personalit jou rn al social psychology

JOU RN Al_ OF

PersonalitySocial Psychology

Volume 37 August 1979 Number 8

A Cybernetic Model of Self-Attention Processes

Charles S. CarverUniversity of Miami

An analysis of the experiential and behavioral consequences of self-directedattention is proposed, based on cybernetic or information-processing ideas. Theproposed model includes the following assumptions: When attention is directedto environmental stimuli, those stimuli are analyzed and categorized accordingto the person's preexisting recognitory schemas. Self-directed attention oftenleads to a similar analysis of ^//-information; experientially, such a state ofself-focus comprises an enhanced awareness of one's salient self-aspects. Insome cases categorization—either of one's context or of some self-element—elicits a response schema, which constitutes a behavioral standard. If a priorcategorization has evoked such a behavioral standard, subsequent self-attentionengages an automatic sequence in which behavior is altered to conform moreclosely with the standard. This matching-to-standard is construed as the occur-rence within a psychological system of a test-operate-test-exit unit or a nega-tive feedback loop—a cybernetic construct applicable to many different phe-nomena both within and outside of psychology. More specifically, self-focuswhen a standard is salient is seen as the "test" or comparison phase of thatfeedback loop. If the matching-to-standard process is interrupted by one ofseveral specified classes of events, subsequent behavior depends on an outcomeexpectancy judgment. A favorable outcome expectancy causes a return to thematching-to-standard attempt; an unfavorable expectancy results in behavioralwithdrawal. Research support for the model is discussed, and comparisons aremade with self-awareness theory, Bandura's theory of fear-based behavior,helplessness theory, and social comparison theory.

Psychological theorizing in this century hasbeen characterized by the occasional emer-gence of quite general theories of behavior(such as psychoanalytic theory and learningtheory), with intervening periods in which nogeneral theory has seemed universally ap-plicable and problem-specific theories insteadhave predominated. In recent years yet an-other general approach to understanding be-havior has begun to coalesce, an approach

usually identified with the terms information-processing, cybernetic, or control theory (e.g.,Apter, 1970; Buckley, 1968; Powers, 1973;Singh, 1966; Wiener, 1948). In the minds ofmany theorists and researchers, information-processing ideas provide an ideal frameworkwithin which to confront a wide variety ofproblems. Indeed, it is a framework that, byvirtue of its abstractness, many believe mayprovide the best basis yet attainable for a

Copyright 1979 by the American Psychological Association, Inc. 0022-35U/79/3708-1251J00.75

1251

1252 CHARLES S. CARVER

comprehensive model of behavior. This articlerepresents an attempt to apply such ideas toan area of interest to social psychologists thathas not been treated before in these terms.That area is the experiential and behavioralconsequences of self-directed attention.

The article is divided into four major sec-tions. The first section is intended to providesome background to the theory that is pre-sented later. This background material in-cludes a brief summary of Duval and Wick-lund's (1972) theory of objective self-aware-ness and a basic introduction to the uses ofthe terms cybernetic, control, and informationprocessing. In the second section a theoreticalmodel is presented, along with some evidencein support of its propositions. This sectionrepresents the central portion of the article.Because of the unique relationship betweenthis theory and Duval and Wicklund's (1972)self-awareness theory, a third section is in-cluded, which addresses specific points of con-flict between the two models. The presenttheory also has some broader implications,however. These implications are consideredin the final major section. That section in-cludes comparisons between the present modeland each of the following: Bandura's recentanalysis of fear-based behavior (Bandura,1977), learned helplessness theory (e.g.,Abramson, Seligman, & Teasdale, 1978), andsocial comparison theory (Festinger, 19 SO,19S4).

Background

Self-Awareness Theory

In 1972 Duval and Wicklund proposed atheory of self-awareness that consisted of two

My most sincere gratitude is expressed to ClydeHendrick, Hazel Markus, Paul Neumann, JohnRosen, Michael Scheier, Neil Schneiderman, ShelleyTaylor, Robert Wicklund, and particularly PaulBlaney, all of whom generously commented on oneor more drafts of this article; to Mark Bickhard,who gave me my introduction to the concepts ofcybernetics; and to my wife, Nora de la Garza, whoremained unfailingly supportive throughout the longand arduous process of shaping the manuscript.

Requests for reprints should be addressed toCharles S. Carver, Department of Psychology, P.O.Box 248185, University of Miami, Coral Gables,Florida 33124.

central assumptions. First, they suggested thatthe objects of conscious attention could bedichotomized: Attention can be directed out-ward to the environment, or attention can bedirected inward to the self (self-awareness).1

Self-awareness is increased by stimuli thatremind a person of himself or herself (e.g., acamera or a mirror); self-awareness is de-creased by stimuli that distract attention fromthe self (e.g., perceptual-motor tasks, such asa pursuit rotor).

Duval and Wicklund's (1972) second as-sumption was that self-focus leads to a self-critical evaluation process such that the self-attentive person compares himself or herselfwith some standard on whatever behavioraldimension happens to be salient (see alsoWicklund, 197Sa).2 Those authors furtherargued that self-directed attention is there-fore usually aversive, because in most casesthe person's actual behavior or state is worse

1 Duval and Wicklund (1972) originally termedself-focus "objective" self-awareness and environ-ment focus "subjective" self-awareness. In recentyears this rather cumbersome convention has beenwidely abandoned in favor of such alternatives ashigh versus low self-awareness or self-focus versusenvironment focus.

2 It will be noted that many potential standardsof comparison exist at any given moment, evenwith respect to a single behavioral dimension. How-ever, in Duval and Wicklund's theory it was assumedthat at any given time the following two conditionsobtain: One standard is more salient than others,and the person adopts that standard as his or hercomparison value, at least temporarily. The sameassumption will be made throughout the presentdiscussion. Cases in which the individual's attentionis drawn to a potential standard that is not adoptedas a comparison value (because it is seen as irrel-evant to one's behavior or for some other reason)are explicitly excluded from the analysis. This issuewill be taken up in somewhat greater detail below.

Perhaps the distinction between standards anddimensions should also be clarified at this point. Thedistinction is similar to that commonly made be-tween the terms value and variable. A behavioraldimension (or a variable) such as aggressiveness orpoliteness is defined by some quality that may bepresent in one's behavior to a greater or lesserdegree. A standard is a point on that behavioraldimension, for example, nonaggressiveness or mod-erate aggressiveness. The existence of a behavioralstandard, of course, always presupposes the existenceof a behavioral dimension.

CYBERNETIC MODEL OF SELF-ATTENTION PROCESSES 1253

than the standard of comparison. This aver-siveness presumably leads the person to at-tempt to escape the self-aware state (Wick-lund, 197Sa, 197Sb). If it is not possible toavoid self-awareness-inducing stimuli (andthus the aversive state), the self-focused per-son may attempt to alter his or her behaviorso that it conforms more closely to the stan-dard, as a way of reducing the aversivenessof self-awareness. That is, the aversiveness ofself-attention was seen as the motivator forbehavioral alterations. Thus self-awarenesstheory, as proposed by Duval and Wicklund(1972), was one of a larger class of drivetheories—one in which the presence or ab-sence of drive was determined by the directionof one's focus of attention.

This theory has been provocative, and ithas led to a good deal of imaginative research.However, evidence has begun to accumulatethat its assumptions are incorrect. The re-mainder of this article will be devoted to theassertion that the elements of self-awarenesstheory that have been supported by researchfindings can be most usefully construed ascomponents of a more elaborate model of so-cial cognition, and further, that such a modelis most usefully framed in information-pro-cessing or cybernetic terms.

Information Processing, Cybernetics,and Control

Definitions. Some amplification on theuses of the terms information processing, cy-bernetic, and control is in order at the outset.The term information processing is used intwo related but distinguishable ways by psy-chologists. At a micro level, the term refers toanalyses of the exact bases by which informa-tion is encoded in and retrieved from memory.At the macro level, information processingapplies to descriptions of control systems andis more synonymous with the term cybernetic.Cybernetics is the science of control and com-munications systems (Apter, 1970; Singh,1966; Wiener, 1948, 1954). The term control,as used here, refers to the sequencing that isimplicit in a set of instructions, each of whichawaits and depends on the execution of the

EXIT

(Incongruity)

(Congruity)

OPERATE

Figure 1. The TOTE unit, as presented and discussedby Miller, Galanter, and Pribram (1960).

previous instruction (as in a computer pro-gram).

A specific illustration may clarify the con-cept of control. Miller, Galanter, and Pribram(1960) introduced to psychologists a hypo-thetical construct called a TOTE unit (Figure1); the word TOTE is an acronym for test-operate-test-exit. This construct is knownmore widely among cyberneticists and engi-neers as a negative feedback loop (negativebecause it reduces discrepancies between anexisting state and a standard).

The diagram of a TOTE unit (Figure 1)portrays a self-regulatory system. The arrowsin the diagram convey three things simul-taneously. They represent energy, that is, theactivation of a physical mechanism to do somesort of work; they represent informationtransfer, that is, the information that con-gruity between two values has or has not beenattained; and they represent control, the factthat one process logically depends on an out-come of the previous process and then leadsto a subsequent process. Any given process issaid to control another when the executionof the first triggers the execution of the sec-ond. In effect, any decision-making flow chartdescribes a control sequence.

In general, the present use of the terminformation processing is meant to conveythe latter (macro-level) meaning of the term.Though there are differences in implicationbetween the terms information-processing,cybernetic, and control theory, the differencesare relatively subtle, and the terms will beused interchangeably here.

Applications. Many people tend to thinkof the word cybernetic as applying solely toelectronic computers. However, cybernetics as


a science was seen from its infancy as beingapplicable to both living and nonliving sys-tems. To most cyberneticists, the living-non-living dichotomy is not terribly useful (cf.Turing, 1950; Wiener, 1948). Because of thefact that the TOTE unit or feedback loop hassome particularly important implications forthe discussion to follow, let us consider theliving-nonliving distinction as it applies to theTOTE unit (Figure 1).

In the control system of the TOTE unit,three mechanisms are activated in an orderlysequence. The first of these—test—is a mech-anism by which a comparison is made betweensome input and a standard. If the input is notcongruous with the standard, the secondmechanism—operate—is activated. Operatealters the existing state of affairs in some way.Following operate, test recurs. If the inputnow is congruous with the standard, the finalmechanism—exit—occurs, deactivating thesystem or freeing it for other applications. If,on the other hand, the second test reveals thatincongruity still exists, operate is reactivatedand continues to sequence alternately withtest until congruity between input and stan-dard is attained 3

The most commonly used illustration of theoperation of a TOTE unit is the behavior ofa nonliving system: the room thermostat. Thethermostat senses existing temperature, com-pares that existing temperature with a presetstandard, and activates a furnace when thetemperature falls discriminably below thestandard or activates an air conditioner whenthe temperature rises above the standard.However, this control structure is certainlynot limited to electrical systems. Indeed, inthe view of cyberneticists, the feedback loopmay occur in virtually any type of physicalsystem (cf. Buckley, 1968; Kuhn, 1974). Thesame control structure that is realized elec-trically in the thermostat is realized biolog-ically in innumerable forms: for example, thehomeostatic mechanisms that maintain normalbody temperature, maintain appropriate levelsof oxygen and nutrients in the blood and tis-sues, and regulate hormonal balances (cf.Wiener, 1948, especially p. 114). Moreover,the same control structure can be realized inpsychological systems. For example, it is im-

plicit in a child's self-correcting attempts toreach out to touch a nearby object, and in theadult's checking and adjusting his or herclothing in a mirror.

Until recently, cybernetic constructs havenot been widely applied to subjects of interestto social psychologists. The theorizing pre-sented below is intended, in part, as a stepin that direction.4

The Model

The control-theory model of self-attentionprocesses proposed here subsumes many find-ings generated by researchers in self-aware-ness theory. It does not, however, incorporateall of the same assumptions as does thattheory. The difference in assumptions be-tween the two allows the present model toaccommodate existing data with greater in-ternal consistency and to make predictionsthat seem not to be easily derivable from self-awareness theory as framed by Duval andWicklund (1972). The model consists of thefollowing propositions, which will be pre-sented along with some amplification and sup-porting evidence. (Figures 2 and 3 providea flow-chart illustration of the interrelation-ship among propositions.)

1. A dichotomy may be imposed on the po-tential objects of conscious attention, such

3 This alternation may be quite rapid, of course,depending on the characteristics of the physical sys-tem in which it occurs. (Indeed, some theorists, e.g.,Powers, 1973, hold that this information flow isbetter conceptualized as being continuous at allstages of the feedback loop.) Thus, although we caneasily separate the component elements logically, forother purposes it is more convenient to think of theentire feedback loop as the unit of analysis.

* Control theorists seem to have been more awareof the potential applications of such constructs tosocial behavior than have social psychologists. Forexample, MacKay (1963) has noted that "an artifactcapable of receiving and acting on information aboutthe state of its own body can begin to parallel manyof the modes of activity we associate with self-consciousness" (p. 227). An important exception tothis general rule is the work of Powers (1973, 1978),whose background includes training in both psychol-ogy and physics. Even Powers's writing, however, hasbeen concerned predominantly with psychologicalprocesses other than those involved in social be-havior.


that attention may be said to be directedeither outward toward the environment orinward toward the self.

This seemingly simple statement (which isidentical to the first assumption made byDuval and Wicklund, 1972) requires severalkinds of qualification and elaboration. First,it should be clear that the dichotomy betweenself-focus and environment focus is an imper-fect one (as in some respects is the distinctionbetween what is self and what is nonself).When one's attention is focused on the en-vironment, one is attending to immediate per-ceptual input from the distance receptors: theeyes and ears. Self-focus, in comparison, in-cludes a wider variety of possibilities. Whenattention is self-directed, it sometimes takesthe form of focus on internal perceptualevents, that is, information from those sen-sory receptors that react to changes in bodilyactivity (for example, the autonomic andproprioceptive activity that contributes to thesubjective experience of emotion). Self-focusmay also take the form of an enhanced aware-ness of one's present or past physical be-havior, that is, a heightened cognizance ofwhat one is doing or what one is like. Alterna-tively, self-attention can be an awareness ofthe more or less permanently encoded bits ofinformation that comprise, for example, one'sattitudes. It can even be an enhanced aware-ness of temporarily encoded .bits of informa-tion that have been gleaned from previousfocus on the environment; subjectively, thiswould be experienced as a recollection or im-pression of that previous event.

It is sometimes difficult to maintain theclarity of this dichotomy, as is illustrated bythe following example. To look at a part ofone's own body is, initially, to focus on theenvironment. The visual perception is a per-ception of a thing out in the world, until thatpercept is categorized as an aspect of self.Because the visual stimulus in question is apart of one's body, however, it is likely thatit will ultimately be categorized as a compo-nent of self. At that point the experience maybecome one of self-attention, as one mentallyconsiders that self-aspect. It is this mechanismby which experimental manipulations such asmirrors presumably act to heighten self-focus:

by inducing visual perceptions that remindsubjects of themselves in a very general sense,that is, remind them of self-dimensions otherthan their visual representations.

It should also be clear that the term height-ened self-awareness and its synonyms as theyare used in this article do not necessarilyconnote a long, deep examination of the self.Rather, such terms imply an increase in theprobability or the frequency with which atten-tion is momentarily directed to some aspectof the self. In research, such increases in self-focus are caused by stimuli that remind sub-jects of themselves. It is assumed, however,that attention normally shifts back and forthbetween self and environment to some de-gree, whether such reminders are present ornot.

2. When attention is directed to the en-vironment, incoming stimulus information isprocessed, leading to classification of that in-put. The process of classifying incoming stim-uli has been studied extensively by cognitivepsychologists. Posner (1969; see also Posner& Rogers, 1978), Franks and Bransford(1971), Reitman and Bower (1973), andNeumann (1974) are among the theoristswho have developed models to account for theclassificatory process in various ways. Thoughthese models vary in other respects, they haveas a unifying idea the notion that, over a suc-cession of perceptual experiences, a singlerepresentation in memory is abstracted from aseries of relatively similar percepts. Thissingle representation, then, stands for a classof perceptual events. These abstracted repre-sentations—sometimes termed recognitoryschemas or recognitory prototypes—are storedin memory and are subsequently used to clas-sify and interpret new inputs.

Evidence for the existence and use of recog-nitory prototypes in classifying stimuli in theenvironment comes from a vast accumulationof research in cognitive psychology. In cogni-tive research, the prototypes are typicallygenerated experimentally by presenting sub-jects with a series of discrete training stimulithat vary along several dimensions. Aftertraining, subjects are found to classify newstimuli according to the organization implicitin the training series. Presumably the catego-


FOCUS/S

ENVIRONMENT SELF

ANALYSISYIELDS NO CUEIMPLICATIONFOR BEHAVIOR

ANALYSISYIELDS CUEIMPLICATIONFOR BEHAVIOR

111

i N

r >̂ 4

NO BEHAVIORALSTANDARDSALIENT

ANALYSIS ANALYSISYIELDS CUE YIELDS NO CUEIMPLICATION IMPLICATIONFOR BEHAVIOR FOR BEHAVIOR

IfBEHAVIORAL !STANDARD '

HAS BEEN EVOKED ^BY PRIOR

CATEGORIZATION

(TO FIGURE 3)

Figure 2. Flow-chart description of the cybernetic model of the consequences of self-directedattention. (Numbers in the figure refer to comparable propositions in the text.)

rizational structures that we carry with us inour day-to-day lives similarly reflect our ex-periential histories, but histories that aremuch longer and more complex.5

In the restricted settings of cognitive re-search, the subject categorizes only one stim-ulus at a time. However, in the broader con-text of social behavior one doubtless catego-rizes the nature of the interaction setting as awhole, as well as categorizing single stimuli.The kinds of information contributing to thisoverall identification undoubtedly includegestural, postural, and other nonverbal cuesemitted by persons in the visual field; sig-nificant physical elements in the situation(e.g., presence of objects with identifiableuses); and verbally transmitted information,including inferences that the listener makesfrom the verbalization. Each cue that is en-coded presumably contributes something tothe resultant categorization of the context.

The work of a number of environmentalpsychologists (e.g., Barker, 1968; Frederick-sen, 1974; Gump, 1971) would seem to fit

very easily into this line of thought. Thoseresearchers have emphasized that behaviorsettings vary along many discriminable di-mensions, and they have worked toward thedevelopment of taxonomies of the situationsin which behaviors take place. Presumablytheir efforts in categorization are similar tothe processes that we all use continuously,albeit less consciously. In a similar vein, Can-tor and Mischel (1977) have recently foundevidence that personality-trait terms act verymuch as recognitory prototypes in the classi-fication of other persons.

The classification of environmental inputs,which is assumed in the present model to

B The specific processes by which such schemas areformed and altered, although obviously very impor-tant, are outside the scope of this article. Thus, theexperimental literature of that area is treated quiteperipherally here. Inasmuch as this article is intendedto deal primarily with consequences of self-directedattention, discussion will emphasize research withdirect relevance to the propositions that deal withself-focus.


(from Figure 2 }

SELF-FOCUS WITHI BEHAVIORAL STANDARD SALIENT I

STANDARD HASNEGATIVE VALENCE:BEGIN DISCREPANCYENLARGEMENT (TOTE)

SEQUENCE

STANDARD HASPOSITIVE VALENCE:BEGIN DISCREPANCY

REDUCTION (TOTE)SEQUENCE

(from this point,the possibilitiesparallel thoseIllustratedto the right)

BEHAVIOR:RETURN TO

MATCHING-TO-STANDARD

LOW EFFICACYPERCEPTIONS:

AFFECTASSOCIATED

WITHENVIRONMENT

HIGH EFFICACYPERCEPTIONS:

AFFECTASSOCIATED

WITHSELF

LOW EFFICACYPERCEPTIONS:

AFFECTASSOCIATED

WITHSELF

HIGH EFFICACYPERCEPTIONS:

AFFECTASSOCIATED

WITHENVIRONMENT

Figure 3. Flow-chart description of the cybernetic model of the consequences of self-directedattention. (Numbers in the figure refer to comparable propositions in the text.)

occur whenever attention is directed outwardto the environment, is presumed to be aninteractive process. That is, it involves a com-parison of the incoming information withpreexisting recognitory schemas." Indeed,many believe that the resulting percept is asmuch a function of the previously encodedinformation as it is a function of the inputinformation itself (cf. Neisser, 1976). Thefact that this categorization involves such aninteraction between input and schema sug-gests another qualification that must be maderegarding the proposed dichotomy betweenself-foms and environment focus. Specifically,if processing and classification of even a verysimple sensory input occurs by reference topreexisting recognitory prototypes, it wouldappear to be an oversimplification to say thatattention is ever directed "at" the environ-

ment alone. The prototypes are, in a verybasic sense, an aspect of oneself. The process-ing of environmental information thus seemsto involve the simultaneous utilization of anaspect of the self.

At our level of analysis, however, the di-chotomy still remains useful and meaningful.When internally stored schemas are accessedfor comparison with environmental input dur-ing normal perception, this accessing occurspreattentively. It is the resulting percept thatis represented in consciousness, not the evok-ing of the schemas. With regard to consciousexperience, then, it seems quite reasonable forus simply to ignore the role played by the

6 The perceptive reader may have noted that thisactivity in itself represents a TOTE unit at a lowerlevel of analysis than is being emphasized in thispaper.


schemas in generating the percept. Moreover,it bears repeating that the function of thesequence that leads to the percept is quiteobviously the categorization of the stimulusinput—and the input is an event whose pointof initiation is outside the self. Because thisis the goal of the activity, it thus seems rea-sonable to refer to it with such terms as focuson the environment and processing of incom-ing information. In both of these respects, theproposed dichotomy can be viewed as a verbalshorthand that is useful in classifying con-scious experience at a macro level (wheremost of our present concerns are located)even while recognizing that at a micro levelmuch more complicated events are occurring.

3. In some cases, response-prototype infor-mation either constitutes part of a recognitoryprototype or is directly implied by the recogni-tory prototype; in other cases, this informa-tion is not contained or implied (see Figure2) . Saying that a recognitory schema used toidentify one's behavioral context does specifyresponse-prototype information is exactlyequivalent to saying that the context evokesin the person a behavioral standard. As anexample, most people in our culture have beentaught to dampen the intensity of their be-havior—that is, to speak and walk quietly—in places that are classified as religious in na-ture. When a person with this behavioralschema classifies a behavioral setting as beingreligious, quietness is immediately evoked asa behavioral standard.

Recognitory schemas do not always containthis kind of response-specifying information,however. In social psychological terms, manysituations exist in which no particular stan-dard of appropriate behavior is salient, whereguidelines for action are not implicit in one'srecognition of the nature of the context.Whether a situation has or has not evokeda response prototype as a behavioral standardhas important implications for what occurswhen attention is self-directed.

4. // a behavioral standard has not alreadybeen evoked by classification of one's context,attention directed to the self leads to increasedawareness of salient self-elements (Figure 2).Stated another way, the analysis and catego-rization that were postulated above to occur

for environmental information when attentionis directed outward (Proposition 2) occurwith regard to self-information when attentionis directed inward. The underlying compo-nents of this process presumably are the sameas those discussed above. That is, comparisonsare being made between information and re-cognitory schemas. The difference is that theinformation now being compared with theschemas is self-information—information fromwithin (e.g., recollections of prior behaviorsor present bodily feedback activity). Experi-entially, in this type of circumstance, self-attentive persons simply become increasinglycognizant of themselves and their own salientcharacteristics, much as they might examineany external stimulus.

In such cases, self-focus is not evaluativeexcept in the minimal sense that preattentiveanalysis "evaluates" in order to categorize.More specifically, this state of self-focus is notphenomenologically aversive.

The notion that self-directed attention mayin some cases lead simply to heightenedawareness of salient self-elements rather thanto a seli-critical examination is a major de-parture from the assumptions on which Duvaland Wicklund (1972) based their theory. Yetthere is a good deal of evidence to support thisassertion. This evidence will be reviewed inthe following paragraphs.

Duval and Wicklund (1973) themselvesdemonstrated that more self-attentive subjectsm'ake greater self-attributions for hypothet-ically experienced outcomes than do less self-attentive subjects. This outcome has sincebeen replicated (Buss & Scheier, 1976), al-though other research suggests some limita-tions on the effect (Hull & Levy, 1979). Itmay be that the mere heightened salience ofself as an entity led subjects in the Duval andWicklund (1973) and Buss and Scheier(1976) studies to weight the self more heavilyin the judgments they made. Alternatively,perhaps the salient aspect of self in this at-tribution-making context is one's potential asa causal agent (cf. Duval & Wicklund, 1973).

The reasoning that salient self-aspects maybe more fully represented in consciousnesswhen self-focus is high than when it is lowhas also been extended to the internal experi-


ence of emotional states (Scheier & Carver,1977). In that research, it was proposed thatwhen one's attention is self-focused in a con-text in which there is affect, one may attendto that affect as being the salient componentof self. Thus, in two of Scheier and Carver'sstudies, subjects who had experienced positiveand negative mood inductions subsequentlyreported feeling more elation and depression,respectively, when more self-attentive tharfwhen less so. In two additional studies, sub-jects reported experiencing greater attractionand repulsion toward pleasant and unpleasantslides, respectively, when self-attentive thanwhen not. Others (Borkovec & O'Brien, 1977)have found similarly that directing a subject'sattention to his or her bodily responses leadsto an increase in the self-reported intensity ofthe emotion being experienced.

The notion that heightened self-focus leadsto enhanced awareness of one's internal statesuggests another interesting possibility: thatself-focus can lead to heightened awarenessof the absence of anticipated internal activity.This hypothesis was recently tested (Gibbons,Carver, Scheier, & Hormuth, 1979), in astudy in which some subjects were led toexpect arousal symptoms from a pill, whereasothers were not. In all cases the pill was aplacebo. Among subjects who expectedarousal, fewer symptoms subsequently werereported by self-focused subjects than bythose with less self-focus.

The Gibbons et al. (1979) finding appearsto indicate that when led to expect a differentinternal state than is actually present, theself-attentive person has greater access to theveridical internal state than does the less self-attentive person. Two additional studies(Scheier, Carver, & Gibbons, in press-a) haveinvestigated further this heightened awarenessof internal experience among self-attentivepersons by assessing whether it can serve toreduce other kinds of suggestibility phenom-ena. In the first of these studies, subjects wereexposed to stimuli of moderate sexual attrac-tiveness, which they were instructed to eval-uate on the basis of their own bodily reactions.Subjects were led, however, to anticipateeither highly arousing or very nonarousingstimuli. Subjects with experimentally height-

ened self-attention were less misled by thesecues than were subjects in whom self-focushad not been increased. The second study ex-tended the reasoning to the experience oftaste. Subjects were led to expect either anincrease or a decrease in a flavor intensity(relative to a previous sample); they thenreceived either a slightly stronger or a slightlyweaker solution. The intensity judgments sub-sequently made by highly self-attentive sub-jects were less in line with manipulated antic-ipations and more in line with actual flavorintensities than were judgments made by lessself-attentive subjects. Thus, again, self-focusminimized suggestibility, apparently by in-creasing awareness of an actual internal ex-perience.

Nor is the increased accuracy of self-reportthat occurs under high self-focus limited toreports of internal perceptual experiences.Other studies (Pryor, Gibbons, Wicklund,Fazio, Si Hood, 1977; Scheier, Buss, & Buss,1978) indicate that self-focus enhances peo-ple's accuracy in reporting on their habitualbehavioral tendencies. Thus, for example, self-reports of sociability that were made underconditions of heightened self-attention werefound to be highly correlated with unobtru-sive measures of actual social behavior; re-ports made by less self-attentive subjects, onthe other hand, poorly represented their actualbehavior (Pryor et al., 1977). Presumably,self-attention provides more veridical or morethorough access to the memories that arerelevant to such self-reports. Additional evi-dence that self-focus increases the activationof self-relevant memory areas has been pro-vided by Geller and Shaver (1976).7

7 In addition to determining ease of access to pre-viously stored information, the direction of one'sattention probably also has implications for theaccess one has to schemas for use in encoding newstimuli (whether they be self-stimuli or environ-mental stimuli). Thus, for example, when one isself-focused, one may not have good access to theschemas that are typically used for encoding informa-tion about the environment (cf. Vallacher, 1978).Indeed, easy access to a schema may increase its use,even in situations where such use would not other-wise be expected. This may be the mechanism behindthe "top of the head" phenomena discussed by Taylorand Fiske (1978), in which some particularly salient


A final illustration of the fact that self-attention can lead simply to greater cog-nizance of the self comes from research con-ducted by Carver and Scheier (1978) asvalidation of self-awareness manipulations.Subjects in that research completed a self-focus sentence completion blank (Exner,1973) developed and validated earlier as ameasure of egocentrism. This instrument wascompleted either in an empty room or in thepresence of a self-awareness-inducing stim-ulus. As was expected, proportionally moreself-focus sentence completions were emittedin conditions of heightened self-awarenessthan under control conditions.

Salience. One final issue, implicit in muchof the above discussion, probably should beaddressed at this point. What determineswhich aspect of the self is "salient" whenattention is self-directed and which aspect ofthe environment is salient when attention isdirected to the environment? A full and com-plete answer to this question cannot yet begiven. Doubtlessly the ultimate answer will becomplex. However, one determinant of sali-ence probably derives directly from the inter-active nature of the categorization process de-scribed above. As was noted, classification ofthe environment takes place by comparison ofinput against recognitory prototypes. If theprototypes to which the environmental inputscorrespond have cue implications for somespecific self-information, the self-informationis probably accessed (in a preliminary pre-attentive fashion) by the very process ofclassifying the environmental input. Theaspect of self that is represented by that self-information presumably will be more salientthan other self-aspects when attention is sub-sequently directed inward. Another way ofsaying this is that when we attend to the self,we are often examining an aspect of self thathas been suggested by some cue in our en-vironmental context. For example, an environ-mental stimulus that is categorized as typ-ically emotion-inducing may cue a search for

stimulus exerts an inordinate influence on subjects'behavior. This notion might also have some relevanceto the recent finding that imagining an event in-creases the degree to which a person expects theevent to occur (Carroll, 1978).

bodily arousal as a salient self-dimensionwhen attention is self-directed. Indeed, such acue implication may serve in vivo as an im-petus to shift focus from the environment tothe self.

Similarly, the self-aspect that has just beenunder examination may be a major determi-nant of the aspect of the environment that issalient when attention is directed outward.<For example, consider a case in which self-attention has had as its object an existingstate of bodily arousal for which there is norecognized antecedent. In such a case, it seemslikely that attention subsequently directedoutward would be oriented specifically tosearching for aspects of the environment thatare categorizable as potentially arousing. Thusa self-aspect can have a cue implication forsome aspect of the environment, leading toheightened salience of that aspect when one'sfocus turns outward. More simply, when weattend to the environment, we are often focus-ing on aspects of the environment that havebeen suggested by some cue within ourselves.

5. Just as categorization of environmentalstimuli sometimes evokes a behavioral proto-type as a standard, a categorization of a sali-ent self-aspect may in some cases evoke a be-havioral standard (see Figure 2). As anexample, a salient emotion might evoke aprototypic response to that emotion. The re-search presented above regarding awarenessof affect was limited to reports of subjectiveemotional experiences. Other research hasdemonstrated, however, that heightenedawareness of an affect may also lead to in-creased behavioral responsivity to the affect.For example, Scheier (197.6) found that pro-voked aggression was more intense amongmore self-focused than among less self-focusedsubjects.8 Other research (Scheier, Carver,

8 Scheier (1976) found that subjects' personalattitudes toward aggression did not correspond tothe aggression they actually emitted. However, anattitude is only one type of standard that is poten-tially applicable in that context. The issue of varietiesof standards will be addressed more fully below. Theposition taken here with regard to the Scheier (1976)data is that awareness of anger evoked a stereotypicretaliatory response prototype and that the proto-type was used as a comparison value for subsequentbehavior.


Schulz, Glass, & Katz, 1978) has demon-strated that feelings of sympathy are ex-pressed to a greater degree among more self-attentive than among less self-attentive per-sons. Finally, yet other studies (Scheier,Carver, & Gibbons, in press-b) have shownthat strong fear leads to greater avoidanceamong more self-attentive than among lessself-attentive persons.

6. A standard can have either a positive ora negative valence. A positive valence impliesthat the standard is taken as a "desired" goalstate. A negative valence exists when a stan-dard of comparison is taken as an "undesired"goal state. Valences presumably are based onprevious learning and generalization and ontemporary encoding through imitation andverbal transmission. Multiple or componentvalences can easily exist with respect to anyspecific standard, of course. Consider, for ex-ample, the teacher's behavioral standard ofgoing from the office to teach a class. On agiven day the teacher may feel that he or sheshould go to the class because teaching is hisor her vocation (positive valence); theteacher may also feel like not going becausethe teacher does not enjoy the class (negativevalence). One important component valence,of course, occurs as a function of an obligationto do the behavior in question (cf. Vickers,1973).

It seems likely that any given componentvalence can also vary in salience over time.Its salience is likely to depend on the salienceof the environmental cues to which the val-ence is attached. For example, if bright, inter-ested students have just been in the teacher'soffice, a positive valence may be salient. Pre-sumably the accumulation of these componentvalences (weighted by salience) determinesthe overall valence of the standard as dis-cussed here.

7. Standards and behavioral dimensions canalso vary in importance. There are at leasttwo ways in which importance may be con-strued in informational terms. The first de-pends on association of positive or negativeevents with the stimulus class during the per-son's prior experiences. A priority label maybecome part of the categorization as a func-tion of the intensities of such events. High

intensity events (e.g., strong punishment forpoor behavior with respect to a standard)would lead to a high priority tag or "impor-tance"; low intensity events would lead to atag of less priority.

A second possibility concerns the centralityor interrelatedness of a behavioral dimensionwith respect to other dimensions. Importancein this case would be defined according to thenumber of cross-reference tags attached to agiven behavioral dimension that refer to otherbehavioral dimensions. For example, theamount of work that one person does may beweighed in his or her mind as important, butits importance does not extend to other areasof the person's life. The work, in effect, is leftat the office. In contrast, another person maynot only see his or her work output as im-portant but may also view it as having im-plications for other parts of his or her life, forexample, interpersonal relationships. Thework output dimension thus would be effec-tively more important for the second personthan for the first, by virtue of its intercon-nectedness with other dimensions.9

8. // a behavioral standard has been evokedeither by environmental input or by a salientself-element, subsequent self-focus leads to acomparison between self (one's present be-havior or characteristics) and the previouslyevoked standard. Comparison between selfand a positively valenced standard leads todiscrepancy reduction; comparison betweenself and a negatively valenced standard leadsto discrepancy enlargement (see Figure 3).The comparison between self and standard isregarded as representing the test phase of apsychological TOTE unit. It thus leads to theengagement of one or the other of two possibleadjustment and feedback sequences: either anegative feedback loop (discrepancy reduc-tion) or a positive feedback loop (discrepancyenlargement).

The notion that heightened self-focus leads

9 Standards can also vary in their degree of ac-ceptedness, which is logically distinguishable fromimportance, even though the two in reality are oftenconfounded. A standard that one accepts half-heartedly will be less reflected in behavior than oneaccepted enthusiastically.


to increased conformity of one's behavior to asalient standard was, of course, a main tenetof self-awareness theory, although for quite adifferent reason than is assumed here. Self-awareness theory held that discrepancy reduc-tion was one of several potential responses tothe postulated aversiveness of self-attention.The present model assumes that the discrep-ancy-reduction impulse is affect free and is theprepotent response to the recognition of adiscrepancy. These distinctions are importantones and will be elaborated more fully in alater section of the paper.

Discrepancy reduction. Whatever its basis,the prediction that self-directed attentionleads to heightened conformity to the salientbehavioral standard has received ample re-search support. For example, a series ofstudies of instrumental aggression showed thatsubjects matched their behavior to salientstandards (which varied from study to study)to a greater degree when made more self-aware than when less self-aware. In the firstof these studies (Scheier, Fenigstein, & Buss,1974), self-focus caused male subjects to con-form to a "chivalry" norm when shocking afemale victim who had not provoked them. Ina second study (Carver, 1974), female sub-jects shocked at higher levels when self-focuswas increased, in line with an experimentallyinduced standard in which high shock wasportrayed as desirable. In yet other studies(Carver, 197S), subjects' aggressive behaviorwas more consistent with their previously ver-balized attitudes when the subjects were moreself-attentive than when less so.

Moreover, demonstrations of the fact thatself-focus leads to increased conformity tostandards are by no means limited to studiesof aggression. For example, in an experimentin which speed of copying prose was madesalient as a standard of comparison (Wick-lund & Duval, 1971), more prose was copiedwithin a given time span by subjects in whomself-focus was heightened than by those inwhom it was not. In addition, compared toless self-aware controls, self-aware subjectshave been found to behave more consistentlywith their previously assessed stage of moralreasoning (Froming, in press) and more con-

sistently with their previously assessed levelsof sex guilt (Gibbons, 1978).

Discrepancy enlargement. The notion thatstandards can have negative valuation andthat focus on those standards leads to discrep-ancy enlargement has received much lessattention than has the matching-to-standardprocess. However, this notion is logically con-sistent with an older idea in social psychology.Newcomb (1958) wrote of the "negative ref-erence group," a group with which one doesnot identify, but with which one actively com-pares oneself. Such a comparison allows one toemphasize, and indeed to further increase, theexisting differences. Newcomb coined the termnegative reference group in the context ofsocial comparison theory (Festinger, 1950,1954). Yet the nature of the process thatNewcomb proposed is strikingly similar to theprocess proposed above for cases in whichsome negatively valenced behavioral standardis salient. The difference is that in the presentmodel the comparison value need not be theattitudes or behavior of a particular group ofpeople. It can be any negatively valencedvalue that is taken up for purposes of com-parison. (The relationship between the pres-ent theorizing and social comparison theorywill be examined in more detail in a later sec-tion of the article.)

Another class of behavior that can beviewed as discrepancy enlargement is a phe-nomenon that is usually analyzed in terms ofpsychological reactance (Brehm, 1966; Wick-lund, 1974). Researchers have demonstratedrepeatedly that coercive attempts to persuadelead to resistance to persuasion or even to atti-tude change in a direction opposite to theposition being advocated (e.g., Brehm &Brehm, 1966; Snyder & Wicklund, 1976).Such effects may be interpreted as attemptsto distance oneself from the standard of com-parison (i.e., the position that is advocated inthe communication). In the present modelthat standard would acquire a negative va-lence in very much the same manner as was

.postulated by Brehm (1966)—that is, byvirtue of the threatening elements containedin the associated communication. Consistentwith the reasoning that distancing-from-stan-dard might occur under these circumstances


in response to self-focus, Carver (1977) hasfound that attitude reversal in response to acoercive communication was greater amongself-focused subjects than among less self-focused subjects. Similar results have occurredin two additional studies (Carver & Scheier,in press).

Standards. Standards for self-comparisoncan vary widely, both in terms of their spe-cific nature and in terms of their source. Onereasonable distinction in this regard would bebetween the type of a standard and its con-tent. For example, smoking in response tostress is different in content from eating, orpacing in response to stress. Yet they are allsimilar in one respect: They are of the typecalled habit. A personal-preference attitude isa different type of standard. A third type isan attitude based on more complex reasoning,for example, moral reasoning. Similarly, aninstruction from another person that isadopted and conformed to is different in typefrom all these, as is the temporarily assumedsocially desirable behavior that others nearbyare exhibiting. With these multiple possibil-ities of source, type, and content, what deter-mines which is evoked in a given situation andis thus taken as the comparison value whenone is self-focused?

Specific applications of this general ques-tion abound in personality and social psychol-ogy. For example, the literature of moraldevelopment is replete with illustrations ofthe fact that moral reasoning is a judgmentalcapability rather than a typical way of behav-ing. Under which conditions will a personrespond to a moral choice by exerting his orher psychological capabilities to their fullest,and under which conditions will the person'sbehavior be dictated 'by other considerations?The answer to this question is by no meansclear, for the following reason. Althoughcategorization processes and the use of catego-rizational schemas to classify and interpret in-formation have recently begun to receive at-tention from social psychologists (e.g., Cantor& Mischel, 1977; Kuiper & Rogers, 1979;Markus, 1977; Rogers, Kuiper, & Kirker,1977; Rogers, Rogers, & Kuiper, 1979), thusfar virtually no work has been aimed ex-plicitly at understanding the process of ex-

tracting behavioral prototypes in response toenvironmental cues. Which standard becomessalient in a given experimental situation hasusually been more a matter of good intuitionson the part of the researcher who is preparingthe situation than a matter of precise analysisof the determinants of salience. Quite ob-viously, this entire question is an importantarea for future work. For the present, let ussimply note that matching-to-standard effectsappear to have been demonstrated with regardto all of the following: habits10 (Liebling,Seiler, & Shaver, 1974), personal-preferenceattitudes (Carver, 1975), moral reasoningcapabilities (Froming, in press), and experi-mentally provided instructions (Carver,1974).

9. Matching-to-standard is the normal con-sequence of self-attention when a behavioralstandard is salient, until and unless the pro-cess is impeded in some way. (Although thisand the following propositions will be framedin terms of matching-to-standard, it should beclear that the same logic is applicable to dis-tancing-from-standard phenomena; see alsoFigure 3). Throughout a discrepancy-reduc-tion sequence, self-focus represents a recur-rence of the test mechanism of the psycholog-ical TOTE unit. Test will continue to alter-nate with one or more operate mechanismsuntil behavior or present state is brought intoline with the standard. As in any TOTE se-

10 The fact that matching-to-standard apparentlycan occur with regard to habits as standards raisesthe additional question of how "conscious" theprocess is in such cases. Matching to a habit standardmay feel very unconscious, phenomenologically, forthe following reason. The degree to which executionof a behavioral sequence seems conscious depends onthe salience of its component acts and their conse-quences; this in turn depends on the degree to whichthose acts are monitored while being carried out. Inhabitual activity, the behavioral components havepreviously been encoded as a well-connected chain(cf. Schank & Abelson, 1977, discussion of cognitive"scripts"). To use TOTE terminology, a single op-erate consists of a behavioral sequence rather thana small bit of behavior change. Therefore, fewer testsare required over the course of the behavior thanis the case for other, less well-known behavioralsequences. The relative paucity of attention devotedto monitoring this behavior suggests that it may beexperienced less completely in consciousness. •


quence, if successful completion of matching-to-standard occurs, it is followed by disen-gagement.

This proposition stands in direct contradic-tion to the assumption in recent statements ofself-awareness theory (Wicklund, 197Sa,197Sb) that the prepotent response to self-focus in a standard-salient context is avoid-ance of self-awareness rather than discrepancyreduction. As was noted above, this contradic-tion has important ramifications that will beconsidered further in a later section of thepaper.

10. // something impedes the matching-to-standard process, behavior is interrupted andan assessment process is evoked (see Figure3). This assessment entails the further pro-cessing of relevant information, yielding anoutcome expectancy: an estimate of the likeli-hood of being able to more closely approxi-mate the standard, based on the nature of thesituation and on the behaviors available tothe person. This assessment process can occurprior to the initiation of behavior if, for ex-ample, the person has foreknowledge that thebehavior is going to be difficult to execute suc-cessfully. Alternatively, it can occur duringthe matching-to-standard sequence if the per-son encounters difficulty in either selectingor executing the appropriate operation tomove toward the standard.

The assessment of outcome expectancy maybe an evaluation of the likelihood of attainingthe overall goal, given the context and one'sresources. Alternatively, if the matching-to-standard sequence involves separable steps, itmay be an assessment of the likelihood of at-taining the next step. In either case, theassessment process has two products: One isbehavioral, the other affective. These will beconsidered in turn.11

11 (a). // the assessment process leads to afavorable outcome expectancy, the behavioralresponse is a return to the matching-to-stan-dard sequence (Figure 3).

ll('b). // examination of the context andone's resources reveals a low subjective prob-ability of being able to alter behavior appro-priately, the behavioral consequence is with-drawal. The latter would appear to representa disengagement from the dimension in ques-

tion, a kind of going-out-of-the-field phenom-enon (cf. Lewin, 1935). In practice, thiscould take the form of a mental dissociationfrom the dimension (refusal to consider fur-ther) or could occur as a behavioral with-drawal from contextual elements that eitherheighten self-attention or make the standardsalient.

There are several types of events that canpotentially prevent or disrupt the matching-to-standard process. One possibility is that thebehavior in question is fixed in the past. Ifthere is no opportunity available to repeat,reattempt, or undo the behavior, it is thusrendered unchangeable. In such a case, dis-crepancy reduction cannot even begin. Anexample of this state of affairs occurred in astudy conducted by Duval, Wicklund, andFine (1972) . Some subjects in that studywere informed that their scores on a pre-viously administered test of intelligence andcreativity were very discrepant from the levelsthat they had anticipated and desired. More-over, it was implicit that the test was notgoing to be repeated, nor was any other op-portunity to be provided for reducing thediscrepancy (thus all subjects would havehad low outcome expectancies with regard tomatching the behavior to standard). In thisstudy, subjects who were to wait for anotherexperimenter after receiving the negativefeedback withdrew sooner if the room con-tained a self-attention-heightening stimulusthan if it did not.

Results of a more recent study (Steenbarger& Aderman, in press) confirm that this with-drawal tendency depended on the fact thatthe behavioral discrepancy was portrayed asunalterable. This study made use of a para-digm derived from that of Duval et al.(1972), but one in which there was ostensiblyto be a second task. Steenbarger and Aderman(in press) found that when subjects experi-

11 This proposition and the two following proposi-tions, though derived independently, have a gooddeal in common with Stotland's (1969) analysis ofhope as a psychological construct. Stotland used theterm hope to refer to the perception of a high prob-ability of attaining a goal and the term anxiety torefer to the perception of low probability of attain-ing a goal.


enced an inflexible discrepancy—that is, whenthe impression was created that subjects couldnot expect to alter their performances ap-preciably—self-awareness hastened with-drawal. When the impression had been createdthat improvement was a real possibility, onthe other hand, no increase in withdrawaloccurred as a function of self-attention.

Intrapersonal deficit. A second set of con-ditions that may lead to behavioral interrup-tion stems from the possibility that the execu-tion of the behavior in question requires spe-cific skills, abilities, or characteristics that theperson is not certain he or she possesses. Theimplication that these skills are required maycause the discrepancy-reduction attempt to besuspended and cue the assessment process. Anexample of the importance of this possibilityis provided by recent studies of fear-basedbehavior conducted by Carver and Blaney(1977a, Experiment 3; 1977b). In thosestudies, some subjects were chosen explicitlyon the basis of their doubts about their abilityto cope with fear, whereas others were chosenas believing that they could cope with theirfear. All of these subjects, however, reportedhaving identical (moderate) degrees of fear.It seems reasonable that signs of fear whileattempting to approach and pick up thefeared stimulus (in this case, a snake) shouldcue a self-assessment process in all such sub-jects. In the Carver and Blaney (1977b)study, the subjects who doubted their abil-ities to overcome their fear (i.e., who hadchronically low outcome expectancies) re-acted to the perception of rising fear, inducedvia false heartbeat feedback, by avoiding at-tending to the behavior-goal comparison (ac-cording to self-reports) and by withdrawingbehaviorally. The latter effect replicated aprevious finding (Carver & Blaney, 1977a,Experiment 3). In contrast, those subjectswho had expressed more confidence aboutbeing able to go through with the behavior(i.e., who had chronically higher outcome ex-pectancies) responded to the perception ofanxiety by focusing more on the behavior-goal comparison and attempting—successfully—to match the one with the other.

The Carver and Blaney (1977a, 1977b)studies investigated the behavioral conse-

quences of variations in expectancy, but thosestudies did not explicitly incorporate experi-mental manipulations of self-focus. More re-cently, however, a similar study has beenconducted using a conventional self-awarenessmanipulation (Carver, Blaney, & Scheier,1979). In that experiment, confident subjectswere found to be undeterred 'by heightenedself-awareness during the approach task, butdoubtful subjects withdrew earlier in theapproach sequence when self-focus was highthan when it was lower. Postexperimental self-reports formed a pattern that was consistentwith the present model, that is, that height-ened self-attention had led to increased aware-ness of fear among all subjects and thus toself-assessment during the approach sequence;that self-assessment, in turn, had led to con-tinued approach attempts among confidentsubjects and to early withdrawal amongdoubtful subjects.

This result has recently been conceptuallyreplicated (Carver, Blaney, & Scheier, inpress) in quite a different 'behavioral domain:i.e., responses to failure. All subjects in thatresearch were confronted with a failure ex-perience on an intellectual task, in order tocreate a large self-versus-standard discrep-ancy. Subjects then undertook a second task,ostensibly bearing on the same intellectualskill. Rather than assessing subjects' chronicexpectancies, however, as had been the casein the Carver et al. (1979) study, an experi-mental manipulation of outcome expectancywas introduced. Subjects were led to believeeither that they could potentially do quitewell on the second task, or that they wouldprobably do quite poorly. In reality, the sec-ond task was a measure of persistence. Aspredicted, subjects in whom unfavorable ex-pectancies had been induced were less per-sistent when self-focus was high than when itwas low. Also as predicted, subjects withfavorable expectancies were more persistentwhen self-focus was high than when it waslow.

The tasks in all of the above research pro-vided subjects with escapable situations. Thatis, it was possible for subjects engaged in taskattempts in those studies to withdraw phys-ically from the behavioral context. This pos-


sibility does not always exist, however, whena perceived self-deficiency threatens one's be-havioral adequacy. For example, consider theperson with severe test anxiety who must takean unavoidable test. While in the testingsituation, such persons probably are verysusceptible to being interrupted from theirperformance attempts by stimuli that cue out-come assessment. In many test situations, thisassessment yields unfavorable expectancyjudgments. Because they cannot withdrawbehaviorally, however, these persons may befrozen in the self-assessment phase of thesequence, where they repeatedly reconfrontthe evidence of their own inadequacy. At-tempts at task-relevant activity are sporadicbecause outcome assessment indicates thatsuch attempts will be unsuccessful, and theattempts are short-lived because the assess-ment process is likely to be re-evoked. Thusthe test-anxious person is caught in a cogni-tive loop in which self-assessment and cogni-tive disengagement from the task predom-inate, rather than renewed effort.

This portrayal is consistent with Wine's(1971) characterization of the test-anxiousperson as being chronically self-focused insituations where the adaptive response is tobe task focused. The present model seemsmore adequate than is Wine's, however, intwo respects. Though Wine's analysis wouldnot necessarily predict that self-focus canfacilitate performance among persons low intest anxiety, the present model clearly does.Such performance facilitation has been dem-onstrated elsewhere (Wicklund & Duval,1971). The present model also implies thatself-focus can facilitate performance evenamong highly test-anxious persons, if circum-stances lead those persons to have positiveoutcome expectancies. This possibility doesnot seem derivable from Wine's (1971)theory. Recent research (Slapion & Carver,Note 1) has yielded evidence that this canoccur as well.

Environmental constraints. Yet a third setof conditions that would lead to behavioralinterruption is the possibility that some fea-ture of the environment may prevent the suc-cessful execution of the appropriate behavior(a circumstance that fits the usual opera-

tional definition of frustration). To use asomewhat trivial example, if one goes shop-ping at a time when stores are closed, onetypically withdraws from the attempt. Thiswithdrawal does not occur, however, untilone's attempt has been interrupted and therealization has occurred that the desired out-come is unlikely (i.e., low outcome probabil-ity). The notion that environmental con-straints lead either to withdrawal or toattempts to circumvent them seems a very ob-vious point. This case is discussed here largelyin order to point out that the psychologicalprocesses implicit in this class of events arethe same as those that occur in the other eventclasses discussed above. This indicates thehigh degree of generality of the processesunder consideration.

Novelty. The analysis presented here ofprocesses leading to withdrawal also bearssome resemblance to Berlyne's (1963) discus-sion of the somewhat perplexing fact thatnovelty sometimes leads to approach, some-times to withdrawal. He proposed that thepresence of some degree of "collative" (i.e.,comparison-requiring) properties in a stimulus(novelty, for example) engages exploration,movement toward the stimulus in order toassimilate it. Too much of the same proper-ties, however, leads to withdrawal. Accordingto Berlyne, the reason for this withdrawal isthat the organism assesses the alterations re-quired to assimilate the stimulus as being be-yond its present capabilities. In effect, out-come expectancy with regard to assimilationis negative, and withdrawal is the conse-quence.

12(a). The judgment that one cannot alterone's behavior in the direction appropriate tothe standard may also lead to negative affectin proportion to the importance of the be-havioral dimension or the standard and theperceived magnitude of the discrepancy.12

This negative affect remains present as long asthe person focuses on the judgment that he

12 It seems worth reemphasizing that importancehere refers to importance as defined by oneself. Astandard that others view as important but thatseems irrelevant to oneself is for present purposesnot an important standard.


or she cannot reduce the discrepancy. Evi-dence for this proposition comes from a study(Steenbarger & Aderman, in press) that wasdiscussed above in connection with the as-sumption that self-focus leads to withdrawalonly when discrepancy reduction is preventedin some way. That same study also showedthat subjects who had been led to believe thatthey could not reduce the discrepancy experi-enced more negative affect when self-focuswas high than when it was lower. This effectdid not occur among subjects who expectedto be able to reduce the discrepancy.

12(b). Engagement of the assessment pro-cess can also lead to positive affect, if assess-ment leads to the perception of a favorableoutcome probability. This is comparable toSimonov's (1970) argument that emotion be-comes positive when "there is a surplus ofinformation available as compared with theinformation necessary to satisfy a sufficientlystrong need" (p. 147). This is the elation ex-perienced, for example, when one knows thatone knows how to arrive at the solution of aproblem, even though the problem has notyet been solved.13

The assumption of two potential responses(behavioral and affective) to the assessmentof outcome expectancy raises a question as totheir relationship. The two may be parallelresponses. For example, both the cessation ofbehavioral attempts and the experience ofnegative affect may be cued by a recognitionthat nothing can be done to improve one'spresent state. This would be similar in somerespects to LeventhaPs (1970) parallel re-sponse model of reactions to fear communica-tions. An alternative possibility is that be-havioral withdrawal may occur as a responseto the existence of negative affect. It is notpossible to specify the relationship unequi-vocally at this time. It seems clear, however,that environmental constraints on potentialbehavior will have a large role in determiningwhich of the two response modes predom-inates, particularly in cases of unfavorableoutcome expectancy. That is, when the en-vironment permits withdrawal, withdrawalshould predominate. When the environmentdoes not permit withdrawal, negative affectshould predominate.

13. Affect that results from assessment ofoutcome expectancy can take one of twoforms, depending on the cognitions associatedwith the affect (see Figure 3). What cogni-tions are linked to negative affect will be de-termined by the person's perception of thelocus of the barrier to 'behavior alteration.If the person has a low efficacy expectancy—that is, if he or she perceives that behavior isbeing impeded by a self-deficit—the affectwill be associated with the self. It may be re-flected in such ways as reduced self-esteem. Ifthe person has high efficacy expectancy andperceives that his or her behavior is impededby some environmental constraint, the affectwill be associated with those constraining ele-ments. It may be reflected by such things asexpressions of resentment.14 Similarly, whatcognitions are associated with positive affectwill be determined by the perceived locus ofthe cause of favorable outcome expectancy.The affect will be associated with the self ifpersonal efficacy is seen as having caused thegood outcome to be likely; it will be asso-ciated with the environment if the positiveoutcome expectancy is ascribed to a benignenvironment. Although the logic behind thisproposition is relatively straightforward, thesepredictions remain to be tested.

Comparisons With Self-Awareness Theory

At least three questions deserve further con-sideration with regard to differences betweenthe present model and self-awareness theoryas proposed by Duval and Wicklund (1972)

13 This analysis may help to account for the ob-servation that the anticipation of a pleasant eventoften seems more pleasurable than the event itself,as if knowing that the outcome will occur is betterthan its occurrence. Indeed, this would be consistentwith the frequently expressed observation amongpsychologists that "the fun is over" once an idea hasbeen conceived and a study to test it is designed.

14 It may be this variable that distinguishes, forexample, between depression attributed by a clientto environmental contingencies and depression at-tributed to inadequacies of the self. That is, insuf-ficiently positive outcome expectancies lead to de-pressed affect, but whether cognitions of self orenvironment are associated with the negative affectmay vary with efficacy expectancies.


and amended by Wicklund (197Sa). They arethe following: (a) Under what conditions isself-attention aversive? (b) What instigatesor energizes the matching-to-standard pro-cess? (c) Finally, what is the prepotent re-sponse to self-focus when a behavioral stan-dard is salient—discrepancy reduction orwithdrawal? These issues will be discussed inthe following three sections.

Aversiveness of Self-Focus

One important area of conflict between thepresent model and that of Duval and Wick-lund concerns whether negative affect is aninevitable consequence of self-attention, andif not, what other conditions are necessaryfor it to occur. Duval and Wicklund (1972)and Wicklund (197Sa, 197Sb) have arguedthat any time a discrepancy exists such thatone's present state or behavior is worse thanthe standard of comparison, self-attention isaversive.15 "It is assumed that the objectivelyself-aware person will not simply react to him-self impartially and in a neutral manner, butthat he will come to evaluate himself as soonas the objective state occurs" (Duval & Wick-lund, 1972, p. 3). "Certainly objective self-awareness is postulated to be an aversive,motivational state" (Wicklund, 197Sb, p. 80).Indeed, as was stated at the beginning of thepresent article, the postulated aversiveness ofself-attention was that theory's crux. In con-trast, it has been proposed here that self-directed attention leads to negative affectonly when the person perceives that he or shecannot alter his or her present state in thedirection of the standard. (This discussionwill be framed in terms of discrepancy reduc-tion, but it is of course applicable to discrep-ancy enlargement as well.)

It is instructive to note that the studiescited as evidence of the unpleasantness ofself-focus were invariably cases in which awithin-subjects discrepancy existed and inwhich subjects were prevented from alteringtheir present state or behavior. For example,in the first demonstration that subjects wouldattempt to escape from self-focus (Duval etal., 1972), the subject's behavior on the rele-vant dimension was anchored in the past and

thus could not be altered in the direction ofthe standard. Similarly, though self-esteemhas been found to be reduced when subjectswere made self-aware (Ickes, Wicklund, &Ferris, 1973), it is unclear to what degreethat finding depended on the nature of thestudy's dependent measure. The dependentmeasure was a comparison between subjects'self-rated real selves and ideal selves alongseveral dimensions. Subjects in that researchhad no way to alter their real selves. Theywere simply reminded of the real-ideal dis-crepancy to a greater degree when self-aware-ness was heightened than when it was not.

Aside from these examples in which sub-jects' behavior or characteristics were un-changeable, there appears to be no evidencethat self-attention itself is phenomenologicallyaversive. To the contrary, Steenbarger andAderman (in press) found an increase innegative affect only when self-focus was com-bined with a nonreducible discrepancy; whenthe discrepancy was flexible, the oppositetendency occurred. Similarly, Carver andScheier (1978), in demonstrating that thepresence of either a mirror or an audienceincreases self-focus, were unable to find anyevidence that self-attention led to negativeaffect. A similar conclusion has also beenreached by other researchers (Davis & Brock,1975; Hull & Levy, 1979).

Affect and valuation. There is only onerestricted sense in which the present modelimplies the existence of something resemblingnegative affect whenever self-attention is di-rected to a behavior-standard comparison. Itmay be useful in this regard to draw a dis-tinction between the phenomenological ex-perience of affect and the function of valua-tion (cf. Diggory, 1966, p. 70). The eventthat initiates the matching-to-standard se-quence according to the present model is thefact that the test of the TOTE unit assesses

15 Wicklund (1975a) has also pointed out that theexistence of a positive discrepancy (i.e., self betterthan the standard) is a different case, leading in hisview to positive affect. The present discussion, how-ever, will be limited to instances of negative dis-crepancy, as has implicitly been the case throughoutthe article.


the existence of a discrepancy between a pres-ent state and an accepted goal state. Inas-much as the function of the behavior change(operate) is to reduce the discrepancy, it ispossible to construe the recognition of the dis-crepancy as being (on some level) a nega-tively evaluated event. However, there seemsto be no reason to assume a concomitantphenomenological experience of negative emo-tion. As was noted in the preceding para-graphs, however, there is substantial reason tobelieve that negative affect occurs in caseswhen one is prevented from reducing thosediscrepancies (Duvaletal., 1972; Steenbarger& Aderman, in press). Thus the data seemconsistent with the assumption in the presentmodel that self-focus has phenomenologicallyaffective consequences only following outcomeassessment.

To the degree that negative affect is experi-enced in contexts in which behavior change is(objectively) not difficult, it may be seen as avariation of the inability-to-alter experience.For example, a workman who repeatedlyreaches to the wrong spot for tools may feelnegative affect, not because of failure to reachthe tool, but because the inability to correctthe initial reaching impulse is beginning toalter his perceptions of self-efficacy with re-gard to tool handling. Similarly, a man at aformal dinner who suddenly realizes that hehas been eating his salad with the wrong forkcan easily alter his behavior. Any negativeaffect he experiences is caused by concern thathis behavior may be regarded by others asuncultured and the possibility that he will notbe able to alter the first impression that hehas thus created. Finally, the perception ofthe need for a discrepancy reduction that willtake an extraordinarily long time can lead tonegative affect. However, this should occuronly if there are salient doubts as to the ulti-mate outcome. If there is a positive outcomeexpectancy—hope, in Stotland's (1969) terms—positive affect should ensue, even if manymonths of behavior remain between the personand his or her goal.

Instigation of Matching-to-Standard

There is a second important distinction be-tween the present model and Duval and

Wicklund's (1972) theory that is intimatelybound to the question of whether self-focus isaversive. This second distinction concerns thequestion of how behavior change is instigatedand energized when self-focus occurs and abehavioral standard is salient. As was men-tioned above, Duval and Wicklund (1972)devised their model from a drive-theory per-spective. By postulating that self-focus isaversive any time a within-self discrepancyexists, those authors could assume that subse-quent behavior change was energized by adrive state and was instigated by that drivestate's aversiveness.

In the present model, in contrast, thematching-to-standard sequence that is en-gaged when self-focus occurs in the presenceof a salient behavioral standard is seen as therealization within a psychological system of anegative feedback loop. The fact that self-attention is required in order to engage thematching-to-standard sequence is interpretedin control-theory terms in the following way.Self-focus in a standard-salient context repre-sents the test phase of the TOTE unit: anassessment of whether a discrepancy exists.The behavioral response to self-focus (thealteration of behavior in the direction of thestandard) is the operate phase. The orderingof the control sequence of the feedback loopdictates that operate cannot take place beforetest reveals a discrepancy. Thus, focus on adiscrepancy is required to engage the TOTEsequence. In phenomenological terms, thisawareness simply corresponds to the realiza-tion that one is in a context in which there isa standard to match with one's behavior. Achange in behavior depends on that realiza-tion.

Inasmuch as both models assume a com-parison process between self and standard, thecritical difference between them in this con-text seems to be the presence or absence of adrive postulate. This issue introduces con-siderable theoretical complexity, stemmingfrom two facts: first, that a great many vari-ants of drive as a construct have been pro-posed over the years, and second, that thereis far from universal agreement on what mea-surable state, if any, corresponds to a theo-retical state of heightened drive. A completely


adequate treatment of this complexity is wellbeyond the scope of the present article. Theinterested reader is referred to Appley's(1970) review article for a more detailedcritique of drive and arousal constructs. Con-sideration of the issues here will be muchmore limited.

Does self-attention heighten drive, or doesit not? One might at first view this as an em-pirical question, although it will be arguedbelow that this appearance may be illusory.However, let us consider the existing evidenceon the relationship between self-focus anddrive. Although agreement on the nature ofdrive is often difficult to obtain, as was notedabove, many researchers have assumed thatdrive is either equivalent to, or reflected by,autonomic arousal.10 To the limited degreethat evidence on this point is available, self-attention appears not to lead to an increasein physiological arousal. If one were willingto equate such arousal increases with height-ened drive, it would thus appear that self-focus is not drive inducing. Evidence on thisissue comes from two sources.

Salience oj nonarousal. One set of datacomes from a study (Gibbons et al., 1979)that was discussed earlier as evidence thatself-attention when no standard is salientmerely makes one more conscious of one'ssalient characteristic. Subjects in that studywere led to anticipate that arousal symptomswould occur as a function of a pill (actually aplacebo). Subjects later reported the extent oftheir symptoms, with self-attention heightenedor not. If self-directed attention per se werearousal inducing, greater arousal should havebeen experienced, and thus reported, by sub-jects in the high self-awareness condition thanby control subjects. Instead, the opposite oc-curred. Self-attentive subjects apparently be-came more conscious of 'the absence of arousal(consistent with the present informationalanalysis) and consequently reported fewersymptoms.

One difficulty with applying this finding tothe drive question, however, is the fact that inthe early statement of their theory, Duval andWicklund (1972) largely ignored the pos-sibility that a standard of comparison mightnot always be present. Wicklund (197Sa)

later concluded that self-focus was aversiveonly if one's present state was worse than asalient standard of comparison. However, heargued at the same time that one was almostalways worse than some standard and that onewould ultimately become conscious of thatdiscrepancy. Nevertheless, if self-attentionshould be drive inducing only when a discrep-ancy exists between self and standard, and ifit were possible to assume a case within Duvaland Wicklund's framework in which no stan-dard is salient, then those authors might thusanticipate that self-focus would not increasedrive in the Gibbons et al. (1979) case.

Physiological data. A second set of data onthe arousal question does exist, however(Paulus, Annis, & Risner, 1978). Moreover,this finding is not liable to the caveat that wasapplied above to the Gibbons et al. finding.Paulus et al. gave subjects a task (copyingprose) with a clear standard (copyingquickly). As subjects prepared to attempt thetask, either before a self-focusing stimulus orwith no such stimulus present, the experi-menter took a physiological measure called thepalmar sweat index (Johnson & Dabbs, 1967).Increased palmar sweat values have com-monly been believed to reflect increasedarousal (cf. Geen & Gange, 1977; Martens,1969a, 1969b). However, in the Paulus et al.study, lower values on this index were foundamong more self-aware than among less self-aware subjects. The meaning of the palmarsweat index is open to considerable questionon other grounds (see the following para-graphs). But if that measure were acceptableas an index of drive, the finding of Paulus etal. would seem to contradict Duval and Wick-lund's assumption that self-attention is driveincreasing.17

18 Though it certainly oversimplifies the case toimply that all theories utilizing a drive constructcontain this assumption, it is probably fair to saythat it is implicit in the large majority of drivemodels in social psychology.

17 The data of Paulus et al. were derived in thecontext of a social facilitation paradigm. Thepresent theoretical model has rather extensive im-plications for social facilitation phenomena, althoughconsideration of those implications is beyond thescope of this article. These implications are, how-


It was noted above that the arousal or driveissue is not entirely an empirical one. Thereare also two important logical considerationsthat render the entire arousal controversysomewhat less meaningful.

The meaning of physiological indices.First, contemporary researchers in physiolog-ical psychology tend to recoil at the meresuggestion that autonomic arousal is a unitaryphenomenon. Such psychologists are inclinedto look at autonomic indicants in terms ofwhat specific functions 'they imply rather thanwhether the organism is aroused. For example,Sokolov (1963) has proposed that one patternof physiological responses comprises an "ori-enting" response, whereas another patterncomprises a "defensive" response. The orient-ing response is held to facilitate greaterknowledge of an external stimulus by makingthe organism more receptive to input informa-tion from that stimulus. The defensive re-sponse, in contrast, involves an inhibition ofsensitivity to input information (cf. also Hare& Blevings, 1975). This latter function pre-sumably protects the organism from over-stimulation.

This argument is easily taken one step far-ther, by assuming that such function-specificphysiological response patterning is a normalconcomitant of different kinds of consciouslycontrolled behavioral involvement. Williams,Bittker, Buchsbaum, and Wynne (1975) havediscussed this possibility in some detail (seealso Lacey, 1967). They reasoned -that experi-mental tasks vary in the types of attentionaldemands they make (as opposed to the degreeof demand). Some tasks (e.g., a visual decod-ing task) require sensory intake; others (e.g.,mental arithmetic) require sensory rejection.Conceptually, the sensory intake patternseems similar to Sokolov's (1963) orientingresponse; the sensory rejection pattern,though not particularly "defensive," seems toembody many functional components of Soko-lov's defensive reaction, in that there is a

ever, taken up elsewhere (Carver & Scheier, Note 2).That paper also presents further data on the roleof drive in social facilitation and self-awarenessphenomena.

suppression of sensory input. In the Williamset al. analysis, however, this pattern simplyreflects the attempt to concentrate focus in-ward and prevent distraction. In a test of thisreasoning, Williams et al. (1975) were able toshow that subjects differed as predicted in car-diovascular response patterns as a functionof the type of task being performed.

It is instructive to note, in this regard, thatinitial theorizing about the palmar sweat in-dex (Dabbs, Johnson, & Leventhal, 19,68)concerned attentional issues fully as much asit concerned arousal. Dabbs et al. (1968)argued that increased palmar sweat was asso-ciated with readiness to engage the environ-ment, and that decreased values reflected self-directed attention and an attempt to concen-trate. These descriptions seem a good deal likethe sensory intake and sensory rejection pat-terns of Williams et al. (1975). Taken to-gether, these two descriptions converge on twonotions, one of them general and the othermore specific. The general one is that at leastsome physiological indices may be more use-fully construed as providing informationabout attentional and information-processingphenomena than as providing informationabout arousal. The more specific one is this:that the finding of decreased palmar sweat inthe presence of a self-attention-inducing stim-ulus (Paulus et al., 1978) may most reason-ably be interpreted in attentional terms, asreflecting inward focus of attention and thesimultaneous suppression of environmental in-put.

Cybernetics and the energizing of behavior.In addition to constraints imposed by thecomplexity of the meaning of autonomic re-sponses, another issue of logic should be ad-dressed. This issue concerns the nature ofcybernetic theory.

Among persons accustomed to think of be-havior as being "energized" and thus impelledby some kind of drive, the idea that a control-theory model of motivation is truly adequateto explain variations in the intensity of be-havior will seem difficult to accept. This istrue even though a number of motivationalpsychologists (e.g., Guilford, 1965; Hunt,1965; Taylor, 1960; Vickers, 1973) have fora number of years characterized the TOTE


unit as the most promising model of motivesleading to intentional activity.

The sticking point for many people seemsto be the obvious fact that it takes energy todo work. Thus a drive model would seem tobe preferable to a controMheory or a cyber-netic one. But this view fails to recognize thefact that physiological changes that occur inthe course of executing a behavior are in noway incompatible with cybernetic assump-tions. Cybernetic theory would simply relegatethe observed physiological change to a sec-ondary causal role. As Guilford (1965) hasnoted, "In automated devices, as in livingorganisms, there is a clear distinction betweentwo functions of energy. There is the energythat carries out the various operations, wherework must be done, and there is the energythat has control functions, where triggeringand release of much greater quantities ofenergy are involved" (p. 320). Cyberneticistswould tend to emphasize the importance ofthe energy that has control functions, eventhough it is the work-doing energy that ismost readily observed in terms of changes inphysiological state. Thus Wiener argued longago (1948) that the then-current focus onarousal was misguided, that the body is infact very far from being a conservative systemwith limited energy available. He argued thatthe appropriate type of "bookkeeping" forevents in the nervous system is not one thatis energy based. Instead, the critical notionswere those of message, quantity of informa-tion, coding technique, and so on.

It should be repeated, however, that thefact that physical work systems are subsidiarycomponents of the body's control system hasthis important implication: Covariation ofphysiological change with behavioral changein no way constitutes a threat to a control-theory analysis of behavior regulation. In suchan analysis, a change in physiological activityleads merely to the question of what specificfunctions are being performed, rather thanthe more misleading question of whetherarousal exists. Indeed, for this reason it isarguable (see also Carver & Scheier, Note 2)that control theory provides a more subtleunderstanding of physiological events thandoes drive theory.

Prepotent Response Tendency

There is a third distinction between theDuval and Wicklund (1972) model and thepresent model that should also be addressed.It bears on the following question. In a con-text in which a behavioral standard is salient,which response is prepotent when self-focusoccurs: matching-to-standard or withdrawal?In the model proposed here, the matching-to-standard sequence is prepotent, as part of thehuman being's normal self-regulatory system.Withdrawal occurs only after an outcomeassessment and only if that assessment yieldsa negative expectancy. In contrast, althoughDuval and Wicklund (1972) originally feltthat it was not possible to know which ofthese two modes would be preferred, Wick-lund later argued for an ordering opposite tothat assumed in the present model. "Certainlya successful averting of self-focused attentionwould eliminate the negative affect, howevertemporarily; thus an individual's immediatereaction to objective self-awareness should bean avoidance of self-focusing stimuli and/orefforts to find distractions" (Wicklund, 197Sa,p. 236). Thus, according to Wicklund, dis-crepancy reduction occurs only when self-focus cannot be avoided.

Results of two recent studies (Carver,Blaney, & Scheier, in press; McDonald, inpress) suggest that the present analysis is themore accurate of the two. In McDonald's re-search, subjects in whom a large negativewithin-self discrepancy had been made salientwere more persistent on a subsequent task(for which persistence had been defined asappropriate) when self-focus was increasedthan when it was not. Moreover, there wasalso evidence that this effect was exaggeratedby instructions that the second task wasrelevant to the dimension on which the dis-crepancy had been created. The possibilitythat subjects had simply been distractingthemselves from self-focus by burying them-selves in their task was considered and re-jected by McDonald, because of the fact thata much simpler and quicker means of avoidingself-focus was readily available but was notused, that is, announcing "I'm finished" andleaving the context. Thus McDonald con-


eluded that discrepancy reduction is prepotentunder self-focus when a means of reducing thediscrepancy is available.

The Carver et al. (in press) research wasdiscussed earlier in this article, in the sectionon outcome expectancy. Some subjects in thatresearch were led to have favorable expec-tancies of being able to reduce a large within-self discrepancy, others were led to have un-favorable expectancies. Subjects with favor-able expectancies were more persistent whenself-attention was high than when it was low,consistent with the McDonald (in press) re-sults. Carver et al. rejected the possibilitythat this persistence represented attempteddistraction for the same reason as had Mc-Donald, that is, that there was an easier wayto avoid self-focus readily available. Indeed,this point is further emphasized in the Carveret al. research by the fact that one group ofsubjects did utilize that way of avoiding self-focus—specifically, subjects with negative out-come expectancies. Thus those results appearto 'be unequivocal in supporting the presenttheory.

Other Theoretical Comparisons

The model proposed in this article alsobears certain similarities to models in socialpsychology other than self-awareness theory.Three of the similarities are striking enoughthat they deserve some separate attention.The first of these concerns Bandura's recent(1977) analysis of fear-based avoidance be-havior; the second concerns recent analysesof human helplessness (Abramson et al.,1978; Wortman & Brehm, 1975); the thirdconcerns social comparison theory (Festinger,19SO, 1954). These similarities will be dis-cussed in the following sections.

Self-Efficacy and Avoidance

The fact that the approach-withdrawal de-cision process postulated here has been framedin terms of outcome and efficacy expectanciessuggests that some commonality may exist be-tween the present model and Bandura's recentanalysis of the role of self-efficacy expec-tancies in avoidance behavior (Bandura,

1977; Bandura, Adams, & Beyer, 1977). Ban-dura's analysis and the present one do sharesome common ground, but they also differ inimportant respects.

Terminology. The first difference is one ofterminology (see Figure 4). Bandura used theterm outcome expectancy to refer to "a per-son's estimate that a given behavior will leadto certain outcomes" (Bandura, 1977, p.193). His primary concern was in distinguish-ing that belief from the expectancy that onecan execute the behavior. This he labeledefficacy expectancy. Clinically, such a distinc-tion is quite an important one. In many casespeople know what behavior a situation re-quires but do not believe themselves capableof doing the behavior. Because efficacy expec-tancy was most clinically relevant and thusmost important to Bandura, he devoted rela-tively little attention to defining and discuss-ing outcome expectancy. Implicitly, however,outcome expectancy in Bandura's terms isbased only on the presence or absence ofknowledge about the normal consequences ofa behavior. In the present model, in contrast,the term outcome expectancy is used to denoteexpectancy about an event's likelihood of oc-currence. This usage quite explicitly includesa variety of inputs: knowledge of the normalconsequences of task-appropriate behavior,constraints imposed by external forces or bythe passage of time, and efficacy expectancy—one's judgment about whether or not one canexecute the requisite behavior (Figure 4).

In the present model, therefore, outcomeexpectancy is the direct determinant of theperson's subsequent behavior and affect(Propositions 11 and 12). Efficacy expectancyis merely one input into outcome expectancy,but one determining what cognitions are asso-ciated with affect experienced as a function ofthat outcome expectancy. In Bandura's model,in contrast, the person's efficacy judgmentis the direct determinant of his or her subse-quent behavior. Outcome expectancy can beseen as a partial—though often trivial—de-terminant of efficacy expectancy in Bandura'smodel (Figure 4) in that one cannot feelcapable of doing a behavior unless one knowsthe behavior and what its ordinary conse-quences are.


KNOWLEDGE of BEHAVIOR'SCONSEQUENCES

OUTCOME EXPECTANCY

(Bandura)EFFICACY EXPECTANCY »• BEHAVIOR

(Carver)

KNOWLEDGE of BEHAVIOR'SCONSEQUENCES

•»• BEHAVIOR

Figure 4. Relationships among outcome expectancy, efficacy expectancy, and behavior, as con-ceptualized by Bandura (1977) and Carver.

The models are in somewhat closer agree-ment with regard to the nature and determi-nants of efficacy expectancy. Bandura con-ceived of efficacy expectancy as being aproduct of four classes of information: priorperformance accomplishments, vicarious 'ex-periences, verbal instructions, and emotionalarousal. An additional set of relevant infor-mation would seem to be the degree to whichone perceives that one can separate a behaviorinto components, each of which may be man-ageable by itself. These various sources of in-formation (involving both previously encodedstimuli and presently experienced stimuli) areprocessed together, yielding a judgment ofefficacy expectancy.

Role of emotional arousal. Even with re-gard to the determinants of efficacy expec-tancy, however, there is a point of conflict be-tween the models: specifically, the role that isassumed to be played by the perception ofemotional arousal. As was noted in the pre-ceding paragraph, Bandura (1977) held thatfear arousal was one kind of input into effi-cacy expectancy. Although this position hasbeen moderated to take into account attribu-tional contingencies (Bandura et al., 1977),there still appears to be a difference betweenBandura's position and the position taken inthe model proposed here. The present model

holds that the role of perceptions of feararousal in an approach attempt (initially, atleast) is to cue the self-assessment process(see Figure S). This assessment results in anoutcome expectancy judgment, which for fear-based behavior is largely determined by effi-cacy expectancy (which derives, in turn, fromthe other informational sources enumeratedby Bandura, 1977). This expectancy deter-mines whether the person subsequently with-draws or returns to the approach attempt.

This analysis has received empirical sup-port in three studies. Carver and Blaney(1977a, Experiment 3; 1977b) showed thatpersons who differed only in their chronicefficacy expectancies responded to false feed-back of arousal in two quite different ways.Confident subjects focused on their approachattempt and were not debilitated behavior-ally; doubtful subjects avoided focusing onthe approach attempt and withdrew morequickly, compared to doubtful subjects givenfeedback of nonarousal. This interactive influ-ence of fear perceptions on focus of attentionand behavior—a set of findings 'that has sub-sequently 'been conceptually replicated usinga self-awareness manipulation to heightenawareness of veridical fear (Carver et al.,1979)—does not seem easy to predict fromBandura's assumptions. The findings are, how-


ever, clearly predicted from the present modelof self-attention processes.

Generality. A final useful comparison be-tween the theories concerns their relativegenerality. Bandura's (1977) analysis wasaimed quite explicitly at fear-related behaviorand the therapeutic change of such behavior.In applying the two models to situations inwhich the likelihood of a positive outcome isentirely dependent on intrapersona\ factors,which is the case in fear-based behavior, thetwo are functionally equivalent. This is truebecause outcome and efficacy expectancies asdefined in the present model are perfectly con-founded with each other in this particularclass of situations. A consideration of broadercontexts reveals, however, that the presentmodel is more general in two respects than isBandura's. First, because outcome expectancyis not held to be determined solely by knowl-edge of the task-appropriate behavior, thepresent model can generate predictions for

two classes of behavioral interruptions notconsidered by Bandura: incomplete or inade-quate behavior that occurred in the past andbehavior constrained by environmental factors(see Figure 3). Secondly, since efficacy ex-pectancy is held to exert an influence on be-havior that is independent of outcome expec-tancy, the present model explicitly predictsthat different affective experiences will beassociated with similar behavioral events,based on differences in efficacy expectancy.That is, Proposition 13 above holds that ifthere is negative outcome expectancy, thepresence of high efficacy perceptions predictsthat affect will occur as resentment toward theenvironment; presence of low efficacy percep-tions predicts negative affect associated withthe self. If outcome expectancy is favorable,low efficacy perceptions predict that positiveaffect will be associated with the environment;high efficacy perceptions predict that theaffect will be associated with the self. These

ATTEMPTBEHAVIORALCONFORMITY

TO STANDARD

Figure 5. Information-processing sequence following self-attention in a fear-provoking behavioralcontext. (This sequence represents a specialized application of the model of behavioral regulationthat was presented in Figure 3.)


predictions, which are now being tested, donot seem to be derivable from Bandura's(1977) theory.

Helplessness Theory

The present theoretical model also appearsto have some implications for the area of re-search and theory known as "learned helpless-ness." Helplessness is a performance deficitthat reflects presumed motivational, cognitive,or learning deficits resulting from fairly ex-tended exposure to uncontrollable outcomes.There is at least one important similarity be-tween aspects of the present model and thekind of cognitive analysis developed in recentyears to account for helplessness effects amonghumans (e.g., Abramson et al., 1978; Wort-man & Brehm, 1975). Abramson et al.(1978), for example, have proposed that theimpact of uncontrollability on subsequent per-formance depends upon the development of anexpectation oj future noncontingency. Sim-ilarly, Wortman and Brehm (197S) haveargued that helplessness occurs when the per-son has an expectancy oj no control for a taskbeing undertaken. Both of these characteriza-tions are quite similar to what has beentermed in the present article an unfavorableoutcome expectancy regarding the subsequenttask.

Abramson et al. held that this expectancy isinfluenced largely by attributions concerningthe reason for the initial failure. In fact, thisattributional aspect of their model has re-ceived by far the greatest amount of attentionamong researchers in that area. Despite this,however, it bears emphasizing that the directdeterminant of performance in their theory, asin Wortman and Brehm's, is the person's ex-pectancy, not the attributions. If one has anunfavorable expectancy, regardless of the rea-son, helplessness should result. If one's ex-pectancy is favorable, there should be no help-lessness.

The central emphasis accorded to expec-tancy fcy these theorists is quite consistentwith the model presented in this article. In-deed, one study conducted to test the model(Carver et al., in press, discussed in detailearlier) could be regarded quite easily as a

helplessness experiment. Subjects underwenta failure pretreatment, and some of them laterdisplayed reduced persistence, which is oftentaken as a sign of helplessness. In providingsupport for the present model, that researchthus also appears to offer support for othermodels of helplessness that rely on postulatesabout expectancies.

However, that research also suggests tworespects in which the present model adds tosuch theories of helplessness. The more ob-vious contribution of this model stems fromthe fact that no other analysis of helplessnesseffects includes any consideration of 'the roleof self-directed attention in promoting sucheffects. As is demonstrated in the Carver et al.(in press) research, that role may be quite animportant one.

The second contribution of the presenttheory is its suggestion that the impulse towithdraw is basic to a wide variety of help-lessness effects. The assumption of a with-drawal impulse is inherent in the presentmodel. And it was overt withdrawal that wasdisplayed in the Carver et al. (in press) re-search. In contrast, most studies in the help-lessness tradition do not explicitly allow sub-jects this behavioral option. As was suggestedearlier in this article, however, when physicalwithdrawal is prevented, the result may be acognitive withdrawal—a mental dissociationfrom task attempts. This characterizationseems consistent with typical helplessness ef-fects, which often reflect an apparent unwill-ingness or inability to utilize task-relevantcues. An interesting possibility that should beexamined further is that all these effects maystem from a thwarted impulse to remove one-self from the behavioral context.

Social Comparison Theory

Earlier in this article it was noted thatthere is also considerable similarity betweensome aspects of the present theory and socialcomparison theory (Festinger, 1950, 1954).18

Social comparison theory can be viewed as

18 Although this similarity exists for Duval andWicklund's theory, as well as for the present model,this discussion will focus on the present model.


having three central assumptions. The first isthat we tend to define much of reality—espe-cially our social reality—by comparing ourown opinions, reactions, characteristics, andcapabilities with those of other people, usuallytermed reference groups. The second assump-tion is that the tendency to do this is exag-gerated by conditions of ambiguity. The thirdassumption is that once consensus has beenreached and some normative value implicitlyestablished, then there is pressure towardconformity to this value among the membersof the group.

Holding in abeyance the role of perceivedambiguity in the instigation of these pro-cesses, the elements of social comparisontheory thus seem to fulfill two functions: first,to establish some value as a standard of com-parison (by means of an implicit social con-sensus), and second, to increase behavioralconformity to that standard. These two func-tions are, of course, precisely the same asthose assumed in a control-theory model ofbehavioral self-regulation such as the one pro-posed here. That is, a control-theory approachto motivation assumes two kinds of informa-tion-processing systems. The first of these isthe system that analyzes and categorizes per-ceptual input, yielding a behavioral standard.The second system—a TOTE unit—regulatesbehavior with regard to that standard.

One way in which social comparison theoryis similar to the model proposed here concernsthe first stage of social comparison, in whicha standard is established by implicit socialconsensus. This aspect of that model wouldseem to represent one important way in whicha person can extract a behavioral standardthrough categorization of his or her environ-mental context (Propositions 2 and 3 of thepresent model). This may indeed be one of themost commonly used methods of choosing abehavior among human adults. In principle,however, it represents only a subset of alarger class of potential ways to determine be-havior. Thus, it would seem that this facetof social comparison theory could be sub-sumed under the more general process ofcategorizing the nature of one's context (aprocess that certainly deserves much more

attention than it has received in the presentarticle) ,19

The second stage of the social comparisonprocess, in which conformity pressure occurs,is also functionally similar to a major facet ofthe present model. The difference between thetwo is in the locus assumed for the impetusto conform. In the present model, the con-formity tendency (via the normal matching-to-standard sequence) is construed as inter-nally based. Discussions of social comparisontheory, in contrast, often seem to assume theexistence of external pressure to conform.

Finally, it seems worthy of 'brief note thata subsidiary aspect of social comparisontheory holds that if a person cannot conformto the group's standard, there is a tendencyfor either the group or the person to withdrawfrom the other. This seems not unlike thepresent proposition that when one cannotmatch one's behavior to the standard—that is,when outcome expectancy is unfavorable—there is an impulse to withdraw from the at-tempt.

Summary and Concluding Comment

In the preceding pages, a cybernetic modelof self-attention processes was presented,along with support for its propositions. Sum-marized briefly, self-focus in a context whereno behavioral standard is salient leads toheightened cognizance of salient self-elements.Self-focus when a behavioral standard doesexist leads to an automatic matching-to-stan-dard sequence. This sequence is regarded asthe realization within a psychological systemof a negative feedback loop. If matching-to-standard is interrupted, an outcome assess-ment ensues. A favorable outcome perception

19 The assumption in social comparison theory thatambiguity increases the tendency to undertake socialcomparison suggests another possibility not pre-viously addressed in this article. That is, it may bea more general rule that ambiguity leads to increasedinformation search, even when the context is non-social, in order to determine better what behavior isappropriate. This possibility certainly should receivemore attention. If confirmed, it would further in-crease the degree to which these two analyses over-lap.


leads to positive affect and/or to a return tomatching-to-standard. An unfavorable out-come perception leads to negative affect and/or behavioral withdrawal.

Following presentation of the model, threespecific points of conflict between this theoryand self-awareness theory were addressed. Ineach of those cases, evidence was reviewedthat indicated the present analysis to be themore internally consistent and parsimonious ofthe two. Comparisons also were undertakenbetween the present model and three othertheories: Bandura's analysis of cognitive pro-cesses underlying fear-related behavior, help-lessness theory, and social comparison theory.Each of those comparisons revealed the pos-sibility for considerable integration betweenmodels.

There are many other potential links thatcould be developed between the present theoryand information-processing ideas currentlyunder investigation elsewhere in social andcognitive psychology. A fairly obvious ex-ample is the similarity between the proposi-tions of the present model that concern sa-lience of oneself versus the environment andresearch conducted on other kinds of saliencephenomena (see, e.g., Taylor & Fiske, 1978).Although there simply is not the space to pur-sue those interrelationships here, 'the existenceof such commonalities among theories suggeststhat the ideas on which the present model isbased have a good deal of integrative poten-tial. Indeed, this seems apparent even if oneconsiders only the material discussed in thispaper. The model of self-attention processespresented here has proven to be applicable toa wide range of phenomena, topics as diverseas taste and test anxiety, in a way that is in-ternally consistent. It is to be hoped, more-over, that this perspective will prove to beuseful in generating additional hypotheses,beyond those considered in the precedingpages. Thus might the specific ideas advancedhere, and the concepts of control theory moregenerally, come to be seen as important toolsfor the analysis of a broad range of long-standing problems in human behavior.

Reference Notes

1. Slapion, M., & Carver, C. S. Self-directed attentionand facilitation of intellectual performance among

persons high in test anxiety. Manuscript submittedfor publication, 1979.

2. Carver, C. S., & Scheier, M. F. The self-attention-induced feedback loop and human motivation: Acontrol-systems analysis of social facilitation. Man-uscript submitted for publication, 1979.

References

Abramson, L. Y., Seligman, M.-E. P., & Teasdale,J. D. Learned helplessness in humans: Critique andreformulation. Journal of Abnormal Psychology,1978, 87, 49-74.

Appley, M. H. Derived motives. In P. H. Mussen &M. R. Rosenzweig (Eds.), Annual review of psy-chology (Vol. 21). Palo Alto, Calif.: Annual Re-views, 1970.

Apter, M. J. The computer simulation of behavior.New York: Harper Colophon, 1970.

Bandura, A. Self-efficacy: Toward a unifying theoryof behavioral change. Psychological Review, 1977,84, 191-215.

Bandura, A., Adams, N. E., & Beyer, J. Cognitiveprocesses mediating behavioral change. Journal ofPersonality and Social Psychology, 1977, 35, 125-139.

Barker, R. G., Ecological psychology: Concepts andmethods for studying the environment of humanbehavior. Palo Alto, Calif.: Stanford UniversityPress, 1968.

Berlyne, D. E. Motivational problems raised by ex-ploratory and epistemic behavior. In S. Koch(Ed.), Psychology: A study of a science (Vol. 5).New York: McGraw-Hill, 1963.

Borkovec, T. D., & O'Brien, G. T. Relation of auto-nomic perception and its manipulation to the main-tenance and reduction of fear. Journal of AbnormalPsychology, 1977, 86, 163-171.

Brehm, J. W. A theory of psychological reactance.New York: Academic Press, 1966.

Brehm, J. W., & Brehm, M. L., In J. W. Brehm(Ed.), A theory of psychological reactance. NewYork: Academic Press, 1966.

Buckley, W. Modern systems research for the be-havioral scientist. Chicago: Aldine, 1968.

Buss, D. M., & Scheier, M. F. Self-awareness, self-consciousness, and self-attribution. Journal of Re-search in Personality, 1976,10, 463-468.

Cantor, N., & Mischel, W. Traits as prototypes: Ef-fects on recognition memory. Journal of Personal-ity and Social Psychology, 1977, 35, 38-48.

Carroll, J. S. The effect of imagining an event on ex-pectations for the event: An interpretation in termsof the availability heuristic. Journal of Experi-mental Social Psychology, 1978,14, 88-96.

Carver, C. S. Facilitation of physical aggressionthrough objective self-awareness. Journal of Ex-perimental Social Psychology, 1974, 10, 365-370.

Carver, C. S. Physical aggression as a function ofobjective self-awareness and attitudes toward pun-ishment. Journal of Experimental Social Psychol-ogy, 1975, 11, 510-519.

Carver, C. S. Self-awareness, perception of threat, and


the expression of reactance through attitudechange. Journal of Personality, 1977, 45, 501-512.

Carver, C. S., & Blaney, P. H. Avoidance behaviorand perceived arousal. Motivation and Emotion,1977,1, 61-73. (a)

Carver, C. S., & Blaney, P. H. Perceived arousal,focus of attention, and avoidance behavior. Journalof Abnormal Psychology, 1977, 86, 154-162. (b)

Carver, C. S., Blaney, P. H., & Scheier, M. F. Focusof attention, chronic expectancy, and responses toa feared stimulus. Journal of Personality and SocialPsychology, 1979, 37, 1186-1195.

Carver, C. S., Blaney, P. H., & Scheier, M. F. Re-assertion and giving up: The interactive role ofself-directed attention and outcome expectancy.Journal of Personality and Social Psychology, inpress.

Carver, C. S., & Scheier, M. F. Self-focusing effectsof dispositional self-consciousness, mirror presence,and audience presence. Journal of Personality andSocial Psychology, 1978, 36, 324-332.

Carver, C. S., & Scheier, M. F. Self-consciousness andreactance. Journal of Research in Personality, inpress.

Dabbs, J. M., Jr., Johnson, J. E., & Leventhal, H.Palmar sweating: A quick and simple measure.Journal of Experimental Psychology, 1968, 78, 347-350.

Davis, D., & Brock, T. C. Use of first person pro-nouns as a function of increased objective self-awareness and prior feedback. Journal of Experi-mental Social Psychology, 1975, 11, 381-388.

Diggory, J. C. Self-evaluation: Concepts and studies.New York: Wiley, 1966.

Duval, S., & Wicklund, R. A. A theory of objectiveself-awareness. New York: Academic Press, 1972.

Duval, S., & Wicklund, R. A. Effects of objective self-awareness on attribution of causality. Journal ofExperimental Social Psychology, 1973, 9, 17-31.

Duval, S., Wicklund, R. A., & Fine, R. L. Avoidanceof objective self-awareness under conditions of highand low intra-self discrepancy. In S. Duval & R. A.Wicklund (Eds.), A theory of objective self-aware-ness. New York: Academic Press, 1972.

Exner, J. E., Jr. The self-focus sentence completion:A study of egocentricity. Journal of PersonalityAssessment, 1973, 37, 437-455.

Festinger, L. Informal social communication. Psycho-logical Review, 1950, 57, 271-282.

Festinger, L. A theory of social comparison processes.Human Relations, 1954, 7, 114-140.

Franks, J. J., & Bransford, J. D. Abstraction of visualpatterns. Journal of Experimental Psychology, 1971,90, 65-74.

Fredericksen, N. Toward a taxonomy of situations.In P. M. Insel & R. H. Moos (Eds.), Issues in so-cial ecology: Human milieus. Palo Alto, Calif.: Na-tional Press Books, 1974.

Froming, W. J. The relationship of moral judgment,self-awareness, and sex to compliance behavior.Journal of Research in Personality, in press.

Geen, R. G., & Gange, J. J. Drive theory of socialfacilitation: Twelve years of theory and research.Psychological Bulletin, 1977, 84, 1267-1288.

Geller, V., & Shaver, P. Cognitive consequences ofself-awareness. Journal of Experimental Social Psy-chology, 1976, 12, 99-108.

Gibbons, F. X. Sexual standards and reactions topornography: Enhancing behavioral consistencythrough self-focused attention. Journal of Personal-ity and Social Psychology, 1978, 36, 976-987.

Gibbons, F. X., Carver, C. S., Scheier, M. F., & Hor-muth, S. Self-focused attention and the placeboeffect: Fooling some of the people some of thetime. Journal of Experimental Social Psychology,1979, 15, 263-274.

Guilford, J. P. Motivation in an informational psy-chology. In D. Levine (Ed.), Nebraska Symposiumon Motivation (Vol. 20). Lincoln: University ofNebraska Press, 1965.

Gump, P. V. The behavior setting: A promising unitfor environmental designers. Landscape Architec-ture, 1971, 61, 130-134.

Hare, R. D., & Blevings, G. Conditioned orientingand defensive responses. Psychophysiology, 1975,12, 289-297.

Hull, J. G., & Levy, A. S. The organizational func-tions of the self: An alternative to the Duval andWicklund model of self-awareness. Journal of Per-sonality and Social Psychology, 1979, 37, 756-768.

Hunt, J. McV. Intrinsic motivation and its role inpsychological development. In D. Levine (Ed.),Nebraska Symposium on Motivation (Vol. 20).Lincoln: University of Nebraska Press, 1965.

Ickes, W. J., Wicklund, R. A., & Ferris, C. B. Objec-tive self-awareness and self-esteem. Journal of Ex-perimental Social Psychology, 1973, 9, 202-219.

Johnson, J. E., & Dabbs, J. M., Jr. Enumeration ofactive sweat glands: A simple physiological indica-tor of psychological changes. Nursing Research,1967, 16, 273-276.

Kuhn, A. The logic of social systems. San Francisco:Jossey-Bass, 1974.

Kuiper, N. A., & Rogers, T. B. Encoding of personalinformation: Self-other differences. Journal of Per-sonality and Social Psychology, 1979, 37, 499-514.

Lacey, J. I. Somatic response patterning and stress:Some revisions of activation theory. In M. H. Ap-pley & R. Trumbull (Eds.), Psychological stress.New York: Appleton-Century-Crofts, 1967.

Leventhal, H. Findings and theory in the study offear communications. In L. Berkowitz (Ed.), Ad-vances in experimental social psychology. (Vol. 5).New York: Academic Press, 1970.

Lewin, K. A dynamic theory of personality. NewYork: McGraw-Hill, 1935.

Liebling, B. A., Seller, M., & Shaver, P. Self-aware-ness and cigarette-smoking behavior. Journal ofExperimental Social Psychology, 1974, 10, 325-332.

MacKay, D. Mindlike behavior in artefacts. In K. M.Sayre & F. J. Crosson (Eds.), The modeling ofmind: Computers and intelligence. Notre Dame,Ind.: University of Notre Dame Press, 1963.

Markus, H. Self-schemata and processing informationabout the self. Journal of Personality and SocialPsychology, 1977, 35, 63-78.

Martens, R. Palmar sweating and the presence of an


audience. Journal of Experimental Social Psychol-ogy, 1969, S, 371-374. (a)

Martens, R. Audience effects on learning and per-formance. Journal of Personality and Social Psy-chology, 1969, 12, 252-260. (b)

McDonald, P. J. Reactions to objective self-aware-ness. Journal of Research in Personality, in press.

Miller, G. A., Galanter, E., & Pribram, K. H. Plansand the structure of behavior. New York: Holt,Rinehart & Winston, 1960.

Neisser, U. Cognition and reality. San Francisco:Freeman, 1976.

Neumann, P. G. An attribute frequency model for theabstraction of prototypes. Memory & Cognition,1974, 2, 241-248.

Newcomb, T. M. Attitude development as a functionof reference groups: The Bennington study. InE. E. Maccoby, T. M. Newcomb, & E. L. Hartley(Eds.), Readings in social psychology. NewYork: Holt, Rinehart & Winston, 19S8.

Paulus, P. B., Annis, A. B., & Risner, H. T. Ananalysis of the mirror-induced self-awareness effect.Bulletin of the Psychonomic Society, 1978,12, 8-10.

Posner, M. I. Abstraction and the process of recogni-tion. In G. H. Bower & J. T. Spence (Eds.), Thepsychology of learning and motivation (Vol. 3).New York: Academic Press, 1969.

Posner, M. I., & Rogers, M. G. K. Chronometricanalysis of abstraction and recognition. In W. K.Estes (Ed.), Handbook of learning and cognitiveprocesses (Vol. 5). Hillsdale, N.J.: Erlbaum, 1978.

Powers, W. T. Behavior: The control of perception.Chicago: Aldine, 1973.

Powers, W. T. Quantitative analysis of purposivesystems: Some spadework at the foundations ofscientific psychology. Psychological Review, 1978,85, 417-435.

Pryor, J. B., Gibbons, F. X., Wicklund, R. A., Fazio,R. H., & Hood, R. Self-focused attention and self-report validity. Journal of Personality, 1977, 45,513-527.

Reitman, J. S., & Bower, G. H. Storage and laterrecognition of exemplars of concepts. CognitivePsychology, 1973, 4, 194-206.

Rogers, T. B., Kuiper, N. A., & Kirker, W. S. Self-reference and the encoding of personal information.Journal of Personality and Social Psychology, 1977,35, 677-688.

Rogers, T. B., Rogers, P. J., & Kuiper, N. A. Evidencefor the self as a cognitive prototype: The "falsealarms effect." Personality and Social PsychologyBulletin, 1979, 5, S3-S6.

Schank, R., & Abelson, R. Scripts, plans, goals andunderstanding. New York: Halsted, 1977.

Scheier, M. F. Self-awareness, self-consciousness, andangry aggression. Journal of Personality, 1976, 44,627-644.

Scheier, M. F., Buss, A. H., & Buss, D. M. Self-con-sciousness, self-report of aggressiveness, and aggres-sion. Journal of Research in Personality, 1978, 12,133-140.

Scheier, M. F., & Carver, C. S. Self-focused attentionand the experience of emotion: Attraction, repul-

sion, elation, and depression. Journal of Personalityand Social Psychology, 1977, 35, 624-636.

Scheier, M. F., Carver, C. S., & Gibbons, F. X. Self-directed attention, awareness of bodily states, andsuggestibility. Journal of Personality and SocialPsychology, in press, (a)

Scheier, M. F., Carver, C. S., & Gibbons, F. X. Self-focused attention and reactions to fear. Journal ofResearch in Personality, in press, (b)

Scheier, M. F., Carver, C. S., Schulz, R., Glass, D. C.,& Katz, I. Sympathy, self-consciousness, and reac-tions to the stigmatized. Journal of Applied SocialPsychology, 1978, 8, 270-282.

Scheier, M. F., Fenigstein, A., & Buss, A. H. Self-awareness and physical aggression. Journal of Ex-perimental Social Psychology, 1974, 10, 264-273.

Simonov, P. V. The informational theory of emotion.In M. Arnold (Ed.), Feelings and emotions: TheLoyola Symposium. New York: Academic Press,1970.

Singh, J. Great ideas in information theory, language,and cybernetics. New York: Dover, 1966.

Snyder, M. L., & Wicklund, R. A. Prior exercise offreedom and reactance. Journal of ExperimentalSocial Psychology, 1976, 12, 120-130.

Sokolov, Y. N. Perception and the conditioned reflex.New York: Macmillan, 1963.

Steenbarger, B. N., & Aderman, D. Objective self-awareness as a non-aversive state: Effect of an-ticipating discrepancy reduction. Journal of Per-sonality, in press.

Stotland, E. The psychology of hope. San Francisco:Jossey-Bass, 1969.

Taylor, D. W. Toward an information-processingtheory of motivation. In M. R. Jones (Ed.),Nebraska Symposium on Motivation (Vol. 8). Lin-coln: University of Nebraska Press, 1960.

Taylor, S. E., & Fiske, S. T. Salience, attention, andattribution: Top of the head phenomena. In L.Berkowitz (Ed.), Advances in experimental socialPsychology (Vol. 11). New York: Academic Press,1978.

Turing, A. M. Computing machinery and intelligence.Mind, 1950, 59, 433-460.

Vallacher, R. Objective self-awareness and the per-ception of others. Personality and Social PsychologyBulletin, 1978, 4, 63-67.

Vickers, G. Motivation theory—a cybernetic contribu-tion. Behavioral Science, 1973, 18, 242-249.

Wicklund, R. A. Freedom and reactance. Potomac,Md.: Erlbaum, 1974.

Wicklund, R. A. Objective self-awareness. In L.Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 8). New York: Academic Press,1975. (a)

Wicklund, R. A. Discrepancy reduction or attempteddistraction ? A reply to Liebling, Seiler, and Shaver.Journal of Experimental Social Psychology, 1975,11, 78-81. (b)

Wicklund, R. A., & Duval, S. Opinion change andperformance facilitation as a result of objectiveself-awareness. Journal of Experimental Social Psy-chology, 1971, 1, 319-342.


Wiener, N. Cybernetics: Control and communicationin the animal and the machine. Cambridge, Mass.:M.I.T. Press, 1948.

Wiener, N. The human use of human beings: Cy-bernetics and society. New York: Avon Books,1954.

Williams, R. B., Bittker, T. F., Buchsbaum, M. S., &Wynne, L. C. Cardiovascular and neurophysiologiccorrelates of sensory intake and rejection. I: Effectof cognitive tasks. Psychophysiology, 1975, 12, 427-433.

Wine, J. D. Test anxiety and direction of attention.Psychological Bulletin, 1971, 76, 92-104.

Wortman, C. B., & Brehm, J. W. Responses to un-controllable outcomes: An integration of reactancetheory and the learned helplessness model. In L.Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 8). New York: Academic Press,197S.

Received August 10, 1978 •

personalit jou rn al social psychology

Documents