Leary, M. R., & Miller, R. S. (1986). Social psychology and dysfunctional behavior: Origins, diagnosis, and treatment, (pp. 147-163). New York: Springer-Verlag Inc.
Chapter
2
Attributional Processes
Imagine that you are a counselor meeting a client who is newly divorced. When you ask about the circumstances, he has a ready explanation: Although he partly blames his ex-wife, he shoulders most of the responsibility himself. With somber affect, he declares that his wife's faithlessness was surely a result of his own inadequacy, that his failure with her shows he is not cut out for marriage, and that, indeed, he cannot imagine anyone ever being attracted to him again. For him, the evidence is clear: He simply does not measure up in the interpersonal marketplace.
Your depressive client would make clear two important points. First, people seek explanations for important (and for trivial) events in their lives; they spontaneously engage in causal analyses of their own and others' behavior, seeking to understand why it occurred (Weiner, 1985; Winter, Uleman, & Cunniff, 1985). Thus, any newly divorced person would be likely to have a set of working hypotheses explaining why he or she had become divorced (Hill, Rubin, & Peplau, 1976). Second, these explanations, or attributions, have important psychological consequences. One's emotional reactions, self-esteem, expectations for the future, and judgments of oneself and others are all influenced by one's causal beliefs. How well a person copes with diabetes (Tennen, Affleck, Allen, McGrade, & Ratzen, 1984), an industrial accident (Brewin, 1984), an abortion (Major, Mueller, & Hildebrandt, 1985), or a relationship's end (Hill et al., 1976), for example, will depend on how the person explains the traumatic event. Attribution theories in social psychology have attempted to describe the manner in which people decide what factors underlie and explain some event. In addition, attribution theorists have studied the effects these attributions have on subsequent feelings and behavior (cf. Antaki, 1982).
There are a variety of possible explanations for most of the events in our lives, and the causes we identify can be described by several different dimensions. Causes may be relatively internal, reflecting our own personalities, abilities, and efforts, or they may be more external, reflecting the influence of the situations (and other people) surrounding us. For instance, students who do well on exams usually attribute their performance to internal causes (e.g., their preparation and talent), whereas students who do poorly blame their performances on external factors (e.g., "ambiguous" tests). Causes may also be rather stable and lasting, as our abilities are, or unstable and impermanent, such as moods that come and go. In addition, causes may be relatively global, affecting many situations in our lives, or specific, affecting only a few. These three dimensions-internality, stability, and globality-have been used most frequently in the research we will discuss (e.g. , Abramson, Seligman, & Teasdale, 1978; Weiner, Russell, & Lerman, 1978), but more recently a fourth dimension has been identified: We perceive some events as caused by factors we can control, others by factors that are uncontrollable. As we will see later, attributions to uncontrollable factors may be especially related to dysfunctions such as shyness and depression (Anderson & Arnoult, 1985).
In general, events are attributed to the apparent cause with which they seem to covary (Kelley, 1973). However, it is possible for two perceivers to formulate quite different attributional explanations for the same event from the same objective information, and this possibility makes attributional processes particularly relevant to the clinician and counselor. A person who attributes a moment of embarrassment to lasting, unchangeable, and broad personal deficiencies is likely to be much more humiliated than one who shrugs it off by seeing temporary, specific, manageable, environmental causes at work. Indeed, maladaptive attributions may create psychological problems, and more adaptive attributions can minimize them. We will later find that some practitioners suggest that most therapies involve some sort of attributional analysis and reinterpretation of a client's problems (Brewin & Antaki, 1982; Frank, 1973; Sloane, Staples, Cristol, Yorkston, & Whipple, 1975). This and the next chapter will address these issues, seeking to explicate the role of attributional processes in dysfunctional behavior.
Maladaptive
Attributions
Molehills
can appear to be mountains if we look at them the wrong way. Our explanations
for an event can exaggerate its importance, and our attributions for small
problems can make them seem much worse than they are. In the first half of this
chapter we consider three examples of self-defeating attributional processes.
First, we examine the manner in which attributional judgments may exacerbate
dysfunctional behavior, creating sizeable problems out of trivial concerns.
Second, we will find that some people employ damning judgmental patterns that
are chronic enough to be termed "styles" of attribution. Finally, we
investigate the mistakes, or misattributions, that complicate and confuse our
perceptions of our own internal states.
The Emotional Exacerbation of
Dysfunctional Behavior
Skilbeck (1974) described the case of Nancy N., a 21-year-old student who reported to a university counseling center in a state of agitation. She had been having trouble concentrating and felt she was doing badly in her abnormal psychology course. She attributed her woes to emerging psychopathology; her father and sister had undergone psychiatric hospitalization (and had recently visited her), and Nancy felt that her recent distraction was an indication that she, too, must be emotionally disturbed. Skilbeck reported, however, that Nancy's "symptoms" began with the announcement of an exam in her dreaded psychology course. In fact, she seemed to be experiencing apprehension and anxiety just like that of her classmates-but explaining it very differently. Skilbeck's intervention encouraged Nancy to explain her feelings not as free-floating pathology, but as a reasonably normal response to the anxiety-arousing test. With this less threatening explanation in hand, Nancy's feelings seemed much less troublesome, and her anxiety greatly diminished. It appears that Nancy's original attribution for her anxietythinking it indicative of incipient psychopathology-had actually added to her distress, making her even more anxious and distraught.
Indeed, Storms and McCaul (1976), elaborating an idea introduced by Storms and Nisbett (1970), argued that certain attributions can sometimes inflame unwanted behavior, making it worse. For many psychological conditions, an internal or personal attribution implies personal inadequacy and/or a lack of control that is frustrating, threatening, or embarrassing. Such attributions increase the sufferer's emotionality or "anxiety," and thereby fuel the original problem. This may occur because of increased physiological arousal (which would exacerbate insomnia, for example) or enhanced drive states (which would potentiate habitual responses, and increase stuttering), or because of self-deprecating cognitions that interfere with successful coping (as with shyness and test anxiety). Whatever the reason, the unwanted behavior is exacerbated by damning self-blame that sets the cycle in motion (see Figure 2-1).
Consider a male who, after drinking too much, finds himself impotent. If, instead of rightly blaming the alcohol, he formulates an internal attribution and begins to worry that there is something wrong with him, he may enter the exacerbation cycle that Storms and McCaul (1976) described. He may ruminate about his inadequacy to the
Figure 2-1. The emotional exacerbation of psychological dysfunction. Based on material in "Attribution Processes and Emotional Exacerbation of Dysfunctional Behavior" by M. D. Storms and K. D. McCaul. In J. H. Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New Directions in Attribution Research, Vol. 1, 1976, Lawrence Erlbaum Associates, Inc.
extent that, when he next finds himself in a sexual encounter, he is so tense and preoccupied with the possibility of another sexual failure that he is again unable to maintain an erection. One can imagine his dismay and his certainty that his is a serious problem; yet had he dismissed the first occurrence as an isolated, drug-related event, the "problem" would have never developed.
Storms and McCaul suggested that this exacerbation model is applicable to any behavioral syndrome in which the primary symptoms are increased by anxiety or emotionality. Thus, the model helps explain some instances of" insomnia, shyness, phobia, depression, and even schizophrenia. As Storms and McCaul (1976) demonstrated, the model is certainly applicable to stuttering. They taped interviews with 44 males and then informed each of them that he had displayed a high number of speech dysfluencies (stammers, pauses, and repetitions). Half of the men were told that this was a normal result of experimental participation, but the remaining subjects were informed that the errors were attributable to their personal (and, apparently, deficient) speech patterns and abilities. When they were later asked to make a public speech, the self-attribution subjects evidenced a significantly greater increase in stammering than did those blaming their awkwardness on the unusual situation. The concern generated by their internal attributions made their dysfluencies worse, a finding that closely fits the experiences of real stutterers (e.g., Van Riper, 1971).
The exacerbation model is also relevant to the plight of insomniacs who, frustrated at their sleeplessness, become so restless that their insomnia lasts even longer. Van Egeren, Haynes, Franzen, and Hamilton (1983) randomly sampled the thoughts of subjects awaiting sleep and found that, unlike the rest of us, insomniacs focus on their insomnia, fretfully ruminating about their sleep problems. Moreover, as Storms and McCaul (1976) suggested, the more strongly insomniacs attribute their problems to internal causes, the more concerned they are about their sleeplessness, and the more elusive sleep becomes.
These findings suggest that insomniacs would profit by avoiding attributions that exacerbate their insomnia. Lowery, Denney, and Storms (1979) suggested two ways this might be done. In their study, some insomniacs received a placebo, which, they were told, would arouse them; participants in this "pill-attribution" treatment could logically attribute their restlessness to the pill, escaping any troublesome self-blame (cf. Storms & Nisbett, 1970). Other subjects in a second treatment were told that they were personally responsible for their insomnia, but that they should not worry; they were shown bogus. physiological data indicating that their arousal levels were higher than average but were still completely normal. Lowery et al. found that, compared to no-treatment controls, subjects in both treatments found it easier to get to sleep; in addition, they found that subjects in the second, "nonpejorative" self-attribution group reported falling asleep more quickly.
Thus, an exacerbation cycle may be blocked either at its inception, by changing the individual's self-attribution, or by ameliorating the emotionality and anxiety a self-attribution would cause. However, as our later discussion will indicate, some self-attributions are easier to change than others. Moreover, internal or personal attributions for undesirable behavior are not necessarily disadvantageous. A self-attribution that precipitates exacerbation probably involves characteristics that are not only internal, but stable and uncontrollable as well (cf. Brewin & Antaki, 1982; Van Egeren et al., 1983). In contrast, a self-attribution that emphasizes the person's ability to control and change unwanted behavior is likely to be beneficial. For these reasons, the precise nature of the judgments that both trigger and prevent exacerbation warrants close consideration.
Interestingly, the second strategy of reducing a client's emotionality is, in one sense, already a component of established therapies. For example, Ascher and Efran (1978) reported that paradoxical intention is successful in treating insomnia. Urging an insomniac to try to remain awake, they believe, reduces the performance anxiety and frustration that make the insomnia worse. In addition, the nonpejorative analysis Lowery et al. (1979) employed bears a fundamental similarity to Ellis' (1977a) rational-emotive therapy; in both, the exaggerated importance clients attach to certain troublesome events is disputed.
The
exacerbation model, then, emphasizes the role of a client's interpretation of
his or her condition in exacerbating or minimizing some disorder. It clearly
does not apply to all psychological dysfunction, but it may substantially affect
some problems that have an emotional basis. Importantly, since exacerbation
cycles are triggered by internal attributions, they may affect some people more
often than others; people differ in their tendencies to make dispositional
"self" -attributions. It is to these individual differences in
attribution, or attributional styles, that we now turn.
Attributional Styles
It is well known that depressed people make "logical errors" in their thinking, magnifying the importance of negative events and minimizing whatever successes they do achieve (Beck, Rush, Shaw, & Emery, 1979). In fact, people, who believe that they are chronic failures may be using a maladaptive attributional style that keeps them thinking so. A style of attribution is a "habitual way of answering questions about causality" (Layden, 1982, p. 64). Such a habit could lead people to blame themselves routinely for failure even when they are not at fault; it could even become an attributional rut that contributes to dysfunctions like depression (see Peterson, Schwartz, & Seligman, 1981, and chapter 3).
In fact, chronic patterns of attribution appear to be rather commonplace. Weiner and Kukla (1970) have noted that persons high in achievement motivation take more personal responsibility for their successes than do those with lesser motivation. Ickes and Layden (1978) discussed the tendency of people with low self-esteem to externalize success and internalize failure to a greater extent than those high in self-esteem. Shy (Teglasi & Hoffman, 1982) and lonely (Anderson, Horowitz, & French, 1983) people, like depressives, attribute failures in their interactions with others to internal, stable defects in themselves. Even children show evidence of using specific attributional styles (Seligman et al., 1984).
Attributional styles and dysfunctional behavior. Such attributional styles may put some individuals at risk for emotional and behavioral problems. When a shy woman attributes another's antagonism or disinterest to her own personal inadequacy, she reinforces her shyness, perpetuating her inhibition and social anxiety. By seeing herself as the cause of another person's grumpy mood, she paints herself as inept and unskilled and makes future interactions all the more threatening. She may in fact be socially maladroit, and the unfavorable responses she receives from others may partially be of her own making. Still, her inability to distinguish her own awkwardness from other, external causes of uneasy interaction is likely to make her shyness more intractable.
In fact, Anderson (1983a) showed that certain attributional styles can lead to motivational and performance deficits in social situations. Using a pretest that assessed subjects' attributional styles, Anderson selected subjects who tended to blame interpersonal. failures either on their own imperfect characters (with attributions to stable, uncontrollable abilities and traits) or on their inappropriate, but changeable, behavior. When he then asked his subjects to persuade others to donate blood (a task in which several failures were likely), the character-attribution group expected less success, displayed less motivation, improved less quickly, and-as one would guess-actually enjoyed fewer successes than those who blamed their failures on inappropriate behavior. Their habit of blaming failure on deficient traits appeared to rob the character-attribution group of the motivation and interest that would help them to improve.
In addition, Anderson (1983a) tried to convince some of his subjects that success at this novel task was due to the "basic persuasiveness of the caller"-that some people were simply good at recruiting blood donors, and others simply were not. By contrast, other subjects were led to believe that success was a matter of strategy and that good recruiters had merely found the most persuasive tactics. The subjects were thus led to anticipate success as a result of either their abilities and traits or their strategies and effort, respectively. The two groups behaved very differently. Whatever their natural. attributional style, subjects who were led to make ability/trait attributions expected and enjoyed fewer successes than those who considered strategy and effort more important determinants of success.
Thus, Anderson found that naturally occurring attributional style differences are associated with actual behavioral differences, as studies of shy, lonely, and depressed persons would suggest. Importantly, he also demonstrated that experimentally changing an individual's attributional outlook can change one's motivation and performance as well.
Anderson's work clarifies the question of whether maladaptive styles precede and help create dysfunctional behavior, or whether they emerge as a result of the dysfunction (and then help perpetuate it). Metalsky and Abramson (1981) suggested, for instance, that attributional styles may exist either because people are constantly confronted with the same attributional data or because they rely on strong, generalized beliefs about causality that are sometimes inappropriate. In the first case, an internalizing, character-blaming style could result because a person really is inept and is accurately assessing his or her failures. In the second case, however, a person's mistaken beliefs could lead to chronic patterns of attribution that unfairly underestimate his or her skills. Anderson's (1983a) study provided evidence that belief-based styles can exist independently of any behavioral deficit: When equipped with a more desirable attributional set (in the strategy/effort condition), characterological self-blamers behaved just as effectively as other subjects. Their characteristically lower motivation and performance appear to have resulted more from their chronic style of cognition than from any lack of social skill. Thus, maladaptive attributional styles may often reflect existing behavioral deficits, but they can also help create dysfunctional behavior that would otherwise not exist.
The measurement of attributional style. Several different instruments have been used to measure attributional styles, but as we review them two related points should be kept in mind. A first consideration is that maladaptive attributional styles are often quite specific. Cutrona, Russell, and Jones (1984) found scant evidence for broad cross-situational styles of attribution, and, indeed, the self-defeating styles examined in other studies are usually limited to particular problem areas. Shy and lonely people, for instance, display unique patterns of attribution only for interpersonal failure situations; their judgments of personal performances that do not involve interaction with others and their perceptions of interpersonal successes do not differ from those of nonshy and nonlonely people (Anderson & Arnoult, 1985; Teglasi & Hoffman, 1982; cf. Watson & Dyck, 1984). Nevertheless, the apparent specificity of maladaptive attributional styles may be partly due to the use of measurement scales that assess respondents' attributions for hypothetical outcomes. When people are asked to explain actual events in their lives over a period of time, broad, cross-situational patterns of attribution may emerge (Zautra, Guenther, & Chartier, 1985).
A second point is that habitual use of particular attributions may be more damaging than others. In an examination of this issue, Anderson and Arnoult (1985) assessed subjects' depression, shyness, and loneliness and then measured their attributional styles with the Attributional Style Assessment Test developed by Anderson et al. (1983). Regression analyses indicated that subjects who routinely attributed events in their lives to causes they could not control were particularly likely to be shy, lonely, or depressed. People who blamed themselves for failure were also likely to suffer these dysfunctions, but whether people made stable or unstable, global or specific attributions did not matter much. Thus, certain types of attributions may be more debilitating than others. In particular, our (perceived) ability to control the events in our lives seems to be of fundamental impact, a matter we will consider in more detail in chapter 3. For now, let us note that efforts to measure (and, perhaps, to change) maladaptive patterns of attribution need to address the particular importance of uncontrollable, internal attributions as correlates of dysfunction.
Three different scales have been used to measure attributional style, and though they are conceptually similar each has a unique format. The first was developed by Ickes and Layden (1978); it consists of 24 hypothetical events and their positive or negative outcomes (e.g., "You got an 'A' on a class project"), which the respondent is asked to explain. Each item is followed by four possible causes, which are internal or external and stable or unstable (e.g., "You worked hard to prepare this project," or "The project was relatively easy"), and the perceived influence of each is rated on 5-point scales. The scale thus assesses only the internality and stability of respondents' explanations, but it has been successful in identifying chronic patterns of attributions (e.g., Sober-Ain & Kidd, 1984).
The Attributional Style Questionnaire (ASQ) developed by Peterson et al. (1982) expands on Ickes and Layden's (1978) effort by distinguishing interpersonal situations (e.g., "You go out on a date and it goes badly") from achievement situations (e.g., "You become very rich"). Again, half of the items describe positive events, half negative. Subjects are asked to envision the situation and write down the one major cause that comes to mind; that freely chosen cause is then rated on separate 7-point scales assessing its internality, stability, and globality. The ASQ has the advantage of not constraining the subject's judgments of causality, and several studies (e.g., Seligman et al., 1978) point to its usefulness, but its psychometric properties have been recently criticized (Cutrona et al., 1984; Johnson, Petzel, & Sperduto, 1983).
Anderson and his colleagues (1983a; Anderson et al., 1983) have since developed their own Attributional Style Assessment Test (ASAT). It is the most elaborate of the three, consisting of 36 interpersonal or noninterpersonal successes or failures, which, in the latest version of the scale (Anderson & Arnoult, 1985), are judged using the same response format as that of the ASQ. Unlike the ASQ, however, the ASAT asks for ratings of each cause's controllability in addition to ratings of its internality, stability, and globality. Given the importance of the controllability dimension (Anderson & Arnoult, 1985), the ASAT thus seems to be the most useful of the three scales. Nevertheless, it is likely that they have all enjoyed some success because they all use items that contain minimal situational information and so are causally ambiguous (Metalsky & Abramson, 1981). For instance, one ASAT item describes this interpersonal failure: "You have just attended a party for new students and failed to make any new friends." The lack of detail forces respondents to - rely on their own generalized beliefs or past histories, allowing their personal patterns of attribution to emerge.
Modifying attributional styles. To the extent that a shy, lonely, or depressed person actually lacks the social skills necessary for rewarding interactions with others, some self-blame is appropriate, and a behavioral therapy that teaches the needed skills is desirable. Still, as we have seen, a maladaptive style can create dysfunction in the absence of any behavioral deficit, keeping sufferers from recognizing and appreciating the successes that they already enjoy. In this case, as Layden (1982) notes, it may be useful to attempt to change their attributional habits, asking them "to scan success situations for internal causes, to focus on abilities and personality traits they normally ignore, and to make realistic estimates of the amount of effort they expend" (p. 72).
This may be easier said than done, however. Convincing a client that his or her perceptions of events are inaccurate may be quite difficult. Layden (1982) suggests that clients frequently deny that they have an attributional habit, or, if they do recognize the patterns in their judgments, insist that their perceptions are accurate. They may have to be shown why it would be to their advantage to change, and even then they may consider it immodest not to blame themselves for their troubles. Indeed, Ickes and Layden (1978) found that merely getting low self-esteem subjects to practice making attributions that were inconsistent with their styles (i.e., encouraging them to internalize success and externalize failure) did not consistently change their styles. Ickes and Layden asked their subjects to list as many possible causes-of the appropriate type-they could think of for 15 positive and 15 negative specific events occurring during the 5-week study. The subjects were not given a rationale for this procedure and were not shown how some attributions were better than others. Perhaps as a result, this minimal intervention did not produce consistent effects; still, when subjects' styles did change, there was a concomitant increase in their self-esteem.
Thus,
interventions designed to change a client's attributional style probably should
strive to demonstrate why such changes are both feasible and desirable, thereby
providing the client with needed incentive and purpose. In support of this,
Sober-Ain and Kidd (1984) compared Ickes and Layden's (1978)
alternative-attribution treatment with a condition in which an active,
supportive experimenter consistently rewarded "appropriate"
attributions, stressed the advantages of such attributions, and boosted the egos
of selfblaming subjects. This "supportive suggestion" treatment
appeared to be more effective than Ickes and Layden's strategy in changing the
subjects' styles, but the changes, although beneficial (cf. Layden, 1982), came
grudgingly. It appears that attributional styles can be so habitual that clients
may need considerable help in altering them. Indeed, Sober-Ain and Kidd warned
that, "since attributional style is a pervasive, and possibly a
dispositional, characteristic of the person, global change is apt to take place
slowly, if at all" (p. 134).
Misattributions: Mistaken
Causality
As the preceding discussion of attributional styles implies, our judgments regarding the causes of our-and others'-feelings and behaviors are not always accurate. We do not always know why we do what we do, and we may occasionally believe that we are influenced by particular factors that in fact have little effect. In short, misattributions, or mistaken perceptions of causality, are possible. Moreover, they can help develop or maintain dysfunctional behavior.
Johnson, Ross, and Mastria (1977) presented a case study that exemplifies this point. A 37-year-old male was institutionalized with the belief that he was being sexually aroused by an unseen "warm form." He never masturbated but had been brought to orgasm several times by this mysterious entity. Understandably, he was quite concerned. When Johnson et al. encouraged him to attribute his sensations to normal sexual desire (he had no steady partner) and inadvertent masturbation through leg movements, his delusions disappeared and he had no further problems. In this case, the misattribution to a fictional cause was the problem.
Misattribution can inflame existing dysfunctions as well. For instance, Borkovec, Wall, and Stone (1974) provided speech-anxious subjects with bogus heart rate feedback, which indicated that a brief speech had caused their pulse rates to increase. When later they gave another speech, they become more discombobulated than they had been earlier, becoming more anxious and displaying more speech dysfluencies. In addition, they were more uneasy than other subjects who had heard feedback suggesting either that their heart rates had decreased or that they had not changed. The mistaken belief that they had been physiologically aroused by the prior talk apparently led them to suppose that their speech anxiety was even more serious than they had thought. Indeed, one of the present authors has forsworn caffeinated coffee because he realized, in retrospect, that he sometimes misattributed his caffeine-induced arousal to performance anxiety in stressful situations.
In a related vein, alcoholics who believe they are drinking vodka-and-tonics, but who are actually drinking straight tonic water, develop the same craving for additional liquor that they do when drinking real vodka. Similarly, tonic water gets male subjects just as aggressive as vodka does (alcohol increases aggression), if the men think there is vodka in the drink (Marlatt & Rohsenow, 1980). Examples like these illustrate that what we think does influence what we feel, and our wrongful perceptions can create problems where none would otherwise exist.
Mistaking the level of arousal. Two broad types of misattribution are possible. In the first, one misperceives one's level of arousal in response to some stimulus. In studies of this phenomenon, subjects are led to believe that they are or are not being influenced by surrounding events (manipulating, in Kelley and Michela's [1980] terms, selective "arousal" or "quiesence"). The first studies to manipulate either arousal or quiesence were both performed 20 years ago by Valins (1966; Valins & Ray, 1967). In his first study, Valins showed men photographs of female models while they listened to an amplified rhythmic sound that they believed was their own heartbeat. For half of the photographs, subjects heard the "pulse rate" change, either increasing or decreasing; in truth, the subjects heard a tape recording that could be controlled by the experimenter. The subjects' heart rates had not actually changed, and different photos were associated with the pulse change for different men. Nevertheless, when they were later asked to judge the photos, the men generally liked best whichever pictures they could reasonably assume had "turned them on." The bogus information regarding their internal arousal had affected their choices of which photos were the most compelling.
Valins and Ray (1967) soon followed with a provocative study that had more obvious clinical implications. They attempted to convince participants that they were not being greatly affected by a stimulus that actually did arouse them. They recruited subjects with mild phobias toward snakes and showed them slides of reptiles while they listened to bogus heart rate feedback. The snake photos were intermixed with photos of the word "shock," and when "shock" appeared, the viewers were actually given a mild electric shock. The subjects found that their "heart rates" increased when they were confronted by "shock" but remained steady when snakes appeared. They were thus led to believe that the snakes did not affect them particularly much, and compared to a control group who believed the "heartbeat" rhythm to be extraneous noise, the misattribution subjects seemed less wary of snakes, becoming more likely to approach and even touch live snakes.
Influencing a person's perceived level of arousal, then, may minimize, as well as exacerbate, dysfunctions. Koenig (1973), for instance, provided testanxious subjects with manipulated information about their emotionality, confronting them with a galvanic skin response meter that indicated high, average, or low arousal. As the preceding discussion suggests, subjects' belief that they were strongly aroused debilitated their subsequent performance on math problems; attributing their apparent anxiety to the testing situation made it an even more threatening stimulus. By contrast, subjects given low-arousal feedback, indicating relative calm and relaxation, did better than average, answering more problems correctly and working more quickly. The belief that they were not experiencing their usual anxiety apparently facilitated their performance, largely obviating, for the moment at least, their very real problem.
Mistaking the cause of arousal. Thus, changing a person's perceptions of how much he or she is reacting to some event can have desirable as well as undesirable outcomes. There is, however, a second broad type of misattribuion that was once believed to have even greater therapeutic potential. People may misunderstand not only the level but the source of their arousal as well. Studies of this kind of misattribution induce subjects to attribute troublesome emotional reactions not to their actual cause but to less threatening, emotionally neutral sources. In this way, fear or anxiety can be reinterpreted is innocuous "stimulation" or "activation," thereby making one's feelings less worrisome. Unlike Valins' paradigm, which might try to convince subjects that they are not much aroused, this approach attempts to get subjects thinking that their obvious arousal is not aversive.
Dienstbier and Munter (1971) provided a good example of this technique. All of their subjects took a placebo pill, and half of them were told to expect noticeable arousal as a result. The subjects then encountered failure on a contrived vocabulary test and were given an opportunity to change their answers covertly to improve their scores. Dienstbier and Munter suggested that the tension and guilt that usually accompany cheating normally inhibit wrongdoing. However, half of their subjects could reasonably assume that any nervousness was being caused by the pill rather than by personal guiltiness, and with that misattribution cheating would be less aversive. In fact 49% of the pill/arousal group cheated by changing their answers, versus 27% of a control group. The results suggested that "even naturally occurring emotion is subject to this apparent ease of reinterpretation" (P. 213).
Studies like these imply that we are not always certain of our feelings and that misattributions of the cause and extent of our emotions are possible. Moreover, whereas certain misattributions may create or exacerbate dysfunctions, other attributional mistakes may have a beneficial effect. To complete our discussion of misattributions, therefore, we need to address their potential therapeutic uses and begin consideration of the adaptive applications of attributional processes.
Adaptive
Attributions
Misattributions
Under
the impetus of provocative studies like that of Valins and Ray (1967),
misattribution research flourished in the 1970s, and the term attribution
therapy was introduced (Ross, Rodin, & Zimbardo, 1969). It was thought that
misattribution procedures might be readily applied to clinical settings as
treatments; in particular, interventions that led people to misattribute
internal arousal to external sources often seemed to have therapeutic effects. A
well-known study by Storms and Nisbett (1970) exemplified this approach.
Insomniacs, it was argued, exacerbated their sleeplessness by attributing it to
internal causes and making it a threatening, anxiety-arousing event. If they
could be supplied a plausible cause for their restlessness that did not Thus,
despite the excitement and interest generated by misattribution techniques, the
adaptive uses of misattributions now appear to be limited, for two interrelated
reasons: First, it is difficult to convince subjects that strong emotions that
have troubled them for some time either are not particularly potent or are
largely caused by innocuous factors they had not considered. This can be done,
of course, given enough experimental control, but such misattributions are
unlikely to last (and this is our second reason) because they are inherently
deceptive. The beneficial belief that one's insomnia is being caused by a
stimulant pill depends on both the presence of the pill and the fragile fiction
that, without it, one's insomnia would be reduced. Such misattributions are
difficult to maintain in a person's normal environment simply because they are
falsehoods. Indeed, therapists may be reluctant to employ them at all for this
reason. Misattribution may be quite useful for treatment of infrequent events
and novel symptoms (Fincham, 1983), but at bottom, the alternative use of
adaptive reattributions probably has wider applicability.
Reattributions
Instead of trying to mislead people, however kindly, about the level or the source of their symptoms, why not try to equip them with an accurate but nevertheless more desirable attribution for their feelings? In providing reattributions, one suggests explanations for a person's experience that point to (at least potentially) veridical causes that are less threatening than those the person has already identified. To promote coping, the person's dysfunction is interpreted as less severe and more manageable than it now seems. Reattributions follow the same lines as, but are usually much more specific than, efforts to retrain a person's broad attributional style; whereas style interventions must break an enduring habit of pejorative self-attribution, reattributions are most often designed to facilitate coping with a particular delimited problem. For instance, Valins and Nisbett (1972) described the distress encountered by sufferers of the -FNG [for "fucking new guy"] syndrome" in Vietnam. New arrivals in the war zone were often met with suspicion and hostility by field veterans, a situation that greatly disturbed many of the new men. To deal with the syndrome, the newcomers were urged to replace the attribution "They hate me" with the more tolerable explanation "They hate the ING.'' Reattributing the others' hostility to situational role requirements instead of some personal deficiency made the adjustment period easier to endure.
The concept of reattribution is quite similar to processes already at the heart of many psychotherapies. Whether one uses psychoanalysis or behavior modification, for instance, one relies on new causal interpretations of a client's dysfunction in order to motivate and direct his or her coping responses (Sloane et at., 1975). Strong (1978) has argued that, "changing . . . causal attributions is the therapist's first order of business" (p. 27), and Hoffman and Teglasi (1982) suggested that all counseling therapies provide new interpretations and frames of reference for the client. In fact, Hoffman and Teglasi found that providing shy clients with any causal attribution for their shyness increased their motivation, involvement, and expectations of success beyond those of clients who were not provided some explanation of their problem (cf. Strong, Wambach, Lopez, & Cooper, 1979). Thus, social psychological reattribution is not a new idea. However, as we are about to find, most reattribution training explicitly addresses the preferred, use of particular attributional dimensions, and is thus more specific than most "interpretations" (cf, Forsterling 1985).
A well-known study by Dweck (1975) is a good example. She identified 12 children with extreme, helpless reactions to failure and placed each child into one of two 25-day treatments. Half of the children encountered constant success on arithmetic problems (and were given personal credit for their achievements), whereas the other subjects met with both success and failure. When the latter children failed, however, they were implicitly urged to attribute their failure to a lack of effort: "You should have tried harder." When all the children were later given difficult (i.e., failure-producing) problems, the failure-reattribution children persisted in their attempts to solve the problems, but the success-only children quickly gave up. Learning to attribute failure to controllable, unstable factors such as effort promoted perseverance and achievement better than did getting a taste of success.
Similar results have been obtained with children's reading behavior (Chapin & Dyck, 1976) and perceptual. reasoning (Andrews & Debus, 1978), and two provocative studies by Wilson and Linville (1982, 1985) have even extended these effects to college freshmen. In their studies, students who were instructed that low first-term grades were due to unstable causes (and thus were likely to improve) actually achieved higher grades in subsequent semesters and were somewhat less likely to drop out of school than were "untreated" control subjects.
Importantly, such reattributions may have more profound and lasting effects than mere persuasive advice, especially when they influence a person's self-perceptions. Miller, Brickman, and Bolen (1975) compared the effectiveness of an attribution treatment stressing ability or effort (e.g., "You really work hard in arithmetic") to persuasive exhortations (e.g., "You should work harder") and positive reinforcements (e.g., a "math award") in improving student achievement in mathematics. The attribution treatments clearly had more effective and lasting impact, leading to significantly more long-term improvement. Leading the students to internalize desirable behavior by identifying the personal, controllable causes at work was thus a much more valuable approach than either instructing them how to act or rewarding them for their successes. I
Indeed, a series of studies by Haemmerlie and Montgomery (1982, 1984; Haemmerlie, 1983; Montgomery & Haemmerlie, in press) has successfully ameliorated heterosocial anxiety in college students not by teaching them new skills, but by leading them to think differently about the ones they already have. In this procedure, anxious subjects engage in "purposefully biased" 10-minute interactions with each of six different opposite-sex peers. These "interaction assistants" need not know the purpose behind these meetings, and are merely instructed to carry on pleasant conversation (without being negative, discussing sex, or making any dates). Whether or not the subjects think they are getting "therapy" and whether their expectations are positive or negative, the procedure seems to be quite beneficial, substantially reducing the subjects' anxiety. It is much more effective, in fact, than an imaginal therapy in which subjects envision successful interactions without actually engaging in them (Haemmerlie & Montgomery, 1984). Admittedly, this treatment is unlikely to work with persons who have real socialskills deficits. Nevertheless, for reasonably skillful people, the procedure may lead to a reattribution of the occasional successes they do enjoy; subjects are implicitly taught not only that they can have enjoyable interactions with others, but that they are more adept than they might have believed. In this and other reattribution research, learning how to explain successes (and failures) more profitably appears to be as important as experiencing the successes themselves.
Recent research does suggest that personal and contextual variables can influence the effectiveness of adaptive reattribution, however. Forsyth and Forsyth (1982), for example, found that a counseling session that invited subjects to attribute their social anxiety to internal and controllable (and thus manageable, changeable) causes was more helpful for subjects with an internal, rather than an external, locus of control (cf. Brewin & Shapiro, 1985). Moreover, the timing of the reattribution intervention may be important (Altmaier, Leary, Forsyth, & Ansel, 1979). Forsyth and Forsyth were led to conclude that, "an even-handed analysis suggests that a guided exploration of the causes of behavioral and psychological problems may be helpful for only (unfortunately) some of the people some of the time" (1983, p. 457).
Still,
when they are wisely chosen and are applied to selected individuals (they are
obviously unlikely to palliate paranoia, for example), reattributions may have
considerable beneficial impact. Storms and his colleagues have now forsworn
misattribution therapies for insomnia in favor of veridical reattributions that
explain the problem as normal, albeit high, levels of autonomic arousal (Storms,
Denney, McCaul, & Lowery, 1979). The procedure is ethical and effective and
often leads to improvement over time. Indeed, as Storms et al. noted, adaptive
attributions can also facilitate the maintenance of desirable change in behavior
over long periods of time, well beyond the end of any organized therapy.
The Maintenance of Behavior
Change
The long-term success of a therapeutic intervention may depend on the perceptions with which a client leaves therapy. The client who judges external factors (such as an ingenious therapist or a calming drug) to be the primary cause of his or her improvement may find it hard to maintain his or her gains when the external agents are withdrawn. If a person suffering from sleeplessness thinks 1) "I'm an insomniac, but, " 2) "this medication helps me sleep," imagine his or her dismay on finding the medicine bottle empty. By contrast, if the client sees himself or herself as personally capable of managing the dysfunction, the treatment gains may be more likely to last. When faced with sleeplessness, the client might think, "I can deal with this." This approach to maintenance touches on issues we will also address in chapters 3 (perceptions of control) and I I (self-efficacy). Nevertheless, the specific manner in which a person attributes desirable change seems to be a central concern.
Again consider insomnia. In one of the first tests of the attributional maintenance hypothesis, Davison, Tsujimoto, and Glaros (1973) provided insomniac subjects a treatment package in which they scheduled a specific time for settling down to sleep, received relaxation training, and took 1,000 mg of a hypnotic (chloral hydrate) each night before bedtime. For a week thereafter the subjects reported the latency and quality of their sleep, and for most the treatment was effective; most of them fell asleep in less than half their usual time. At the end of that week, however, the subjects were divided into two groups: Some were told that they had been taking an optimal dosage of the drug, making it "a very effective sleep aid"; the others were informed they had been taking a minimal, "very ineffective" dosage. The subjects were thus led to perceive the drug either as an important active agent in their improvement or as a largely inconsequential adjunct. They then went home without the drug and were told to rely on their scheduling and relaxation for the next several days. The optimal-drug group slipped badly, requiring an average 45 minutes longer to get to sleep than they had needed the preceding nights. By contrast, the minimal-drug group regressed less precipitously, taking only 13 minutes longer and, unlike their drug-oriented counterparts, retaining some of their treatment gains. A check of the subjects' attributions showed that the two groups had indeed made different attributions for their prior improvement: The optimal-drug group believed that the drug had been mostly responsible, whereas the minimal-drug subjects credited themselves. As a result, Davison et al. asserted that the favorable self-attributions of the minimal-drug subjects had facilitated the maintenance of their behavioral change (cf. Davison & Valins, 1969).
Their argument received further support from subsequent studies that extended this framework to methods of weight reduction and smoking cessation. Behavioral weight control techniques are often successful, but they are plagued by clients' failures to maintain their new lower weights (Stunkard & Penick, 1979). Jeffrey (1974) provided suggestive evidence indicating that these failures could be reduced if clients were taught a greater sense of selfcontrol. In his study, obese subjects were placed in one of two treatment groups. For the first group, the therapist controlled the rewards they received (which consisted of cash refunds from money they had deposited with the program) for meeting weight-loss goals. The other participants doled out their own rewards, and their personal responsibility for managing their weight was stressed. Everybody lost weight during the 7-week treatment, but only the self-control subjects maintained their progress over a 6-week follow-up period; on average, the external-control subjects regained 55% of the weight they had lost. In a partial replication, Sonne and Janoff (1979) specifically assessed subjects' attributions for their weight loss and even more obviously supported the attributional maintenance hypothesis: The more personal control over their weight subjects perceived, the more likely they were to continue to lose weight over a 3-month follow-up period. Colletti and Kopel (1979) obtained similar results in a smoking cessation study over an even longer posttreatment period of 1 year. Indeed, they flatly argued that, "superior maintenance is associated with a greater degree of self-attribution (p. 616; cf. Chambliss & Murray, 1979).
The size of the behavioral differences that are obtained between externalattribution and self-attribution subjects in studies like these, although statistically significant, is sometimes small (e.g., Sonne & Janoff, 1979). This fact has not been lost on some critical observers (Grimm, 1980). Nevertheless, the wide applicability of the attributional maintenance model-to alcoholism (Davies, 1982) and drug therapies with hyperactive children (Whalen & Henker, 1976) among other examples-and the weight of the empirical evidence clearly point to its usefulness and importance. This is also true in that a person's sense of self-responsibility and self-control in a specific therapeutic treatment is usually modifiable, and can be enhanced by a judicious therapist.
There
are several ways to do this. Kopel and Arkowitz (1975) suggested using the
"least powerful reward or punishment" that will work in any behavior
modification procedure. Bowers (1975) echoed this idea with his concept of
"subtle control." The external response contingencies facing a person
should not be so overwhelming that self-attributions for desirable change are
undermined. Similarly, Kopel and Arkowitz advocate allowing a client to play an
active role in the planning and execution of his or her therapy whenever
possible. Both of these tactics are likely to enhance the client's sense of
behavioral freedom (Sonne & Janoff, 1982), maximizing his or her feelings of
self-direction and self-control. Sonne and Janoff also suggested showing clients
how capable they are through videotape replays of their own behavioral
successes. At the very least, these clinicians agree, therapists should be alert
to the potential value of appropriate self-attribution in helping maintain
therapeutic change.
Conclusions
Our perceptions of why we behave the way we do appear to be centrally involved in the development and, perhaps, the amelioration of dysfunctional behavior. It seems to be advantageous to consider one's problems change able and controllable, and one's causal attributions either facilitate or preclude that view. Indeed Brewin and Antaki (1982) raised the provocative argument that the usefulness of the various etiological models used in psychotherapy (that is, whether one's problems are described in psychoanalytic, behavioral, or other terms) depends largely on how well they help clients reattribute their problems in desirable directions. This is not to imply, of course, that modification of a person's causal attributions is always readily accomplished. We have seen that general attributional styles are difficult to change, and Peterson (1982) has wisely reminded us that we should not approach attributional beliefs "as if they were removable like psychological tumors" (p. 107). Nevertheless, attributional processes remain a key focus of the social/clinical interface because of their primary importance: "attributions affect our feelings about past events and our expectations about future ones, our attitudes toward other persons and our reactions to their behavior, and our conceptions of ourselves and our efforts to improve our fortunes" (Kelley & Michela, 1980, p. 489).