Journal ol Personality and Social Psychology 1975, Vol. 32, No. 1, 92-101

Choosing to Suffer as a Consequence of Expecting to Suffer: Why Do People Do It? Ronald Comer and James D. Laird Clark University

In three previous studies, most subjects who anticipated having to perform an unpleasant task (e.g., eat a worm) and who were then offered a choice of that or a neutral task chose the unpleasant task. It was hypothesized that this "choosing to suffer" effect was produced by the subjects' attempt to make sense of their anticipated suffering, through one of three kinds of change in their conception of themselves and the situation: (a) they deserved the suffer- ing, (b) they were brave, or (c) the worm was not too bad. Subjects antici- pating a neutral task did not change on these measures, while subjects eating a worm changed significantly, indicating that anticipated suffering did cause the predicted reconceptualizations. Changes in these three kinds of conception predicted subject's choice or not of eating the worm, indicating that these changes produced the subsequent choice. Finally, subjects who saw them- selves as either more brave or more deserving also chose to administer painful electric shocks to themselves, while subjects who had changed their concep- tions of the worm did not, indicating that these conceptual changes can affect behavior in situations different from that which initially caused the conceptual changes.

In a few more or less unrelated studies, a man & Radtke, 1970). Clearly then, some- particularly paradoxical bit of behavior has thing happens during the interval between a been observed, that subjects often "choose" subject's assignment to a negative event and to suffer as a consequence of having expected his subsequent choice which influences most to suffer. For example, when subjects were subjects to choose to perform the negative assigned an unpleasant task (e.g., eat a cater- event though they no longer must. pillar) and given a short time to wait in this Foxman and Radtke (1970) proposed an state of "negative expectancy," approximately "anxiety adaptation" hypothesis, in which the 80% chose this unpleasant task when given subjects were presumedly becoming more the opportunity to perform a neutral task accustomed to feeling anxious and hence did instead (Aronson, Carlsmith, & Darley, 1963; not mind choosing the caterpillar. This expla- Foxman & Radtke, 1970). Moreover, such nation suffers from being rather vague about "negative expectancy" subjects tend to shock the mechanisms involved and, indeed, could be themselves more during the expectancy period taken as no more than a description of what than do other subjects (Walster, Aronson, & happened: that a previously noxious stimulus Brown, 1966). became less so. In addition, it does not really Neutral expectancy control subjects who make a great deal of sense to posit some were assigned to perform a neutral task be- mechanism, similar presumedly to sensory have as common sense suggests: No neutral adaptation, such that not only does the stimu- expectancy subject in any of the studies has lus lose some of its potency, but, in fact, ever chosen a negative task. In addition, sub- "reverses" and now becomes positive. In the jects who are given a negative expectation, other two studies of this sort (Aronson et al., but given no period of time alone to await 1963; Walster et al., 1966), the explanations the negative event, also do not choose the offered were post hoc (in fact opposite predic- negative event (Aronson et al., 1963; Fox- tions had been made by Walster et al.), tenta- tively offered hypotheses drawn from cogni- Requests for reprints should be sent to James D. tive dissonance theory and, indeed, both of Laird, Department of Psychology, Clark University, these authors confess to uncertainty and dis- Worcester, Massachusetts 01610. satisfaction with their own explanations. 92 CHOOSING TO SUFFER AS A CONSEQUENCE OF EXPECTING To 93

The present study represents an attempt were simply trying to make sense of their im- to understand these effects of negative expec- pending suffering and, somehow, the sense tancies in terms of the kinds of cognitive they made of it led them to then go on and changes which might be expected to occur in choose to suffer when the opportunity to subjects in these situations. The theoretical escape suddenly became available. Of course, perspective adopted here has been called "self- it in fact made no sense for them to try to perception theory" (Bern, 1972) and "self- make sense of their suffering since their as- attribution theory" (Kelley, 1967). However, signment to the negative conditions was it is clear that cognitive dissonance theory clearly described as due to chance. They (Festinger, 1957) could as easily encompass "ought" to have thought merely that it was the results reported here. The use of the self- unfortunate that they would have to endure attribution language should be understood as this unpleasantness and, then, have leaped at representing the authors' preferences, rather the chance to avoid it. Obviously, however, than any suggestion that one or the other they did not follow this line of thought and perspective is more adequate in this context. action. In the self-attribution view, the experiences There is a clearly parallel phenomenon in we all have of our "inner, mental states" are the realm of other attribution. In a series of instead taken to be interpretations of our studies, Lerner and his associates have dem- behavior and the context of its occurrence. onstrated that subjects who are exposed to Feelings or awareness of our attitudes, emo- another person purportedly suffering, but tions, motives, etc. are seen, then, as informa- chosen for the suffering quite by chance, tion about our behavior or our behavioral dis- apparently are unable to "believe" in the ef- positions and not as causes of the particular fects of chance and instead treat the sufferer ongoing behaviors. These bits of "informa- as if he deserved his fate (Lerner & Simmons, tion" are presumed to have the same kinds of 1966). Since the root assumption of self- effects on subsequent behaviors that any attribution theory is that people understand other information might have. Thus, we might and know themselves in precisely the same assume that the sequence of events begins way in which they know others, it would not with an individual doing something, noticing be surprising to find that a similar failure to what he is up to and interpreting his behavior accept the workings of chance occurs with in the form of attributing some property to respect to self. himself and then subsequently acting in the There seems to be three major ways of ways in which he believes someone who pos- "making sense" of the negative expectancy sesses that property would or should behave. situation. One has already been mentioned; As an example, consider a man who is con- as suggested by Lerner's work, the person fronted by some intense frustration. He re- may conclude that he deserves his fate, that sponds with physiological arousal, a scowl, he is a bad person. A second alternative is clenched fists, and a harsh tone of voice. He that the person may attribute to himself some then becomes aware of these behaviors in a "positive" qualities, for instance, that he is a highly integrated way, which he experiences brave person or a person who is willing to simply as the feeling of "anger." At that sacrifice for the sake of science. This possibil- point, he may then try to inhibit his "angry" ity was implied by one of Walster et al.'s behavior or find an appropriate target, or (1966) dissonance explanations of the nega- assess his chances of winning a fight, or any tive expectancy phenomenon. A third possibil- number of other things. The point is that the ity is that an individual may not change his feeling of anger was a product of one set of conception of himself at all, but instead behavior, in this case mostly automatic, and change his conception of some aspect of the subsequently affected some other quite dif- external circumstances, particularly of the ferent kinds of behavior (Laird, 1974). negative task. For example, the negative task Applying this theory to the phenomenon may be seen as not as unpleasant as it first of negative expectancy, the general assump- seemed, so that, for instance, the subject tion is that the subjects in these experiments might have concluded that eating caterpillars 94 RONALD COMER AND JAMES D. LAIRD

was really not so bad after all. This notion themselves as opposed to causes in their situ- has been involved indirectly in some of the ations. Individuals who tended to assign the dissonance explanations offered by Aronson causes of their behavior to themselves might et al. (1963). be expected to employ either the "deserving- The central hypothesis of this study was ness" or the "bravery" options in the negative that the paradoxical effects of expecting to expectancy situation, while people who see suffer on later choices were due to subjects the causes of behavior as residing more in changing their conceptual systems in one or situations might be expected to change their another of these three ways. The general form evaluation of the worm. of the study was, then, to introduce subjects The second group of hypotheses and expec- to a negative expectancy situation, measure tations in this experiment involved the rela- changes in their conceptions of themselves and tionship of subjects' attributions to their sub- the situation, and to relate these changes to sequent choice behavior. First, insofar as the their subsequent behavior. subjects' attributions are seen as the cause of The negative expectancy format employed the subsequent negative choice behavior, it was similar to that used by Aronson et al. was expected that experimental subjects who (1963) and Foxman and Radtke (1970). A did not later choose the negative event would subject was told he will eat a worm, sat wait- be individuals who had not reconstrued them- ing for the negative event for 10 minutes, then selves or the negative situation. was given a choice between the negative event Second, if one decides that the worm one is and a neutral event. Measures of the subjects' going to eat is not really so bad, then one has conceptions of themselves and the situation made a relatively small change in his view of were obtained near the end of the waiting the world, one that has implications only for period prior to the subsequent behavioral a restricted range of future actions, all in- choice. These measures were then compared to volving worms. The individual who changes identical measures obtained a month earlier to his conception of himself, on the other hand, assess changes in either self-conception or has made a change of much broader potential conceptions of the situation which might have importance. If he deserves this piece of suf- occurred during the expectancy period. With- fering, perhaps he also deserves other kinds of in these measures there were separate sub- distress. If he is brave and sacrificing, he must scales measuring each of the domains which be so in all situations. The empirical implica- seemed to offer some opportunity for "making tion of this distinction is that a person who sense" of the anticipated suffering. changes his view of himself should choose to Two additional groups of hypotheses de- suffer, whatever the suffering and its relation- pended on the observation of these reconcep- ship to his original expectations. The person tualizations. The first concerns the relation- who changes his conception of the specific ships among the alternative strategies of con- negative event, on the other hand, should ceptual change. While discussion thus far had perhaps be willing to choose that event later, assumed the likelihood that one of the three but there is no reason for him to choose some available directions (negative self-attribution, new negative experience. This line of thinking positive self-attribution, or external attribu- was assessed by the addition of another ex- tion) would be preferred, this assumption ob- perimental group that was given the original viously required testing. Then, if it appeared worm-eating expectation, but then following that individuals do, in fact, employ only one the waiting period were offered the choice of of the possible alternatives, these behaviors the neutral task and a different negative task, could be related to other indices of "attribu- giving themselves painful electric shocks. The tional style." In a recent study (Laird & specific hypothesis was that those who had Berglas, in press) a scale, the Locus of changed their views of their own bravery or Causal Attribution Scale, was developed deservingness would also choose the shock, which reflected the degree to which individu- whereas those who had only changed their als attribute their behaviors to causes within views of the worm would not. CHOOSING TO SUFFER AS A CONSEQUENCE OF EXPECTING To 95

METHOD there was no expectation that the subjects would be Subjects aware of the change in their own evaluation, they would have a reasonably accurate idea of how Fifty subjects (27 males and 23 females) partici- others must feel and would be able to compare how pated in the study. All were freshmen and sophomore they felt now with that standard. Fortunately, the students in introductory psychology courses. Sub- experiment contains within it some check on these jects were assigned to one of three conditions (worm assumptions. One would predict, if these assump- expectancy and worm choice, 15 subjects; worm tions were true, that the negative expectancy and expectancy and shock choice, 20 subjects; neutral neutral expectancy groups would not differ in their expectancy control condition, IS subjects). The male- assessments of where the general "other" was, and female distribution was approximately equal for each that among the neutral expectancy group there condition. No differences between the sexes were would be little if any difference between the evalu- observed in any of the dependent measures. ations of self and other people. Among the negative expectancy group there would be expected to be at Measures least some individuals who showed marked discrep- The scale to measure hypothesized changes in con- ancies between their judgments of themselves and of ception of self consisted of 22 statements about the other. oneself, each of which was to be rated on a 14-point A scale to measure characteristic styles of re- continuum ranging from "not at all true" to "very sponsibility assignment was also administered (Locus true." Six statements were expected to reflect nega- of Causal Attribution Measure). It was devised tive self-attribution (deservingness). These state- (Laird & Berglas, in press) from a study by Mc- ments were modifications of those used on a scale Arthur (1972) and was administered to subjects in measuring guilt (Coopersmith, 1967) and included order to premeasure the general direction to which a subject attributes responsibility (i.e., to persons or items such as: "I wish that I could make up for 1 something(s) that I've done recently." situations) . Two statements were expected to reflect positive At the conclusion of the experiment, before de- self-attribution. They were derived from "disso- briefing, subjects were administered a brief question- nance" interpretations in the Walster et al. (1966) naire to ascertain the degree to which they had study. The positive self-attribution statements em- believed and accepted the "cover story" of the study. ployed in the questionnaire were: "I am a brave person" and "I am a sacrificing person." Procedure Dembo (1960) has found that suffering or un- At the time subjects were recruited, they were told pleasantness is frequently interpreted to be desir- that the experiment was concerned with the rela- able. Accordingly, four statements were included in tionship between various physiological measures and the survey to ascertain the degree to which a changes in personality variables following the per- subject may "generally" view a negative situation in formance of "certain tasks." Thus, the subjects a positive manner, including as an example, "I would spend 10 minutes now completing a pretask want to suffer sometimes to put things in proper personality survey (actually the "deservingness" and perspective." "bravery" premeasures) and 40 minutes next week The remaining 10 statements were filler items. All performing an experimental task and then completing statements were randomly mixed for presentation on the personality survey again. In fact, each subject the questionnaire, which was presented to the sub- was not contacted until 1 month later and an ap- jects as a personality survey. Each subject filled out pointment was made for the second part of the the survey twice (before and during expectancy study. period). When the subject entered the experimental room, The measure of change in "worm-eating pleasant- he was greeted at a table; on one side of the table ness" could not be so straightforward. Asking an were eight covered cups which weighed different individual to rate how pleasant or unpleasant eating amounts (neutral weight discrimination task); a worm would be is unfortunately a very reactive across the table was a plate with a dead worm on measure. Once you have asked a subject that un- it, a cup of water, a napkin, and a fork (worm- usual question, he is likely to remember the ques- eating task); and in Experimental Condition B on tion and his answer for a long time. Thus, it was a third side was an inductorium system for shocking not possible to ask subjects for a preassessment and oneself. postassessment of the desirability of eating a worm. The experimenter immediately read a statement Instead a different strategy was employed. Subjects describing the psychology department's policies on were asked at the time of the postmeasure on the participation in research and reminding the subjects self-conception questionnaire to rate on a scale of they were "free to refuse or to terminate partici- 16 the pleasantness of the worm-eating task and then to rate how they thought other people would rate the same thing. The latter was interpreted as Hidden Figures Test-form V (copyright reflecting what the subjects' ratings of the worm- Educational Testing Service) was also administered, eating would have been at the beginning of the but no relationships were predicted and none were experiment. The rationale for this was that although observed. 96 RONALD COMER AND JAMES D. LAIRD

paling at any time in a study if, for any reason, you did not dispute their assignment to the condi- find it difficult to participate." tion of negative expectancy and, at least as Subjects in all conditions then performed the weight discrimination task so as to familiarize them revealed in the postexperimental questionnaire with the neutral task. The experimenter then ex- and interviews, fully expected to eat the plained that the subjects had been randomly as- worm, and both they and the control subjects signed to perform one of the three tasks, following believed the purpose of the study as described which a number of physiological measurements to them. would be obtained, as well as a second "personality survey." The experimenter then briefly described the experimental tasks, leafed through his papers, an- Negative Task Choices nounced to the subject his condition, seated the sub- ject at the table in front of his task, and left the The results with respect to "choosing" to room. suffer replicated those of Aronson et al. The experimenter returned in 10 minutes, explained (1963) and Foxman and Radtke (1970). All there would be a slight additional delay and said, of the neutral expectancy subjects chose the "While you're waiting, I was thinking, it's been neutral task when the choice was offered. In quite a while since you filled out the pretask person- ality survey—I don't even know if it's valid any- the worm-choice negative expectancy condi- more. Would you just fill it out again now before tion, 12 of the IS subjects chose to eat the you eat the worm [discriminate the weights] ? That worm or said that they did not care either way, I'll be able to judge the one you fill out after way. In the other negative expectancy condi- the task a lot better." In this way the subjects were induced to provide tion, 10 of the 20 subjects chose to give them- the second measure of deservingness, bravery, etc. selves painful electric shocks. Differences be- under conditions which led the subject to believe tween the two negative expectancy conditions that the focus was on change from this second mea- and the control condition were tested by Fish- sure to a third (actually never administered) mea- er's Exact Test (Siegel, 1956) and yielded sure. The experimenter returned in 8 minutes and asked significant differences (p = .00001 and p — subjects to evaluate the tasks (worm eating and .002, respectively). weight discrimination). The format then varied only with respect to the negative choice offered. To con- Validation Analyses of Worm-Attractiveness trol subjects and "worm-choice" subjects, the choice was between weight discrimination and eating a Measure worm, while for the shock-choice condition, shock The pattern of results with respect to the was substituted for worm eating. While the subject "worm-attractiveness" measure fits very filled out the task evaluation sheet, the experimenter began leafing through his papers and then said, "Oh, closely the pattern required to support the an error has been made. You weren't exactly assigned interpretation of this as a measure of change to the right condition. You actually are supposed to in the attractiveness of eating a worm. The choose which task you will perform, between eating neutral and negative expectancy groups did the worm '[giving yourself a shock on the hand] or discriminating the weights. That is, you choose not differ significantly in their judgments of either the worm [shock] or the weight discrimina- other people's opinions (mean negative ex- tion. It's up to you; which will you perform?" If pectancy group = 2.8, mean neutral group = the subject chose the worm or shock, the experi- 1.9). Furthermore, the reports of the neutral menter then said, "Actually in light of the misrepre- expectancy group's own opinions of the worm's sentation at the beginning, it would be best if you perform the weight discrimination task." palatability and their judgments of the opin- The subject then responded to a questionnaire ions of other people were very nearly identi- designed to determine whether he perceived the cal, the largest discrepancy being 2 units on actual purpose and focus of the experiment and then the 16-point scale. In the negative expectancy was totally debriefed concerning the focus, purpose, and slight deceptions of the study. groups, on the other hand, many subjects showed considerably greater discrepancies, the RESULTS largest being 10. Thus, the preliminary analy- In the experimental condition, 3 subjects ses all support the use of this measure. refused to eat a worm even before the choice between the negative event and the neutral Changes in Conceptions of Self and Situation event was revealed to them. They were ex- Since the experimental conditions were cused. The remaining 35 experimental subjects identical for both experimental groups until CHOOSING TO SUFFER AS A CONSEQUENCE OF EXPECTING To 97

after these measures were obtained, the two Fourteen of the negative expectancy sub- groups were pooled for these analyses. jects had change scores of 3 or larger and Comparing all of the subjects on each of the were designated "worm attributors." These three critical measures separately, no signifi- subjects were presumed to have "made sense" cant differences between the experimental of their negative expectancy by changing their subjects and the neutral expectancy subjects opinion of how bad it would be to eat a worm were found either in initial scores or in the and would be expected not to have change hypothesized changes. A lack of differences in scores on the self-attribution measures above these overall analyses would be expected, those of the control group. This was, in fact, however, since if there were consistent ten- the case. The mean total self-attribution dencies to employ one, but only one of these change score of the worm-attributor group strategies, these analyses which compare all was 7.43, nonsignificantly below the control of the subjects on each of the three measures group mean of 8.60 (t < 1.0) both of which underestimate the degree of change on each. were significantly lower than the mean of the If, for instance, a subject had decided that remaining negative expectancy subjects of worm eating was not so bad, he would be 12.90 (worm-attributor vs. other negative unlikely to also decide that he was a hero or expectancy subjects, t = 2.38, p < .05; con- deserving of suffering, since he had recon- trol vs. other negative expectancy subjects, strued the situation such that he was no longer t = 1.90, p < .10). This latter group was then as much of a hero or sufferer as he might have designated as the "self-attributors." been. Among the self-attributors, the change Accordingly, an index combining all three scores for the bravery and deservingness mea- of these possible strategies was developed. sure were highly negatively correlated (r — Change scores for both experimental and con- -.65, N = 2l, /K.01). Subjects who trol subjects on each of the variables were changed in one way did not change in the converted into standard scores and then the other. They did not come to think of them- total of the three for each subject was ob- selves as both more brave and more deserving tained. The mean of the sum of these stan- of suffering. Instead they tended to think of dard scores for the subjects in the two experi- themselves as more deserving of suffering and mental conditions was .36, while the mean less brave or the converse. Since these changes sum of standard scores for the subjects in were presumedly due to attempts to make the control condition was —.93. This differ- sense of the negative expectancy situation, ence was significant, ^(48) = 2.5S, p < .02. one would not expect similar correlations In this more appropriate analysis, it is clear among either the worm attributors, who had then that the negative expectancy situation made sense of it another way, or the neutral had affected subjects' constructions of them- expectancy subjects who had nothing to make selves or their situation. sense of. This also was confirmed, in that the To assess properly the role of these three correlations between the bravery and the de- alternative strategies and their interrelation- servingness scores among both these groups ships, it was necessary to "sort" subjects ac- were positive, .31 and .43, respectively, rather cording to their dominant mode of response to than negative. Although neither of these cor- the negative expectancy situation. The major relations are significant, they suggest that in dimension of variation was between those sub- these groups changes on these scales reflected jects who reconstrued their conceptions of some general differences among subjects in themselves and those who changed their opin- the variability of their responses. ion of the worm. As noted above, among the In sum, the negative expectancy subjects control subjects the largest change in "worm subdivide fairly neatly into three groups. The preference" was 2, while among the negative two major divisions were: (a) the worm at- expectancy subjects much larger changes oc- tributors whose increases in liking for the curred. Accordingly, this largest change ob- worms were all greater than the greatest such served among the control subjects was chosen increases in the control group, and whose as the cutting score. changes in conceptions of themselves were RONALD COMER AND JAMES D. LAIRD

essentially the same as those of the control tion change of subjects in the worm-choice group; (b) the self-attributors, whose worm- condition who did or did not choose to eat liking scores had changed no more than the the worm. This composite index of attribution changes observed in the control group, but change is described above and consists for who had changed their self-conceptions sig- each subject of the sum of his normalized nificantly more than either the control group scores on each of the three alternative ways or the worm attributors. This latter group, of reconstruing the situation. Since only the self-attributors, consisted of subjects who three subjects did not choose to eat the worm, had employed one or the other of two highly parametric procedures were not appropriate. negatively correlated strategies, coming to However, the three who did not choose the view themselves as deserving of suffering and worm had the smallest, next smallest, and less brave ("negative self-attributors"), or fourth from smallest change scores in this those who had come to view themselves as group. The probability that this array of heroic and helpful and less deserving of suf- scores would occur by chance was assessed by fering ("positive self-attributors"). the Randomization Test for Two Independent Samples (Siegel, 19S6). The exact probabil- Correlates of Attribution Strategy ity, two-tailed, of this outcome is .009. It is In order to test the expectation that a clear, then, that the degree to which the sub- subject's attribution focus (self or worm) jects changed their conceptions, either of would reflect a general tendency to make themselves or of the worms, predicts very sense of the world in terms of locating causes powerfully whether they would choose the in persons or situations, the Locus of Causal negative event subsequently. Attribution (LCA) scores of self-attributors In the worm-choice condition, in which sub- were compared with those of worm attributors. jects were choosing whether to submit to the Self-attributors did indeed have significantly negative event they had previously expected more "person" oriented LCA scores than to have to endure, there were no differences did worm attributors. The mean for the self- between the compliance rates of the worm attributors was 7.1, while the mean for the attributors and the self-attributors. Of the IS worm attributors was 5.6, where scores could subjects in that condition, 7 were assigned to range from 14, exclusively person oriented, to the worm-attribution group and 8 were as- 0, exclusively situation oriented. This differ- signed to the self-attributor category. Six of ence was significant (t = 2.21, p < .05, two- each of these groups chose to eat the worm. tailed). In the shock-choice condition, of the 20 One additional measure of change assessed subjects, 7 met the worm-attributor criterion whether suffering in general might have come and of these only 1 chose to shock himself, to be seen as in some way good in itself, for whereas, of the remaining 13 self-attributors, example, a kind of induced masochism. How- 9 chose to shock themselves. The difference ever, there were no significant changes in this between the choice rates in the shock versus variable, in any of the groups, nor any dif- worm conditions among the external attribu- ferences between the neutral expectancy group tors is significant, as tested by Fisher Exact and any of the experimental groups. Test (P = .03). The differences between the choice rates of the "self-attributor" group and Relationships oj Changes in Conception of the worm-attributor group in the shock-choice Self and Worm to Subsequent Behavior condition is marginally significant, again as It is clear at this point that the ngeative tested by Fisher Exact Test (P = .057). In short, then, it seems that not only does expectancy led some subjects to change their the amount of change subjects make in their conceptions of themselves or of the worm they understanding of the negative expectancy were to eat. The relationship between these situation relate to the likelihood they will changes in conceptions and the subsequent later choose to perform the negative event, behavioral choices was assessed by comparing but the kind of change they make in their the scores on the composite index of attribu- conception of the situation relates to the kinds CHOOSING TO SUFFER AS A CONSEQUENCE OF EXPECTING To 99 of subsequent choices of negative experiences One possible explanation of the simple they will make. choice to eat a worm or shock oneself is that subjects simply chose the task they had been DISCUSSION assigned to initially, or the task that was most As anticipated, people apparently chose to like it. However, that kind of "task inertia" "suffer" after expecting to suffer because explanation does not explain why subjects during the waiting period they changed their changed their conceptions of themselves or view of themselves and the situation in one or the situation, or why these changes were another of three ways. Some people seem to related to subsequent choices. It especially have made the choice because they came to does not explain the differential rates of see themselves as at least to some degree choosing among the worm attributors and the heroic figures who willingly would endure dis- self-attributors in the shock-choice condition, comfort for the good of science, while others since any inertia should be identical for both. seem to have come to believe that they de- A third alternate explanation has been that served the suffering to come, while still a the results are really due to experimenter third group simply decided that if they were "demand" or bias (Rosenthal, 1966). Experi- going to endure it, it could not really be that menter demand might explain the fact that bad after all. subjects chose to eat a worm in the worm- Furthermore, it appears that individuals choice condition, and in the previous expect- tended to adopt only one of these strategies ancy studies, in which the comparisons were and that the nature and degree of the re- always between groups of which the experi- conceptualization an individual performed menter was well aware. However, the focus in had important effects on his subsequent be- this study was not on the simple fact of havior. Any one of the three was sufficient choice, but on the conceptual changes which to induce him to freely choose to endure the produce these choices. The measures of these expected experience, one which everyone in conceptual changes were scored after the our culture would agree was extremely un- study was completed so that the experimenter pleasant. And if his response was either of could not have known consciously or other- the modes of change in self-conception, then wise at the time which subjects to influence it would also lead him to choose other, to choose the negative event and which not to. unrelated kinds of unpleasantness. It seems clearly impossible to explain the A number of possible alternative explana- results in these terms then. tions could be suggested for these results There may be other possible explanations, to avoid the conclusion that people act as particularly for parts of these findings, but "irrationally" as this interpretation implies. no alternative seems to encompass all of them For instance, it might be objected that the as comfortably as that which generated the subjects may not have believed that they study in the first place. It appears that the would really have to eat the worm. In this "irrational" decisions to perform a disgusting study the strongest evidence against this point or painful action were guided by the prior, is that no subject revealed at any time, either in many senses even more irrational, ways on the postexperimental questionnaire or in in which people had attempted to make sense the interviews after debriefing, that he ex- of a situation which, in fact, made no sense. pected not to have to eat the worm. It seemed That is, the subjects could not deal with the clear at the time that the squeamishness of fact that chance and chance alone had led the experimenters was the only obstacle to the them into an extremely unpleasant situation. subjects' consumption of the worms and, in Even though they had been told that a ran- fact, one subject had already cut his worm up dom number table had selected them for the into bite-size pieces during the waiting period. negative rather than the neutral condition, Furthermore, Foxman and Radtke (1970) they still "made sense" of this impending un- apparently let their subjects eat the cater- pleasantness. Although it is superficially pillars which were their stimulus objects and striking to observe people choosing to eat a the subjects apparently did eat them. worm, the more impressive aspect of these 100 RONALD COMER AND JAMES D. LAIRD findings is the degree to which people will the worm and their scores on the Locus of change their conceptual system to make sense Causal Attribution Scale. The Locus of Causal of the random events of their lives. Attribution Scale has previously been found This is not meant to imply that they made (Laird & Berglas, in press) to relate to con- these changes intentionally, or even that they sistent stable differences in how people re- realized how they had changed. In fact, in no spond to a forced-compliance situation. Thus, case did any subject reveal any awareness of there is some justification for believing that the kinds of changes in his view of things that the Locus of Causal Attribution Scale does had, in fact, taken place. A more common measure enduring differences in the way peo- response was to deny them stoutly. In this ple organize their understanding of the behav- respect these observations closely parallel ior of persons, both others and themselves. what seems to take place during the attitude Since this kind of variation is also consistent "change" that occurs following counteratti- with the relationship observed here, it seems tudinal behavior. Subjects do not seem to be reasonable to conclude at least tentatively aware that any change has taken place at all that people may well differ consistently in nor that there is any discrepancy between whether they make sense of the adversities their behavior and previous attitudes (Bern that befall them in terms of their own & McConnell, 1970; Berglas, Note 1). In attributes or those of the adversity. short, the subjects in this study were making Finally, it is not clear to what extent we major changes in their beliefs about them- might expect to see the particular strategies selves or some aspect of their world, without of conceptualization that were observed here realizing that such changes were occurring and in operation in real life. We have probably then, worse yet, acting on these new concep- all encountered, in both ourselves and others, tions. We must, it would seem, question the the tendencies to take credit in some way for appearance of continuity and consistency outcomes which "really" happened by chance which all of us experience about our lives, or to decide that the bad things that befall us since it would seem that this may be as likely may be all for the good somehow. It seems to have been constructed after the fact as much more improbable that we should decide to have been what actually happened. to derogate ourselves because we have fallen One point that should not be overlooked on difficult times. However, in fact, pre- here is that some of the subjects did not make cisely that kind of behavior seems to be char- this "attribution error" and subsequently also acteristic of those people who are labeled did not choose to suffer. It may be these are "depressed." Beck (1967) has pointed out simply the more sensible people, but perhaps that depressed individuals blame and dislike they just do not get involved in the experi- themselves for the widest variety of events. ment in quite the right way. At this juncture Similarly, it is not uncommon to observe that we have no way of knowing what distin- some people suffering terminal illnesses or guished those "sensible" subjects from the other catastrophes increase their sufferings by others and whether they were really more their feeling that they "must have done sensible. something wrong to deserve this." The discussion thus far has implicitly as- In short, then, this study seems to have sumed that the particular reconceptualization demonstrated that the previously observed strategies employed by the subjects were ex- but not at all clearly understood effects of emplars of a general style of "making sense" "expecting to suffer" were due to the subjects' of the world. Of course, this is in no way attempts to justify to themselves this antici- justified by these data, which do not provide pated suffering. To the extent that we can the necessary sampling of reconceptualiza- generalize to the world at large, this study tions. However, there is at least one piece of suggests that much of the apparent stoicism data that does support, albeit not conclu- of the world's victims may be due to their sively, the notion that these are general reconstruction of their views of themselves styles. This is the relationship found between and the world, so that their sufferings are seen whether subjects reconstrued themselves or as either not so bad or no more than appropri- CHOOSING TO SUFFER AS A CONSEQUENCE OF EXPECTING To 101 ate for people such as themselves. As a short- Dembo, T. A theoretical and experimental inquiry term solution to their immediate problems, into concrete values and value systems. In S. Wapner & B. Kaplan (Eds.), Perspectives in psy- such reconstructions are obviously helpful and chological theory. New York: New International might be described as simply "cooperating Universities Press, 1960. with the inevitable." However, if that suffer- Festinger, L. A theory of cognitive dissonance. Stan- ing becomes no longer inevitable, then the ford, Calif.: Stanford University Press, 1957. results of this study must make us wonder if Foxman, J., & Radtke, R. Negative expectancy and the choice of an aversive task. Journal of Person- these adaptations themselves may now be- ality and Social Psychology, 1970, IS, 253-257. come impediments to the amelioration of the Kelley, H. H. Attribution theory in social psychology. suffering. In D. Levine (Ed.), Nebraska Symposium on Motivation (Vol. 15). Lincoln: University of Nebraska Press, 1967. REFERENCE NOTE Laird, J. D. Self-attribution of emotion: The effects 1. Berglas, S. The experience of dissonance and of expressive behavior on the quality of emotional "dissonance-induced" change. Paper presented at experience. Journal of Personality and Social Psy- the meeting of the Eastern Psychological Associa- chology, 1974, 29, 475^86. tion, Boston, April 1972. Laird, J. D., & Berglas, S. Individual differences in the affects of engaging in counterattitudinal be- havior. Journal of Personality, in press. REFERENCES Lerner, J. J., & Simmons, H. Observer's reaction to Aronson, E., Carlsmith, J. M., & Darley, J. M. The the "innocent victim": Compassion or rejection? effects of expectancy on volunteering for an un- Journal of Personality and Social Psychology, pleasant experience. Journal of Abnormal and 1966, 4, 203-210. Social Psychology, 1963, 66, 220-224. McArthur, L. A. The how and what of why: Some Beck, A. T. Depression. New York: Harper & Row, determinants and consequences of causal attribu- 1967. tion. Journal of Personality and Social Psychology, 1972, 22, 171-193. Bern, D. J. Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental social psychology Rosenthal, R. Experimenter effects in behavioral research. New York: Appleton, 1966. (Vol. 6). New York: Academic Press, 1972. Siegel, S. Nonparametric statistics for the behavioral Bern, D. J., & McConnell, H. K. Testing the self sciences. New York: McGraw-Hill, 1956. perception explanation of dissonance phenomena: Walster, E., Aronson, E., & Brown, Z. Choosing to On the salience of premanipulation attitudes. Jour- suffer as a consequence of expecting to suffer: An nal of Personality and Social Psychology, 1970, unexpected finding. Journal of Experimental Social 14, 23-31. Psychology, 1966, 2, 400-406. Coopersmith, S. The antecedents of self-esteem. San Francisco: W. H. Freeman, 1967. (Received March 8, 1974)