This article is for use only of students enrolled in GPS 520: Advanced Social Psychology for purposes associated with this course and may not be retained or further disseminated.

Leary, M. R., & Miller, R., S. (1986).  Social psychology and dysfunctional behavior: Origins, diagnosis, and treatment.  New York: Springer-Verlag.

Chapter 3

Attributions, Perceived Control, and Depression

Many theories in psychology are based on the assumption that an individual's sense of psychological well-being is augmented by a belief that he or she has some degree of control over personally relevant events. As early as 1930, the neo-Freudian Alfred Adler observed that the need for control is "an intrinsic necessity of life" (Adler, 1930, p. 398)-a point that has been reiterated by many other theorists (e.g., Bandura, 1977; deCharms, 1968; Langer, 1983; Lefcourt, 1973; White, 1959; Woodworth, 1958). As usually defined, control is said to exist when an individual's outcomes are dependent upon his or her responses (Seligman, 1975). Perceived control is more important than control per se, however. As we shall see, people's beliefs about how much control they have are more closely associated with functional and dysfunctional reactions than is their objective level of control.

Perceived control fosters a sense of personal well-being for many reasons. Control over one's environment allows the individual to achieve desired goals while avoiding outcomes that are aversive. As a result, it is rewarding to feel that one is in control of personality relevant events, and frustrating or threatening to feel that one is not.

Further, studies show that the mere belief that one can control one's outcomes has positive psychological effects, even when one's perception of control is mistaken and, in reality, no control exists. For example, people are less anxious about experiencing painful stimuli (such as electric shocks) when they think they can control their onset, and aversive stimuli are perceived as less painful when people think they can control them, even when no control exists or they choose not to exert control (Bowers, 1968; Glass & Singer, 1972; Kanfer & Seidner, 1973). Similarly, information that leads surgery patients to believe they can control their pain through cognitive means reduces the need for painkillers and sedatives (Langer, Janis, & Wolfer, 1975). In another intriguing line of research, Pennebaker and his colleagues have found that perceived failure to control one's environment leads people to report an increased number of physical symptoms (Penneba ker, Burnam, Schaeffer, & Harper, 1977). As these studies demonstrate, perceived control over one's outcomes generally has positive effects, whereas perceived lack of control has dysfunctional consequences.

The effects of perceived control have been of particular interest in realworld settings that lead people to conclude that they have little control over their lives. Langer and Rodin (1976) explored the impact of perceived control upon elderly residents in a nursing home, an environment conducive to feelings of helplessness (see also Schulz, 1976). An administrator gave two different talks to samples of the residents; one talk emphasized the residents' responsibility for themselves, admonishing them to make their own decisions about their lives in the facility, whereas the other talk emphasized the staff's responsibility for the residents. In both cases, the responsibility and options stressed by the administrator were already available to the residents, so the groups differed chiefly in the degree to which their freedom, responsibility, and choice were explicitly stressed.

Compared to the other residents, those who heard the talk that emphasized their personal control and responsibility became more active and alert, rated themselves as happier, and became more involved in activities within the facility. The nursing staff rated them as more interested, sociable, self-initiating, and vigorous than other residents, and follow-up data collected 18 months after the intervention showed long-term psychological and physical effects (Rodin & Langer, 1977). Overall, the residents who experienced greater personal control showed a greater improvement in health and a lower mortality rate.

As this brief review demonstrates, people's beliefs regarding their personal effectiveness and control have important implications for their psychological and physical well-being. Not surprisingly, a great deal of attention has been devoted to identifying the factors that foster a sense of personal control, and to understanding the emotional and behavioral consequences of losing perceived control. In this chapter, we examine dysfunctional reactions that can occur when people feel they have lost control, focusing on how these problems are affected by the kinds of attributions they make.

Learned Helplessness

As hard as they might try, people cannot always control what happens to them. Despite their best efforts, people become ill; lose their jobs; have accidents; fail in business, school, and love; and experience a host of other negative events that they try hard to avoid. Because their responses to such events affect the quality of people's lives, a great deal of research has explored how people react to instances in which they lose control over aspects of their lives.

Interestingly, the earliest work on this topic was conducted not on human beings, but on dogs. In 1967, Overmeier and Seligman showed that dogs that had been exposed to electric shocks that they were unable to escape later had great difficulty learning to escape shock in a new situation. Unlike animals that had earlier experienced escapable shock or no shock at all, dogs that had previously suffered uncontrollable shock failed to exert control even when avoidance of the shock became possible. This debilitating effect of uncontrollable events, termed learned helplessness, has since been demonstrated in a variety of species, including cats, rats, fish, pigeons, and humans. This extensive research is far beyond our capacity to review here, and so we will focus our attention on only one portion of the literature, that dealing with helplessness and depression in humans, and suggest that the interested reader see Garber and Seligman (1980) for a more extensive review of the helplessness literature.

Learned Helplessness and Depression

In an early statement of his theory of learned helplessness, Seligman (1975) noted the similarities between the behavioral consequences of learned helplessness in animals and the concomitant features of reactive depression in humans. Specifically, he observed that depression and learned helplessness share several common features, including passivity in the face of aversive outcomes, difficulty in learning that outcomes are controllable, and affective correlates involving depression and/or anxiety. On the basis of these similarities, Seligman (1974, 1975) suggested that reactive depression in humans, like learned helplessness in animals, may result from the perception that one has no control over important outcomes.

In his original statement of the theory, Seligman maintained that helplessness occurs when an organism learns to expect that its behavior will not influence the likelihood of certain outcomes. The expectancy that one's outcomes are independent of one's responses (or "noncontingent") was assumed to produce the motivational, cognitive, and affective consequences associated with both learned helplessness and depression.

However, Seligman's (1975) original theory had difficulty accounting for some aspects of learned helplessness in depressed humans (Abramson et al., 1978). For example, the theory did not specify the conditions under which the perception of noncontingent outcomes leads to helplessness. As Abramson et al. (1978) noted, many events in life that people regard as uncontrollable do not depress them. Second, the original theory did not explain why depression is often accompanied by low self-esteem. Third, the theory did not easily account for the paradoxical finding that depressed individuals tend to blame themselves for events over which they think they have no control (Abramson & Sackheim, 1977). Perhaps most importantly, the original model did not address either the chronicity or the generality of depression; the time course of depression can vary from minutes to years, and encompass all of a person's life or only part.

The Attributional Model of Helpless Depression

In a reformulation of learned helplessness theory, Abramson et al. (1978) proposed that helplessness effects in humans are mediated by the causal attributions people make when they find themselves unable to control important outcomes. Specifically, the reformulated model posits that learned helplessness occurs when individuals make attributions that lead them to expect that they will be unable to control their outcomes in the future. Further, using three of the attributional dimensions introduced in chapter 2 (internalexternal, stable-unstable, global-specific), Abramson et al. showed how a person's attributions might affect the cognitive, motivational, and affective aspects of depression.

Internal-external. According to Abramson et al. (1978), attributing one's helplessness to internal factors-characteristics of oneself-results in "personal helplessness," in which the person believes that his or her lack of control is due to a personal deficiency. As a result, personal helplessness is often accompanied by lowered self-esteem. However, when a person makes an external attribution, explaining helplessness in terms of the situation or other people, "universal helplessness" results. Because an external attribution implies that no one would be able to control the outcome, self-esteem is not affected, although helplessness and depression may still occur.

Stable-unstable. Some causes of helplessness are perceived as stable and unchangeable (such as a physical deformity or an incurable disease), whereas others are regarded as unstable and potentially changeable (e.g., temporary fatigue). The attributional reformulation of learned helplessness posits that attributions to stable factors result in deeper depression and in greater motivational decrements than do attributions to unstable factors, because the individual expects that the helplessness will persist over time. When the cause of one's difficulty is thought to be temporary or modifiable, the individual may continue to work toward control and show few helplessness deficits. Thus, the chronicity or time course of a depressive episode may be related to the stability of the perceived causes.

Global-specific. The perceived causes of one's helplessness may be ones that affect many areas of one's life (i.e., global) or just a few (i.e., specific). The more global the apparent causes of one's helplessness are, the deeper the depression and the more it generalizes to many areas of life.

Thus, from the standpoint of the individual confronting an uncontrollable situation, the most dysfunctional attributions are those that emphasize internal, stable, and global causes, and the least dysfunctional involve factors that are external, unstable, and specific.

Research on the Attributional Model

Attributional style and depression.   Abramson et al. (1978) speculated that certain people may be more likely to become depressed than others because they have a "depressive attributional style" characterized by a tendency to attribute negative events to internal, stable, and global factors. Persons who tend to make these sorts of attributions should be particularly prone to general and chronic helpless depression that is accompanied by lowered self-esteem.

In fact, research prior to the appearance of the attributional model of learned helplessness did show that depressed subjects are more likely to make internal attributions for their failures than are nondepressed individuals (e.g., Klein, Fencil-Morse, & Seligman, 1976; Kuiper, 1978; Rizley, 1978). The first study explicitly designed to test the relationship between attributional style and depression, however, was conducted by Seligman, Abramson, Semmel, and von Baeyer (1979). In this study, depressed and nondepressed college students completed the Attributional Style Questionnaire (ASQ), which assessed the internality, stability, and globality of their attributions (see chapter 2). As predicted by the model, depressed students' attributions for negative outcomes were more internal, stable, and global than the attributions made by nondepressed students. Further, depressed students made more external and unstable attributions for positive outcomes than did less depressed students. Consistent with the theory, then, certain patterns of attributions were associated with higher levels of depression.

Subsequent research has generally supported the hypothesized link between attributional style and depression using a variety of populations, including children (Seligman & Peterson, in press), depressed patients in a Veterans Administration hospital (Raps, Peterson, Reinhard, Abramson, & Seligman, 1982), lower-class women (Navarra, 1981, cited in Peterson & Seligman, 1984), and, of course, college students (Metalsky, Abramson, Seligman, Semmel, & Peterson, 1982; Peterson, Schwartz, & Seligman, 1981). Further, support has been obtained using both cross-sectional and longitudinal designs, and employing both the ASQ and open-ended measures of attributional style (see Peterson & Seligman, 1984, for a review).

Importantly, however, although each of these studies has obtained support for one or more predictions derived from the attributional model, few of them have confirmed the importance of all three attributional dimensions within a single study. For example, Metalsky et al. (1982) found that depression correlated with the tendency to make internal and global (but not stable) attributions, whereas Hammen and Cochran (1981) obtained effects only on the globality dimension. Although this is not an indictment of the theory per se, it highlights the need to explore other variables that may mediate between perceived lack of control and depression (see Ickes & Layden, 1978; Peterson & Seligman, 1984). It may be that the perceived controllability of the cause, rather than its internality, stability, or globality, may best predict was due to "the kind of person you are" (a characterological attribution) and how much to "what you did" (a behavioral attribution). Her data showed that, although depressed and nondepressed subjects did not differ in the degree to which they made generally internal attributions, depressed subjects were more likely than less depressed subjects to make characterological attributions for bad events. In addition, she found that depressed subjects made stronger attributions to luck and chance, and scored higher in external locus of control (the belief that one's outcomes in life are largely beyond one's control). In short, depressed subjects were more likely to attribute negative events to aspects of their own character and other uncontrollable factors than were nondepressed subjects, a finding that is consistent with the attributional model.

Other studies have replicated the positive correlation between characterological self-blame and depression, as well as demonstrating that behavioral self-blame is negatively correlated with depression (Anderson et al., 1983; Peterson et al., 1981). Together, these studies suggest that whether individuals make internal attributions per se is less important than whether their internal attributions emphasize stable, characterological deficiencies or unstable, behavioral factors, such as the momentary use of an inappropriate strategy. Even so, the research on characterological and behavioral selfblame provides further evidence that attributions are related to the experience of depression.

Depression and Perceptions of Control

Another implication of the learned helplessness approach to depression is that some people may be particularly prone to depression either because they have difficulty detecting contingencies between their actions and outcomes or because they tend to underestimate the degree of contingency that does exist. Such perceptual tendencies may lead people to conclude, erroneously, that they have little control over their outcomes, the results being helplessness and depression. However, as plausible as this notion sounds, research evidence clearly refutes it. Not only is there no reason to suspect that depression-prone individuals are deficient in their ability to assess behavior-outcome contingencies, but there is strong evidence that depressed people are better at doing so than nondepressed, "normal" individuals.

Support for this point comes from research on the "illusion of control," or the tendency for people to perceive uncontrollable events as controllable (Langer, 1975). Several studies have demonstrated that people sometimes act as if they can influence the occurrence of chance events, such as the roll of dice and the random selection of lottery winners (Langer, 1983). Research has shown that nondepressed subjects are more susceptible to the illusion of control than are depressed subjects; that is, normals are more likely to overestimate the degree of control they have over uncontrollable events (Alloy & Abramson, 1979, 1982; Golin, Terrell, & Johnson, 1977; Golin, Terrell, Weitz, & Drost, 1979). Put another way, depressed people are more accurate in assessing contingencies between their behavior and outcomes than are nondepressed subjects, leading Alloy and Abramson (1979) to characterize them as "sadder but wiser."

At first glance, it would appear that learned helplessness theory has difficulty accounting for these results, because depressed people do not appear to underestimate their control over events. It is possible, however, that depressed individuals maintain the perception that they have little control over their outcomes even though they accurately assess behavior-outcome contingencies (see Alloy & Abramson, 1979). For example, depressed persons may recognize that certain behaviors lead to certain outcomes, but doubt that they can execute the requisite behaviors. Phrased in terms of Bandura's (1977) self-efficacy theory (see chapter 11, this volume), depressives may accurately assess outcome expectancies but underestimate efficacy expectancies, or the probability that they can successfully execute the behavior needed to produce the outcome. Further research is needed to assess this possibility (see Kanfer & Zeiss, 1983).

Alternatively, the research on illusion of control suggests that depression results not so much from perceived lack of control as nondepression results from perceived control. In other words, the tendency to overestimate one's degree of control provides a buffer against depression, and people who do not overestimate their control are more prone to depression. This raises the provocative possibility that the incidence of depression would be higher if everyone accurately perceived the minimal control they have over many objectively uncontrollable events.

Critique of the Attributional Model of Depression

Despite a fair amount of empirical support for the attributional model of helpless depression, questions remain regarding the usefulness of the attributional model as an explanation of both helplessness effects and naturally occurring depression.

First, virtually all of the research that has studied attribution and depression has been correlational in nature. Thus, at best, the data demonstrate that certain kinds of attributions and depressive symptoms covary, but we are unable to conclude that certain attributions actually cause depression as the attributional model suggests (Abramson et al., 1978). An alternative explanation of the attribution-depression relationship is that depression leads people to make "helpless" attributions. This possibility is strengthened by research showing that depressed subjects whose moods are temporarily raised by an experimental induction succumb to the illusion of control (i.e., their perception of control increases), whereas depressed subjects whose moods are not raised do not (Alloy, Abramson, & Viscusi, 1981). Similarly, there is evidence that, although depressed people engage in more characterological self-blame for negative events, characterological self blame does not appear to cause depression directly (Peterson et al., 1981; see also Danker-Brown & Baucom, 1982). In all likelihood, the relationship between attribution and depression is reciprocal: Certain attributions lead to depression, and a dysphoric mood increases the likelihood that the person will make pessimistic, "helpless" attributions. This possibility may explain why it is difficult to "think" oneself out of depression (see Kuiper, Derry, & MacDonald, 1982). In any case, more research is needed that explores the nature of the relationship among attribution, expectancy, and depression.

Second, both Seligman's (1975) original theory and the reformulation offered by Abramson et al. (1978) assume that, for helplessness effects to occur, the individual must first become consciously aware that outcomes are not contingent upon behavior. Recent research by Oakes and Curtis (1982) challenges this assumption. In their study, subjects shot a "light gun" at a target that had a photoreceptor cell in the bull's-eye. Because of the lighting in the room, subjects were unable to determine their accuracy visually whenever they hit the bull's eye (thereby allowing the researchers to experimentally manipulate subjects' perceptions of their accuracy). Results showed that actual noncontingency (when accuracy feedback was truly independent of subjects' performance) produced helplessness effects, whereas subjects' perceptions of noncontingency did not. In other words, learned helplessness seemed to occur even when subjects did not consciously realize that their behavior and outcomes were independent. Further, cognitive measures, such as attributions and moods, were unrelated to helplessness effects. Other failures to demonstrate a mediating role of perceptions and attributions in learned helplessness have been reported by Termen, Gillen, and Drum (1982) and Termen, Drum, Gillen, and Stanton (1982). The latter study showed that neither perceived uncontrollability of outcomes nor the internality, stability, and globality of attributions mediated learned helplessness effects.

What are we to make of such findings? For psychologists who assume that cognitions mediate between stimuli and behavior, findings such as these are difficult to interpret. In a response to these studies, Alloy (1982) pointed out that they do not adequately address two important aspects of learned helplessness theory. First, as noted earlier, expectations of future responseoutcome noncontingency, rather than attributions per se, are assumed to mediate helplessness effects. Although certain patterns of attributions should lead to such expectations, "helpless" expectations may occur for other reasons. Thus, it is possible for a person to hold a low responseoutcome expectancy without having made relevant attributions about previous performance. Because none of these studies assessed expectations directly, they do not adequately test the theory.

Second, Abramson et al. (1978) explicitly stated that the expectation of future uncontrollability is a sufficient, but not necessary condition for helplessness effects. Thus, behavioral deficits that characterize learned helplessness may occur for other reasons, and it is possible that the effects observed in these studies fall into this category. In short, it is not clear that studies such as these do much damage to the attributional model of learned helplessness, although they do raise several important questions that the theory does not explicitly address (see Alloy, 1982).

However, even if these particular studies do not provide evidence against the learned helplessness model, they raise an important question that psychologists interested in attributional processes have not adequately addressed: Do people spontaneously make attributions that resemble those elicited in research settings? It is quite possible that, although people provide researchers with attributions when asked to so, they do not spontaneously make such attributions as a matter of course (Hanusa & Schulz, 1977; Wortman & Dintzer, 1978).

To investigate this question, Peterson, Bettes, and Seligman (1982) asked subjects to describe the worst two events that happened to them during the past year. Even though the subjects were not prompted to explain or account for the events, each description included at least one explanation. Further, many of these explanations could be classified along the internality, stability, and globality dimensions. Thus, it appears that people do make spontaneous attributions, at least in regard to particularly traumatic events.

A remaining question regards the attributional dimensions that are most important in producing helplessness and depression. The Abramson et al. (1978) model posits three-internality, stability, and globality-and research attests that these dimensions are related to helplessness and depression. However, are they the most important? Wortman and Dintzer (1978) suggested that whether the individual believes the factors responsible for his or her state of helplessness are controllable may be far more important than whether they are seen as internal, stable, or global.

Recent work by Anderson and Arnoult (1985) supports this notion, showing that controllability is the single most important attributional predictor of depression. Further, they found that, once the controllability dimension is considered, the three attributional dimensions posited by Abramson et al. (1978) add little to predicting one's level of depression. The importance of attributions of controllability is also suggested by Janoff-Bulman's (1979) work on characterological and behavioral self-blame. As we saw earlier, internal attributions to characterological factors (which are uncontrollable) are more strongly associated with depression than are attributions to behavioral and strategy factors (which are more controllable).

One way to reconcile the Abramson-Seligman-Teasdale model with more recent research is to postulate an attributional hierarchy. According to this view, certain "lower order" attributional dimensions, such as stability and globality, may be subsumed by "higher order" dimensions such as controllability. Put another way, whether people's attributions involve controllable or uncontrollable factors may depend upon the kinds of "lower order" attributions they make. In any case, taken together, research attests to the importance of attributions-particularly those relevant to controllability-in the development of reactive depression.

Therapeutic Implications of Learned Helplessness Theory

As noted earlier, the attributional theory posits that helpless depression occurs when the attributions people make lower their expectancies of control in the future. Thus, broadly speaking, helpless depression should be reduced by changing depressed individuals' perceptions of powerlessness (Beach, Abramson, & Levine, 1981). Such an approach may be used either to attenuate an ongoing depressive episode or to reduce the individual's vulnerability to future bouts of depression.

It should be noted, however, that directly modifying clients' attributions (as discussed in chapter 2) is only one possible way to do this. Other steps may be taken to give clients a greater sense of control over their lives. For example, it may be necessary to lower their unrealistically high goals (e.g., Bandura, 1969; Ellis, 1962; Rehm, 1977), to provide training in skills in the domain in which they feel helpless (Bellack & Hersen, 1979), or to improve clients' ability to detect behavior-outcome contingencies. Beach et al. (1981) argued that practitioners should carefully assess the specific locus of a depressed client's difficulties, remembering that feelings of helplessness may arise from a number of sources. The interested reader should refer to Beach et al. (1981) for a detailed discussion of the implications of the reformulated attributional model for the treatment of depression.

Egotism as an Alternative Explanation of Helplessness

Any psychological phenomenon is open to multiple interpretations and explanations, Although the attributional model of helpless depression enjoys a great deal of popularity and empirical support, Frankel and Snyder (1978) have offered an alternative explanation of helplessness effects based on a self-esteem maintenance or "egotism" model. According to this approach, the perception that one is unable to control important outcomes threatens the individual's self-esteem, resulting in a sense of failure and, often, depression. Further, when faced with an opportunity to reexert control, the individual may fear a second failure and another loss of self-esteem. To avoid such damning feedback, the individual may exert minimal effort in subsequent attempts to exert control. By doing so, one can attribute a continuing failure to control one's outcomes to inadequate effort rather than personal incompetence, thereby preserving self-esteem. (Many readers will recognize this esteem-maintaining strategy as a type of "self-handicapping" [Berglas & Jones, 1978; Jones & Berglas, 19781, a topic covered in the next chapter.) Thus, the decreased motivation observed after depressing failures may result from people's attempts to avoid future failures, rather than from perceived helplessness.

Two studies have supported this explanation of helplessness effects (Frankel & Snyder, 1978; Snyder, Smoller, Strenta, & Frankel, 1981). In the experiment by Snyder et al. (1981), subjects first worked on unsolvable problems, and then worked on solvable problems in the presence of music that they thought would either hinder or facilitate their performance. Learned helplessness theory predicts that subjects should experience greater helplessness when an allegedly debilitating factor is present, thereby lowering their expectations regarding performance. An egotism model, on the other hand, posits that distracting music provides subjects with an external attribution for failure, thereby allowing them to exert full effort on the task without risking their self-esteem should they fail. Consistent with the egotism hypothesis, but contrary to learned helplessness theory, subjects solved more problems when the noise during the second task was ostensibly distracting.

The egotism explanation for helplessness and depression is similar to one suggested by Rothbaum, Weisz, and Snyder (1982), who distinguish between two ways of exerting control. Rothbaum et al. suggested that, when attempting to control events, people may try either to change the environment to fit their desires ("primary" control) or to change themselves to fit the environment ("secondary" control). When people believe they are unable to exert primary control over events, they may become passive and withdrawn in order to avoid the disappointment and possible loss of self-esteem that would result from repeated attempts to control seemingly uncontrollable events. Thus, rather than representing helplessness, decreased motivation in uncontrollable situations may be an attempt to exert "secondary" control by managing one's cognitive and affective reactions to such events.

Although promising, the egotism and secondary control models are underdeveloped as explanations of helplessness and depression and have been tested only in relation to performance on laboratory tasks. Whether they can be extended fruitfully to an analysis of naturally occuring helpless depression has yet to be seen.

Reactance as a Reaction to Loss of Control

Although there is little question that helplessness is a common reaction to perceived loss of control, another, opposite tendency is often observed. Reactance theory (Brehm, 1966, 1972) proposes that when a person's freedom of behavioral choice is threatened, he or she will experience "reactance" and become motivated to restore the freedom. This effect is easily observed in the "lure of the forbidden fruit"; when people find they cannot have something they may become more strongly motivated to attain it.

A moment's thought should show that reactance is precisely the opposite of learned helplessness. In helplessness, loss of control decreases one's motivation to pursue the goal; in the case of reactance, loss of control heightens this motivation. Importantly, the same conditions that produce helplessness, such as noncontingent outcomes and internal attributions for failure, have been found to produce an increase in performance in some studies (Hanusa & Schulz, 1977; Roth & Kubal, 1975).

In a paper addressing the apparent discrepancy between the learned helplessness and reactance formulations, Wortman and Brehm (1975) suggested that when a person expects to be able to influence a certain outcome but finds his or her control and freedom threatened, reactance is the initial response as the individual first tries harder to exert control. However, if after continued effort the individual becomes convinced that further attempts will not produce the outcome, helplessness results. The results of a study by Roth and Kubal (1975) directly support this integration of the theories. In their study, one instance of noncontingent feedback produced an improvement in subsequent performance (i.e., reactance), whereas three instances of noncontingent feedback produced a performance decrement (i.e., helplessness).

The Wortman-Brehm (1975) model highlights the need for an overriding theoretical framework that encompasses both reactance and helplessness effects. One possibility is that, when people first encounter events they have difficulty controlling, they find it hard to assess the stability, globality, and controllability of the problem. Because it is more facilitative to assume initially that the cause of the difficulty is unstable and specific, they may exert increased effort in an attempt to reexert control. However, if after repeated attempts to exert control they are still unable to do so, they may begin to assume the outcome is truly uncontrollable and experience helplessness.

Individual Differences in Response to Loss of Control

Whether people experience helplessness or reactance in response to uncontrollable outcomes may also be related to certain personality characteristics. Although individual differences in reactions to loss of control have not received a great deal of attention, two lines of research seem particularly promising.

Gender Differences

A great deal of research has shown that women become depressed more frequently than men (Weissman & Klerman, 1977). This fact suggests that there may be differences in the ways in which men and women tend to respond to uncontrollable outcomes. Specifically, sex differences in response to helplessness-producing situations may be mediated by so-called "sex roles," or the degree to which an individual possesses instrumental (i.e., masculine) and expressive (i.e., feminine) personality attributes (see chapter 7).

Baucom (1983) predicted that an instrumental sex-role orientation would be associated with attempts to regain control under conditions that normally produce helplessness. To test this, female subjects took what they believed was an intelligence test and received feedback that either was or was not contingent upon their performance. They were then told that the study involved a problem-solving task in which two people would work together, and they were given a choice regarding their participation on this task. They were told that they could take part in the task but not have control over the team's decision, take part in the task and have control over the team's decision, or not take part in the task at all.

As expected, results showed that, whereas women low in instrumentality avoided control of the group task, a great majority of the highly instrumental women chose to be in control. To the degree that helplessness results in depression, these results suggest that women who are low in instrumentality may be particularly predisposed to depression because they do not attempt to regain control in situations in which they feel they have lost it. The mediating effects of traditional socialization on depression in women deserve additional research attention.

Desire for Control

People's reactions to a loss of control may also be mediated by the degree to which they are motivated to exercise control over their lives. Burger (1985; Burger & Cooper, 1979) has introduced the concept of "desire for control," which he views as a stable personality characteristic reflecting the extent to which individuals are motivated to control the events in their lives.

Individuals who score high on the Desirability of Control Scale (Burger & Cooper, 1979) persist longer and exhibit greater effort when confronted by impediments (Burger, 1985). Thus, people who have a high desire for control may be more likely to experience reactance when their sense of control is first threatened. However, such people may also be more susceptible to depression, possibly because they want to have more control over events than is often possible (Burger & Arkin, 1980). Thus, such individuals try harder to exercise control but are more often frustrated or depressed at their inability to do so.

In short, the available data suggest a link between an individual's desire for control and reactions to loss of perceived control. Further work is needed, however, to fully explore the relationship between this characteristic and dysfunctional reactions such as helplessness and depression.

Negative Reactions to Control

Throughout this chapter, perceived lack of control has been consistently associated with dysfunctional reactions, including lowered motivation and increased depression. Evidence that perceived lack of control has deleterious effects is quite extensive, but there are also circumstances in which people react negatively to having control. For example, Rodin, Rennert, and Solomon (1980) found that subjects who were given control over a simple laboratory task subsequently showed decrements in self-esteem compared to subjects who did not have control. Likewise, Miller (1980) showed that subjects who had control in a reaction-time experiment were more hostile and anxious than those who relinquished their control to a partner.

Two different explanations of this effect have been suggested. Miller (1980) and Rodin et al. (1980) suggested that personal control carries a burden of personal responsibility that people sometimes find aversive. However, Burger, Brown, and Allen (1983) proposed that having control is aversive only when it raises concerns that the individual will be unable to demonstrate his or her competence. Faced with the possibility of appearing incompetent to themselves and others, people often prefer not to be in control and are willing to relinquish control to others. When control is voluntarily relinquished to others in order to avoid negative consequences, helplessness and depression should not occur.

Conclusions

People generally prefer to believe that they are able to control events that affect them, and they experience negative psychological and physical effects such as stress, depression, and feelings of helplessness when they perceive that they are unable to do so. The mediating role of attributions and expectancies in these phenomena has been explored extensively, and although the data are not unequivocal, most would agree that people's reactions to uncontrollable events are affected by the attributions they make. However, despite widespread interest in attributional processes among social psychologists, we think that the utility of attributional models of dysfunctional behavior has not yet been fully explored. Existing attributional models need to be elaborated, the critical attributional dimensions identified, measures of attributional style refined and validated, and more research conducted using clinical and subclinical populations. Only then will we be able to assess the full value of attributional models for understanding and treating dysfunctional reactions.