S1 Bandura 1978 Self System In Reciprocal Determinism

havior is an interacting determinant, not simplyan outcome of a "person-situation interaction."The methodology used to evaluate the precedingconceptualizations relies heavily on factorial designs in which responses of different individualsare measured under varying situational conditions.The data are then analyzed to determine howmuch of the variation in behavior is due to personal characteristics, how much to situational conditions, and how much to their joint effects. Theattention of researchers working within this framework centers mainly on the dispute over which ofthe components—persons, situations, or Person XSituation—accounts most for variation in behavior.However, the basic weakness in the conceptualscheme (i.e., treating behavior as a dependentrather than as an interdependent factor) goeslargely unnoticed.In the social learning view of interaction, whichis analyzed as a process of reciprocal determinism(Bandura, 1977b), behavior, internal personal factors, and environmental influences all operate asinterlocking determinants of each other. As shownin Figure 1, the process involves a triadic reciprocal interaction rather than a dyadic conjoint ora dyadic bidirectional one. We have already notedthat behavior and environmental conditions function as reciprocally interacting determinants. Internal personal factors (e.g., conceptions, beliefs,self-perceptions) and behavior also operate as reciprocal determinants of each other. For example,people's efficacy and outcome expectations influence how they behave, and the environmental effects created by their actions in turn alter theirexpectations. People activate different, environmental reactions, apart from their behavior, bytheir physical characteristics (e.g., size, physiognomy, race, sex, attractiveness) and socially conferred attributes, roles, and status. The differential social treatment affects recipients' self-conceptions and actions in ways that either maintain oralter the environmental biases.The relative influence exerted by these threesets of interlocking factors will vary in differentindividuals and under different circumstances. Insome cases, environmental conditions exercise suchpowerful constraints on behavior that they emergeas the overriding determinants. If, for example,people are dropped in deep water they will allpromptly engage in swimming activities, howeveruniquely varied they might be in their cognitiveand behavioral repertoires. There are times whenbehavior is the central factor in the interlocking346 APRIL 1978 AMERICAN PSYCHOLOGISTsystem. One example of this is persons who playfamiliar piano selections for themselves that createa pleasing sensory environment. The behavior isself-regulated over a long period by the sensoryeffects it produces, whereas cognitive activitiesand contextual environmental events are not muchinvolved in the process.In other instances, cognitive factors serve asthe predominant influence in the regulatory system. The activation and maintenance of defensivebehavior is a good case in point. False beliefsactivate avoidance responses that keep individualsout of touch with prevailing environmental conditions, thus creating a strong reciprocal interaction between beliefs and action that is protectedfrom corrective environmental influence. In extreme cases, behavior is so powerfully controlledby bizarre internal contingencies that neither thebeliefs nor the accompanying actions are muchaffected even by extremely punishing environmental consequences (Bateson, 1961).In still other instances, the development andactivation of the three interlocking factors are allhighly interdependent. Television-viewing behavior provides an everyday example. Personal preferences influence when and which programs, fromamong the available alternatives, individuals chooseto watch on television. Although the potentialtelevised environment is identical for all viewers,the actual televised environment that impinges ongiven individuals depends on what they select towatch. Through their viewing behavior, theypartly shape the nature of the future televisedenvironment. Because production costs and commercial requirements also determine what peopleare shown, the options provided in the televised environment partly shape the viewers' preferences.Here, all three factors—viewer preferences, viewing behavior, and televised offerings—reciprocallyaffect each other.The methodology for elucidating psychologicalprocesses requires analysis of sequential interactions between the triadic, interdependent factorswithin the interlocking system. Investigations ofreciprocal processes have thus far rarely, if ever,examined more than two of the interacting factorssimultaneously. Some studies analyze how cognitions and behavior affect each other in a reciprocal fashion (Bandura, 1977a; Bandura & Adams,1977). More often, however, the sequential analysis centers on how social behavior and environment determine each other. In these studies ofdyadic exchanges, behavior creates certain condi-

tions and is, in turn, altered by the very conditionsit creates (Bandura, Lipsher, & Miller, 1960; Patterson, 1975; Raush, Barry, Hertel, & Swain,1974; Thomas & Martin, 1976).From the perspective of reciprocal determinism,the common practice of searching for the ultimateenvironmental cause of behavior is an idle exercise because, in an interactional process, one andthe same event can be a stimulus, a response, oran environmental reinforcer, depending on wherein the sequence the analysis arbitrarily begins.Figure 2, which represents a sequence of reactionsof two persons (A and B), shows how the sameevents change their status from stimuli, to responses, to environmental reinforcers, at differententry points in the flow of the two-way interaction. For example, event A2 is an environmentalstimulus in the third point of entry, a response inthe second analysis, and an environmental reinforcer in the first analysis. One cannot speak of"behavior" and its "controlling environmental conditions" as though these two factors were fundamentally different events.The preceding analysis focused only on the dependencies among acts, and how they change fromresponses to environmental events in the flow ofinteraction. However, regulatory processes are notgoverned solely by the reciprocal influence of antecedent and consequent acts. While behaving,people are also cognitively appraising the progression of events. Their thoughts about the probableeffects of prospective actions partly determine howacts are affected by their immediate environmentalconsequences. Consider, for example, investigations of reciprocal coercive behavior in an ongoingdyadic interaction. In discordant families, coercive behavior by one member tends to elicit coercive counteractions from recipients in a mutualescalation of aggression (Patterson, 1975). However, coercion often does not produce coercivecounteractions. To increase the predictive powerof a theory of behavior, it is necessary to broadenthe analysis to include cognitive factors that operate in the interlocking system. Counterresponsesto antecedent acts are influenced not only by theirimmediate effects but also by judgments of laterconsequences for a given course of action. Thus,aggressive children will continue, or even escalate,coercive behavior in the face of immediate punishment when they expect persistence eventually togain them what they seek. But the same momentary punishment will serve as an inhibitor ratherthan as an enhancer of coercion when they expectAA2ls'— RstB2A3 sreinf. Rst sreinf . R sreinf.Figure 2. Illustration of how the same behavioral event can be an antecedent stimulus, aresponse, or a reinforcing consequent depending onwhere one arbitrarily begins the analysis in theflow of a social interaction. The A's are successiveresponses by one person, and the B's are successiveresponses by the second person in the dyadic interaction. S* represents stimulus; R representsresponse; and S1'''1"* represents reinforcer.continuance of the aversive conduct to be ineffective.The predictive value of momentary reciprocalconsequences derives partly from people's expectations of how their actions are likely to changefuture consequences over the course of sequentialinterchanges. Findings from several lines of research document how cognitive factors alter thefunctional relationship between actions and outcomes. The degree to which behavior is influencedby its momentary effects depends on people's beliefs about action-outcome contingencies (Baron,Kaufman, & Stauber, 1969; Estes, 1972; Kaufman, Baron, & Kopp, 1966; Spielberger & DeNike,1966), the meaning they attribute to the outcomes (Dulany, 1968), and their expectations thatpersistence in a given course of behavior will eventually alter people's reinforcement practices (Bandura & Barab, 1971).In the above studies, cognitive influences serveas controlling rather than controllable factors.But cognitions do not arise in a vacuum, nor dothey function as autonomous determinants of behavior. In the social learning analysis of cognitive development, conceptions about oneself andthe nature of the environment are developed andverified through four different processes (Bandura,1977b). People derive much of their knowledgefrom direct experience of the effects produced bytheir actions. Indeed, most theories of cognitivedevelopment, whether they favor behavioristic, information-processing, or Piagetian orientations,focus almost exclusively on cognitive changethrough feedback from direct experimentation.However, results of one's own actions are not thesole source of knowledge. Information about theAMERICAN PSYCHOLOGIST APKIL 1978 347

nature of things is frequently extracted from vicarious experience. In this mode of verification,observation of the effects produced by somebodyelse's actions serves as the source and authentication of thoughts.There are many things we cannot come to knowby direct or vicarious experience because of limited accessibility or because the matters involvemetaphysical ideas that are not subject to objectiveconfirmation. When experiential verification iseither difficult or impossible, people develop andevaluate their conceptions of things in terms ofthe judgments voiced by others. In addition toenactive, vicarious, and social sources of thoughtverification, all of which rely on external influences,logical verification also enters into the process,especially in later phases of development. Afterpeople acquire some rules of inference, they canevaluate the soundness of their reasoning and derive from what they already know new knowledgeabout things that extend beyond their experiences.External influences play a role not only in thedevelopment of cognitions but in their activationas well. Different sights, smells, and sounds willelicit quite different trains of thought. Thus,while it is true that conceptions govern behavior,the conceptions themselves are partly fashionedfrom direct or mediated transactions with the environment. A complete analysis of reciprocal determinism therefore requires investigation of howall three sets of factors—cognitive, behavioral, andenvironmental—interact reciprocally among themselves. Contrary to common misconception, sociallearning theory does not disregard personal determinants of behavior. Within this perspective,such determinants are treated as integral, dynamicfactors in causal processes rather than as statictrait dimensions.Self-RegulatorySelf SystemFunctions of theThe differences between unidirectional and reciprocal analyses of behavior have been drawn mostsharply in the area of self-regulatory phenomena.Exponents of radical behaviorism have always disavowed any construct of self for fear that it wouldusher in psychic agents and divert attention fromphysical to experiential reality. While this approach encompasses a large set of environmentalfactors, it assumes that self-generated influenceseither do not exist or, if they do, that they have348 APRIL 1978 AMERICAN PSYCHOLOGISTno effect upon behavior. Internal events aretreated simply as an intermediate link in a causalchain. Since environmental conditions presumablycreate the intermediate link, one can explain behavior in terms of external factors without recourse to any internal determinants. Through aconceptual bypass, cognitive determinants are thusexcised from the analysis of causal processes.In contrast to the latter view, internal determinants of behavior are gaining increasing attention in contemporary theorizing and research. Indeed, self-referent processes occupy a central position in social learning theory (Bandura, 1977b).As will be shown later, self-generated events cannot be relegated to a redundant explanatory link.In the triadic reciprocal system, they not onlyoperate as reciprocal determinants of behavior butthey play a role in the perception and formationof the environmental influences themselves.Self influences have traditionally been conceptualized in terms of the self-concept (Rogers,1959; Wylie, 1974). In these approaches, selfconceptions are measured by having people ratein one way or another evaluative statements thatthey consider apply to themselves. The principalthesis that self-conceptions determine psychological functioning is then tested by correlating selfconcepts or disparities between actual-ideal selveswith various indexes of adjustment, attitudes, andbehavior.One can point to several features of self theoriesof this type that detract from their explanatoryand predictive power. For the most part, they areconcerned with global self-images. A global viewof what people think of themselves cannot possibly account for the wide variations they typicallyshow in their self-reactions under different situational circumstances, on different activities, andat different times. A postulated internal determinant cannot be less complex than its effects. Another limitation of self theories is that they failto specify in sufficient detail how self-conceptsregulate specific actions.In social learning theory, a self system is not apsychic agent that controls behavior. Rather, itrefers to cognitive structures that provide reference mechanisms and to a set of subfunctions forthe perception, evaluation, and regulation of behavior. Before proceeding to a reciprocal analysisof self influences, the processes by which peopleexercise some control over their own behavior willbe reviewed briefly.

SELF-OBSERVATIONPERFORMANCE DIMENSIONSJUDGMENTAL PROCESSPERSONAL STANDARDSSELF-EVALUATIVE REACTIONSQUALITYMODELING SOURCESRATEREINFORCEMENT IVEREFERENTIAL PERFORMANCESSTANDARD NORMSTANGIBLE SELF-APPLIED CONSEQUENCESREWARDINGSOCIAL COMPARISONPUNISHINGPERSONAL COMPARISONCOLLECTIVE COMPARISONNO SELF-RESPONSEVALUATION OF ACTIVITYREGARDED HIGHLYNEUTRALDEVALUEDPERFORMANCE ATTRIBUTIONPERSONAL LOCUSEXTERNAL LOCUSFigure 3. Component processes in the self-regulation of behaviorby self-prescribed contingencies.COMPONENT PROCESSES IN SELF-REGULATIONFigure 3 summarizes the different component processes in the self-regulation of behavior throughself-prescribed contingencies. Behavior typicallyvaries on a number of dimensions, some of whichare listed in the self-observation component. Depending on value orientations and the functionalsignificance of given activities, people attend selectively to certain aspects of their behavior andignore variations on nonrelevant dimensions.Simply observing variations in one's performances yields some relevant information, but suchdata, in themselves, do not provide any basis forpersonal reactions. Behavior produces self-reactions through a judgmental function that includesseveral subsidiary processes. Whether a givenperformance will be regarded as commendable ordissatisfying depends upon the personal standardsagainst which it is evaluated. Actions that measure up to internal standards are appraised favorably; those that fall short are judged unsatisfactory.For most activities, there are no absolute measures of adequacy. The time in which a givendistance is run, the number of points obtained onan achievement test, or the size of charitable contributions often do not convey sufficient information for self-appraisal even when compared withan internal standard. When adequacy is dennedrelationally, performances are evaluated by comparing them with those of others. The referentialcomparisons may involve standard norms, the performances of particular individuals, or the accomplishments of reference groups.One's previous behavior is continuously used asthe reference against which ongoing performance isjudged. In this referential process, it is self-com-parison that supplies the measure of adequacy.Past attainments influence performance appraisalsmainly through their effects on standard setting.After a given level of performance is attained, itis no longer challenging, and new self-satisfactionsare often sought through progressive improvement.Another important factor in the judgmentalcomponent of self-regulation concerns the evaluation of the activities. People do not much carehow they perform on tasks that have little or nosignificance for them. And little effort is expendedon devalued activities. It is mainly in areas affecting one's welfare and self-esteem that favorableperformance appraisals activate personal consequences (Simon, Note 2).Self-reactions also vary depending on how peopleperceive the determinants of their behavior. Theytake pride in their accomplishments when theyascribe their successes to their own abilities andefforts. They do not derive much self-satisfaction,however, when they view their performances asheavily dependent on external factors. The sameis true for judgments of failure and blameworthyconduct. People respond self-critically to inadequate performances for which they hold themselvesresponsible but not to those which they perceiveare due to unusual circumstances or to insufficientcapabilities. Performance appraisals set the occasion for self-produced consequences. Favorablejudgments give rise to rewarding self-reactions,whereas unfavorable appraisals activate negativeself-reactions. Performances that are judged tohave no personal significance do not generate anyreactions one way or another.In the social learning view, self-regulated incentives alter performance mainly through theirmotivational function (Bandura, 1976). Contin-AMEKICAN PSYCHOLOGIST APRIL 1978 349

gent self-reward improves performance not because that a response had been executed (Rachlin,it strengthens preceding responses. When people 1974). The assumptions on which such redefinimake self-satisfaction or tangible gratifications tions are based and the contravening evidence areconditional upon certain accomplishments, they discussed at length elsewhere (Bandura, 1976) andmotivate themselves to expend the effort needed need not be repeated here.to attain the desired performances. Both the anThe second, and more commonly used, solutionticipated satisfactions of desired accomplishments is to execute a regress of causes. By locating aand the dissatisfactions with insufficient ones pro- remote environmental factor that might affect selfvide incentives for actions that increase the likeli- reactions, self-generated influences are therebyhood of performance attainments.converted into simple operants. As Stuart (1972)Much human behavior is regulated through self- succinctly put it, "The behaviors commonlyevaluative consequences in the form of self-satis- ascribed to self-control can be functionally anafaction, self-pride, self-dissatisfaction, and self- lyzed as a special subset of operant responsescriticism. The act of writing is a familiar example which are, in fact, under situational control" (p.of a behavior that is continuously self-regulated 130). The organism thus becomes simply a rethrough evaluative self-reactions. Writers adopt a pository of self-control responses waiting to bestandard of what constitutes an acceptable piece externally activated, but otherwise it possesses noof work. Ideas are generated and rephrased in capacity to generate guides and incentives for itsthought before they are committed to paper. Pro- own actions. But causal regression is a no morevisional contructions are successively revised until convincing disposal of self-generated influencesauthors are satisfied with what they have written. than is renaming, because for every environmentalThe more exacting the personal standards, the cause that is invoked, one can find a prior permore extensive are the corrective improvements.sonal cause of that environment.People also get themselves to do things theySome conceptual regressions of self-generated inwould otherwise put off by making tangible out- fluences to situational causes treat reciprocal incomes conditional upon completing a specified level fluences as rivalrous or even as confounding facof performance. In programs of self-directed tors. This view is exemplified by Jones, Nelson,change, individuals improve and maintain behavior and Kazdin (1977), who consider external influon their own over lon

Social learning theory (Bandura, 1974, 1977b) analyzes behavior in terms of reciprocal determin-ism. The term determinism is used here to signify the production of effects by events, rather than in the doctrinal sense that actions are completely determined by a prior sequence of c