reported believing that the purpose of the experiment was to measure empathy, and none reported attempting to control his psychophysiological responses voluntarily. Responses to the posttest questionnaire indicated that the reward-punishment manipulation was effective. Subjects in the two high-affect groups reported believing that the performer experienced more pleasure while the reward light was on than subjects in the two low-affect groups (M = 8.0 vs. 6.25, on a 9-point scale), t ( 5 6 ) = 2.97, p < .0025, for the one-tailed prediction).3 Subjects from the high-affect groups also reported believing that the performer experienced less pleasure while the shock light was on (M — 2.7 vs. 4.9), t = 6.946, p < .0001. There were no significant differences between the responses of subjects in the high-similar and high-dissimilar groups (M = 8.2 vs. 7.8 for reward and 2.6 vs. 2.8 for punishment) or lowsimilar and low-dissimilar groups (M = 6.3 and 6.2 for reward and both 4.9 for punishment). It should be noted that although subjects in the low-affect group believed that the performer felt quite neutral (4.9 on a 9-point scale) while waiting to make a reflex reaction, they unexpectedly attributed considerable pleasure to the performer as he waited 8 Unless otherwise reported, all t tests followed from a 2 X 2 analysis of variance. The degrees of freedom associated with the t score are the same as those associated with the within-subjects mean square error of the analysis of variance, which was 56 (Lindquist, 19S3). One-tailed tests were employed unless otherwise reported.
1139
for a chip to fall (6.2). The difference between the two reactions was statistically significant (£ = 2.84, p< .025, two-tailed). Responses to the posttest questionnaire also supported the assumption that the similarity manipulation was generally effective. Subjects in the two similar groups rated the performer as significantly more similar to them than subjects in the two dissimilar groups (M = 5.2 vs. 2.7), t = 5.234, p < .0001. There were no significant differences between perception of similarity in the high-similar versus low-similar group (both Ms =5.2). However, subjects in the high-dissimilar group unexpectedly viewed themselves as more similar to the performer than subjects in the low-dissimilar group (Ms = 3.5 and 2.0, respectively), / = 2.21, p < .05, twotailed). Measurement of Empathy In all, 14 indices of empathy were tapped: 3 subjective reports and 11 psychophysiological reactions. In all cases, subjects from the high-similar group were expected to emit the strongest response. The one-tailed tests that follow compare the mean responses of subjects in the high-similar group with the mean response of subjects in the other three groups combined. When subjects from the high-similar group did not evidence the greatest response, the prediction was disconfirmed. Subjective measures of empathy. Subjects in the high-similar group reported identifying with the performer to the greatest extent M = 6.1 vs. 4.8 for the other three groups), t = 2.15, p < .025. Subjects in the highsimilar group reported feeling the worst of all subjects while waiting for the performer to receive a punishment (M = 3.5 vs. 5.0), t = 2.54, p < .01. Self-report measures relating to affective reactions while waiting for the performer to receive a reward failed to differ among groups (F = 1.94). The mean response of subjects in the high-dissimilar group was slightly higher than the mean response of subjects in the high-similar group (6.3 vs. 5.9). The posttest interview indicated that the machine that dispensed chips was a source of great amusement and pleasure to
1140
DENNIS KREBS
subjects in all conditions, and as reported earlier, subjects from all groups believed that the performer enjoyed waiting for a chip to fall. Measurement oj psychophysiological responses. Lacey (1949) and others have shown that changes in galvanic skin response and blood pulse volume generally correlate quite highly with the level of the base from which they originate. (Changes in heart rate are relatively independent of their base). Therefore, in the present study, electrodermal activity was measured in ohms and was converted to log skin conductance. Responses were scored as a proportion of their base. The basal level of skin conductance was denned as the most stable level of skin conductance in the 6 sec prior to the onset of a stimulus. The average correlation between the basal level of skin conductance and the proportion of change to the lights that comprised the main experimental stimuli was .022. Blood pulse volume was measured as a proportion change of raw mV score with the basal level denned as the mean of the maximum and minimum blood pulse volume in the 6 sec prior to the onset of a stimulus. The average correlation between base rate and change to the lights was .066. The measure of arousal associated with a change in blood pulse volume is termed vasoconstriction. Heart rate was measured as the average increase or decrease in beats per minute from a basal level that was defined as the mean of 1.050 i— HIGH AFFECT-SIMM AR
HIGH A F F E C T - D I S S I M I L A R
LOW AFFECT-SIMILAR LOW AFFECT-DISSIMILAR
FIGURE 1. Proportion of change in log skin conductance to instructions.
1.900 1.800
i.roo u o < o
1.600 1.500
o 1.400 \° 1.300 o a-°~
HIGH AFFECT-DISSIMILAR
1.200 LOW AFFECT-SIMILAR 1.IOO 1.000
A LOW A F F E C T - D I S S I M I L A R
.900 FIGURE 2. Proportion of change in blood pulse volume to instructions.
the maximum and minimum beats per minute in the 6 sec prior to the onset of stimulus (cf. Craig, 1968). The mean correlation between raw heart rate (number of beats per minute) and heart rate change to the onset of the three lights was .152. The relationship between heart rate and vicarious arousal is more complex than the relationship between skin conductance and vasoconstriction and vicarious arousal. Heart rate acceleration is associated with heightened arousal when arousal occurs over a period of time, but heart rate deceleration is associated with attentiveness during the immediate intake of information. Thus, the appropriate measure of empathy is heart rate deceleration immediately after stimuli signaling that another will be rewarded or punished or heart rate acceleration while waiting for another to receive rewards or punishments (cf. Craig, 1968; Craig & Lowery, 1969). Psychophysiological reactions to the instructions. Figures 1-3 display the mean changes in basal level of skin conductance, vasoconstriction, and heart rate from the final habituation trial to the first conditioning trial, the time during which subjects from the high-affect groups learned that the performer could win money or receive shocks and subjects from the low-affect groups learned that the performer would perform a less
EMPATHY AND ALTRUISM
HIGH AFFECT-SIMILAR
LOW AFFECT-SIMILAR
HIGH AFFECT-DISSIMILAR LOW A F F E C T - D I S S I M I L A R
FIGURE 3. Beats-per-minute change in heart rate instructions.
arousing task. Observers in the high-affect groups were expected to input stronger emotional responses to the performer (especially fear of receiving a shock) than subjects in the low-affect groups. Subjects in the high-similar group were expected to be most prone to experience the responses vicariously and therefore to evidence the strongest psychophysiological reactions. Subjects from the high-similar group evidenced the greatest increase in skin conductance (t = 2.27, p < .02), vasoconstriction (£ = 2.80, p < .005), and
1141
heart rate acceleration (t = 1.12, p < .12) to the instructions. Subjects in the high-dissimilar group emitted the second strongest responses. Psychophysiological reactions to the reward light. Figures 4-6 display the mean psychophysiological reaction of subjects to the light that signaled a reward/chip outcome. Subjects in the high-similar group evidenced the greatest increases in skin conductance (t = 2.01, p < .025, heart rate deceleration (t = 1.65, p = .06), and heart rate acceleration (t = 1.30, p< .10). In all cases, skin conductance and vasoconstriction were denned as the maximum increase during the 12 sec in which the lights were on, heart rate deceleration was denned as the maximum deceleration during the 6 sec after the onset of lights, and heart rate acceleration was denned as the maximum acceleration during the final 6 sec (in anticipation of the outcome). There were no significant differences between groups in vasoconstriction. Subjects from the high-dissimilar group emitted the second strongest responses to the light that signaled reward. Psychophysiological reactions to the shock light. Figures 4-6 also show that subjects in the high-similar group tended to react most strongly to the light signaling that the per-
O HIGH AFFECT-SIMILAR • HIGH AFFECT-DISSIMILAR A LOW AFFECT-SIMILAR A LOW AFFECT-DISSIMILAR
REWARD/CHIP LIGHT
SHOCK/PUNISH LIGHT
NO OUTCOME LIGHT
FIGURE 4. Proportion of change in log skin conductance to lights.
DENNIS KREBS
1142
O HIGH AFFECT-SIMILAR
1.3200
• HIGH A F F E C T - D I S S I M I L A R A LOW AFFECT-SIMILAR
1.2800
A LOW AFFECT-DISSIMILAR
1,2400
5 1.2000 p 1.1600 cc 1.1200 ce 1.0800 1.0400 1.0000 REWARD/CHIP LIGHT
SHOCK/REFLEX LIGHT
NO OUTCOME LIGHT
FIGURE 5. Proportion of change in blood pulse volume to lights.
former would receive a shock. They evidenced the greatest increase in skin conductance (t — 2.35, p < .02) and heart rate deceleration (£ = 2.38, p
