Changing the World and Changing the Self: A Two-Process ......Weiner, 1974) but is consistent with the position of other locus of control theorists who see limited ability attri-butions

Changing the World and Changing the Self:A Two-Process Model of Perceived Control

Fred RothbaumTufts University

John R. WeiszUniversity of North Carolina at Chapel Hill

Samuel S. SnyderNorth Carolina State University

There is extensive evidence that people strongly value and are reluctant to re-linquish the perception of control. Yet, both helplessness and locus of controltheorists interpret various "inward" behaviors (passivity, withdrawal, and sub-missiveness) as signs of relinquished perceived control. As evidence, they notethat inward behavior frequently is accompanied by causal attributions to limitedability, chance, and powerful others—all attributions that suggest uncontrolla-bility. In contrast, we claim,that these attributions and the behaviors to whichthey relate may often reflect a type of perceived control that is generally over-looked. People attempt to gain control not only by bringing the environment intoline with their wishes (primary control) but also by bringing themselves into linewith environmental forces (secondary control). Four manifestations of secondarycontrol are considered: (a) Attributions to severely, limited ability can serve toenhance predictive control and to protect against disappointment. Passive andwithdrawn behaviors reflect the attempt to inhibit unfulfillable expectations, (b)Attributions to chance can reflect illusory control, since people often construechance as a personal characteristic akin to an ability ("luck"). Individuals whomake attributions to chance may exhibit passivity and withdrawal in skill situ-ations, reserving energy and emotional investment for situations that allow themto capitalize on their perceived strength—that is, being lucky, (c) Attributionsto powerful others permit vicarious control when the individual identifies withthese others. Submission to a powerful leader, a group, or a deity sometimesenables the individual to join in their power, (d) All of the preceding attributionsmay foster interpretive control, in which the individual seeks to understand andderive meaning from otherwise uncontrollable events in order to accept them.When perceived control is recognized in both its primary and secondary forms,a broad range of inward behaviors can be seen as efforts to sustain rather thanrelinquish the perception of control.

Factor analytic studies with children (re-viewed in Achenbach & Edelbrock, 1978)and systematic observations of adults (e.g.,Beck, 1967; Marks, 1977) have identified a

This article was funded in part by National Instituteof Mental Health Grant 1-RO1-MH-34210-01.

The authors are indebted to Camille Wortman forher unusually thorough review of a draft of this article.Her comments added depth and clarity to many of theCentral points.

Requests for reprints should be sent to Fred Roth-baum, Department of Child Study, Tufts University,Medford, Massachusetts 02155.

set of interrelated problem behaviors includ-ing passivity, withdrawal, and submissive-ness (hereafter referred to as inward be-havior). The popularity of the learnedhelplessness model (Abramson, Seligman,& Teasdale, 1978; Seligman, 1975) is duelargely to its lucid account of at least someof these seemingly maladaptive behaviors(especially passivity and withdrawal). Ad-herents of this model maintain that percep-tions of uncontrollability are frequent an-tecedents of inward behavior. In support ofthis view, a number of studies indicate that

Journal of Personality and Social Psychology, 1982, Vol. 42, No. 1, 5-37Copyright 1982 by the American Psychological Association, Inc. 0022-3514/82/4201-0005$00.75

F. ROTHBAUM, J. WEISZ, AND S. SNYDER

experiences designed to induce perceptionsof uncontrollability lead to decreased learn-ing, decreased persistence, and depressedaffect—key components of inward behavior.

The locus of control literature also seemsto link perceived uncontrollability and in-ward behavior. Persons with an external lo-cus of control—those who see circumstances,chance, and powerful others as controllingoutcomes—are, in general, more likely tomanifest such inward behaviors as passivity,withdrawal, compliance, conformity, anddepressive symptomatology than are personswith an internal locus of control—those whosee their own effort or ability as controllingoutcomes (for reviews see Lefcourt, 1976,1980; Phares, 1976; Rothbaum, Wolfer, &Visintainer, 1979).1 Also relevant here arestudies in which perceptions of uncontrol-lability have been manipulated and locus ofcontrol has been assessed (Cohen, Rothbart,& Phillips, 1976; Diener & Dweck, 1978;Dweck, 1975; Gregory, Chartier, & Wright,1979; Hiroto, 1974; Pittman & Pittman,1979). While the findings entail complexitieswhich will be dealt with later in the article,in general they indicate that externality andmanipulations designed to decrease per-ceived control are both associated with in-ward behavior (also see Coyne, Metalsky,& Lavelle, 1980; Janoff-Bulman & Brick-man, 1980). Because of their considerableoverlap in this respect, the helplessness andlocus of control positions are referred to col-lectively as the uncontrollability model ofinward behavior.

In addition to their focus on perceptionsof control, uncontrollability theorists alsoemphasize the motivation for perceived con-trol. The general assumption, supported byresearch, is that perceptions of uncontrol-lability and decreased motivation for controlfuel one another in a downward spiral(DuCette & Wolk, 1973; Roller & Kaplan,1978). The resulting inward behavior is seen,in part, as a motivational deficit, that is, asa consequence of relinquishing the motiva-tion for control (Seligman, 1975).

Uncontrollability theorists' emphasis onthe motivation for perceived control hasmany roots in earlier theorizing, dating atleast from Groos's (1901) observation thatchildren derive pleasure from opportunities

to control stimulation. Theorists from a widevariety of perspectives have been emphaticabout the importance of control: Adler(1956) depicted control as an intrinsic "ne-cessity of life"; Malinowski (1955) notedamong primitives a "desperate" need to be-lieve they could master the world; White(1959), in a now classic review of the liter-ature on "effectance motivation," refers toa pervasive, intrinsic need to exercise controlover the environment; DeCharms (1968)argued that the urge to be effective in chang-ing the environment is "man's primary mo-tivational propensity." Evidence of motiva-tion for perceived control can be found in awide variety of research areas. It is promi-nent, for example, in research on (a) help-lessness and locus of control (briefly re-viewed in this article), (b) perceived "con-tingency" in infancy (e.g., Finkelstein &Ramey, 1977; Gunnar-Vongnechten, 1978;Watson, 1967), (c) effectance and masterymotivation in childhood (reviewed by Harter,1978), (d) intrinsic motivation, both in chil-dren (e.g., Lepper, Greene, & Nisbett, 1973)and adults (Deci, 1975), (e) the self-servingbias in children (Weisz, 1980, 1981) andadults (Snyder, Stephan, & Rosenfield, 1978;Zuckerman, 1979), (f) feelings of compe-tence in the elderly (Langer & Rodin, 1976;Schultz, 1980), and (g) the attractivenessand adaptive value of different types of per-ceived control, such as predictability (Av-

1 Rotter (1966), in original formulation of the locusof control construct, defined internality in terms of bothself-contingency and self-causality. This definition leadsto confusion in the case of certain attributions, such asattributions to limited ability, which imply an absenceof self-contingency (independence between the self's re-sponse and the environmental outcomes) but the pres-ence of self-causality (the self s inability is the ascribedcause of the environmental outcome). In this article weequate internality with contingency and externality withnoncontingency. This position conflicts with the positionof the locus of control theorists who depict limited abilityas an internal attribution (e.g., Abramson et al., 1978;Weiner, 1974) but is consistent with the position of otherlocus of control theorists who see limited ability attri-butions as contributing to the belief that effort does notmake a difference—a belief that exemplifies externality(e.g., Cohen, Rothbart, & Phillips, 1976; Diener &Dweck, 1978; Lefcourt, 1976). For a fuller discussionof this topic, see Janoff-Bulman (1979) and Rothbaum(1981).

A TWO-PROCESS MODEL

erill, 1973; Burger & Arkin, 1980; Lefcourt,1973).2

Motivation for perceived control and itscorollary, aversiveness of perceived uncon-trollability, are cornerstones of the uncon-trollability formulation (Lefcourt, 1976; Se-ligman, 1975). Yet ironically, uncontrol-lability theorists may be shortchanging thegenerality and potency of this motive. In thisarticle we argue that the motivation to feel"in control" may be expressed not only inbehavior that is blatantly controlling butalso, subtly, in behavior that is not. In somecases, inward behavior may reflect a relin-quishing of the powerful motive for per-ceived control. In other cases, however, suchbehavior may be initiated and maintainedin an effort to sustain perceptions of control.This effort is particularly likely when theinward behavior helps prevent disappoint-ment, when it leads to a perception of align-ment with forces such as chance or powerfulothers, and when it is accompanied by at-tempts to derive meaning from a situation.The uncontrollability model does not explainany of these phenomena.

A related problem is that the uncontroll-ability model does not account for the per-sistence of some inward behavior. This per-sistence is evident in research on depressives'negative response set (Beck, 1967) and inresearch on obsessive compulsives' persev-erative thoughts and actions (Rackman &Hodgson, 1980). There is even evidence ofpersistence in the research on helplessness.For example: dogs exposed to uncontrollableshock subsequently have to be dragged intoa safety area (Seligman, 1975); individualssubjected to uncontrollable social situationsmanifest increased perseverative behavior(Goetz & Dweck, 1980); and individualsexposed, to unsolvable tasks go out of theirway to find reasons why they cannot succeed(Diener & Dweck, 1978; Seligman, 1975).These findings, and others reviewed later inthe article, indicate that persons sometimesactively undermine their own chances of suc-cess. In all of these cases, inward behaviorpersists well beyond the time during whichit is adaptive by rational standards, and itpersists despite attempts by others to alterit. It is largely this quality of persistence thatprompts analytically oriented theorists to

speculate about deeper motives and to labelthe behavior "neurotic" (e.g., Kessler, 1966;Wolman, 1965). Such persistence suggeststhat the behavior does satisfy some under-lying purpose and that it does not merelyreflect a "deficit" in motivation .

These phenomena, we suspect, can best beunderstood by broadening our conception ofcontrol and how it is manifest in everydaylife. Simply stated, our proposal is this: Be-cause control is so valued, the quest for itis rarely abandoned; instead, individuals arelikely to shift from one method of strivingfor control to another. The uncontrollabilitymodel and the present formulation, then,disagree as to how frequently motivation forcontrol is abandoned. This disagreementgoes hand in hand with a related disagree-ment about the pervasiveness of perceiveduncontrollability. Many situations that ac-cording to uncontrollability theorists indi-cate a perception of uncontrollability aresituations in which we believe perceived con-trol is actually a major dynamic (cf. Janoff-Bulman & Brickman, 1980). In large part,these disagreements regarding motivationand perception hinge on different concept-ualizations of control. A reasonable nextstep, then, is to contrast uncontrollabilitytheorists' definition of control with the def-inition we propose.

Definitions of Control

Helplessness and locus of control theoristsconceptualize perceived control similarly.Both place a major emphasis on the contin-gency between action and outcome (Rotter,1966; Seligman, 1975). Related constructssuch as perceived choice and competencealso figure prominently in both helplessnesstheory (e.g., Langer & Benevento, 1978)

2 Recently, some authors claimed that the motivationfor control is not as pervasive as has typically been as-sumed (Brehm & Brehm, 1981; Rodin, Rennert, &Solomon, 1980). They take issue with the notion of ageneralized motivation to seek control and with the no-tion that control is always desirable. As already noted,we believe there is considerable formal and informalevidence of a generalized motivation to seek control, butwe also recognize that much more research is neededto resolve this matter. Later in this article we brieflyconsider the evidence indicating that control sometimesappears undesirable (see pp. 15-16).

F. ROTHBAUM, J. WEISZ, AND S. SNYDER

and locus of control theory (e.g., Lefcourt,1976; Rotter, 1966; see also Weisz & Stipek,in press). In each case the individual's abilityto change the environment to fit the self sneeds is emphasized. Seligman's definitionof contingency, for example, makes it clearthat the individual's actions must lead tochange in the environment. Even White, whohas identified myriad expressions of control,emphasizes throughout his formulation theindividual's ability to affect the environ-ment.

Unlike most uncontrollability theorists,we conceive of control as a two-processrather than a one-process construct. The firstprocess involves attempts to change theworld so that it fits the self s needs. Whenthis process is salient, we use the expressionprimary control. It is our belief that uncon-trollability theorists have equated controlwith primary control and have all but ig-nored a second process we wish to high-light—attempts to fit in with the world andto "flow with the current." We refer to thisprocess as secondary, and when it is salient,we employ the expression secondary control.

The attempt to change things that offerresistance, which characterizes primary con-trol, is apt to produce some satisfying suc-cesses but also some disappointing failures.The attempt to adjust to resistances, whichcharacterizes secondary control, is apt to beexperienced as safer and to lead to less ex-treme highs and lows. Neither process isthought to exist in pure form; often bothprocesses are intertwined, as when personsnegotiate and compromise (Lazarus, 1981).Often there is vacillation between the twoprocesses, as when persons shift betweenphases of encounter and retreat in copingwith a physical disability (Shontz, 1975;Silver & Wortman, 1980). Thus, the differ-ences between primary and secondary con-trol should be thought of as differences inemphasis.

A rough analogy can be drawn betweenthis primary/secondary relationship and thecomplementary processes of assimilation andaccommodation as described by Piaget(1970) in his theory of cognitive develop-ment. Assimilation denotes the tendency toperceive the environment in ways consistentwith one's existing cognitive structures. In

extreme cases of both assimilation and pri-mary control, the individual, in a sense,places his or her desires above the demandsof reality. In complementary fashion, ac-commodation is the tendency to modify one'scognitive structures in an attempt to effecta better fit with reality. As is true of sec-ondary control, in extreme accommodationthe demands of reality overwhelm the self sdesires (cf. Lazarus & Launier, 1978, p.314). Averill (1973) highlights this accom-modative process when he notes that controlincreases to the extent that the individual isable to "accommodate to necessity" (cf.Chein, 1972; Kelly, 1955). Piaget and othercognitive developmental theorists argue thatthe goal of development is equilibration—the coordination of assimilation and accom-modation—so as to optimize the individual'sadaptation to the environment. Similarly, inthe model proposed here, optimal adaptationis defined as the coordination of primary andsecondary control.

The terms primary and secondary areadopted for several reasons. First, they re-flect the fact that one process (primary) hasgenerally received more attention than theother (secondary). Second, the terms distin-guish between the presence and absence ofmore powerful controlling agents: If the selfis the most powerful agent, then control isprimary; if more powerful agents are ac-knowledged (e.g., a task that one's sev-erely limited ability makes insurmountable,chance, or a powerful other person), theself s control is secondary. Another virtue ofusing thb terms primary and secondary isthat they capture differences in the temporalsequence of the control processes they delin-eate. Secondary control is most likely to oc-cur after attempts at primary control havefailed. Naturalistic evidence on the temporalsequence of the two types of control indicatesthat behaviors reflecting secondary control(passivity, withdrawal, and submissiveness)are a late stage of reaction to stress; typicallythey are preceded by a stage of anger andprotest—behaviors that we associate withprimary control (for evidence regardingstages see Baum, Aiello, & Calesnick 1978;Klinger, 1975; Mann & Dashiell, 1975;Rothbaum, 1980; Wortman & Brehm, 1975).However, the evidence is far from clear-cut

A TWO-PROCESS MODEL

(Silver & Wortman, 1980), and we suspectthat vacillation between primary and sec-ondary control is common.

The primary/secondary distinction wepropose resembles in certain respects theoffensive/defensive distinction frequentlyemployed in the psychological literature.One problem with the latter is that it hasbeen used to refer to such a broad varietyof phenomena that its meaningfulness is di-minished. Secondary control differs fromdefensiveness in that it does not refer to allattempts to neutralize instinctual urges (asdoes use of the term defensive in psycho-analytic writings), and it does not refer toattempts to maintain self-esteem by forget-ting or denying the occurrence of failure (asdoes use of the term defensive in attributiontheory literature). In fact, we would gener-ally regard the protection of self-esteemthrough denial as an attempt to maintainbelief in the eventual effectiveness of pri-mary control—the ability to change the en-vironment.

Proof of Uncontfollability

Differences between our definition of con-trol and that advocated by uncontrollabilitytheorists lead to differences in the criteriaused to infer uncontrollability. In this sectionwe attempt to explain these differences.

The Uncontrollability Model

Uncontrollability theorists rely on severalkinds of information in deciding whether ornot a given situation is perceived as uncon-trollable. First, the individual's behavior it-self may indicate perceived uncontrollabil-ity. Passive, withdrawn, and to a lesserextent, submissive behaviors often do notserve instrumental ends (e.g., to obtain adesired goal, solve a problem, or terminatean aversive event). In such cases, uncon-trollability theorists typically infer an aban-doning of attempts to control. They are par-ticularly likely to draw this inference whenthere is a cluster of the above behaviors.However, uncontrollability theorists recog-nize a variety of exceptions to this rule, andthey do not hinge their proof of uncontrol-

lability on this evidence alone. More com-pelling, they believe, are the causal at-tributions that are frequently found inconjunction with these inward behaviors.Both helplessness theorists (Hiroto & Selig-man, 1975; Klein & Seligman, 1976; Miller& Seligman, 1975) and locus of control the-orists (Lefcourt, 1976; Phares, 1976; Rotter,1966) claim that the tendency to attributefailure to severe limitations in ability,3 tochance, and to powerful others relates to thetendency to engage in inward behavior (cf.Wortman & Brehm, 1975), especially whenthe attributions are global and stable(Abramson, Seligman, & Teasdale, 1978).Because these theorists equate control withthe ability to change the environment, theymaintain that these attributions reflect per-ceived uncontrollability.

Probably the most widely cited evidenceof perceived uncontrollability is that theabove attributions and inward behaviorsboth seem to result from manipulations de-signed to induce perceived uncontrollability.The demonstration of these causal connec-tions is a major objective of helplessness re-search. Manipulations such as noncontin-gent aversive events (e.g., loud noise, electricshock) and unsolvable problems have re-peatedly been associated with the above at-tributions. Zuckerman (1979), for example,found that failure experiences led to an in-creased occurrence of external attributionsin 27 of the 38 studies he reviewed. Simi-larly, persons high in externality on locus ofcontrol measures tend to have socializationhistories indicating repeated experiences offailure (Lefcourt, 1976; Crandall, Note 1).

Manipulations of uncontrollability havebeen shown to lead to inward behavior, par-ticularly passivity (e.g., Benson & Kennelly,1976; Cohen, Rothbart, & Phillips, 1976;Glass & Singer, 1972; Griffith, 1977; Hir-

3 Although locus of control theorists refer to attri-butions to circumstances rather than to attributions tolimited ability, the major circumstance they consider istask difficulty. Thus, there is substantial overlap in theseformulations. The qualifier "severe" is added becauseof findings that attributions to slight to moderate lim-itations in ability lead to noninward behavior (Pittman& Pittman, 1979; Roth & Kubal, 1975; Tennen & Eller,1977).

10 F. ROTHBAUM, J. WEISZ, AND S. SNYDER

oto, 1974; Hiroto & Seligman, 1975; Roth& Kubal, 1975; Tennen & Eller, 1977;Thornton & Jacobs, 1971; for a review seeRoth, 1980). There is also evidence thatwithdrawn and submissive behaviors are as-sociated with the aforementioned attribu-tions and manipulations. Because previousreviews of the uncontrollability model havegiven less attention to the latter behaviors,a brief consideration of them is appropriatehere.

Most uncontrollability theorists wouldprobably regard withdrawal as a manifes-tation of perceived uncontrollability (e.g.,Goetz & Dweck, 1980). As Carver (1979)suggested, the "impulse to withdraw is basicto a wide variety of helplessness effects" (p.1276). However, Carver also noted thatmany helplessness paradigms have failed toinclude opportunities for physical with-drawal. When physical withdrawal is pre-vented, he argued, subjects may engage incognitive withdrawal—a mental dissociationfrom the task. The decreased learning foundin many helplessness studies could be re-garded as evidence of this cognitive with-drawal (Carver, 1979).

Submissiveness is another behavior thathas been linked to perceived uncontrollabil-ity. Locus of control theorists explicitly drawconnections between submissiveness and per-ceived uncontrollability (Lefcourt, 1976).Although helplessness theorists have failedthus far to elaborate this connection, thehelplessness model is relatively new and opento extension in this direction. Seligman(1975) seemed favorably inclined towardjust such an elaboration of the helplessnessmodel when he reported anecdotal evidenceof a connection between helplessness andsubmissiveness (p. 25). In at least one study,helplessness was operationalized as a rela-tively submissive "going to get the experi-menter" as opposed to taking more indepen-dent action (Roth & Bootzin, 1974). Findingsfrom this study indicated greater submis-siveness following manipulations of uncon-trollability (see also Langer & Benevento,1978). A link between submissiveness, help-less behavior, and perceived uncontrollabil-ity has also been noted in studies involvingretarded children exposed to recurring fail-

ure experiences (Floor & Rosen, 1975;Weisz, 1979, in press; Yando & Zigler,1971).

To summarize, uncontrollability theoristsrely on several types of evidence in substan-tiating their claims: (a) The inward behav-iors in and of themselves suggest perceiveduncontrollability, especially when they co-occur, (b) These behaviors are associatedwith causal attributions indicating perceiveduncontrollability. (c) Both the behaviors andthe attributions have been shown to resultfrom experiences designed to induce per-ceived uncontrollability. Although uncon-trollability theorists to date have focusedmore on passivity than on withdrawal andsubmissiveness, the evidence reviewed heredemonstrates that the model has been usedto explain the latter behaviors as well.

A Two-Process Model

The uncontrollability model and the po-sition we propose lead to different criteriaof uncontrollability. We claim that attemptsto fit more effectively with the world con-stitute a process of secondary control; if weare right, attributions and behaviors citedabove do not necessarily provide proof ofperceived uncontrollability. There is a defi-nitional difference here, but its importanceshould not be exaggerated. Whether onechooses the term control when the secondaryprocess is salient is not crucial. Rather, atissue is how the attributions and behaviorsreviewed above may be most usefully ex-plained. We maintain that underlying theseattributions and behaviors is a powerful mo-tivation to fit in with the environment. Incontrast, uncontrollability theorists high-light the absence of attempts to change theworld and the absence of motivation for con-trol (cf. Wortman & Brehm, 1975). This isthe point of contention.

In four subsequent sections we review re-search involving causal attributions to se-verely limited ability, to chance, and to pow-erful others, along with the respective inwardbehaviors to which they frequently relate.These combinations of attribution and be-havior, we argue, often reflect a perceptionof secondary control. To support our two-

A TWO-PROCESS MODEL 11

process model, we will attempt to show thatthese attributions and behaviors are accom-panied by forms of persistence (e.g., under-mining success, increasing effort on chancetasks, seeking out powerful others, and in-creased attributional activity) that indicatea striving for secondary control. To intro-duce the two-process model, we first providea brief overview.

Attributions to severely limited ability canafford the individual control in the form ofmaking the best of a situation that is per-ceived as difficult or impossible to alter. Thiscontrol stems from the ability to predict theuncontrollable outcome (or its recurrence)and from the adjusting of the self s expec-tations accordingly. Adjustment of expec-tations is particularly important because un-fulfilled expectations undermine perceivedcontrol; obversely, fulfilled expectationsheighten perceived control. Because Averill(1973), Burgess and Holmstrom (1979),Lazarus (1966), Miller and Grant (1980),and other theorists concerned with this kindof control all emphasize predictability, werefer to it as predictive control.

Attributions to chance permit what is re-ferred to here as illusory control. Personssometimes attempt to associate themselveswith chance so as to share in the power ofthis larger force. Chance is viewed as a prop-erty of the individual; often persons speakof luck as an ally that they can "rely on,"and of being "born lucky," seemingly re-garding luck as a type of control. Throughvarious means—superstitious behavior, rit-uals, and, more simply, investing energy inchance-determined situations rather than inskill-determined situations—people attemptto align themselves with chance.

Attributions to powerful others are similarto attributions to chance in that they permitassociation with a more powerful force. Al-though such vicarious control may some-times serve instrumental ends, it often ap-pears to function as an end in its own right.That is, persons sometimes associate withothers simply for the sake of sharing psy-chologically in the others' control. Vicariouscontrol, then, is similar to the phenomenaof identification and deindividuation, inwhich individuals submerge a sense of self

in order to enhance a sense of close associ-ation with a more powerful entity.

Finally, all of the above attributions en-able the self to derive meaning from other-wise uncontrollable experiences. We refer tothe ability to interpret events so as to betterunderstand and accept them as interpretivecontrol (Averill, 1973; Frankl, 1963; Gar-ber, Miller, & Abramson, 1980).

The predictive, illusory, vicarious, and in-terpretive control described above are all in-stances of what we label secondary control.In each case, a tendency to change the selfso as to fit more effectively with the envi-ronment overshadows a tendency to changethe environment so that it fits the self sneeds. We believe that failure to considerthese and other instances of secondary con-trol underlies uncontrollability theorists'mistaken conclusion that perceived uncon-trollability and abandoning the motive tocontrol are so prevalent.

Although uncontrollability theorists mightacknowledge the existence of predictive, il-lusory, vicarious, and interpretive control,they would probably emphasize their pri-mary aspects. Most likely, predictive controlwould be defined as a belief in the ability topredict events and thereby plan a successfultask outcome; illusory control would be de-fined as an irrational belief that persons can,through effortful action, influence chance-determined (as well as skill-determined) out-comes (e.g., superstitious behavior throughwhich they erroneously believe their actionsdetermine chance outcomes); vicarious con-trol would be defined as a belief that one cangain control through powerful others, eitherby manipulating their actions or by imitatingthem; and interpretive control would be de-fined as a belief that one can find the mean-ing of a problem and thereby solve it. (SeeTable 1 for a summary of the primary andsecondary manifestations of predictive, il-lusory, vicarious, and interpretive control.)The key difference between primary and sec-ondary control is that in the former case thegoal is to change the environment whereasin the latter case the goal is to fit in with theenvironment. The above instances of pri-mary control are dissimilar to instances ofsecondary control in that they are unlikely


, to accompany attributions to severely lim-ited ability, to chance, or to powerful othersand they are unlikely to accompany nonin-strumental behaviors of a passive, with-drawn, or submissive kind.

In the following four sections, we attemptto show that many situations previouslythought to involve perceived uncontrollabil-ity are ones in which persons strive for sec-ondary control. These sections are organizedby the four kinds of secondary control justreviewed—predictive, illusory, vicarious, andinterpretive. Because we agree with uncon-trollability theorists that consideration ofbehaviors, attributions, and precipitatingevents is essential, evidence regarding all

three factors will be treated in each section.To support our claims we attempt to providebehavioral and self-report evidence of striv-ing for secondary control in situations inwhich persons (a) manifest passive, with-drawn, and submissive behavior; (b) attrib-ute outcomes to severely limited ability,chance, or powerful others; and (c) have justexperienced manipulations designed to in-crease perceived uncontrollability.

Predictive Control

When an attribution to severely limitedability occurs in conjunction with passive ofwithdrawn behavior, and when the situation

Table 1Primary and Secondary Processes

Type ofcontrol

Salientprocess Description

Predictive

Vicarious

Primary

Secondary

Primary

Secondary

Attempt to predict events so as to succeed at them.Active behavior and attributions to the self s effortand ability are likely, especially in tasks of moderatedifficulty.

Attempt to predict events so as to avoid disappointment.Passive and withdrawn behavior, especially in tasks ofmoderate difficulty. Active, persistent behavior inextremely easy or difficult situations. Attributions toseverely limited ability are likely.

Illusory Primary

Secondary

Attempt to influence chance-determined outcomes.Active behavior in chance and skill situations andattributions to the self s effort and ability are likely.

Attempt to associate with chance. Active behavior inchance situations but passive and withdrawn behaviorin skill situations and attributions to chance are likely.

Attempt to manipulate powerful others or imitate theirpower or ability. Instrumental submissive behavior ormanipulative behavior and attributions to the self seffort and ability are likely.

Attempt to associate with powerful others.Noninstrumental, submissive behavior and attributionsto others are likely.

Interpretive Primary

Secondary

Attempt to understand problems so as to be able tosolve them or otherwise master them. Active behaviorand attributions to the self s effort and ability arelikely.

Attempt to understand problems so as to derive meaningfrom them and to accept them. Passive, withdrawn,and submissive behavior and attributions to severelylimited ability, chance, and powerful others are likely.


is one in which effort would likely producea desired outcome, uncontrollability theo-rists typically conclude that the individualhas given up—has abandoned the attemptto exercise control. In contrast, we believethat in certain situations the combination ofan attribution to severely limited ability andpassive or withdrawn behavior allows forpredictive control. In the present context, theexpression predictive control refers to theability to predict aversive events to avoiddisappointment (cf. Averill, 1973; Lazarus,1966; Miller & Grant, 1980).

Individuals who have repeatedly experi-enced failure sometimes avoid tasks com-pletely or participate with minimal effort,even when they realize that they have a goodchance of success, In justifying their seem-ingly self-defeating behavior, these individ-uals often maintain that their ability to avoiddisappointment is their last line of defense.The av6rsiveness of disappointment was viv-idly demonstrated on a national scale in theUnited States during the Iranian hostagecrisis of 1980-1981. Many Americans wouldhave echoed the feelings of a parent of oneof the hostages: "We've been disappointedso many times. Now I won't allow myselfto believe in the eventual success of the ne-gotiations until I see them [the hostages]step off the plane." Disappointment is soaversive, we believe, because it adds the in-sult of secondary uncontrollability (not beingprepared for the outcome) to the injury ofprimary uncontrollability (not being able toalter the outcome). We believe that individ-uals actively strivfe to avoid disappointmentand that a major objective in doing so is togain an increased sense of secondary control.To support our claim, we will provide evi-dence that individuals exert effort and makesacrifices to avoid disappointment and thattheir objective is an increased sense of ef-fectiveness. We will rely primarily on threebodies of literature: research directly con-cerned with disappointment, research onachievement motivation, and research on re-actions to unpredictable stress.

Mettee (1971) provided the clearest em-pirical support for the position that disap-pointment is actively avoided. He arguedthat persons avoid success when it is per-

ceived to increase the probability of unful-fillable expectations and thus of eventualdisappointment. To test this notion, Metteeexposed subjects first to failure, then to suc-cess, and finally gave them the opportunityto accept or reject the success—that is, hereadministered the test at which the subjectshad just succeeded, thus providing them theopportunity to change their correct answers.Between failure and success experiences,some subjects were informed that an inter-view very likely confirming their initial fail-ing performance would occur at the conclu-sion of the study; other subjects were assuredthat no further information was forthcom-ing. As predicted, the former group, forwhom acceptance of success implied a greatervulnerability to disconfirmed expectations^was more rejecting of the success experience.To create a third condition, some of the sub-jects in the "future jnterview" conditionwere told that the forthcoming feedbackwould inevitably confirm their prior failure.Mettee hypothesized that when future fail-ure was assured, disappointment would notbe possible and that, as a consequence, sub-jects would accept success (i.e., not changetheir correct answers). His findings sup-ported this hypothesis. Subjects acceptedsuccess in this condition, reasoned Mettee,because it was irrelevant to their future ex-pectations. Subsequent studies in which dif-ferent manipulations were employed to makesuccess irrelevant have yielded similar find-ings (Brickman, Linsenmeier, & McCareins,1976; Maracek & Mettee, 1972).

Additional evidence for the view that per-sons actively strive to avoid disappointmentcomes from Wortman, Costanzo, and Witt(1973), who gave subjects a "social percep-tiveness task" with predetermined feedback.Regardless of whether feedback was positiveor negative, subjects who thought they mightbe tested further were less likely to perceivethemselves as capable than were subjects ledto believe there would be no further testing.One interpretation of these findings is thatsubjects wanted to mitigate the possibilityof disappointment in the new situation. At-tributions to severely limited ability in thesecases offer a distinct advantage: They min-imize the possibility of the kind of discon-


firmed expectation that is most disappoint-ing. To expect success but instead to fail isa double defeat; one has failed not only asa performer of the task but also as a pre-dictor of the outcome.

In an earlier pair of studies similar toMettee's except that the possibility of sub-sequent feedback was not mentioned, Aron-son and Carlsmith (1962) also found thatsubjects undermine unexpected success. Theyinterpret these results as evidence for a de-sire to avoid inconsistency. Although therehave been a few successful replications ofthe above results (Brock, Edelman, Ed-wards, & Schuck, 1965; Cottrell, 1965; Hai-mowitz & Haimowitz, 1966), most of thereplication attempts have not been support-ive (see Archibald, 1974, for a list of 10 suchstudies). Mettee's model provides an alter-native explanation for both the effect andits equivocal support: Subjects sometimesreject success in order to avoid disappoint-ment, but the degree to which success is re-jected depends upon the degree to which fac-tors contributing to fears of disappointment(e.g., threats of later feedback) are present.

Other consistency theorists have shownthat subjects with low self-esteem preferevaluators who provide mostly negative feed-back to evaluators who provide mostly pos-itive feedback (e.g., Backman & Secord,1959; Deutsh & Solomon, 1959). Hereagain we take issue with the desire-for-con-sistency explanation and instead suggest thatlow self-esteem persons fear disappointment.Consistent with our view, Jones's (1973) ex-amination of the relevant literature suggeststhat subjects in these studies often expect tobe given another test in the future.

The reliance on avoidance of disappoint-ment as a conscious strategy was investi-gated more directly in a study by Sidle,Moos, Adams, and Cady (1969). Personswho reported that they expect the worst asa means of coping tended to view themselvesas passive, slow, changeable, and low in self-esteem. These are precisely the perceptionsthat have led previous researchers to inferan absence of perceived control.

A number of other findings, which do notdirectly address the issue of disappointment,can nonetheless be seen as relevant to this

issue. In general, the findings indicate thatpersons low in self-esteem and persons ex-posed to failure show low preference for andlow effort on tasks that offer moderatechances of success but high preference forand high effort on extremely easy or ex-tremely difficult tasks. Before consideringhow avoidance of disappointment may use-fully explain these findings, a brief reviewof the relevant literature is in order.

Atkinson's (1957, 1964) findings onachievement motivation are the best knownin this area. He found an overall tendencyfor subjects to prefer moderately difficulttasks because they maximize the possibilityof a meaningful success experience. Atkin-son noted, however, that low self-esteem("failure avoidant") persons tend to avoidmoderate-level tasks (see also Kukla, 1974).The low self-esteem person, Atkinson be-lieved, seeks to attenuate failure, and thiscan best be accomplished by maximizing theprobability of success (very easy tasks) orby minimizing the shame of failure (verydifficult tasks). Subsequent research by Kar-abenick and Yousseff (1968) showed thatpersons chronically concerned about failureperform better at tasks described as verydifficult. Similarly, Shaban and Jecker(1968) showed that subjects exposed to afailure manipulation prefer evaluators whoevaluate almost everyone positively or neg-atively to more moderate evaluators (whoevaluate about half of the people positively).

Following in this line of research, Frankeland Snyder (1978) exposed subjects to un-solvable problems and subsequently assessedtheir passivity and withdrawal on either ex-tremely difficult or moderately difficult tasks.According to Frankel and Snyder, helpless-ness theorists maintain that passivity andwithdrawal result from perceived uncontroll-ability, and, by this logic, more of these be-haviors would be expected on the extremelydifficult tasks. Their own self-esteem/defen-sive-effort model, in contrast, leads to theconclusion that passivity and withdrawal re-flect an attempt to excuse failure and thatthe need for such excuses is greater withmoderately difficult tasks. Their results, in-dicating that passivity and withdrawal weregreatest in the moderately difficult task con-


dition, supported the self-esteem/defensive-effort model.

Thus, there have been numerous studies,spanning over 20 years, indicating that per-sons with low self-esteem and persons ex-posed to failure tend to avoid tasks of mod-erate difficulty. Frankel and Snyder'sexplanation that these subjects are seekingan excuse for possible failure is an appealingone (see Snyder & Wicklund, 1981, for anelaboration of this position). We believe,however, that avoidance of disappointmentprovides an equally appealing explanation.Subjects are unlikely to be disappointed byperformance at very easy tasks because theywill almost definitely succeed and are un-likely to be disappointed by performance atvery difficult tasks because their expecta-tions of success are very low.4 Moreover,there are two pieces of evidence that castdoubt on Frankel and Snyder's interpreta-tion: (a) The subjects who exerted little ef-fort on the moderately difficult tasks scoredlow in self-reports of esteem, (b) These sub-jects did not make a greater number of effortattributions for their failure (e.g., "I didn'ttry"). Neither finding fits with Frankel andSnyder's hypothesis; in contrast, both find-ings are consistent with our avoidance ofdisappointment hypothesis. We believe thatthe emphasis on the self-esteem/defensive-ness explanation reflects psychologists' per-sistent focus on people's desire for primarycontrol (clinging to the belief that the selfcan, ultimately, change the environment)and their perception that primary control isan unmitigated good (cf. Janoff-Bulman &Brickman, 1980). Rather than clinging toa belief in primary control, we believe thatpersons sometimes actively oppose such be-liefs, especially when these beliefs are seenas opening the floodgates of disappointmentand, thus, as jeopardizing secondary control.

There is other evidence of the aversivenessof disappointment and of people's desire toavoid it through accurate prediction. Glassand Singer (1972) showed that a predictable(fixed interval) aversive event (loud noise)leads to fewer errors on a frustration taskand fewer omissions on a task of caution andattentiveness than does an unpredictable(random) aversive event, even with fre-

quency and intensity of aversive events iden-tical across conditions (for reviews of relatedstudies see Averill, 1973; Burger & Arkin,1980). Although aversiveness per se (inten-sity of noise) has some influence on perfor-mance, its effect pales in comparison withthat of unpredictability. A number of otherinvestigators also report that specific, accu-rate expectations can reduce the adverse ef-fects of stressful procedures (Epstein, 1973;Janis, 1958; Johnson, 1973, 1975; Johnson& Leventhal, 1974; Leventhal, Brown,Shacham, & Engquist, 1979). In interpret-ing some of these findings, Lefcourt (1973,1976) pointed out that prediction allows theindividual to steel him- or herself to the aver-sive event and thus to not be caught offguard—that is, prediction protects againstdisappointment. In a similar vein, Miller(1980) espoused an "internality view of pre-diction" according to which preparing forthe worst enables the individual to "matchhis or her internal state with external events"(cf. research on sensitization cited in Averill,1973; Janis's, 1958, research on the "workof worry"). These authors would probablyagree with our conclusion that predictabilityallows the individual to adjust to the uncon-trollable event, thus making it, in a second-ary sense, "controllable."

Also relevant here are studies showingthat in some instances increased perceptionof opportunities for primary control can bestress inducing. Stress occurs when the avail-able methods of control are ineffective, forexample, when responses intended to ter-minate shock yield inconsistent or negativefeedback (Weiss, 1968; 1971a, 1971b,1971c). In other cases, in which subjects are

"Trope (1975) and Trope and Brickman (1975)found evidence indicating that the major virtue of mod-erately difficult tasks is that they are typically high in"diagnosticity" (i.e., they are especially able to distin-guish the competent from the incompetent). These au-thors suggested that very difficult tasks are appealingto failure-avoidant individuals because of their desireto avoid diagnosticity. This idea resembles our sugges-tion that very difficult tasks enable failure-avoidant in-dividuals to avoid disappointment. Consistent with thediagnosticity view as well as with ours is recent evidencethat persons made anxious about failure subsequentlyshow less preference for highly diagnostic tasks (Schulz& Hanusa, Note 3).


given choices that do affect outcomes, butthey lack the information necessary to in-crease the likelihood of a beneficial result,increased stress is also noted (Rodin, Ren-nert, & Solomon, 1980). In the latter cir-cumstances, increased control (choice) alsoleads to decreased self-esteem. It is inter-esting that when subjects are told that theyare not expected to do well (a manipulationthat is likely to mitigate perceived disap-pointment), the aversive effects of choice arenot obtained. We maintain that in both ofthe above studies there is an inaccurate per-ception of opportunity for primary controlthat leads to an increase in unfulfillable ex-pectations and thus to the likelihood of dis-appointment. The advantage of secondarycontrol is the avoidance of this disappoint-ment.

Summary

The evidence reviewed here suggests thatattributions to severely limited ability andthe passive and withdrawn behavior thatoften accompanies these attributions some-times reflect the individual's striving for pre-dictive control. Individuals who can accu-rately predict uncontrollable events canthereby avoid unfulfillable expectations (i.e.,disappointment) that fuel the. perception ofuncontrollability. This tendency to adjustthe self s expectations and behaviors to makethem compatible with a difficult environ-ment is an example of secondary control.The evidence reviewed here indicates thatthere is an increased desire for predictivecontrol following uncontrollable experiencesand that the striving for predictive controloften takes the form of inward behavior.Moreover, there is evidence that this strivingfor predictive control, as shown by the avoid-ance of moderately difficult tasks, is likelyto accompany attributions to severely lim-ited ability (Kukla, 1974; Maracek & Met-tee, 1972). What is not yet available is ev-idence that persons in the situations describedabove actually perceive themselves as gain-ing predictive control. This evidence is es-pecially needed because of the competingexplanations for the phenomena in question.

This section and the sections that follow

contain evidence of the existence of second-ary control in situations previously thoughtto entail perceived uncontrollability. Severalof the studies cited here (e.g., Frankel &Snyder, 1978; Glass & Singer, 1972; Mettee,1971) employed manipulations of uncon-trollability similar or identical to those em-ployed in helplessness studies. It is possiblethat some subjects who appear to be givingup are in fact abandoning control, whereasother subjects who appear to be giving upare attempting to control disappointment.

Illusory Control

The treatment of illusory control and ofattributions to chance is situated betweenconsideration of attributions to severely lim-ited ability and to powerful others becauseit bears an essential similarity to both.Chance attributions resemble attributions toseverely limited ability (treated in the lastsection) in that both give rise to passivityand withdrawal in situations in which activ-ity is likely to lead to success. Moreover,depression, which is commonly regarded asa behavioral manifestation of perceived un-controllability (Beck, 1967; Seligman, 1975),has been linked both to severely limited abil-ity attributions (see Journal of AbnormalPsychology, February 1978) and to chanceattributions (Janoff-Bulman, 1979). Chanceattributions are also similar to attributionsto powerful others (treated in the next sec-tion) in that both involve association witha powerful external agent (in the case ofchance attributions, the agent is fate). Oneindication of the close connection betweenattributions to chance and attributions topowerful others is the variety of phenomenaincorporating both—belief in supernaturalpowers, witches, demons, fortune-tellers,psychics, mystics, and astrologers. As a con-sequence of their relation to attributions in-volving severely limited ability and powerfulothers, chance attributions are often re-garded as particularly indicative of per-ceived uncontrollability.

Our claim here, as in the preceding sec-tion, is that the attributions involved allowfor a type of perceived control, referred tohere as illusory control. Persons seek to align


themselves with the force of chance so thatthey may share in the control exerted by thatpowerful force. In addition, we maintain thatindividuals who perceive illusory control ina situation nonetheless recognize the chancedetermination of that situation. What we areconcerned with, then, is not simply the trans-formation of a perception of chance into a

, perception of skill. Rather, we believe thatpeople are often aware that chance is op-erating and that they continue to refer toluck and fate in explaining outcomes, butthat they perceive chance as a force withwhich they can align themselves. Moreover,persons who are most familiar with chanceevents (e.g., gamblers) are often most sus-ceptible to perceiving them as controllable(Langer, 1977). Thus, the perception ofchance as a type of control cannot be dis-missed simply as reflecting a lack of famil-iarity with chance.

The Relationship Between Skill ,and Chance

One of the proofs of illusary control isevidence that persons perceive control in pre-dominantly chance-determined situations.Research by Ayerhoff and Abelson (1976),Langer (1975, 1977), Langer and Roth(1975), and Wortman (1975) provides thenecessary evidence. These investigators haveshown that under certain conditions subjectsrespond as if a chance-determined situationwere controllable. Specifically, when factorsnormally associated with skill situations(e.g., choice, effort, task familiarity) are in-troduced into chance situations in whichsubjects cannot influence the outcome, anillusion of control is created (Langer, 1977),Manifestations of this illusory control in-clude increased confidence in success, in-creased risk taking, and self-reported per-ceptions of control. Naturalistic observationsof gambling lead to similar conclusions(Langer, 1977).

The above authors, especially Langer,suggested that under these circumstances,subjects are impressed by the superficial sim-ilarity between chance and skill situationsand, as a consequence, they fail to distin-guish adequately between the two. Support-

ing their interpretation are other findingsthat subjects perceive causal relationships inthe absence of objective contingency (Bruner& Revusky, 1961; Chapman & Chapman,1967; Golding & Rorer, 1972; Smedslund,1963; Starr & Katkin, 1969; Ward & Jen-kins, 1965; Hamilton & Gifford, Note 2).Whereas Langer and her colleagues did notdifferentiate between subjects' sense of con-trol in skill situations and their sense of con-trol in chance situations (except to say thatthe latter is inaccurate), we maintain thatthe former is a perception of primary controland that the latter is, in at least some cases,a perception of secondary control—a senseof participation in the more powerful forceof chance.

The studies described above provide somesupport for the interpretation we propose.First, intelligent and well-educated subjects(Duke University undergraduates in onestudy, Yale undergraduates in another) wereemployed in this research, and the chancenature of the tasks was made explicit. Suchsubjects are likely to perceive the chancenature of the situation accurately. Second,included among the factors that stimulatethe illusion of control is passive involvement:The longer people wait for the outcome ofa lottery to be announced, the greater theirconfidence that they will win. Langer (1975)reasoned that in skill situations passive in-volvement permits time for thought and thusfor the development of strategies that en-hance one's likelihood of success. Such strat-egies, however, are unlikely to emerge whenone contemplates a lottery in which the out-come will be determined via blind drawingby unknown people in an unknown location.Why then does confidence increase the longerone waits for the outcome? We believe thereason is that, over time, "the ticket holderfeels increasingly associated with the forcesof chance that will operate in the situation.The very fact of passivity underscores thesecondary nature of this perceived control.A similar argument applies to Langer's ma-nipulation of subjects' familiarity with thetask. If anything, increased familiarity witha chance task should sharpen one's aware-ness of its chance-determined nature andreduce the likelihood that it will be confused


with a skill task. But increased familiaritymight well increase one's sense of partici-pation in, or alignment with, the task—thatis, one's perception of secondary control.Support for our view would be provided bymore direct evidence that manipulationssuch as passive involvement and task famil-iarity increase illusory control (perceivedprobability of success) under conditions inwhich people are fully aware that no directmanipulation is possible and in which no"strategies" have been developed to influ-ence outcomes.

Although the dynamic we propose is dif-ferent from that proposed by Langer and hercolleagues, the two are not incompatible.Illusory control may involve both an asso-ciation with chance (a secondary process)and a misperception of chance-determinedsituations as skill-determined (a primaryprocess). Because illusory control is perva-sive in the laboratory and in everyday life,both influences are possible.

Locus of Control and Behavior in ChanceSituations

Research on locus of control has repeat-edly established a link between externalityand various forms of inward behavior (seep. 6). These links take on added significanceitr light of evidence on the behavior of ex-ternals in chance situations. In .consideringthis evidence, it is important to keep in mindthat the chance situations are clearly iden-tified as such and that externals, who man-ifest the most illusory control, are also themost likely to attribute outcomes to chance.Indeed, externality is operationalized, inpart, as a tendency to attribute outcomes tochance, luck, and fate (see Lefcourt, 1976;Phares, 1976; Rotter, 1966).

Whereas internals, compared with exter-nals, have consistently been found to preferand perform better on skill-determined tasks,externals have been shown to express agreater preference for chance tasks than dointernals and even to prefer chance tasks toskill tasks (e.g., DuCette & Wolk, 1973;Kahle, 1980; Schneider, 1968, 1972). Rotterand Mulry (1965) found that externals con-sult fortune-tellers and have their horoscopesread more often than do internals, and

DuCette and Wolk (1973) found that exter-nals showed a greater preference for an ex-trasensory perception task than did internals.These authors interpreted their findings asdemonstrating externals' greater preferencefor situations that are congruent with theirgeneralized expectations for control.

Support for the preceding view comes alsofrom findings that externals exhibit "reac-tance" in situations entailing an impersonalthreat, but not in situations entailing a per-sonal threat to control (Cherulnik & Citrin,1974).5 Impersonal threat, operationalizedas denying the subjects a choice of rewarddue to circumstances beyond the control orintent of the experimenter, was designed tofoster attributions to "chance, fate, or theaction of another over which [the subject]has no control" (p. 399). A significant in-teraction between locus of control and theway in which control was eliminated indi-cated that externals, compared with inter-nals, reported greater liking for the elimi-nated reward ("reactance") following theimpersonal, chance-like elimination of thereward (it did not arrive in the mail on time)but less liking for the reward when it waseliminated in a personal way (the experi-menter decided it was not appropriate forthe subject). In explaining their findings,Cherulnik and Citrin suggested that exter-nals expect greater control over impersonal,chance-like rewards than do internals andthat it is the violation of this expectation thatleads to reactance.

Karabenick and Srull (1978) and Srulland Karabenick (1975) reported findingssimilar to those above. In both studies, ex-ternals cheated more than internals on a taskdescribed as chance determined, but an op-posite pattern held when the task was de-scribed as measuring skill (see also Aronson& Mettee, 1968). In both skill and chancetasks, insoluble problems were presented,thereby inducing a loss of control similar tothat induced in the Cherulnick and Citrinstudy. Because cheating is a common reac-tion to perceived loss of control (Brehm,1966; Wicklund, 1974), these findings sug-

5 Cherulnik and Citrin, like most reactance theorists,use the term freedom rather than control, but these twoterms are freely substituted for one another in theirwritings.


gest that externals perceive their failure atchance tasks as a loss of control; that is, theyperceive chance as controllable. Further,self-report findings from the Karabenick andSrull study indicate that externals, but notinternals, perceived the chance task as mea-suring their "ability." This result, the au-thors noted, is consistent with Rotter andMulry's (1965) suggestion that externals re-gard luck as a personal characteristic andare concerned about being lucky. In a morerecent study, Karabenick and Addy (1979)found that in chance situations externalswere much more likely than internals to rateoutcomes as informative of their "disposi-tions to be lucky or unlucky" (p. 227).

Interactions between locus of control andtype of task (chance vs. skill) have also beenfound with performance level (effort andquality) as the dependent variable. The gen-eral pattern in these studies is that externals,compared with internals, try harder and dobetter in tasks described as chance deter-mined. In two studies (Davis & Phares,1967; Rotter & Mulry, 1965), the superiorperformance of externals on chance tasksfailed to reach significance; in three others(Lefcourt, Lewis, & Silverman, 1968; Petzel& Gynther, 1970; Watson & Baumal, 1967)the difference was significant (but seeDuCette & Wolk, 1973, for conflicting re-sults). In a related vein, Pittman and Pitt-man (1979) reported that externals per-formed better than internals under conditionsof high helplessness (60 problem-solvingtrials with noncontingent feedback) but per-formed worse under conditions of low help-lessness (20 trials with noncontingent feed-back) and no helplessness (see Gregory,Chartier, & Wright, 1979, for similar find-ings). Because the high-helplessness condi-tion in this study is intended to accentuatenoncontingency, subjects probably perceivedit as one in which outcomes were chancedetermined. If so, these findings are consis-tent with those described above: Externalsperform better in situations perceived to bechance determined.

The preceding findings, we maintain, aredue to externals' greater preference for andfeelings of competence in chance situations.An alternative explanation, which meritsbrief comment, is that externals might be

particularly likely to perceive contingencyand thus to feel confident in chance situa-tions. Evidence contrary to this view comesfrom Benassi, Sweeney, and Drevno's (1979)finding that internals, compared with exter-nals, give higher estimates of their successand perceive more control when presentedwith a chance-determined task that is intro-duced as being skill determined (a psycho-kinesis' task in which subjects are directedto concentrate so as to affect the outcome).Thus, it appears unlikely that externals over-estimate the role of skill in chance-deter-mined situations. Rather, it is somethingabout the perception of chance per se thatmotivates their performance.

Finally, Maracek and Mettee (1972) rep-licated the findings, discussed earlier, thatpersons who expect future failure resistseeing themselves as successful. If successis an indication of ability, Maracek andMettee reasoned, then the individual ac-cepting success is susceptible to elevated ex-pectations and subsequent disappointment.On the basis of this reasoning, they hypoth-esized that subjects would not discount suc-cess if they could attribute it to luck. Luck,they argued, is an unstable and uncontrol-lable characteristic that will not affect sub-sequent expectations. Although their find-ings supported this hypothesis, there is analternative interpretation. We realize thatmany subjects may see luck as unstable anduncontrollable; but some (externals and per-sons who have experienced acute failure)may be motivated to see it as at least some-what stable and controllable (cf. Weisz &Stipek, in press). In the Karabenick andSrull study cited earlier, externals werefound to consider themselves "luckier peo-ple" (1978, p. 85). These individuals, we sus-pect, perceive primary control as unlikely orimpossible and so adopt stable chance attri-butions (e.g., "I am a lucky person") thatpermit secondary control.

Summary

We have argued in this section that cer-tain individuals neither seek out skill situa-tions nor strive to manipulate the environ-ment; instead, they attempt to align with andrely upon the powerful forces of chance. The


evidence reviewed here indicates that theseindividuals tend to manifest a variety of in-ward behaviors, especially passivity andwithdrawal in skill situations, they tend toseek out chance situations, and they tend tobe more reactant when adverse events occurbecause of chance (their perceived ally) thanwhen adverse events occur because of theirown actions. Taken together, this evidencesuggests that objectively uncontrollable sit-uations, for these individuals, facilitate theperception of and desire for secondary con-trol. Further, this illusory control is revealedin attributions to chance, attributions thatsatisfy a criterion of proof of the two-processmodel outlined earlier.

Secondary control, as noted repeatedly inthis section, is closely related to external lo-cus of control. The differences that do existare, in part, historical ones. Originally, ex-ternal locus of control was assumed to entaila perception of uncontrollability. However,numerous research findings have discon-firmed the perceived uncontrollability as-sumption, and many locus of control theo-rists now espouse a position close to the oneadvocated here. According to the more re-cent congruence hypothesis, people reserveenergy for activities that match the form ofcontrol they feel best able to exercise (withexternals, for example, focusing on chanceactivities). Cherulnik and Citrin (1974) putit well: "Externals do not feel powerless, butsimply pursue rewards in different avenues"(p. 404). Despite the similarity between thishypothesis and ours, the fact remains thatexternal control was initially defined and insome quarters continues to be thought of asan absence of perceived control. For this rea-son we prefer the primary/secondary dis-tinction. Indeed, we believe that the confu-sion between the different definitions ofexternality mirrors the confusion betweenperceived uncontrollability and secondarycontrol.

attributions and behavior indicate helpless-ness in the sense that individuals do not per-ceive themselves as able to alter the envi-ronment to fit their desires, but individualsare not helpless in that they derive a per-ception of secondary, vicarious control fromassociation with powerful others. This as-sociation is neither a means to an end nora method of fulfilling other objectives; rather,the association is desired for its own sake.By aligning themselves with more powerfulothers, individuals can share in their victo-ries and in their accomplishments—in short,in their control. In a sense they can ride thecoattails of these more powerful others. Butin order to share in this control, the individ-uals must become aligned with the morepowerful other or become a part of the morepowerful group. In other words, they mustsubmit. That such enhanced control can re-sult vicariously from the association withmore powerful others is evidenced in sucheveryday situations as the child who proudlyasserts "my daddy is stronger than yourdaddy," the adolescent cult member whoseperception of control rests squarely in thehands of his or her demagogue leader, theultratraditional housewife who derives asense of power from her union with her hus-band, and the regional, chauvinistic followerof sports heroes. (For example, millions ofAmericans felt a rush of pride when the U.S.Olympic hockey team won the 1980 goldmedal). Most significantly, vicarious controlis evident in religious phenomena, as, forexample, when persons speak of gaining"strength through the Lord."

The notion of vicarious control is relatedto two existing bodies of theory and research.

"One involves the process of identification.The other involves Fromm's theorizing re-garding individuation and also the social-psychological research on self-awareness anddeindividuation. Both literatures are re-viewed in this section.

Vicarious Control

Our explanation of attributions to pow-erful others and submissive behavior is atonce similar to and different from the ex-planation provided by uncontrollability the-orists. According to the proposed model, the

Identification

The construct of identification is similarin several respects to the present notion ofvicarious control. Kelman (1958) describedidentification as having the following prop-erties: (a) It reflects desire for a self-defining


relationship to another person or a groupthat is intrinsically desired (rather thanbeing desired because it leads to specific re-wards or approval or to the avoidance of spe-cific punishments or disapproval), (b) It isnot attributable to the value of the behavioror attitude identified with, (c) It leads tosubmissiveness, even in the absence of sur-veillance.

Another description of identification thatclarifies its relevance to the construct of vi-carious control is provided by Goldschmidt(1974):In every society, each individual identifies himself withone or more groups of individuals.and finds not merelymaterial advantage through such identification, but,more significantly, finds ego gratification (or loss) inresponse to the fortunes of such groups, (p. IS)

The notion of identification has its originsin psychoanalytic theory. According to Freud(1923/1961), the individual takes on char-acteristics of others in order to share in theirexperiences. The psychoanalytic model is sooften used to explain character formationand sex-role development that this under-lying function—the ability to share in thedesirable experiences of others—is some-times neglected. It is perhaps most evidentin Freud's own description of the Oedipalconflict: The son's "identification with hisfather . . . permits the affectionate rela-tionship to the mother to be in a measureretained" (p. 32).

In Freud's view and in related formula-tions (e.g., Bettelheim's, 1943, work on"identification with the aggressor"; Whit-ing's, 1959, 1960, work on "status envy"),the importance ascribed to control is unclear.Bandura (1969), however, reinterpreted theseformulations in a manner consistent with ourposition. He maintained that the incentivesdesired in each case are control and power(see also Maccoby, 1959; Mussen & Distler,1959; Parsons, 1955). In support of hisclaim, Bandura cited evidence indicatingthat people are more likely to identify withothers who possess characteristics suggestingcontrol—competence, expertise, power, anddominance (for a review of these studies, seeBandura, 1969). Another noteworthy fea-ture of Bandura's model is its mention ofidentificatory behavior other than imitation.He observed that powerful models are able

to "extract subservience from others" (p.232)—a clear reference to submissive be-havior, which is of primary concern here.

Direct support for Bandura's "vicariouscontrol" hypothesis was provided in a nowclassic study of the determinants of imita-tion. Bandura, Ross, and Ross (1963) ex-amined situations, designed to simulate thoseoccurring in the nuclear family, in whichchildren could choose to identify with anadult who controlled rewards or an adultwho was the envied competitor for the re-wards. Because children showed greater im-itation of the former adult, Bandura et al.concluded that there is greater support forthe social power explanation than for thestatus envy explanation favored by Whiting.That is, it is the control of rewards, ratherthan their consumption, that is most desired.

While Bandura's addition of the socialpower notion to the construct of identifica-tion serves to link it more closely to the pres-ent construct of vicarious control, his for-mulation differs from ours in one criticalrespect. Although rarely explicitly, Banduraat times suggested that individuals who en-gage in identification do so for instrumentalpurposes—that is, as a means of primarycontrol (see especially Bandura, 1977). (Thisis consistent with his preference for the termimitation over the term identification, whichhe felt has "excess meaning.") Even so, Ban-dura did, to some extent, depart from theinstrumental conceptualization of identifi-cation. This is /evident in the research hecited to support his claims. For example, hecited Hetherington and Frankie's (1967)finding that children are most likely to im-itate the picture preferences of their domi-nant parent (i.e., the parent rated as morepowerful by independent observers) evenwhen it is made clear that their parents willremain unaware of their choices. Such be-havior hardly appears instrumental. Rather,the individual seems to be striving for as-sociation with the more powerful other andfor a form of control that can only be sec-ondary.

iIndividuation

Fromm's position. One of the most in-fluential psycho-philosophical books of the


20th century is Erich Fromm's (1941) Es-cape From Freedom.6 In it, Fromm exploredthe process of individuation, taking as theprototypical case the separation of infantfrom caretaker. The key dynamics in thisprocess, according to Fromm, are an in-creased sense of freedom (similar to the pres-ent notion of increased primary control) andan increased sense of loneliness and a di-minished sense of power (similar to the pres-ent notion of diminished secondary control).Although the separated individual is moreindependent, at the same time the psycho-logical benefits derived from association withmore powerful others are lost. Frommclaimed that persons flee from the individ-uation process in order to regain a sense ofpower. He saw evidence of flight from in-dividuation in neurotics' submissiveness andconformity and in historical phenomenasuch as the Reformation. In the latter,Fromm observed an "emphasis on thewickedness of human nature, the insignif-icance and powerlessness of the individualand the necessity of the individual to sub-ordinate himself to a power outside of him-self (p. 38). In capitalism and fascism, too,Fromm saw the individual as attempting toescape from powerlessness through sacrific-ing the individual self, "either by submissionto new forms of authority or by a compulsiveconforming to accepted patterns" (p. 135).The ever-present "shoulds" and "oughts" ofsubmissive individuals reflect their attemptto subordinate themselves. Motivating thissubmission is "the tendency to give up theindependence of one's individual self and tofuse one's self with somebody or somethingoutside of oneself in order to acquire thestrength which the individual self is lacking"(p. HI).

Fromm viewed masochism as an extremeform of the desire to lose the self. The aimof masochism is symbiosis, "the union of oneindividual self with another self (or any otherpower outside of the own self)" (p. 158).Persons desire "to feel inferior and insignif-icant" out of all proportion to their actualshortcomings. They "belittle themselves,"engaging in "self accusation and self criti-cism" in their aim to make themselves weakand thereby deserving of the association theyso ardently seek: "The goal of masochismis not suffering. . . . [It is] to help the in-

dividual escape his unbearable feeling ofaloneness and powerlessness" (p. 151). Thepreponderance of self-blame among de-pressed individuals, who also perceive ex-tremely little control over outcomes (Abram-son & Sackeim, 1977; Beck, 1967), suggeststhat they too may be attempting to alignthemselves with more powerful others.7

Fromm summarized his thesis in this way:

The annihilation of the individual self and the attemptto overcome thereby the unbearable feeling of power-lessness are only one side of the masochistic strivings.The other side is the attempt to become a part of abigger and more powerful whole outside of oneself, tosubmerge and participate in it. This power can be aperson, an institution, God, the nation, conscience, ora psychic compulsion. By becoming part of a powerwhich is felt as unshakably strong, eternal and glamo-rous, one participates in its strength and glory . . . onegains a new security and a new pride in the participationin the power in which one submerges, (p. 155)

Finally, it should be noted that Frommdid not view alignment with others as aneffective long-term means of. increasing per-ceived power: "The escape into symbiosiscan alleviate the suffering for a time but itdoes not eliminate it" (p. 238). Rather, dein-dividuation leads to a self-perpetuating cycleof greater dependence on others, increasedfear of powerlessness, increased conformity,

6 Fromm's concept of freedom corresponds closely towhat we refer to here as primary control. It is a notionof a lack of external constraints and an opportunity tofunction independently. The reason that people escapeit is that they lack the wherewithal to succeed on theirown. Averill (1973) contrasts this with another type offreedom: "agreement of the individual with whateverconstraints do exist" (p. 299). This type of freedom cor-responds to what we refer to as secondary control.

7 Steinbeck (1963) created a character, Mrs. Trask,who beautifully portrays these dynamics. She is de-scribed as "an inside-herself woman" who

used religion as a therapy for the ills of the world. . . Her . . . god was an expert in punishment. Hedemanded of her a sacrifice. She searched her mindfor some proper egotistical humility and almost hap-pily arrived at the sacrifice—herself. It took her twoweeks to write her last letter with revisions and cor-rected spelling. In it she confessed to crimes she couldnot possibly have committed and admitted to faultsfar beyond her capacity. And then, dressed in a se-cretly made shroud, she went on a moonlight nightand drowned herself in a pond so shallow that she hadto get down on her knees in the mud and hold herhead under water. This required great will power,(p. 15)


even greater dependence, and so on (pp.254-255). This cycle is related to our claimthat individuals think they must maintaintheir submissiveness in order to maintainvicarious control. It helps explain the per-sistence of this form of inward behavior.

Self-awareness and deindividuation.Fromm's case for vicarious control is accom-panied by a kaleidescope of historical, po-litical, economic, religious, sociological, andpsychological evidence. However, researchis needed to substantiate his claims moreadequately. The clearest empirical evidencethat submissiveness results from a desire tofit in with others can be found in the liter-ature on self-awareness and deindividuation.

Manipulations of self-awareness, such asplacing subjects in rooms with mirrors orcameras, exposing them to tape recordingsof their voices, and having them write au-tobiographical sketches, have been found tolead t


uncontrollability. The submissive behaviorthat was noted is not instrumental, and it islikely to be accompanied by the perceptionof others as more powerful than the self.Moreover, self-awareness, which typicallyfocuses attention on negative discrepanciesbetween standards and behavior, has beenlikened to a manipulation of uncontrollabil-ity (Wicklund, 1975). Although there is ev-idence that a type of primary uncontrolla-bility is operative here, we agree with Frommand with self-awareness and deindividuationtheorists that other dynamics are often moreinfluential. In particular, we point to theself s strong desire to submit to others' ex-

. pectations (as opposed to the lack of desireto assert the self) and to increased enjoymentfollowing deindividuation, which can be at-tributed, at least in part, to an increase insecondary control. However, we acknowl-edge that the experimental evidence of vi-carious, secondary control is mostly indirect.To strengthen our claims, more direct evi-dence, including self-reports of increasedperceptions of secondary control followingdeindividuation, is needed.

Further support for our claims may befound in the literature on support systems.Evidence indicating that the family andgroups of a religious, fraternal, ideological,social, or political nature offer their mem-bers "strength," "buttressing," and "in-creased self confidence" (cf. Silver & Wort-man, 1980) suggests the existence of thekind of vicarious, secondary control withwhich we are concerned here. However, thegroups may also provide increased primarycontrol by making the individual more con-fident about himself as an independent agent,so it cannot be assumed that secondary con-trol will prove to be the key-dynamic.

Interpretive Control

Interpretive control refers to the searchfor meaning and understanding. Because itrelates to each of the other kinds of controlalready considered, an analysis of interpre-tive control serves as a summary for the, ar-ticle. All of the attributions treated in theprevious sections are treated again here.These attributions, we argue, help individ-uals find meaning in events and thus accept

them. The claim that interpretation consti-tutes a type of control for which people ac-tively strive has been eloquently made byFrankl (1963), who referred to the will tomeaning, the achievement of meaning, thesense of courage derived from meaning, andthe freedom to choose to accept or reject aparticular meaning. Frankl's writings makeit clear that the search for meaning is par-ticularly intense under conditions in whichthere exists minimal primary control.

Janoff-Bulman and Brickman (1980) pro-vided an excellent example of what is re-ferred to here as interpretive control: ManyEuropean Jews in the early 1930s who hadthe opportunity to escape from Europe in-stead "clung" to the interpretations offeredby the Nazis. What looked like "giving up"and "helplessness," Janoff-Bulman andBrickman argued, was actually "persistenceat accommodation" (p. 24). People workhard at interpreting events so that they canaccept them; in so doing, they may appearto be giving up, but their persistence suggestsotherwise. Because considerable energy isdevoted to interpretation and because a senseof mastery is associated with the understand-ing of and ability to accept aversive events,we maintain that interpretation constitutesa secondary type of perceived control (cf.Averill, 1973; Frankl, 1963; Garber, Miller,& Abramson, 1980).

Several laboratory studies in the area ofattribution theory have indicated that per-sons who perceive a lack of primary controlseek secondary control in the form of reasonsand explanations. When exposed to failure,"helpless" children (those who attribute fail-ure to severely limited ability) have beenshown to engage in a set of interpretationsvery different than those of "mastery-ori-ented" children (those who attribute failureto motivational factors such as effort). Ac-cording to Diener and Dweck (1978), thesehelpless children tend to look to the causesof their failure, typically doubt their ownability, and are likely to concern themselveswith "unmodifiable factors" such as chance.Mastery-oriented children, in contrast, aremuch more concerned with monitoring theirbehavibr via solution-directed verbalizationsand self-instructions. Their interpretationsconcern the possible solutions, not the causes


or reasons, for the problem. Looking tocauses (which entails the perplexing phe-nomenon of self-castigation) has the subtle,secondary benefit of providing the individualwith a reason and thus an acceptable mean-ing for his or her condition. Looking to so-lutions has the obvious, primary benefit ofincreasing the probability of task success.The distinction between interpretations thatexplain the event and thereby enable the in-dividual to accept it and those that are so-lution oriented neatly parallels the distinc-tion between secondary control and primarycontrol.

The Diener and Dweck results also helpto clarify other findings in the attributionliterature. For example, they may explainwhy Alloy and Abramson (1979) found that,following failure, subjects in problem-solv-ing situations engaged in fewer complex, so-lution-oriented hypotheses. The Diener andDweck results suggest that subjects in theAlloy and Abramson study generated fewcomplex solution-oriented hypotheses be-cause they were busy generating many cause-oriented hypotheses (speculations regardingfactors contributing to failure). This sug-gestion closely resembles the conclusionreached by Coyne et al. (1980), whose re-sults indicate that "helpless" subjects "be-come preoccupied with the explanation oftheir inability" (p. 352). Coyne et al. citednumerous other studies in the literature onfailure, repression, anxiety, stress, and frus-tration, which they believe are consistentwith this claim. Perhaps most relevant toDiener and Dweck's findings and to the pres-ent hypothesis are Pittman and Pittman's(1980) results showing increased "attribu-tional activity" following failure manipula-tions. In Pittman and Pittman's study, sub-jects who had experienced failure and whowere subsequently administered an attribu-tion task unrelated to the failure experiencewere more likely to probe for causes under-lying the behavior of persons described inthe attribution task than were persons notexposed to the failure manipulation. Thisgeneralization of the desire for interpretivecontrol is a phenomenon warranting furtherinvestigation.

Whereas most of the evidence reviewedthus far in support of secondary control

comes from experimental laboratory studies,,much of the evidence for interpretive controlcomes from naturalistic field studies (Silver& Wortman, 1980). Bulman and Wortman(1977), for example, showed that paralyzedaccident victims exert considerable effort tomake sense of what has happened to them.All 29 persons interviewed mentioned thatthey had asked the question, "Why me?"and all but one had developed explanationsfor their accidents. The six types of expla-nations they identified—predetermination(fate), probability, chance, God had a rea-son, deservedness, and reevaluation—par-allel in interesting ways the attributionstreated in this article. Similarly, Burgess andHolmstrom (1979) found that many rapevictims had dealt with the "Why me?" ques-tion. They concluded that "coming up withan explanation gives some understanding tothe bizarreness of the act and aids in re-turning some degree of control to the victim"(p. 1270). In contrast to Burgess and Holms-trom's focus on control, Bulman and Wort-man concluded that their data indicate aneed for an orderly and meaningful world(p. 362). These two positions, however, canbe seen as compatible within the present for-mulation: The striving for secondary con-trol—understanding and thereby acceptingthe event—sometimes overshadows the striv-ing for primary control—maintaining thatthe event is or was changeable. That is, bothpositions may be emphasizing the phenom-enon of secondary, interpretive control.

One of the most common explanationsfound in the previously cited study on pa-ralysis victims was to reevaluate the victim-ization in a more positive light (Bulman &Wortman, 1977). These reevaluations wereregarded as explanations because they al-ways incorporated a theory as to why theaccident had occurred. For example, a fo-cusing on positive outcomes, such as slowingdown, learning about other people and life,and choosing their faith, might be accom-panied by the thought "The accident mayhave been for the best." This search for rea-sons and a purpose following tragic eventsis also evident in Burgess and Holmstrom's(1979) finding that victims of rape con-sciously minimize their anxiety. One com-mon strategy is to compare their own rape


with those of others who were killed, sub-jected to perversions, or suffered "greater vi-olence. Through this "cognitive strategy,"Burgess and Holmstrom argued, rape vic-tims are able to render the event more tol-erable and acceptable. Similarly, Langer,Janis, and Wolfer (1975) found that whenpatients were coached to interpret their hos-pitalization experience as an opportunity forsuch positive outcomes as rest, diet, and re-flection, they rated themselves and wererated by hospital staff as manifesting lessstress than did a control group. Like us, theseauthors emphasize the increased sense ofcontrol gained by the coached patients viatheir ability to reinterpret events.

A related aspect of interpretive control issuggested by evidence that people

Documents

Changing the World and Changing the Self: A Two-Process ......Weiner, 1974) but is consistent with the position of other locus of control theorists who see limited ability attri-butions