Psychological Review 1973, Vol. 80, No. 4, 252-283 TOWARD A COGNITIVE SOCIAL LEARNING RECONCEPTUALIZATION OF PERSONALITY J

WALTER MISCHEL 2 Stanford University

Diverse data challenge and undermine ihe central assumptions of the tradi tional trait approach to personality. The implications for conceptions of individual differences and situations in the study of personality are examined. The issues discussed include the nature of behavioral "specificity," the acquired meaning of stimuli, the uses and misuses of traits, and the construction of personality. To move toward a more adequate theoretical approach to per- sons, the following cognitive social learning variables are proposed as basic units for the study of individuals: cognitive and behavioral construction com- petencies, encoding strategies and personal constructs, behavior-outcome and stimulus-outcome expectancies, subjective stimulus values, and self- regulatory systems and plans. The specific interactions between these person variables and psychological situations are analyzed within the framework of a cognitive social learning approach.

There has been a curious—indeed alarm- laboratory and in therapeutic applications, ing—bifurcation between progress in theo- their implications for personality psychol- ries regarding complex social behavior and ogy have not yet been thoroughly explored cognition on the one hand, and in con- and their impact on the basic traditional ceptualizations regarding the basic nature assumptions of personality psychology until of personality on the other. Many of the recently has been limited. therapeutic implications of social learning (social behavior) theories have become evi- During the last 50 years, when basic dent in the last few years. There have concepts were changing rapidly in most been notable advances in treatment tech- fields of psychology, the most fundamental niques as well as significant reconceptuali- assumptions about the nature of personality zations of the treatment process itself (e.g., seem to have been retained with few sub- Bandura, 1969). These developments are stantial modifications. Of course there just starting to be accompanied by com- have been many changes in the names and parable parallel developments in person- particular characteristics of the trait dis- ality theory. In a second direction, there positions advocated by different theore- has been vigorous progress in cognitive ticians and personality researchers in the psychology (e.g., Neisser, 1967). But last few decades. But in spite of the hetero- while cognitive and symbolic processes have geneity of hypothesized dimensions or received increasing attention both in the structures, perhaps the most fundamental assumptions about them have remained 1 Parts of this manuscript are based on the Address almost monolithic until very recently. This of the Chairman, Section III, Division 12, American paper briefly reviews the central assump- Psychological Association, Washington, D. C., September 3, 1971. Preparation of this paper was fa- tions of global dispositional approaches to cilitated by National Institute of Mental Health personality, considers some of the main Grant M-6830 and National Science Foundation misconceptions, issues, and implications Grant GS-32582. Constructive comments have been received from more colleagues and students than arising from recent challenges to those can be listed here; the author is grateful for their assumptions, and finally attempts a re- help. conceptualization of person variables in the 2 Requests for reprints should be sent to Walter Mischel, Department of Psychology, Stanford Uni- light of concepts from the study of cogni- versity, Stanford, California 94305. tion and social learning. 252 RECONCEPTUALIZATION OF PERSONALITY 253

GLOBAL DISPOSITIONAL APPROACHES cannot be summarized adequately here, TO PERSONALITY but several themes emerge. To recapitu- late briefly, impressive consistencies often Assumptions of Traditional Trait have been found for intellective features of Approaches personality and for behavior patterns such It has generally been assumed that as cognitive styles and problem-solving personality dispositions or traits—the basic strategies that are strongly correlated with units of personality study—are relatively intelligence (e.g., Witkin, 1965). Con- stable, highly consistent attributes that sistency also is often high when people exert widely generalized causal effects on rate their own traits, as in questionnaires behavior. Whether one uses the language and other self-reports (e.g., E. L. Kelly, of factors, or of habits, or of basic atti- 1955). Temporal continuity also has been tudes, or of dynamics and character struc- demonstrated often when the individual's ture, this fundamental assumption has been behavior is sampled at different time shared: personality comprises broad under- periods but in similar situations. When lying dispositions which pervasively in- one goes beyond cognitive variables to fluence the individual's behavior across personality dimensions and when one many situations and lead to consistency samples personality by diverse methods and in his behavior (e.g., Allport, 1937).3 These not just by self-report questionnaires, the dispositions are not directly observed but data change and undermine the utility of are inferred from behavioral signs (trait inferring global personality dispositions indicators), either directly or indirectly from behavioral signs, as has been docu- (Mischel, 1968). Guided by this assump- mented in detail (Mischel, 1968): tion, personality research has been a quest Response patterns even in highly similar situations for such underlying broad dimensions, for often fail to be strongly related. Individuals show basic factors, or for pervasive motives, or far less cross-situational consistency in their be- for characteristic life styles. In personality havior than has been assumed by trait-state theories. assessment the trait assumptions regarding The more dissimilar the evoking situations, the less structure are seen in the existence of hun- likely they are to produce similar or consistent responses from the same individual. Even seem- dreds of tests designed to infer dispositions ingly trivial situational differences may reduce cor- and almost none to measure situations. relations to zero. Response consistency tends to be The same belief in global traits that mani- greatest within the same response medium, within fest themselves pervasively is perhaps best self-reports to paper-and-pencil tests, for example, or within directly observed non-verbal behavior. seen in the projective test assumption that Intra-individual consistency is reduced drastically responses to vague or minimal stimuli will when dissimilar response modes are employed. reveal individual differences in fundamental Activities that are substantially associated with generalized dispositions (MacFarlane & aspects of intelligence and with problem solving Tuddenham, 1951). behavior—like achievement behaviors, cognitive styles, response speed—tend to be most consistent [p. 177]. Empirical Status of Assumptions Given the pervasiveness of the con- Psychodynamic Approach to Consistency sistency assumption of dispositional per- sonality theory, its empirical status be- Recognizing both the specificity and com- comes especially important. There have plexity of behavior, psychodynamic theo- been several recent reviews of that evi- rists long ago rejected the idea of broad dence (e.g., Mischel, 1968, 1969, 1971; overt behavioral consistencies across situa- Peterson, 1968; Vernon, 1964). The data tions. Instead, psychodynamic theories

3 emphasize that behavior varies, but diverse In social psychology, the "attitude" has been behavioral patterns serve the same endur- the unit endowed with properties parallel to those assigned to the trait in the field of personality, and it ing and generalized underlying dynamic or appears to be subject to very similar criticisms and motivational dispositions. The search for problems (e.g., Abelson, 1972). dispositions thus rests on a distinction be- 254 WALTER MISCHEL tvveen surface behaviors ("signs" or ments, even when the judges are expert "symptoms") and the motives that they psychodynamicists working in clinical con- serve. This involves the familiar distinc- texts and using their favorite techniques. tion between the "phenotypic" and the Reviews of the relevant research generally "genotypic" and entails an indirect, rather show that clinicians guided by concepts than a direct measurement model (Mischel, about underlying genotypic dispositions 1968). Indeed, the most common argument have not been able to predict behavior for personality consistency in the face of better than have the person's own direct seeming behavioral specificity is the dis- self-report, simple indices of directly rele- tinction between the phenotypic and the vant past behavior, or demographic vari- genotypic. Granted that overt behavior bles (e.g., Mischel, 1968, 1971, 1972). is not highly consistent, might it not be useful to posit genotypic personality dis- MISCONCEPTIONS AND ISSUES positions that endure, although their overt The findings on the specificity-con- response forms may change? This geno- sistency of personality traits and the im- typic-phenotypic model has been at the plications of social behavior theory for the crux of dynamic dispositional theories of psychology of personality may be leading personality (Mischel, 1969). The psycho- to a paradigm crisis in the field (e.g., dynamic model construes behaviors as Fiske, 1973), and hence it is not surprising highly indirect signs of the dispositions that that they are easily misunderstood. These underlie them, because defenses are hy- misunderstandings are evident in repeated pothesized to distort and disguise the true critiques (e.g., Adelson, 1969; Adinolfi, meaning of the observed behaviors. If 1971; Alker, 1972; Craik, 1969; Dahlstrom, basic motives express themselves only in- 1970; Wachtel, 1973) aimed at applica- directly after being distorted by defensive tions of social behavior theory to the do- maneuvers, then their overt behavioral main of personality (e.g., Mischel, 1968, manifestations have to be interpreted sym- 1969) and particularly to the issue of the bolically as indirect signs. Thus, for ex- specificity-generality of behavior. The ample, using the white space of an inkblot thrust of these reactions is that social be- in a percept may be taken as a sign of havior theory, especially in its emphasis on negativistic tendencies, or saying the ink- the discriminativeness ("specificity") of be- blot looks like blood may be interpreted as havior, implies a "personalityless" view of a sign of a psychopathic personality. The man. psychodynamic approach thus shares with the trait approach a disinterest in behaviors Common Misconceptions except as they serve as signs—albeit more indirect signs—of generalized dispositions. The position developed in Mischel's While inherently logical, the utility of (1968) Personality and Assessment has been the indirect sign approacli to dispositions widely misunderstood to imply that people depends on the value of the inferences show no consistencies, that individual dif- provided by the clinical judge. Con- ferences are unimportant, and that "situa- sequently, the reliability and validity of tions" are the main determinants of be- clinicians' judgments become crucial. The havior (e.g., Bowers, 1972). For example, extensive empirical studies on this issue Alker (1972) has thoroughly distorted the have investigated in detail the value of basic issues (as Bern, 1972, has shown), clinicians' efforts to infer broad disposi- guarding the traditional personality para- tions indirectly from specific symptomatic digm against evidence that "behavior varies signs and to unravel disguises in order to from situation to situation." But the fact uncover the motivational dispositions that that behavior varies across different situa- might be their roots. As is now generally tions is not questioned by anyone, includ- recognized, the accumulated findings give ing classical trait theorists. More serious little support for the utility of clinical judg- issues, instead, are the consistency-speci- RECONCEPTUALIZATION OF PERSONALITY 255 ficity with which the same person reacts to misconceptions regarding the status of situations that ostensibly are relatively global traits. similar (i.e., that are selected to evoke the same trait), and most important, the utility Moderator Variables and Person-Situation of predictions based on global trait infer- Interactions ences (Mischel, 1968). In the same vein, Several recent trait studies have in- Wachtel (1973) defended psychodynamic vestigated the relative separate quantita- theory against being forever consigned to tive contributions of persons and situations a "scientific Valhalla" by emphasizing that as well as the variance accounted for by psychodynamic theories in fact recognize the interaction of the individual and the people's responsiveness to variations in environment (e.g., Argyle & Little, 1972; stimulus conditions. Unfortunately, he Endler & Hunt, 1966, 1968, 1969; Endler, ignored the data and challenges that are Hunt, & Rosenstein, 1962; Moos, 1968, relevant, most notably the failure of the 1969). The essential method consists of psychodynamically oriented clinician to sampling the behavior of individuals (by demonstrate the utility of the indirect sign questionnaire and/or by observation) across approach when compared to more parsi- a series of situations and through various monious alternatives (Mischel, 1968, 1972, response modes. On the whole, these 1973b). studies have indicated that the sampled Evidence for the lack of utility of in- individual differences, situations, and re- ferring hypothesized global trait disposi- sponse modes when considered separately tions from behavioral signs should not be tend to account for less variance than misread as an argument for the greater im- does their interaction. portance of situations than persons (Bow- The overall results suggest, as Endler ers, 1972). Is information about in- and Hunt (1969, p. 20) noted with regard dividuals more important than information to their own findings for anxiety, that about situations? The author has per- behavior "is idiosyncratically organized in sistently refrained from posing this ques- each individual. ..." A similar conclu- tion because phrased that way it is un- sion emerges from Moos's (1968) studies answerable and can serve only to stimulate of self-reported reactions by staff and futile polemics. Moreover, in current de- patients to various settings. Consider, for bates on this topic, "situations" are often example, his obtained interactions between erroneously invoked as entities that sup- persons and nine settings with regard to posedly exert either major or only minor "sociable, friendly, peaceful" versus "un- control over behavior, without specifying sociable, hostile, angry" behavior. The what, psychologically, they are or how they results revealed that although different function (Alker, 1972; Bowers, 1972; Wal- individuals reacted differently to the set- lach & Leggett, 1972). But while some tings, a given person might be high on the situations may be powerful determinants dimension in the morning but not at of behavior, others are likely to be ex- lunch, high with another patient but not ceedingly trivial. The relative importance when with a nurse, low in small group of individual differences will depend on the therapy, moderate in industrial therapy, situation selected, the type of behavior but high iti individual therapy, etc. An assessed, the particular individual dif- entirely different pattern might characterize ferences sampled, and the purpose of the the next person. These results and inter- assessment. In later sections, an attempt pretations are totally congruent with the will be made to consider in detail how conclusions emerging from earlier reviews cognitive social learning person variables that emphasize the idiosyncratic organiza- interact with conditions and how "situa- tion of behavior within individuals tions" function psychologically. But first (Mischel, 1968, p. 190). it is necessary to review further, and hope- It would be wasteful to create pseudo- fully to clarify, some of the main issues and controversies that pit person against situa- 256 WALTER MISCHEL

tion in order to see which is more important. his immediately prior experience—the list The answer must always depend on the gets almost endless (Mischel, 1973a). This particular situations and persons sampled; seems to be another way of saying in the presumably, studies could be designed to language of moderator variables and inter- demonstrate almost any outcome. The action terms that what a person does tends interaction studies correctly demonstrated to be relatively specific to a host of vari- that the question of whether individual ables, and that behavior is multiply de- differences or persons are more important termined by all of them rather than being is a fruitless one that has no general the product of widely generalized disposi- answer. The views of Moos (1972, personal tions. Some psychologists may find these communication) regarding the limits of the interpretations more palatable if they are kinds of interaction studies that he and not phrased as reflecting the specificity of Endler and Hunt pioneered seem extremely the acquired meanings of stimuli and the sensible. Moos recognized that these resulting specificity of behavior patterns studies can be designed so that: (Mischel, 1968). Instead, they may prefer to construe the data as highlighting the any result is possible. I think that all one can say is that given relatively real life situations (e.g., uniqueness and complexity of personality. patients on wards or in outpatient psychotherapy, To say that what a person thinks, and does, or your delay of gratification studies) that the major and feels—and hence what he is at any proportion of the variance simply does not appear to moment—depends on many subject and be accounted for by individual difference variables. One could certainly, however, easily design studies condition variables is also to underline in which the major portion of the variance would the complexity and uniqueness of his be accounted for by individual difference variables. behavior. Frankly this is why I have stopped doing studies The foregoing discussion does not imply of this sort. It seems to me that the point has now that predictions cannot be made from sub- been amply demonstrated, and it is time to get on with other matters. ject variables to relevant behaviors, but it does suggest severe limits on the range It is encouraging that recent research on and level of relationships that can be dispositions has started to recognize ser- expected. Consider, as a representative iously the extraordinary complexity of the example, a recent effort to relate individual interactions found between subject vari- differences in young children's expectancies ables and conditions. The concept of about locus of control to their behavior in "moderator variables" was introduced to theoretically relevant situations (Mischel, trait theory to refer to the fact that the Zeiss, & Zeiss, 1973). To explore these effects of any particular disposition gen- interactions, the Stanford Preschool Inter- erally are moderated by such other vari- nal-External Scale was developed as a mea- ables as the subject's age, his sex, his IQ, sure of expectancies about whether events the experimenter's sex, and the character- occur as a consequence of the child's own istics of the situation (Wallach, 1962). action ("internal control") or as a con- When one examines closely the interactions sequence of external forces ("external obtained in research on the effects of control"). Expectancies about locus of dispositions and conditions, the number of control were measured separately for posi- moderator variables required to predict tive and negative events so that scores behavior and the complexity of their inter- reflect expectancies for degree of internal relationships (e.g., McGuire, 1968) tend to control of positive events (I + ), of nega- become most formidable. For example, to tive events (I — ), and a sum of these two predict a subject's voluntary delay of (total 1). Individual differences in I + , gratification, one may have to know how I — , and total I then were correlated with old he is, his sex, the experimenter's sex, the children's ability to delay gratification the particular objects for which he is under diverse working and waiting condi- waiting, the consequences of not waiting, tions. The results provided highly specific the models to whom he was just exposed, but theoretically meaningful patterns of RECONCEPTUALIZATION OF PERSONALITY 257 relationships. To illustrate, relationships particular forms and under particular between total I and overall delay be- conditions. havior were negligible, and 1+ was un- The language of "interactions" and related to I — . As expected, 1+ (but not "moderator variables" provides simply I —) was found to be related to persistence another way of talking about the idiosyn- in three separate situations where instru- cratic organization of behavior and its mental activity would result in a positive dependence upon specific conditions unless outcome; I— (but not 1+) was related to (as Bern, 1972, p. 21, has noted) one can persistence when instrumental activity "predict on a priori grounds which moder- could prevent the occurrence of a negative ators are likely to divide up the world into outcome. useful classes. ..." Demonstrations that The overall findings showed that in- both subject and situational moderators dividual differences in children's beliefs can be used predictively, not merely to about their ability to control outcomes are partial out the variance from each source partial determinants of their goal-directed post hoc, are especially important in light behavior, but the relationships hinge on of the negative conclusions reached by extremely specific moderating conditions, Wallach, one of the main formulators of both with regard to the type of behavior the moderator variable strategy in person- and the type of belief. If such moderating ality research. Commenting on the ex- conditions had not been considered and all tensive results from his decade of work on indices of "delay behavior" had been com- the problem: bined regardless of their positive or nega- tive valence, the actual role of the relevant Further analyses and additional data collection by us and others suggest that not only are findings individual differences would have been ungeneralizable from one sex to the other, but even totally obscured. While the results were when, within sex, one simply tries to duplicate the of considerable theoretical interest, the results of a given study, such attempts do not pan number and mean level of the achieved out. ... we cannoti'sayf.that use of moderators has successfully pinpointed subgroups for whom correlations were not appreciably higher consistency among diverse tests will be predictable. than those typically found in correlational . . . The empirical basis for recommending modera- personality research. Moreover, the ability tors as the answer to the search for consistency thus of these correlations to survive cross- seems more apparent than real [Wallach & Lcggett, validation remains to be demonstrated. 1972, p. 313]. The more moderators required to qualify In regard to this last issue, the inter- a trait, the more the "trait" becomes a action studies of the sort conducted by relatively specific description of a behavior- Endler and Hunt and Moos, unfortunately, situation unit. That is, the more highly leave perhaps the most important question circumscribed, "moderated," and situa- unanswered: once an individual's idiosyn- tion specific the trait, the more it becomes cratic pattern has been identified, can it indistinguishable from a specific behavior- be used accurately to predict consistencies situation description. At its extreme, when in his subsequent behavior later in the many strings of hyphenated moderator same or (even more interestingly) in similar variables are required, the behavioral settings? While the interaction studies "signs" from which the disposition is in- have demonstrated the existence of ex- ferred may become equivalent to the in- tensive Person X Situation interactions, ferred disposition and make the inference they have not yet addressed themselves to the challenge of demonstrating that useful gratuitous. As we increasingly qualify predictions can be made a priori about the description of a person to specify the individual consistencies across a set of exact response modes and conditions in specified conditions. Such demonstrations which a particular behavior will occur, we are particularly necessary in light of the move from characterizing him with general- frequent failures to achieve replications in ized traits to describing his behavior in this domain (e.g., Averill, Olbrich, & 258 WALTER MISCHEL

Lazarus, 1972; Wallach & Leggett, 1972, of noncognitive personality dimensions pp. 313-314). Moreover, the interaction accurately reflects man's impressive dis- studies have not in any sense explained the criminative facility and the inadequency of nature of the obtained interactions. Later the assumption of global dispositions, and sections of this paper attempt to analyze not merely the distortions of measurement the psychological bases for "interaction"; (Mischel, 1968). The term "discrimina- in the absence of such an analysis, an tive facility" seems to fit the data better emphasis on interaction is in danger of than "specificity" and avoids the unfortun- being little more than the proclamation of ate negative semantic connotations of a truism. specificity when applied to persons (e.g., In sum, when interpreting the meaning the implications of inconsistency, in- of the data on Person X Situation inter- sincerity, fickleness, unreliability; see also actions and moderator variables, it has Gergen, 1968). been tempting to treat the obtained inter- Whereas discriminative facility is highly actions as if they had demonstrated that functional (Gibson, 1969) diminished sensi- people behave consistently in predictable tivity to changing consequences (i.e., in- ways across a wide variety of situations. discriminate responding) may be a hall- But demonstrations of the predictive util- mark of an organism coping ineffectively. ity of the moderator variable-interaction In fact, indiscriminate responding (i.e., strategy still lie in the future (e.g., Bern, "consistent" behavior across situations) 1972). The available data on this topic tends to be displayed more by maladaptive, now merely highlight the idiosyncratic severely disturbed, or less mature persons organization of behavior within individuals, than by well-functioning ones (Moos, and hence the uniqueness of stimulus 1968). For example, on the basis of their equivalences and response equivalences for studies of hyperaggressive children under- each person. Such data provide encourage- going therapeutic treatment, Raush, Ditt- ment for idiographic study (Allport, 1937) man, and Taylor (1959) reported: "there but not for the predictive utility of "com- appears to be a trend for social behavior to mon" (nomothetic) personality traits. become more related to situational in- fluences with ego development . . . the "Specificity" or Discriminative Facility? children seem to have gained in the ability Viewed from the perspective of the to discriminate between different situa- traditional personality paradigm, the "spe- tions [p. 368]." Yet although relatively cificity" and "inconsistency" found in be- more "indiscriminate behavior" tends to be havior constitute an embarrassment that found in more immature and/or severely is generally attributed to methodological abnormal persons, its extent should not be flaws and faulty measurements. Thus exaggerated. Even extremely autistic be- empirical evidence concerning the specifi- havior, for example, is highly discrimina- city of the relations between social be- tive when closely analyzed (e.g., Lovaas, havior and conditions usually has been Freitag, Gold, & Kassorla, 1965). interpreted as due to the inadequacies of the tests and measures, faulty sampling, Discrimination, Generalization, and and the limitations of the particular raters Idiosyncratic Stimulus Meanings or clinical judges. These and many other The discriminativeness found in be- similar methodological problems undoubt- havior is not so great that we cannot recog- edly are sources of error and seriously limit nize continuity in people. It is also not the degree of consistency that can be ob- so great that we have to treat each new served (e.g., Block, 1968; Emmerich, 1969). behavior from a person as if we never saw An alternative interpretation, however, anything like it from him before. But the and one favored by a specific interaction findings remind us that what people do in theory of social behavior, is that the any situation may be changed dramatically "specificity" so regularly found in studies even by relatively trivial alterations in RECONCEPTUALIZATION OP PERSONALITY 259

their prior experiences or by slight modifica- response modes are used to sample the tions in the particular features of the im- individual's behavior (e.g., data from mediate situation. Rather than argue questionnaires, from behavior observation), about the existence of "consistency," it consistency is even harder to demonstrate would be more constructive to analyze and (Mischel, 1968). study the cognitive and social learning To the degree that idiosyncratic social conditions that seem to foster—and to undermine—its occurrence. learning histories characterize each per- If expected consequences for the per- son's life, idiosyncratic (rather than cul- formance of responses across situations are turally shared) stimulus equivalences and largely uncorrelated, the responses them- hence idiosyncratic behavior patterns may selves should not be expected to covary be expected. As was noted earlier (Mischel, strongly, as they indeed do not in most 1968, p. 190, italics added): empirical studies. When the probable The phenomena of discrimination and generaliza- reinforcing consequences to the person for tion lead to the view that behavior patterns are cheating, waiting, or working differ widely remarkably situation-specific on the one hand, while across situations depending on the partic- also evokable by diverse and often seemingly ular task or circumstances, the behavior of heterogeneous stimuli on the basis of generaliza- tion effects. The person's prior experiences with others, the likelihood of detection, the related conditions and the exact details of the particular probable consequences of being caught, the evoking situation determine the meaning of the stimuli, frustration induced, the value of success, i.e., their effects on all aspects of his life. Usually etc., impressive generality will not be generalization effects involve relatively idiosyncratic found. Conversely, when similar be- contextual and semantic generalization dimensions and are based on more than gradients of physical haviors are expected and supported in stimulus similarity . . . one must know the properties numerous situations, consistency will be or meaning that the stimulus has acquired for the obtained. subject. If the history is unknown, the response has Because most social behaviors produce to be assessed directly. positive consequences in some situations Idiosyncratic histories produce idiosyn- but negative ones in other contexts, the cratic stimulus meanings. In clinical as- relatively low associations found among sessment of the individual, it is apparent, an individual's response patterns even in for example, that seemingly heterogeneous seemingly similar situations should not be surprising. Consider, for example, the stimuli may come to elicit similar intense intercorrelations among measures intended approach or avoidance patterns accom- to sample dependent behaviors, such as panied by strong arousal (Mischel, 1968). "touching, holding, and being near." If a Because the conditions under which stimuli child has been rewarded regularly at acquire their meaning and power are often nursery school for "touching, holding, and both adventitious and unique, and be- being near" with his teacher but not with cause the dimensions of stimulus and re- his father at home, a high correlation be- sponse generalization tend to be idiosyn- tween dependency measured in the two cratic, it may be futile to seek common situations will not be found and should not underlying dimensions of similarity on the be expected. basis of which diverse events come to The consequences for similar content expressed in different response modes also evoke a similar response pattern for all tend to be drastically different. If on a persons. Especially when the individual's projective test a person tells stories full prior learning history is unknown, and of aggressive themes, he would be judged when he is exposed to multiple and ex- to have a healthy fantasy life, but he would ceedingly complex stimuli as in virtually be jailed if he enacted those themes in his all life situations, it becomes important to relations with other people. It therefore assess the effective stimuli, or "stimuli as should not be surprising that when different coded," which regulate his responses in 260 WALTER MISCHEL particular contexts. These stimuli as coded sent in front of them) into a "color picture should not be confused with the totality in your head," or they can transform the of objective physical events to which he is picture of the objects (presented on a slide exposed. It is hardly novel now to assert projected on a screen in front of them) into that the objective distal stimulus imping- the "real" objects by pretending in im- ing on sense organs does not necessarily agination that they are actually there on a correspond to the "effective" stimulus; plate in front of them (Mischel & Moore, organisms respond selectively to particular 1973b). aspects of the objective stimulus event The results clearly show that what is in (Lawrence, 1959). the children's heads—not what is physically The meaning and impact of a stimulus in front of them—determines their ability can be modified dramatically by cognitive to delay. Regardless of the stimulus in transformations. Such transformations are their visual field, if they imagine the real illustrated in research on the determinants objects as present, they cannot wait long of how long preschool children will actually for them. But if they imagine pictures sit still alone in a chair waiting for a pre- (abstract representations) of the objects, ferred but delayed outcome before they they can wait for long time periods (and signal with a bell to terminate the waiting even longer than when they are distracting period and settle for a less preferred but themselves with abstract representations of immediately available gratification (e.g., objects that are comparable but not rele- Mischel, Ebbesen, & Zeiss, 1972). We vant to the rewards for which they are have been finding that the same child who waiting). Through instructions (adminis- on one occasion may terminate his waiting tered before the child begins to wait) about in less than half a minute may be capable what to imagine during the delay period, of waiting by himself for long times on it is possible to completely alter (indeed, another occasion a few weeks earlier or to reverse) the effects of the physically later, if cognitive and attentional conditions present reward stimuli in the situation and are appropriate (Mischel, 1973a). to cognitively control delay behavior with For example, if the child is left during considerable precision. But while in ex- the waiting period with the actual reward periments the experimenter provides in- objects (e.g., pretzels or marshmallows) structions (which our subjects obligingly in front of him, it becomes extremely followed) about how to construe the stim- difficult for him to wait for more than a ulus situation, in life the "subject" supplies few moments. But through instructions his own instructions and may transform the he can cognitively transform the reward situation in many alternative (unpre- objects in ways that permit him to wait dictable) ways. The ability of individuals for long time periods (e.g., Mischel & to cognitively transform the meaning and Baker, 1973). If he cognitively transforms impact of stimuli in any given situation the stimulus, for example, by thinking (e.g., by self-instructions) makes it even about the pretzel sticks as little brown logs more unlikely that the assessor will dis- or by thinking about the marshmallows as cover a priori broad equivalence classes round white clouds or as cotton balls, he of stimulus meanings for many individuals may wait much longer than our graduate across many situations, unless they all student experimenters. Conversely, if the transform the stimuli in the same way. child has been instructed to focus cogni- Recognition of the idiosyncratic organiza- tively on the consummatory qualities of tion of behavior in each person suggests that the reward objects, such as the pretzel's individually oriented assessments are bound crunchy, salty taste or the chewy, sweet, to have very limited success if they try soft taste of the marshmallows, he tends to label a person with generalized trait to be able to wait only a short time. terms, sort him into diagnostic or type Similarly, through instruction the children categories, or estimate his average posi- can easily transform the real objects (pre- tion on average or modal dimensions (Mis- RECONCEPTUALIZATION OF PERSONALITY 261 chel, 1968).4 Instead, it may be more use- introduced and varied in sampled situa- ful for the clinician to assess the exact tions. To reveal the acquired meanings of conditions that regularly covary with in- stimuli, one must assess what the individual crements or decrements in the problem- says and does when they occur in symbolic producing behaviors for the particular form (e.g., when discussed in interviews) person. For this purpose in a behavioral and more realistically when presented in analysis, one attempts to sample directly hypothetical, role-playing or life situations, the individual's relevant cognitions and as has been discussed in detail (Mischel, behaviors in relation to the conditions of 1968). Considerable evidence suggests particular current concern: that in this assessment enterprise, direct In this sense, behavioral assessment involves an information from the person is the best exploration of the unique or idiographic aspects of source of data (Mischel, 1972). the single case, perhaps to a greater extent than any Some of the clearest examples of the other approach. Social behavior theory recognizes analysis of stimulus conditions influencing the individuality of each person and of each unique behavior are found in efforts to construct situation. This is a curious feature when one con- siders the "mechanistic S-R" stereotypes not in- subjective anxiety hierarchies (e.g., Wolpe, frequently attached by critics to behavioral analyses. 1961). In collaboration with the assessor, A ssessing the acquired meaning of stimuli is the core the individual can identify the specific of social behavior assessment . . . [Mischel, 1968, p. conditions that generate fear in him and 190, italics added]. arranges them on a gradient of severity The above point is often misunderstood. from least to most intense. For one client, For example, Adinolfi (1971, p. 174) asked: items such as "thinks I only did an hour's "How then does the social-behavioral work today," "sitting at the movies," critic of current clinical and personality "going on a casual stroll" and "staying in theory propose to determine the stimulus bed during the day (even though ill)" were conditions to which the observed is respond- some of the events arranged on a subjective ing?" The answer to this question comes continuum of "guilt"-producing stimuli. from actively enrolling the "observed" Such individually oriented assessments person in the assessment process (Mischel, lead naturally to the design of individually 1968). In collaboration with the assessor oriented treatments intended to provide the individual provides hypotheses about the best possible conditions for achieving the conditions that lead to increases and each individual's objectives (Bandura, decreases in his own problematic behaviors. 1969). In the case of the client suffering To elaborate, verify, or modify these hy- from guilt, for example, after the subjec- potheses, the stimulus conditions are intro- tive hierarchy of guilt-inducing stimuli had duced and systematically varied, and their been identified, conditions could be ar- impact on the person is assessed from his ranged to help him make new responses in- self-report and from other changes in his compatible with anxiety when the problem- behavior. In this manner, one can analyze producing stimuli are presented cogni- how changes in the particular stimulus tively through thought-inducing instruc- conditions are correlated with changes in tions. the behavior of interest. The acquired meanings of a stimulus can only be known Uses and Misuses of Traits by determining what the person does with In sum, obviously behavior is not en- it verbally and behaviorally, when it is tirely situation specific; we do not have to 4 It is possible that for each individual there are relearn everything in every new situation, unique but broad classes of subjective stimulus we have memories, and our past predis- equivalences, but these cannot be assessed by com- poses our present behavior in critically paring individuals in situations that are construed important and complex ways. Obviously as equivalent by the assessor. Such subjective equivalences certainly merit attention, but so far people have characteristics and overall the clinician has not demonstrated his ability to "average" differences in behavior between find them reliably (Mischel, 1968). individuals can be abstracted on many 262 WALTER MISCHEL dimensions and used to discriminate among factors helps to clarify when person vari- persons for many purposes. Obviously ables and individual differences are used in knowing how a person behaved before can the everyday formation of impressions. help predict how he will behave again in Person (trait) explanations are invoked similar contexts. Obviously the impact of when the individual's behavior is "dis- any stimulus depends on the organism that tinctive" (Kelley, 1967), that is, when it experiences it. No one suggests that the deviates from others' behavior in the same organism approaches every new situation situation. Thus, behaviors that are at with an empty head, nor is it questioned by variance with relevant group norms (e.g., anyone that different individuals differ success when others fail, failure when markedly in how they deal with most stim- others succeed) are attributed to the per- ulus conditions. What has been questioned son or to "internal causes" (e.g., Frieze & (Hunt, 1965; Mischel, 1968) is the utility Weiner, 1971; Weiner & Kukla, 1970). of inferring broad dispositions from be- Conversely, when a person's behavior is havioral signs as the bases for trying to consistent with the norms in the situation explain the phenomena of personality and (when the person succeeds when others for making useful statements about in- succeed, or fails when others fail), his dividual behavior. The available data do performance is attributed to situational not imply that different people will not act factors such as task difficulty (Weiner et al., differently with some consistency in dif- 1971). ferent classes of situations; they do imply Traits are constructs which are inferred that the particular classes of conditions or abstracted from behavior. When the must be taken into account far more care- relations between the observed behavior fully than in the past, tend to be much and the attributed trait are relatively narrower than traditional trait theories direct, the trait serves essentially as a have assumed, and for purposes of impor- summary term for the behaviors that have tant individual decision making, require been integrated by the observer. People highly individualized assessments of stim- emit behaviors and these are perceived, ulus meanings (Mischel, 1968, pp. 235-280). integrated, and categorized by those who The data also suggest that inferences about observe them, including those who emit global underlying traits and dispositions them. The process of integrating the tend to have less utility for most assess- observed information is receiving much ment efforts to predict or therapeutically study but is still not completely understood modify individual behavior than do more (e.g., Anderson, 1971, 1972). Regardless economical, alternative analyses based on of the exact genesis of trait impressions, more direct data such as the person's past trait labels may serve as summaries (es- behavior in similar situations or his direct sentially arithmetic averages) for categories self-report. of observed behavior (e.g., "dependent on A critique of traits as inadequate causal peers," "physically aggressive with sib- explanations and an indictment of the lings"). For purposes of global character- utility of indirect trait inferences for many izations of salient personal qualities, broad, individually oriented assessment and clin- highly abstract categories may be useful ical purposes (Mischel, 1968) does not with minimal moderators or specific situa- imply a rejection of their other possible tional qualifiers. But for purposes of more uses. The layman as well as the trait specific communication and for prediction psychologist generates and employs trait of specific behavior in relation to specific constructs. The question becomes not "do conditions, careful discriminative limits traits really exist?" but when are trait must be included. constructs invoked and "what are their Estimates of mean past behavior often uses and misuses?" are the best predictors of future behavior Research on the layman's attribution of in similar situations, especially when there causation to dispoeitional versus situational are no other bases for prediction (Mischel, RECONCEPTUALIZATION OF PERSONALITY 263

1968, 1972). The predictive limitations of the importance of the layman's everyday traits become evident, however, when one use of trait categories. How do trait attempts to predict from past behavior to categorizations function for the layman? behavior in different new situations. More- Do they serve him well? For what pur- over, when observers categorize an in- poses might they be used ? In our research dividual's behavior in trait terms, the "sali- my students and I are asking such ques- ent" (central, mean, primary) features of tions now. For example, we find that when the behavior may become the basis for the required to predict a person's behavior and categorization, so that the person becomes given a choice of how to categorize the labeled as "anxious," for example, even if available behavioral information, subjects that term accurately characterizes only a overwhelmingly preferred to organize data small portion of his total social behavior. in terms of traits rather than settings Then the "moderators" become omitted (Jeffrey & Mischel, 1973). But when the and the situation-free trait abbreviations perceiver's purpose was structured as that remain may serve more as global memorizing as much information as pos- stereotypes and broad character sketches sible, setting categories were used. Clearly than as accurate bases for the prediction of the functions of trait constructs for the specific behaviors. layman deserve serious attention and hope- When the consistency issue is viewed in fully will inform us further about the terms of the utility of inferring broad re- psychological uses and abuses of trait sponse tendencies and not in terms of the categorization. more metaphysical question of the exis- tence or validity of personality dispositions, From Behavior to the Construction of it becomes evident that the answer must Personality depend on the particular objective or As Heider (1958) has noted, in the purpose for which the inference is made. psychology of common sense the subject For example, while global trait inferences goes quickly from act to global internalized may have little utility for the prediction disposition. While behavior often may be of the subject's specific future behavior in highly situation specific, it seems equally specific situations or for the design of true that in daily life people tend to con- specific treatment programs, they may strue each other as if they were highly con- have value for the person himself—for sistent, constructing consistent personalities instance, when he must abstract attributes even on the basis of relatively inconsistent to answer such everyday questions as: Is behavioral fragments. your assistant reliable? or What kind of This discrepancy may reflect in part that person is my psychotherapist? or Might people go rapidly beyond the observation this stranger lurking on the next corner be of some consistency which does exist in a murderer? or What are you like? Simi- behavior to the attribution of greater larly, an indictment of the relative lack of perceived consistencies which they con- utility of inferring broad dispositions for struct (e.g., Mischel, 1969; Schneider, purposes of predicting and/or thera- 1973). After these construction systems peutically modifying the individual's be- have been generated, they may be adhered havior does not deny the utility of using to tenaciously even in the face of seemingly such inferences for many other purposes— disconfirmatory data (Mischel, 1968, 1969). such as for gross initial screening decisions Many processes contribute to the con- or for studying average differences between struction and maintenance of consistent groups of individuals in personality research impressions of others. Tversky and Kahne- (Mischel, 1968). mon (1971), for example, contended^that The limitations of traditional personality both sophisticated scientists and naive theories which invoke trait constructs as subjects intuitively but often erroneously the psychologist's explanations for be- interpret small samples of observations as havior should not deflect attention from if they were highly representative. More- 264 WALTER MISCHEL over, after an initial impression of a person information about ourselves and the multi- has been formed, observations of his sub- plicity, variety, and complexity of the sequent behavior are biased toward con- situations we encounter in our own lives, sistency with the initial impression (Hay- whereas we know others in only limited den & Mischel, 1973). Like the clinician contexts and therefore tend to over- (e.g., Chapman & Chapman, 1969), the generalize from their behavior in those layman's impressions may perpetuate con- instances. sistent but invalid "illusory correlations." There even seems to be a substantial bias Traits as Causes versus Traits as of memory for the attributes of behavior Summary Labels in the direction of preexisting cognitive According to the traditional trait para- structures or implicit personality theories digm, traits are the generalized disposi- (D'Andrade, 1970, 1973). Consequently, tions in the person that render many recall-based trait ratings may yield data stimuli functionally equivalent and that that are systematic but unrelated to results cause the individual to behave consistently based on direct observation of ongoing be- across many situations (Allport, 1937). havior as it occurs (Shweder, 1972). The present view, in contrast, construes The overattribution of consistency may the individual as generating diverse be- be something people do unto others more haviors in response to diverse conditions; than to themselves. Jones and Nisbett the emitted behaviors are observed and (1971) noted that when explaining other subsequently integrated cognitively by the people's behavior we invoke their con- performer, as well as by others who per- sistent personality dispositions: Steve is the ceive him, and are encoded on semantic sort of person who puts bumper stickers dimensions in trait terms. Thus while the on his car; Jill tripped because she's traditional personality paradigm views clumsy. But when asked to explain our traits as the intrapsychic causes of be- own behavior we consider specific condi- havioral consistency, the present position tions: "AAA sent me this catchy bumper sees them as the summary terms (labels, sticker in the mail" or "I tripped because codes, organizing constructs) applied to it was dark." Thus Jones and Nisbett observed behavior. In the present view, (1971, p. 58) on the basis of some promis- the study of global traits may ultimately ing preliminary data theorized that "actors reveal more about the cognitive activity tend to attribute the causes of their be- of the trait theorist than about the causes havior to stimuli inherent in the situation of behavior, but such findings would be of while observers tend to attribute behavior great value in their own right. to stable dispositions of the actor." Jones and Nisbett analyzed many possible rea- COGNITIVE SOCIAL LEARNING sons for this seemingly paradoxical state PERSON VARIABLES of affairs, including the tendency to treat every sample of behavior we observe from The previous sections have considered another person as if it were modal or typical the limitations of the basic assumptions of for him. It thus seems as if traits may be traditional global dispositional theories of the consistent attributes that other people personality and some of the main mis- have. When describing other people, we conceptions and issues arising from recent seem to act more like trait theorists, but challenges to those assumptions. Progress when we attempt to understand ourselves in the area of personality will require more we function more like social behaviorists. than criticism of existing positions and Might there be a warning here for clini- hinges on the development of an alterna- cians? Do we pin our clients with consis- tive conceptualization. In this section tent dispositional labels and trait ex- therefore a set of person variables is pro- planations more than we do ourselves? If posed, based on theoretical developments in that is true it may be because we have more the fields of social learning and cognition. RECONCEPTUALIZATION OF PERSONALITY 265

Given the overall findings on the dis- learning. The selections should be seen criminativeness of behavior and on the as suggestive and open to progressive re- complexity of the interactions between the vision rather than as final. These tenta- individual and the situation, it seems rea- tive person variables are not expected to sonable in the search for person variables provide ways to accurately predict broadly to look more specifically at what the person cross-situational behavioral differences be- constructs in particular conditions, rather tween persons: the discriminativeness and than trying to infer what broad traits he idiosyncratic organization of behavior are generally has, and to incorporate in de- facts of nature, not limitations unique to scriptions of what he does the specific trait theories. But these variables should psychological conditions in which the be- serve to demonstrate that a social behavior havior will and will not be expected to approach to persons does not imply an occur. What people do, of course, includes empty organism. They should suggest much more than motor acts and requires useful ways of conceptualizing and study- us to consider what they do cognitively and ing specifically how persons mediate the affectively as well as motorically. impact of stimuli and generate distinctive The proposed cognitive social learning complex molar behavior patterns. And approach to personality shifts the unit of they should help to conceptualize person- study from global traits inferred from be- situation interactions in a theoretical havioral signs to the individual's cognitive framework based on contributions from activities and behavior patterns, studied both cognitive and behavioral psychology. in relation to the specific conditions that The proposed cognitive social learning evoke, maintain, and modify them and person variables deal first with the in- which they, in turn, change (Mischel, 1968). dividual's competencies to construct (gener- The focus shifts from attempting to com- ate) diverse behaviors under appropriate pare and generalize about what different conditions. Next, one must consider the individuals "are like" to an assessment of individual's encoding and categorization of what they do—behaviorally and cogni- events. Furthermore, a comprehensive tively—in relation to the psychological analysis of the behaviors a person performs conditions in which they do it. The focus in particular situations requires attention shifts from describing situation-free people to his expectancies about outcomes, the with broad trait adjectives to analyzing subjective values of such outcomes, and his the specific interactions between conditions self-regulatory systems and plans. The and the cognitions and behaviors of following five sections discuss each of these interest. proposed person variables. While these Personality research on social behavior variables obviously overlap and interact, and cognition in recent years has focused each may provide distinctive information mainly on the processes through which about the individual and each may be mea- behaviors are acquired, evoked, maintained, sured objectively and varied systematically. and modified (e.g., Bandura, 1969; Mis- chel, 1968). Much less attention has been Cognitive and Behavioral Construction given to the psychological products within Competencies the individual of cognitive development and Through direct and observational learn- social learning experiences. Yet a viable ing the individual acquires information psychology of personality demands at- about the world and his relationship to it. tention to person variables that are the As a result of observing events and attend- products of the individual's total history ing to the behavior of live and symbolic and that in turn mediate the manner in models (through direct and film-mediated which new experiences affect him. observation, reading, and instruction) in The proposed person variables are a the course of cognitive development the synthesis of seemingly promising con- perceiver acquires the potential to generate structs in the areas of cognition and social vast repertoires of organized behavior. 266 WALTER MISCHEL

While the pervasive occurrence and impor- Although the exact cognitive processes tant consequences of such observational are far from clear, it is apparent that each learning have been convincingly demon- individual acquires the capacity to con- strated (e.g., Bandura, 1969; Campbell, struct a great range of potential behaviors, 1961), it is less clear how to conceptualize and different individuals acquire different just what gets learned. The phenomena behavior construction capabilities. The to be encompassed must include such enormous differences between persons in diverse learnings as the nature of sexual the range and quality of the cognitive and gender identity (e.g., Kohlberg, 1966), the behavioral patterns that they can generate structure (or construction) of the physical is evident from even casual comparison of world (e.g., Piaget, 1954), the social rules the construction potentials of any given and conventions that guide conduct (e.g., individual with those, for example, of an Aronfreed, 1968), the personal constructs Olympic athlete, a Nobel Prize winner, a generated about self and others (e.g., G. retardate, an experienced forger, or a Kelly, 1955), the rehearsal strategies of the successful actor. observer (Bandura, 1971a). Some theorists The person's behavior construction po- have discussed these acquisitions in terms tential can be assessed readily by introduc- of the products of information processing ing incentives for the most complete con- and of information integration (e.g., Ander- structions that he can render on particular son, 1972; Bandura, 1971a; Rumelhart, performance tasks. In a sense, the assess- Lindsey, & Norman, 1971), others in terms ment conditions here are identical to those of schemata and cognitive templates (e.g., in achievement testing (Wallace, 1966). Aronfreed, 1968). The same strategy can be used to assess The concept of cognitive and behavioral what subjects "know" (i.e., the cognitive construction competencies seems sufficiently constructions they can generate, for ex- broad to include the vast array of psycho- ample, about abstract and physical proper- logical acquisitions of organized informa- ties and relationships as in mathematics tion that must be encompassed. The and geography) and what they are capable term "constructions" also emphasizes the of doing (enacting) in the form of social constructive manner in which informa- behaviors. For example, to assess what tion seems to be retrieved (e.g., Neisser, children had acquired from observing a 1967) and the active organization through model, attractive rewards later were offered which it is categorized and transformed to them contingent upon their reproducing (Bower, 1970; Mandler, 1967, 1968). It the model's behaviors (e.g., Bandura, has become plain that rather than mimick- 1965; Grusec & Mischel, 1966). The re- ing observed responses or returning memory sults showed that the children had acquired traces from undisturbed storage vaults, the a great deal of information from observa- observer selectively constructs (generates) tion of the model which they could recon- his renditions of "reality." Indeed, re- struct elaborately but only when given search on modeling effects has long recog- appropriate incentives. nized that the products of observational For many purposes, it is valuable to learning involve a novel, highly organized assess the quality and range of the cogni- synthesis of information rather than a tive constructions and behavioral enact- photocopy of specific observed responses ments of which the individual is capable. (e.g., Bandura, 1971b; Mischel £ Grusec, In this vein, rather than assess "typical" 1966). The present concept of construc- behavior, one assesses potential behaviors tion competencies should call attention to or achievements. One tests what the per- the person's cognitive activities—the opera- son can do (e.g., Wallace, 1966) rather than tions and transformations that he performs what he "usually" does. Indeed one of the on information—rather than to a store of most recurrent and promising dimensions finite cognitions and responses that he of individual differences in research seems "has." to involve the person's cognitive and be~ RECONCEPTUALIZATION OF PERSONALITY 267 havioral (social) competencies (e.g., White, ables (e.g., socioeconomic class, high school 1959; Zigler & Phillips, 1961, 1962). These graduation) reflect the individual's con- competencies presumably reflect the degree struction capacities and achievements, they to which the person can generate adaptive, also may be expected to predict "adjust- skillful behaviors that will have beneficial ment" and interpersonal competencies, as consequences for him. Personality psy- they often do (e.g., Robbins, 1972). The chology can profit from much greater atten- assessment of competence in response to tion to cognitive and intellectual com- specific problematic situations in the direct petencies since these "mental abilities" manner developed by Goldfried and seem to have much better temporal and D'Zurilla (1969) seems especially promising. cross-situational stability and influence The relative stability of the person's than most of the social traits and motiva- construction capacities may be one of the tions traditionally favored in personality important contributors to the impression research (e.g., Mischel, 1968, 1969). of consistency in personality. The fact The relevance of cognitive-intellective that cognitive skills and behavior-generat- competencies for personality seems evident ing capacities tend to be relatively endur- in light of the important, persistent con- ing is reflected in the relatively high sta- tributions of indices of intelligence to the bility found in performances closely related obtained networks of personality correla- to cognitive and intellectual variables, as tions (Campbell & Fiske, 1959; Mischel, has been stressed before (Mischel, 1968, 1968). In spite of extensive efforts to 1969). The individual who knows how to minimize or "partial out" the role of be assertive with waiters, for example, or intelligence in personality studies, for ex- who knows how to solve certain kinds of ample, cognitive competencies (as tested interpersonal problems competently, or by "mental age" and IQ tests) tend to be who excels in singing, is capable of such among the very best predictors of later performances enduringly. social and interpersonal adjustment (e.g., Anderson, 1960). Presumably, brighter, Encoding Strategies and Personal Constructs more competent people experience more From the perspective of personality interpersonal success and better work psychology, an especially important com- achievements and hence become more ponent of information processing concerns positively assessed by themselves and by the perceiver's ways of encoding and group- others on the evaluative "good-bad" ing information from stimulus inputs. As dimension which is so ubiquitous in trait discussed in earlier sections, people can ratings (e.g., Vernon, 1964). Cognitive readily perform cognitive transformations on achievements and intellective potential, as stimuli (Mischel & Moore, 1973), focusing measured by mental age or IQ tests, also on selected aspects of the objective stim- are receiving a central place in current ulus (e.g., the taste versus the shape of a cognitive-developmental theories (e.g., food object): such selective attention, Kohlberg, 1969) and presumably are an interpretation, and categorization substan- important ingredient of such concepts as tially alter the impact the stimulus exerts "ego strength" and "ego development." on behavior (see also Geer, Davison, & Indeed, it is tempting to speculate that the Gatchel, 1970; Schachter, 1964). Like- pervasive and substantial "first factor" wise, the manner in which perceivers en- found on tests like the MM PI (Block, code and selectively attend to observed 1965), often labeled with terms connoting behavioral sequences greatly influences "adjustment" at the positive end and what they learn and subsequently can do maladaptive character structure at the (Bandura, 1971a, 1971b). Clearly, dif- negative end, reflects to a considerable ferent persons may group and encode the degree the individual's level of cognitive- same events and behaviors in different social competence and achievement. To ways. At a molar level, such individual the degree that certain demographic vari- differences are especially evident in the 268 WALTER MISCHEL personal constructs individuals employ interests and objectives. He selects a (e.g., Argyle & Little, 1972; G. Kelly, category—such as "delay of gratification," 1955) and in the kinds of information to for example—and studies its behavioral which they selectively attend (Mischel, referents. In personality assessment, how- Ebbesen, & Zeiss, 1973). ever, it becomes quickly evident that the The behaviorally oriented psychologist subject (like the psychologist) also groups eschews inferences about global disposi- events into categories and organizes them tions and focuses instead on the particular actively into meaningful units. The lay- stimuli and behaviors of interest. But man usually does not describe his experi- what are "the stimuli and behaviors of ence with operational definitions: he cate- interest?" Early versions of behaviorism gorizes events in terms of his personal con- attempted to circumvent this question by structs (G. Kelly, 1955), and these may simplistic definitions in terms of clearly or may not overlap either with those of the delineated motor "acts" (such as bar press) psychologist or of other individuals. As in response to clicks and lights. As long previously noted (Jeffery & Mischel, 1973), as the behaviors studied were those of observers tend to group information about lower animals in experimenter-arranged persons with dispositional categories (such laboratory situations, the units of "be- as "honest," "intolerant," "freaky," "do havior" and "stimuli" remained manage- gooder"). Skepticism about the utility of able with fairly simple operational defini- traditional trait constructs regarding the tions. More recent versions of behavior subject's broad dispositions in no way theory, moving from cat, rat, and pigeon requires one to ignore the subject's con- confined in the experimenter's apparatus structs about his own and other's character- to people in exceedingly complex social istics. People invoke traits and other dis- situations, have extended the domain of positions as ways of describing and explain- studied behavior much beyond motor acts ing their experience and themselves, just and muscle twitches; they seek to encom- as professional psychologists do, and it pass what people do cognitively, emotion- would be strange if we tried to define out ally, and interpersonally, not merely their of existence the personal constructs and arm, leg, and mouth movements. Now the other concepts, perceptions, and experi- term "behavior" has been expanded to in- ences of the individuals whom we are clude virtually anything that an organism studying. The study of personal construct does, overtly or covertly, in relation to ex- systems (e.g., Little & Stephens, 1973), of tremely complex social and interpersonal implicit personality theories (e.g., Hamil- events. Consider, for example, "aggres- ton, 1971; Schneider, 1973), and of self- sion," "anxiety," "defense," "dependency," concepts (e.g., Gergen, 1968) promises to "self-concepts," "self-control," "self-rein- illuminate an important set of still poorly forcement." Such categories go consider- understood person variables. ably beyond self-evident behavior descrip- Cognitive consistency tends to be en- tions. A category like aggression involves hanced by selective attention and coding inferences about the subject's intentions processes that filter new information in a (e.g., harming another versus accidental manner that permits it to be integrated with injury) and abstractions about behavior, existing cognitive structures (e.g., Norman, rather than mere physical description of 1969). Cognitive processes that facilitate actions and utterances. the construction and maintenance of per- A focus on behavior must not obscure ceived consistency (e.g., U'Andrade, 1970; the fact that even the definition and selec- Hayden & Mischel, 1973) have been men- tion of a behavior unit for study requires tioned earlier and are elaborated elsewhere grouping and categorizing. In personality (Mischel, 1968, 1969). After information research, the psychologist does the con- has been integrated with existing cognitive struing, and he includes and excludes events structures and becomes part of long-term in the units he studies, depending on his memory, it remains available enduringly RKCONCKPTUALIZATION OF PERSONALITY 269

and exerts further stabilizing effects. For sential to consider his specific expectancies example, the individual's subjective con- about the consequences of different be- ception of his own identity and continuity havioral possibilities in that situation. For presumably rests heavily on his ability to many years personality research has remember (construct) subjectively similar searched for individual differences on the behaviors on his part over long time periods psychologist's hypothesized dimensions and across many situations. That is, the while neglecting the subject's own ex- individual can abstract the common ele- pectancies (hypotheses). More recently, ments of his behavior over time and across it seems increasingly clear that the ex- settings, thereby focusing on his more en- pectancies of the subject are central units during qualities. for psychology (e.g., Holies, 1972; Estes, There is considerable evidence that 1972; Irwin, 1971; Rotter, 1954). These people categorize their own personal quali- hypotheses guide the person's selection ties in relatively stable trait terms (e.g., (choice) of behaviors from among the on self-ratings and self-report question- enormous number which he is capable of naires). These self-categorizations, while constructing within any situation. often only complexly and tenuously related On the basis of direct eqaerieace, in- to nonverbal behavior, may be relatively structions, and observational learning, durable and generalized (Mischel, 1968, people develop expectancies about en- 1969). Such stable styles of self-presenta- vironmental contingencies (e.g., Handura, tion and self-description may be reflected 1969). Since the expectancies that are in personality test "response sets" like learned within a given situation pre- social desirability (Edwards, 1957), and in sumably reflect the objective contingencies tendencies to depict oneself in relatively in that situation, an expectancy construct positive or negative terms found in the may seem superfluous. The need for the behavior of so-called "repressers" versus expectancy construct as a person variable "sensitizers" on the Byrne (1961) Repres- becomes evident, however, when one con- sion-Sensitization Scale (Mischel, Ebbesen, siders individual differences in response & Zeiss, 1973). While traditional person- to the same situational contingencies due ality research has focused primarily on ex- to the different expectancies that each ploring the correlates of such self-categor- person brings to the situation. An ex- izations, in the present view they comprise pectancy construct is justified by the fact merely one kind of person variable. that the person's expectancies (inferred from statements) may not be in agreement Behavior-Outcome and Stimulus-Outcome with the objective contingencies in the Expectancies situation. Yet behavior may be generated in light of such expectancies, as seen, for So far the person variables considered example, in any verbal conditioning study deal with what the individual is capable when a subject says plural nouns on the of doing and how he categorizes events. erroneous hypothesis that the experi- To move from potential behaviors to menter is reinforcing them. actual performance, from construction In theories based on lower animal be- capacity and constructs to the construc- havior, the expectancy construct has served tion of behavior in specific situations, re- as a limited heuristic (e.g., Bolles, 1972), quires attention to the determinants of since rats and pigeons cannot tell us their performance. For this purpose, the person expectancies. Fortunately, humans are variables of greatest interest are the sub- not so handicapped and under appropriate ject's expectancies. While it is often in- assessment conditions are willing and able formative to know what an individual can to externalize their expectancies. Hence do and how he construes events and himself, the expectancy construct applied to human for purposes of specific prediction of be- rather than animal learning leads readily havior in a particular situation it is es- to measurement operations and to research 270 WALTER MISCHEL

strategies that can take account directly modify his complex performances (e.g., of the subject's hypotheses. Empirically, Spielberger & DeNike, 1966). As pre- since direct self-reports seem to he one of viously stressed, the essence of adaptive the best data sources about the individual performance is the recognition and ap- (Mischel, 1968, 1972), it should be possible preciation of new contingencies. To cope to fruitfully assess behavior-outcome ex- with the environment effectively, the pectancies by asking the subject. individual must recognize new contin- One type of expectancy concerns be- gencies as quickly as possible and reorganize havior-outcome relations under particular his behavior in the light of the new ex- conditions. These behavior-outcome ex- pectancies. Strongly established behavior- pectancies (hypotheses, contingency rules) outcome expectancies with respect to a represent the "if ; then " relations response pattern may constrain an in- between behavioral alternatives and prob- dividual's ability to adapt to changes in able outcomes anticipated with regard to contingencies. Indeed, "defensive reac- particular behavioral possibilities in par- tions" may be seen in part as a failure to ticular situations. In any given situation, adapt to new contingencies because the the person will generate the response pat- individual is still behaving in response to tern which he expects is most likely to old contingencies that are no longer valid. lead to the most subjectively valuable The "maladaptive" individual is behaving outcomes (consequences) in that situation in accord with expectancies that do not (e.g., Mischel, 1966; Rotter, 1954). In the adequately represent the actual behavior- absence of new information about the be- outcome rules in his current life situation. havior-outcome expectancies in any situa- In the present view, the effectiveness of tion the individual's performance will de- response-contingent reinforcements (i.e., pend on his previous behavior-outcome operant conditioning) rests on their ability expectancies in similar situations. This to modify behavior-outcome expectancies. point is illustrated in a study (Mischel & When information about the response Staub, 1965) which showed that pre- pattern required for reinforcement is con- situational expectancies significantly affect veyed to the subject by instructions, "con- choice behavior in the absence of situational ditioning" tends to occur much more information concerning probable perfor- readily than when the subject must experi- mance-outcome relationships. But the Mis- ence directly the reinforcing contingencies chel and Staub study also showed that new actually present in the operant training information about behavior-outcome rela- situation. For example, accurate instruc- tions in the particular situation may quickly tions about the required response and the overcome the effects of presituational ex- reinforcement schedule to which subjects pectancies, so that highly specific situa- would be exposed exerted far more powerful tional expectancies become the dominant effects on performance than did the rein- influences on performance. forcing contingencies (Kaufman, Baron, & When the expected consequences for per- Kopp, 1966). 'Presumably, such instruc- formance change, so does behavior, as seen tions exert their effects by altering response- in the discriminative nature of responding outcome expectancies. To the extent that which was elaborated in earlier sections and information about new response-reinforce- documented elsewhere (Mischel, 1968). ment contingencies can be conveyed to But in order for changes in behavior-out- motivated human beings more parsimon- come relations to affect behavior sub- iously through instructions or observational stantially, the person must recognize them. experiences than through operant condi- In the context of operant conditioning, it tioning procedures (e.g., Kaufman et al., has become evident that the subject's 1966), an insistence upon direct "shaping" awareness of the behavior-outcome rela- may reflect an unfortunate (and wasteful) tionship crucially affects the ability of failure to discriminate between the animal response consequences (reinforcements) to laboratory and the human condition. RECONCEPTUALIZATION OF PERSONALITY 271

A closely related second type of expec- multitude of learned correlations between tancy concerns stimulus-outcome relations. behavioral signs and outcomes. As noted previously in the discussion of Just as correlational personality research generalization and discrimination, the out- yields a host of validity associations be- comes expected for any behavior hinge on tween behavioral "signs" from persons in a multitude of stimulus conditions that one context and their behavior in other moderate the probable consequences of any situations, so does the perceiver's learning pattern of behavior. These stimuli ("signs") history provide him with a vast repertoire essentially "predict" for the person other of meaningful signs. For example, as re- events that are likely to occur. More pre- search on person perception suggests, cisely, the individual learns (through direct "shifty eyes," "tight lips," "lean and hun- and observational experiences) that certain gry looks," obese body build, age, sex, and events (cues, stimuli) predict certain other an enormous number of even subtler be- events. This concept of stimulus-outcome havioral cues (e.g., regarding the status expectancy is similar to the S-S* expectancy and power of others) come to predict for representing stimulus-outcome contingen- observers other correlated behaviors. If it cies proposed by Bolles (1972) in the con- were possible to compute them, many of text of animal learning. these correlations probably would not Stimulus-outcome expectancies seem es- average more than the .30 "personality pecially important person variables for coefficient" (Mischel, 1968) typically found understanding the phenomena of classical in correlational personality research, but conditioning. For example, through the that may be sufficiently accurate (especially contiguous association of a light and painful on an intermittent schedule) to assure their electric shock in aversive classical condi- persistent use. Some of these stimulus- tioning the subject learns that the light outcome associations presumably reflect predicts shock. If the product of classical the perceiver's idiosyncratic learning his- conditioning is construed as a stimulus- tory and his own evolving personal rules outcome expectancy, it follows that any about stimulus meanings. Many of these information which negates that expectancy associations, however, are likely to be will eliminate the conditioned response. widely shared by members of a common In fact, when subjects are informed that culture and probably depend importantly the "conditioned stimuli" will no longer be on the transcultural semantic associations followed by pain-producing events, their discussed by D'Andrade (1970) and conditioned emotional reactions are quickly Shweder (1971, 1972). An adequate study eliminated (e.g., Grings & Lockhart, 1963). of stimulus-outcome expectancies there- Conversely, when subjects were told that a fore would require attention to the rule particular word would be followed by system of the individual as well as to the shock, they promptly developed condi- shared "sign" grammar of the culture and tioned heart-rate responses (Chatterjee & of the transcultural lexicon structure. Eriksen, 1962). In the same vein, but Both behavior-outcome and stimulus- beyond the conditioning paradigm, if outcome expectancies depend on inferences subjects learn to generate "happy thoughts" about the intentions motivating behavior when faced by stimuli that otherwise would (i.e., its perceived causes). For example, a frustrate them beyond endurance, they can person's reactions to a physical blow from manage to tolerate the "aversive" situa- another will crucially depend on whether tion with equanimity (Mischel, Ebbesen, & it was perceived as accidental or deliberate. Zeiss, 1972). Outside the artificial con- Similarly, whether praise and attention fines of the laboratory in the human inter- produces in the recipient a warm glow (and actions of life, the "stimuli" that predict "conditioning" of his preceding behaviors) outcomes often are the social behaviors of or suspicion (and a rebuff) depends on others in particular contexts. The mean- whether the behaviors are perceived as ings attributed to those stimuli hinge on a sincere or as ingratiating (Jones, 1964). 272 WALTER MISCHEL

Extremely subtle social and interpersonal paper on generalization, discrimination, cues affect the interpretation of the motiva- and idiosyncratic stimulus meanings). tion (and hence the impact) of these com- While behavior-outcome and stimulus- plex human behaviors. outcome expectancies seem viable person Although expectancy constructs often variables, it would be both tempting and have been proposed, some of the main hazardous to transform them into general- formulations have been based entirely on ized trait-like dispositions by endowing animal research (e.g., Bolles, 1972) which them with broad cross-situational con- makes their relevance for human person- sistency or removing them from the context ality remote. Rotter's (1954) "subjective of the specific stimulus conditions on which expectancy" construct was an important they depend. At the empirical level, and theoretically influential exception. "generalized expectancies" tend to be However, it deals only with one type of generalized only within relatively narrow, expectancy (similar to the present "be- restricted limits (e.g., Mischel & Staub, havior-outcome expectancies"); it does not 1965; Mischel, Ebbesen, & Zeiss, 1973). consider stimulus-outcome expectancies. As was noted before in this paper, for ex- Moreover, Rotter's formulation focuses on ample, the generality of "locus of control" "generalized expectancies" which are func- is in fact limited, with distinct, unrelated tionally similar to generalized traits and expectancies found for positive and nega- are not posited in the present approach. tive outcomes and with highly specific In the present view, the person's ex- behavioral correlates for each (Mischel, pecttmcies mediate the degree to which his Zeiss, & Zeiss, 1973). If expectancies are behavior shows cross-situational consist- converted into global trait-like dispositions ency or discriminativeness. When the ex- and extracted from their close interaction pected consequences for the performance of with situational conditions, they are likely responses across situations are not highly to become just as useless as their many correlated, the responses themselves should theoretical predecessors. On the other not covary strongly (Mischel, 1968). As hand, if they are construed as relatively previously noted, since most social be- specific (and modifiable) "if , then haviors lead to positive consequences in " hypotheses about contingencies, it some situations but not in other contexts, becomes evident that they exert important highly discriminative specific expectancies effects on behavior (e.g., Mischel & Staub, tend to be developed and the relatively low 1965). correlations typically found among a per- son's response patterns across situations Subjective Stimulus Values become understandable (Mischel, 1968). Even if individuals have similar ex- lixpectancies also will not become general- pectancies, they may select to perform ized across response modes when the con- different behaviors because of differences sequences for similar content expressed in in the subjective values of the outcomes different response modes are sharply dif- which they expect. For example, given ferent, as they are in most life circumstances that all persons expect that approval from (Mischel, 1968). Hence expectancies tend a therapist depends on verbalizing partic- to become relatively specific, rather than ular kinds of self-references, there may be broadly generalized. Although a person's differences in the frequency of such verbal- expectancies (and hence performances) izations due to differences in the perceived value of obtaining the therapist's approval. tend to be highly discriminative, there Such differences reflect the degree to which certainly is some generalization of expectan- different individuals value the response- cies, but their patterning in the individual contingent outcome. Therefore it is neces- tends to be idiosyncratically organized to sary to consider still another person the extent that the individual's history is variable: the subjective (perceived) value unique. (See the earlier section in this for the individual of particular classes of RECONCEPTUALIZATION OF PERSONALITY 273 events, that is, his stimulus preferences purpose, specific self-report inventories, and aversions. This unit refers to stimuli physiological measures, and direct be- that have acquired the power to induce havior sampling of approach and avoidance positive or negative emotional states in behavior in response to the real or sym- the person and to function as incentives or bolically presented emotional stimulus may reinforcers for his behavior. The sub- all be useful (Mischel, 1968). jective value of any stimulus pattern may The measurement operations for assess- be acquired and modified through instruc- ing stimulus values require considerable tions and observational experiences as well specifity. Just as the probable conse- as through direct experiences (Bandura, quences of any behavior pattern hinge on a 1969). host of specific moderating considerations, Stimulus values can be assessed by mea- so does the affective value (valence) of any suring the individual's actual choices in stimulus depend on the exact conditions— life-like situations as well as his verbal in the person and in the situation—in which preferences or ratings (e.g., Mischel, 1966; it occurs. The many variables known to Mischel & Grusec, 1966). Verbal reports affect the emotional meaning and valence (e.g., on questionnaires) about values and of a stimulus include its context, sequenc- interests also may supply valuable infor- ing, and patterning (e.g., Helson, 1964); mation about the individual's preferences social comparison processes (e.g., Festinger, and aversions, and appear to provide some 1945); and the cognitive labels the person of the more temporally stable data in the assigns to his own emotional arousal state domain of personality (E. L. Kelly, 1955; (Schachter & Singer, 1962). Thus, like Strong, 1955). Alternatively, subjects may instrumental responses, emotional reac- be asked to rank-order actual rewards tions also tend to become far more dis- (Rotter, 1954), or the reinforcement value criminative than dispositional theories have of particular stimuli may be assessed assumed. Lazarus (1963), for example, directly by observing their effects on the has noted the specificity of sexual fears in individual's performance (e.g., Gewirtz & frigid women. For instance, one woman Baer, 1958). could calmly imagine herself engaged in Reinforcement (incentive) preferences certain sexual caresses, but only if they may also be assessed by providing in- occurred in the dark. Or consider the pilot dividuals opportunities to select the out- who became debilitatingly anxious when comes they want from a large array of flying, but only when his plane was higher alternatives, as when patients earn tokens than 9,000 feet (White, 1964), or the young which they may exchange for objects or woman who had asthmatic attacks mostly activities: the "price" they are willing to after she had contacts with her mother pay for particular outcomes provides an (Metcalf, 1956). Good illustrations of the index of their subjective value (e.g., Ayllon analysis of stimulus conditions influencing & Azrin, 1965). The concept that any emotional responses come from attempts to behavior which has a high natural fre- create subjective anxiety hierarchies (e.g., quency of occurrence can serve as a rein- Wolpe, 1961). forcer for other less likely behaviors (Pre- mack, 1965) also suggests that subjective Self-Regulatory Systems and Plans reinforcers may be discovered by asses- While behavior is controlled to a con- sing the individual's naturally occurring siderable extent by externally administered high frequency behaviors in particular consequences for actions, the individual situations (Mischel, 1968). also regulates his own behavior by self- A comprehensive assessment of stimulus imposed goals (standards) and self-pro- values must include attention to stimuli duced consequences. Even in the absence that have acquired strong emotion-eliciting of external constraints and social monitors, powers, as in the conditioned autonomic persons set performance goals for them- reactions seen in intense fears. For this selves and react with self-criticism or self- 274 WALTER MISCHEL satisfaction to their behavior depending on forcement with material rewards is prob- how well it matches their expectations and ably mediated extensively by covert sym- criteria. The concept of self-imposed bolic activities, such as praise and self- achievement standards is seen in Rotter's instructions, as the individual reaches (1954) "minimal goal" construct and in subgoals. When individuals imagine rein- more recent formulations of self-reinforcing forcing and noxious stimuli, their behavior functions (e.g., Bandura, 197Ic; Kanfer, appears to be influenced in the same manner 1971; Kanfer & Marston, 1963; Mischel, as when such stimuli are externally pre- 1968, 1973a). sented (e.g., Cautela, 1971). These covert The essence of self-regulatory systems is activities serve to maintain goal-directed the subject's adoption of contingency rules work until the performance matches or that guide his behavior in the absence of, exceeds the person's terminal standards and sometimes in spite of, immediate ex- (e.g., Meichenbaum, 1971). Progress along ternal situational pressures. Such rules the route to a goal is also mediated by self- specify the kinds of behavior appropriate generated distractions and cognitive opera- (expected) under particular conditions, the tions through which the person can trans- performance levels (standards, goals) which form the aversive "self-control" situation the behavior must achieve, and the con- into one which he can master effectively sequences (positive and negative) of at- (e.g., Mischel et al., 1972; Mischel & taining or failing to reach those standards. Moore, 1973a, 1973b). While achievement Each of these components of self-regulation of important goals leads to positive self- may be different for different individuals, appraisal and self-reinforcement, failure to depending on their unique earlier histories reach significant self-imposed standards or on more recently varied instructions or may lead the individual to indulge in other situational information. psychological self-lacerations (e.g., self- Some of the components in self-regula- condemnation). The anticipation of such tion have been demonstrated in studies of failure probably leads to extensive anxiety, goal setting and self-reinforcement (e.g., while the anticipation of success may help Bandura & Whalen, 1966; Bandura & to sustain performance, although the exact Perloff, 1967; Mischel & Liebert, 1966). mechanisms of self-regulation still require Perhaps the most dramatic finding from much empirical study. these studies is that even young children Self-reactions and self-regulation also are will not indulge themselves with freely influenced by the person's affective state. available immediate gratifications but, Following positive experiences, individuals instead, follow rules that regulate condi- become much more benign both toward tions under which they may reinforce them- themselves and others than after negative selves. Thus, children, like adults, far experiences. For example, after success from being simply hedonistic, make sub- experiences or positive mood inductions, stantial demands of themselves and impose there is greater selective attention to posi- complex contingencies upon their own tive information about the self (Mischel behavior. The stringency or severity of et al, 1973), greater noncontingent self- self-imposed criteria is rooted in the ob- gratification (e.g., Mischel, Coates, & Ras- served standards displayed by salient koff, 1968; Moore, Underwood, & Rosen- models as well as in the individual's direct han, 1973), and greater generosity (e.g., socialization history (e.g., Mischel & Lie- Isen, Horn, & Rosenhan, 1973). bert, 1966), although after they have been In conceptualizing the organization of adopted, the standards may be retained complex self-regulatory behavior, it will with considerable persistence. be necessary to consider the individual's After the standards (terminal goals) for "priority rules" for determining the se- conduct in a particular situation have been quencing of behavior and "stop rules" for selected, the often long and difficult route the termination of a particular sequence of to self-reinforcement and external rein- behavior. The ideas concerning "plans" RECONCEPTUALIZATION OF PERSONALITY 275 as hierarchical processes which control the TABLE 1 order in which an organism performs a SUMMARY OF COGNITIVE SOCIAL LEARNING sequence of operations, proposed by Miller, PERSON VARIABLES Galanter, and Pribram (1960), seem rele- vant. Subjectively, we do seem to generate 1. Construction competencies: ability to construct plans, and once a plan is formed (to go (generate) particular cognitions and behaviors. Related to measures of IQ, social and cognitive on a trip, to marry, to move to a new job, (mental) maturity and competence, ego develop- to write a paper) a whole series of sub- ment, social-intellectual achievements and skills. routines follows. While intuitively plaus- Refers to what the subject knows and can do. ible, the concept of plans has not yet 2. Encoding strategies and personal constructs: units for categorizing events and for self- stimulated the necessary personality-ori- descriptions. ented cognitive research. Promising steps 3. Behavior-outcome and stimulus-outcome ex- toward the study of plans are the concepts pectancies in particular situations. of behavioral intentions (Dulany, 1962), 4. Subjective stimulus values: motivating and intention statements, and contracts (e.g., arousing stimuli, incentives, and aversions. 5. Self-regulatory systems and plans: rules and self- Kanfer, Cox, Greiner, & Karoly, 1973). reactions for performance and for the organization Although self-instructions and intention of complex behavior sequences. statements are likely to be essential com- ponents of the individual's plans and the hierarchical organization of his self-regu- latory behavior, at present these topics to construct the preferred response. For provide perhaps the largest void and the example, due to differences in skill and greatest challenge in personality psy- prior learning, individual differences may chology. arise in interpersonal problem solving, To summarize, a comprehensive ap- empathy and role taking, or cognitive- proach to person variables must take ac- intellective achievements. Response dif- count of the individual's self-regulatory ferences also may reflect differences in how systems. These systems include: the rules individuals categorize a particular situa- that specify goals or performance standards tion (i.e., in how they encode, group, and in particular situations; the consequences of label the events that comprise it) and in achieving or failing to achieve those cri- how they construe themselves and others. teria ; self-instructions and cognitive stim- Differences between persons in their per- ulus transformations to achieve the self- formance in any situation depend on their control necessary for goal attainment; and behavior-outcome and stimulus-outcome organizing rules (plans) for the sequencing expectancies, that is, differences in the ex- and termination of complex behavioral pected outcomes associated with particular patterns in the absence of external sup- responses or stimuli in particular situations. ports and, indeed, in the face of external Performance differences also may be due to hindrances. differences in the subjective values of the outcomes expected in the situation. Overview of Person Variables Finally, individual differences may be due to differences in the self-regulatory systems In sum, individual differences in be- and plans that each person brings to the havior may reflect differences in each of situation. the foregoing person variables and in their interactions, summarized in Table 1. COGNITIVE SOCIAL LEARNING First, people differ in their construction VIEW OF INTERACTION competencies. Even if people have similar expectancies about the most appropriate In this final section, some issues in cur- response pattern in a particular situation rent theorizing about personality will be and are uniformly motivated to make it, reconsidered and interpreted in light of the they may differ in whether or not (and how proposed cognitive social learning person well) they can do it, that is, in their ability variables. These issues concern the role 276 WALTER MISCHEL of individual differences and the specific him is equally likely to be equally ap- interaction of person variables and situa- propriate (i.e., will lead to similar con- tions. sequences), and variance from individual differences will be greatest. Conversely, When Do Individual Differences Make a when subjects expect that only one re- Difference? sponse will be reinforced (e.g., only one From the present viewpoint, the condi- "right" answer on an achievement test, tions or "situational variables" of the only one correct response for the driver psychological environment provide the when the traffic light turns red) and that individual with information which in- no other responses are equally good, and fluences the previously discussed person all subjects are motivated and capable of variables, thereby affecting cognitive and making the appropriate response, then behavioral activities under those conditions. individual differences will be minimal and "Situations" thus affect behavior insofar situational effects prepotent. To the as they influence such person variables as degree that subjects are exposed to power- the individual's encoding, his expectancies, ful treatments, the role of individual dif- the subjective value of stimuli, or the ferences will be minimized. Conversely, ability to generate response patterns. In when treatments are weak, ambiguous, or light of the proposed set of person vari- trivial, individual differences in person ables, it is now possible to return to the variables should exert significant effects. question of when situations are most likely There have been several empirical to exert powerful effects and, conversely, demonstrations of these points. Mischel when person variables are likely to be and Staub (1965) examined some of the most influential. conditions determining the interaction and Psychological "situations" and "treat- relative importance of individual differ- ments" are powerful to the degree that they ences and situations. Adolescent subjects lead all persons to construe the particular were assessed on a measure of their ex- events the same way, induce uniform ex- pectancies for success in ability areas. pectancies regarding the most appropriate Three weeks later, they worked on a series response pattern, provide adequate in- of problems and in one treatment obtained centives for the performance of that re- success, in a second, failure, and in a third, sponse pattern, and instill the skills neces- no information. Next, they had to make sary for its satisfactory construction and many choices, including one between a execution. Conversely, situations and noncontingent but less preferred reward treatments are weak to the degree that they and a more preferred reward whose at- are not uniformly encoded, do not generate tainment was contingent upon their suc- uniform expectancies concerning the de- cessful performance on a task similar to sired behavior, do not offer sufficient in- the one on which they had previously centives for its performance, or fail to either succeeded, failed, or received no provide the learning conditions required for information. On this choice, situational successful construction of the behavior. success and failure had the expected effects: Individual differences can determine subjects who had succeeded chose much behavior in a given situation most strongly more often to work for the contingent when the situation is ambiguously struc- preferred reward than did those who had tured (as in projective testing) so that failed. The effects of situational success subjects are uncertain about how to cate- and failure were so strong that they wiped gorize it and have no clear expectations out the role of individual differences in about the behaviors most likely to be ap- preexperimental expectancy for success. propriate (normative, reinforced) in that But in the "no-information" condition (in situation.* To the degree that the situa- which subjects obtained no feedback about tion is "unstructured," the subject will their performance quality in the situation) expect that virtually any response from preexperimental expectancy was a highly RECONCEPTUALIZATION OF PERSONALITY 277 significant determinant of their choice to edge of stimulus conditions is seen, for work for contingent rewards. Thus situa- example, in predictive studies regarding tional manipulations which provided new posthospital prognosis for mental patients. expectancies minimized the effects of rele- Of special interest are studies which re- vant preexisting individual differences, vealed that the type, as well as the severity, but when situational variables were weak of psychiatric symptoms depended strik- or ambiguous (the no-information about- ingly on whether the person was in the performance condition) the expectancies hospital or in the community, with little that persons brought to the situation consistency in behavior across changing affected their behavior. Similar conclusions situations (Ellsworth, Foster, Childers, come from a recent study investigating the Gilberg, & Kroeker, 1968). Moreover, influence of success and failure experiences accurate predictions of posthospital ad- on subsequent selective attention to in- justment hinged on knowledge of the en- formation about the self (Mischel et al., vironment in which the ex-patient will be 1973). living in the community, such as the avail- The complex social settings of life also ability of jobs and family support, rather may be construed as varying in the degree than on any measured person variables or to which they prescribe and limit the in-hospital behavior (e.g., Fairweather, range of expected and acceptable behavior 1967). In another context, predictions of for persons in particular roles and settings intellectual achievement are greatly im- and hence permit the expression of in- proved if they take account of the degree dividual differences (e.g., Barker, 1966). to which the child's environment supports In some settings the rules and prescriptions (models and reinforces) intellectual de- for enacting specific role behaviors impose velopment (Wolf, 1966). Finally, when narrow limits on the range of possible be- powerful treatments are developed, such as haviors (e.g., in church, at school, in a modeling and desensitization therapies for theatre, at a conference), while in others the phobias, predictions about outcomes are range of possible behaviors is broad and best when based on knowing the treatment often the individual can select, structure, to which the individual is assigned (e.g., and reorganize situations with minimal Bandura, Blanchard, & Ritter, 1969). On external constraints. Because in particular the other hand, when relevant situational settings certain response patterns are information is absent or minimal, or when reinforced while others are not, different predictions are needed about individual settings become the occasion for particular differences in response to the same condi- behaviors in different degrees. Raush tions, or when treatment variables arc (1965), for example, found that in a sample weak, information about person variables of normal American boys, friendly acts led becomes essential. to unfriendly responses in 31% of the in- stances in game situations but in only 4% Specific Interactions between Behavior and of the time at mealtimes. Conditions Person—condition interactions are never Traditionally, trait research has studied static, but environmental stabilities can individual differences in response to the be identified which help to account for "same" situation. But some of the most continuities in behavior and permit useful striking differences between persons may predictions (e.g., Mischel, 1968). While it be found not by studying their responses to would be bizarre to ignore the person in the same situation but by analyzing their the psychology of personality, behavior selection and construction of stimulus often may be predicted and controlled conditions. In the conditions of life out- efficaciously from knowledge about relevant side the laboratory the psychological "stim- stimulus conditions, especially when those uli" that people encounter are neither conditions are powerful (Mischel, 1968). questionnaire items, nor experimental in- The potency of predictions based on knowl- structions, nor inanimate events, but they 278 WALTER MISCHEL involve people and reciprocal relationships about "behavior," it may be more useful (e.g., with spouse, with boss, and with to conceptualize behavior-contingency units children). The person continuously in- that link specific patterns of behavior to fluences the "situations" of his life as well the conditions in which they may be ex- as being affected by them in a mutual, pected. Accurate descriptions require spec- organic two-way interaction. These inter- ifying as precisely as possible the response actions reflect not only the person's reac- mode of the behavior as well as the con- tions to conditions but also his active selec- tingencies in which it is expected to be of tion and modification of conditions through high or low frequency, as was discussed in his own cognitions and actions. earlier sections on situational moderator As the analysis of complex social inter- variables. Thus rather than describe a actions illustrates (e.g., Patterson & Cobb, person as "aggressive," it would be neces- 1971), the person continuously selects, sary to qualify the mode of aggressive be- changes, and generates conditions just as havior (e.g., verbal insults but not physical much as he is affected by them. The attacks) and the specific contingencies (e.g., mutual interaction between person and when criticized for poor athletic perfor- conditions (so easily forgotten when one mance on playground but not in class). Such searches for generalized traits on paper- cumbersome, hyphenated descriptions (e.g., and-pencil tests) cannot be overlooked Mischel, 1969) would lack the "thumbnail when behavior is studied in the inter- sketch" appeal of global trait portraits. personal contexts in which it is evoked, But they would remind us of the discrim- maintained, and modified. inativeness and complexity of the indivi- Generally, changes in behavior toward dual's behavior, its idiosyncratic organiza- others tend to be followed by reciprocal tion, its dependence on conditions, and the changes in the behavior of those others hazards of attempting to abbreviate it (Raush et al., 1959). In Raush's (1965) grossly. studies of naturalistic interactions, for The previously discussed person variables example, "the major determinant of an act should make it plain that, a cognitive social was the immediately preceding act. Thus learning approach does not construe the if you want to know what child B will do, individual as an empty organism buffeted the best single predictor is what child A did entirely by situational forces. Yet it to B the moment before [p. 492]." Con- should be equally apparent that the nature strued from the viewpoint of Child A, and effects of these person variables depend this means that A's own behavior deter- on specific interactions between the in- mines B's reactions to him. In that sense, dividual and the psychological conditions the person is generating his own conditions. of his life. Construction capacities cannot Such subject variables as the person's be adequately understood without linking expectancies, self-regulatory rules, plans, them to the cognitive social learning condi- and constructs presumably guide the situa- tions through which they develop and are tions which he selects, generates, ^lnd struc- maintained and to the behaviors which tures for himself. they yield. Similarly, the study of ex- The proposed cognitive social learning pectancies must not lose sight of their roots approach to person variables emphasizes in the individual's direct and vicarious most strongly the need to study the in- experiences and of their ready modifi- dividual's behavior in specific interaction ability in the light of changes in behavior- with particular conditions. Indeed, the outcome and stimulus-outcome relation- conceptualization of behavior, whether ships. While subjective stimulus values psychologist defined (as in research) or and the individual's preferences and aver- subject defined (as in clinical, individually sions may have a greater degree of sta- oriented assessment) must be embedded in bility, their meaning and impact also relation to the specific conditions in which hinge on the specifics of the conditions the behavior occurs. Rather than talk in which they occur. Self-regulatory RECONCEPTUALIZATION OF PERSONALITY 279 rules, standards, and plans serve to impose perienced. And it recognizes that the additional continuity and consistency upon person's behavior changes the situations of behavior and guide the individual in the his life as well as being changed by them. absence of immediate situational forces. The term "personality psychology" need Yet such standards, rules, and plans also not be preempted for the study of differ- are not situation free, and their flexibility ences between individuals in their con- in response to changing conditions provides sistent attributes: it fits equally well the further testimony to human adaptiveness. study of the individual's cognitive and be- Perhaps substantial immunity to situa- havioral activities as he interacts with the tional changes is shown by some of the conditions of his life. individual's personal constructs. The "theories" formed about behavior (as in Three Perspectives in Personality Study the subject's implicit personality theories The study of persons may be construed about self and others) may be some of the alternatively from three complementary most stable and situation-free construc- perspectives. Construed from the view- tions. That has double-edged conse- point of the psychologist seeking procedures quences; the person's constructs provide a or operations necessary to produce changes measure of perceived stability in an other- in performance, it may be most useful to wise excessively complex, disorganized, focus on the environmental conditions and unstable world, but they also may necessary to modify the subject's behavior become hard to disconfirm. Yet even in and therefore to speak of "stimulus con- the realm of constructs, consistency is far trol," "operant conditioning," "classical from pervasive. For example, Gergen's conditioning," "reinforcement control," (1968) findings reveal that contrary to the "modeling" and so on. Construed from popular belief, when it comes to their the viewpoint of the theorist concerned self-perceptions people do not have a con- with how these operations produce their sistent, unitary self-concept. Indeed, he effects in the subject who undergoes them, concludes with regard to the phenomena of it may be more useful to speak of altera- self-concepts that "inconsistency" rather tions in processed information and specif- than "consistency" seems to be the natural 5 ically in constructs, expectancies, subjec- state of affairs. tive values, rules, and other theoretical The proposed approach to personality person variables that mediate the effects of psychology emphasizes the interdependence conditions upon behavior. Construed from of behavior and conditions, mediated by the viewpoint of the experiencing subject, the constructions and cognitive activities it may be more useful to speak of the same of the person who generates them, and events in terms of their phenomenological recognizes the human tendency to invent impact as thoughts, feelings, wishes, and constructs and to adhere to them as well other subjective (but communicable) in- as to generate subtly discriminative be- ternal states of experience. Confusion haviors across settings and over time. It arises when one fails to recognize that the emphasizes the crucial role of situations same events (e.g., the "operant condi- (conditions) but views them as informa- tioning" of a child's behavior at nursery tional inputs whose behavioral impact de- school) may be alternatively construed from' pends on how they are processed by the each of these perspectives and that the. person. It focuses on how such informa- choice of constructions (or their combina- tion processing hinges, in turn, on the prior tions) depends on the construer's purpose. conditions which the individual has ex- Ultimately, conceptualizations of the field 6 In the same vein, in their analysis of sources of of personality will have to be large enough variance in personal constructs, Argyle and Little to encompass the phenomena seen from (1972 )found that^the average variation attributable to persons was only 16.1%, whereas the percentages all three perspectives. The present cogni- for situations and interaction were 43.6 and 40.2, tive social learning approach to persons respectively. hopefully is a step in that direction. 280 WALTER MISCHEL

REFERENCES ment systems. Journal of Personality and Social Psychology, 1967, 7, 111-116. ABKLSON, R. Are attitudes necessary? In B. T. BANDURA, A., & WHALEN, C. K. The influence of King & E. McGinnies (Eds.), Attitudes, conflict, antecedent reinforcement and divergent modeling and social change. New York: Academic Press, cues on patterns of self-reward. Journal of 1972. Personality and Social Psychology, 1966, 3, 373- ADELSON, J. Personality. Annual Review of 382. Psychology, 1969, 20, 217-252. BARKER, R. The stream of behavior. New York: ADINOLFI, A. A. Relevance of person perception Appleton-Century-Crofts, 1966. research to clinical psychology. Journal of Con- BEM, D. J. Constructing cross-situational con- sulting and Clinical Psychology, 1971, 37, 167-176. sistencies in behavior: Some thoughts on Alker's ALKER, H. A. Is personality situationally specific critique of Mischel. Journal of Personality, or intrapsychically consistent? Journal of 1972, 40, 17-26. Personality, 1972, 40, 1-16. BLOCK, J. The challenge of response sets. New York: ALI.PORT, G. Personality: A psychological interpreta- Appleton-Century-Crofts, 1965. tion. New York: Holt, 1937. BLOCK, J. Some reasons for the apparent incon- ANDERSON, J. The prediction of adjustment over sistency of personality. Psychological Bulletin, time. In I. Iscoe & H. Stevenson (Eds.), Person- 1968, 70, 210-212. ality development in children. Austin: University BOLLES, R. C. Reinforcement, expectancy, and of Texas Press, 1960. learning. Psychological Review, 1972, 79, 394—409. ANDERSON, N. H. Integration theory and attitude BOWER, G. H. Organizational factors in memory. change. Psychological Review, 1971, 78, 171-206. Cognitive Psychology, 1970, 1, 18-46. ANDERSON, N. H. Information integration theory: BOWERS, K. Situationism in psychology: On mak- A brief survey. (Tech. Rep. No. 24) La Jolla: ing reality disappear. Research Reports in Psychol- University of California at San Diego, Center for ogy, No. 37, University of Waterloo, 1972. Human Information Processing, 1972. BYRNE, D. The repression—sensitization scale: ARGYLE, M., & LITTLE, B. R. Do personality rationale, reliability, and validity. Journal of traits apply to social behavior? Journal of Theory Personality, 1961, 29, 334-349. of Social Behavior, 1972, 2, 1-35. CAMPBELL, D. T. Conformity in psychology's ARONFREED, J. Conduct and conscience: The theories of acquired behavioral dispositions. In socialization of internalized control over behavior. I. A. Berg & B. M. Bass (Eds.), Conformity and New York: Academic Press, 1968. deviation. New York: Harper, 1961. AVKRILL, J. R., OLBRICH, E., & LAZARUS, R. S. CAMPBHLL, D., & FISKE, D. W. Convergent and Personality correlates of differential responsive- discriminant validation. Psychological Bulletin, ness to direct and vicarious threat: A failure to 1959, 56, 81-105. replicate previous findings. Journal of Personality CAUTELA, J. R. Covert conditioning. In A. Jacobs and Social Psychology, 1972, 21, 25-29. & [.. B. Sachs (Eds.), The psychology of private AVI.LON, T., & AZRIN, N. H. The measurement and events. New York: Academic Press, 1971. reinforcement of behavior of psychotics. Journal CHAPMAN, L. J., & CHAPMAN, J. P. Illusory cor- of the Experimental Analysis of Behavior, 1965, 8, relations as an obstacle to the use of valid psycho- 357-383. diagnostic signs. Journal of Abnormal Psychology, BANDURA, A. Influence of model's reinforcement 1969, 74, 271-280. contingencies on the acquisition of imitative CHATTERJEE, B. B., & ERIKSEN, C. W. Cognitive responses. Journal of Personality and Social factors in heart rate conditioning. Journal of Psychology, 1965, 1, 589-595. Experimental Psychology, 1962, 64, 272-279. BANDURA, A. Principles of behavior modification. CRAIK, K. C. Personality unvanquished. Con- New York: Holt, Rinehart & Winston, 1969. temporary Psychology, 1969, 14, 147-148. DAHLSTROM, W. C. Personality. Annual Review of BANDURA, A. Analysis of modeling processes. In Psychology, 1970, 21, 1-48. A. Bandura (Ed.), Psychological modeling: Con- D'ANDRADE, R. G. Cognitive structures and judg- flicting theories. Chicago: Aldine-Atherton, 1971. ment. Paper prepared for T.O.B.R.E. Research (a) Workshop on Cognitive Organization and Psycho- BANDURA, A. Social learning theory. New York: logical Processes, Huntington Beach, California, General Learning Press, 1971. (b) August 16-21, 1970. La Jolla: University of Cali- BANDURA, A. Vicarious and self-reinforcement pro- fornia at .San Diego, Department of Anthropology, cesses. In R. Glaser (Ed.), The nature of rein- 1970. forcement. New York: Academic Press, 1971. (c) D'ANDRADE, R. G. Memory and the assessment of BANDURA, A., BLANCHARD, E. B., & RITTKR, B. behavior. Unpublished manuscript, University Relative efficacy of desensitization and modeling of California at San Diego, Department of approaches for inducing behavioral, affective, and Anthropology, 1973. attitudinal changes. Journal of Personality and DULANY, D. H. The place of hypotheses and inten- Social Psychology, 1969, 13, 173-199. tions: An analysis of verbal control in verbal BANDURA, A., & PERLOFF, B. Relative efficacy of conditioning. Journal of Personality, 1962, 30, self-monitored and externally imposed reinforce- 102-129. RECONCEPTUALIZATION OF PERSONALITY 281

EDWARDS, A. L. The social desirability variable in GRINGS, W. W., & LOCKHART, R. A. Effects of personality assessment and research. New York: anxiety-lessening instructions and differential Dryden, 1957. set development on the extinction of GSR. ELLSWORTH, R. B., FOSTER, L., CHILDERS, B. Journal of Experimental Psychology. 1963, 66, GILBERG, A., & KROEKER, D. Hospital and 292-299. community adjustment as perceived by psychi- GRUSEC, J., & MISCHEL, W. Model's characteristics atric patients, their families, and staff. Journal as determinants of social learning. Journal of of Consulting and Clinical Psychology, 1968, 32 Personality and Social Psychology, 1966, 4, 211- (S, Pt. 2). 215. EMMERICH, W. Models of continuity and change. HAMILTON, D. L. Implicit personality theories: Paper presented at the meeting of the Society Dimensions of interpersonal cognition. Paper for Research in Child Development, March 27, presented at the meeting of the American 1969, Santa Monica, California. Psychological Association, Washington, D. C;, ENDLER, N. S., & HUNT, J. McV. Sources of be- September 1971. havioral variance as measured by the S-R in- HAYDEN, T., & MISCHEL, W. Maintaining trait ventory of anxiousness. Psychological Bulletin, consistency in the resolution of behavioral incon- 1966, 65, 336-346. sistency: The wolf in sheep's clothing. Un- ENDLER, N. S., & HUNT, J. McV. S-R inventories published manuscript, Stanford University, 1973. of hostility and comparisons of the proportions HEIDER, F. The psychology of interpersonal relations. of variance from persons, responses, and situa- New York: Wiley, 1958. tions for hostility and anxiousness. Journal of Personality and Social Psychology, 1968, 9, 309- HELSON, H. Adaptation-level theory. New York: 315. Harper & Row, 1964. ENDLER, N. S., & HUNT, J. McV. Generalizability HUNT, J. McV. Traditional personality theory in of contributions from sources of variance in the the light of recent evidence. American Scientist, S-R inventories of anxiousness. Journal of 1965, S3, 80-96. Personality, 1969, 37, 1-24. IRWIN, F. W. Intentional behavior and motivation. ENDLER, N. S., HUNT, J. McV., & ROSENSTEIN, A. New York: Lippincott, 1971. J. An S-R inventory of anxiousness. Psycho- ISEN, A., HORN, N., & ROSENHAN, D. Effects of logical Monographs, 1962, 76 (17, Whole No. 536). success and failure on children's generosity. ESTES, W. K. Reinforcement in human behavior. Journal of Personality and Social Psychology, American Scientist, 1972, 60, 723-729. 1973, 27, 239-247. FAIRWEATHER, G. W. Methods in experimental JEFFERY, K., & MISCHEL, W. The layman's use of social innovation. New York: Wiley, 1967. traits to predict and remember behavior. Un- FESTINGER, L. A theory of social comparison pro- published manuscript, Stanford University, 1973 cesses. Human Relations, 1945, 7, 117-140. JONES, E. E. Ingratiation: A social psychological' FISKE, D. W. The limits of the conventional science analysis. New York: Appleton-Century-Crofts, of personality. Unpublished manuscript, Uni- 1964. versity of Chicago, 1973. JONES, E. E., & NISBETT, R. E. The actor and the FRIEZE, L, & WEINER, B. Cue utilization and at- observer: Divergent perceptions of the causes of tributional judgments for success and failure. behavior. In E. E. Jones et al. (Eds.), Attribu- Journal of Personality, 1971, 39, 591-605. tion: Perceiving the causes of behavior. McCaleb- GEER, J. H., DAVISON, G. C, & GATCHEL, R. K. Seiler, 1971. Reduction of stress in humans through nonveridi- KANFER, F. H. The maintenance of behavior by cal perceived control of aversive stimulation. self-generated stimuli and reinforcement. In Journal of Personality and Social Psychology, A. Jacobs & L. B. Sachs (Eds.), The psychology 1970, 16, 731-738. of private events. New York: Academic Press, GERGEN, K. J. Personal consistency and the 1971. presentation of self. In C. Gordon & K. J. KANFER, F. H., Cox, L. E., GREINER, J. M., & ^' Gergen (Eds.), The self in social interaction. New KAROLY, P. Contracts, demand characteristics York: Wiley, 1968. and self-control. Unpublished manuscript, Uni- GEWIRTZ, J. L., & BAER, D. M. Deprivation and versity of Cincinnati, 1973. satiation of social reinforcers as drive conditions. KANFER, F. H., & MARSTON, A. R. Determinants Journal of Abnormal and Social Psychology. of self-reinforcement in human learning. Journal 1958, 57, 165-172. of Experimental Psychology, 1963, 66, 245-254. GIBSON, E. J. Principles of perceptual learning and KAUFMAN, A., BARON, A., & KOPP, R. E. Some development. New York: Appleton-Century- effects of instructions on human operant behavior. Crofts, 1969. Psychonomic Monograph Supplements, 1966, 1, GOLDFRIED, M. R., & D'ZURILLA, T. J. A be- 243-250. havioral-analytic model for assessing competence. KELLEY, H. H. Attribution theory in social psychol- In C. D. Spielberger (Ed.), Current topics in ogy. In D. Levine (Ed.), Nebraska Symposium clinical and community psychology. Vol. 1. New on Motivation: 1967. Lincoln: University of York: Academic Press, 1969. Nebraska Press, 1967. 282 WALTER MISCHEL

KELLY, E. L. Consistency of the adult personality. MISCHEL, W. Continuity and change in personality. American Psychologist, 1955, 10, 659-681. American Psychologist, 1969, 24, 1012-1018. KELLY, G. The psychology of personal constructs. MISCHEL, W. Introduction to personality. New York: New York: Basic Books, 1955. Holt, Rinehart & Winston, 1971. KOHLBERG, A. A cognitive-developmental analysis MISCHEL, W. Direct versus indirect personality of children's sex-role concepts and attitudes. In assessment: Evidence and implications. Journal E. E. Maccoby (Ed.), The development of sex dif- of Consulting and Clinical Psychology, 1972, 38, ferences. Stanford: Stanford University Press, 319-324. 1966. MISCHEL, W. Processes in delay of gratification. KOHLBERG, L. Stage and sequence: The cognitive- In L. Berkowitz (Ed.), Advances in social psy- developmental approach to socialization. In D. chology. Vol. 7. New York: Academic Press, A. Goslin (Ed.), Handbook of socialization theory 1973, in press, (a) and research. Chicago: Rand McNally, 1969. MISCHEL, W. On the empirical dilemmas of psycho- LAWRENCE, D. H. The nature of a stimulus: dynamic theory: Issues and alternatives. Journal Some relationships between learning and percep- of Abnormal Psychology, 1973, in press, (b) tion. In S. Koch (Ed.), Psychology: A study of a MISCHEL, W., & BAKER, N. Cognitive appraisals science. New York: McGraw-Hill, 1959. and transformations in delay behavior. Un- LAXARUS, A. A. The treatment of chronic frigidity published manuscript, Stanford University, 1973. by systematic desensitization. Journal of Nervous MISCHEL, W., COATES, B., & RASKOFF, A. Effects and Mental Disease, 1963, 136, 272-278. of success and failure on self-gratification. LITTLE, B. R., & STEPHENS, E. D. Psychological Journal of Personality and Social Psychology, construing and selective focusing on content 1968, 10, 381-390. versus expressive aspects of speech. Journal of MISCHEL, W., EBBESEN, E. B., & ZEISS, A. R. Consulting and Clinical Psychology, 1973, in press. Cognitive and attcntional mechanisms in delay of LOVAAS, O. L, FREITAG, G., GOLD, V. J., & KAS- gratification. Journal of Personality and Social SORLA, I. C. Experimental studies in childhood Psychology, 1972, 21, 204-218. schizophrenia: I. Analysis of self-destructive be- MISCHEL, W., EBBESEN, E. B., & ZEISS, A. R. havior. Journal of Experimental Child Psychology, Selective attention to the self: Situational and 1965, 2, 67-84. dispositional determinants. Journal of Personal- MACFARLANE, J. W., & TuDDENHAM, R. D. Prob- ity and Social Psychology, 1973, 27, 129-142. lems in the validation of protective techniques. MISCHEL, W., & GRUSEC, J. Determinants of the In H. H. Anderson & G. L. Anderson (Eds.), rehearsal and transmission of neutral and aver- Projeclive techniques. New York: Prentice-Hall, sive behaviors. Journal of Personality and Social 1951. Psychology, 1966, 3, 197-205. MANDLER, G. Organization and memory. In K. MISCHEL, W., GRUSEC, J., & MASTERS, J. C. W. Spence & J. T. Spence (Eds.), The psychology Effects of expected delay time on the subjective of learning and motivation: Advances in research value of rewards and punishments. Journal of and theory. New York: Academic Press, 1967. Personality and Social Psychology, 1969, 11, MANDLER, G. Association and organization: Eacts, 363-373. fancies and theories. In T. R. Dixon & D. L. MISCHEL, W., & LIEBERT, R. M. Effects of dis- Horton (Eds.), Verbal behavior and general be- crepancies between observed and imposed reward havior theory. Englcwood Cliffs, N. J.: Prentice- criteria on their acquisition and transmission. Hall, 1968. Journal of Personality and Social Psychology, McGuiRE, W. J. Personality and susceptibility 1966, 3, 45-53. to social influence. In E. F. Borgatta & W. W. MISCHEL, W., & MOORE, B. Effects of attention Lambert (Eds.), Handbook of personality theory to symbolically presented rewards upon self- and research. Chicago: Rand McNally, 1968. control. Journal of Personality and Social MEICHENBAUM, D. H. Cognitive factors in behavior Psychology, 1973, in press, (a) modification: Modifying what clients say to them- MISCHEL, W., & MOORE, B. Cognitive transforma- selves. (Research Report No. 25) Waterloo: tions of the stimulus in delay of gratification. University of Waterloo, 1971. Unpublished manuscript, Stanford University, METCALF, M. Demonstration of a psychosomatic 1973. (b) relationship. British Journal of Medical Psy- MISCHEL, W., & STAUB, E. Effects of expectancy chology, 1956, 29, 63-66. on working and waiting for larger rewards. MILLER, G. A., GALANTER, E., & PIUBRAM, K. H. Journal of Personality and Social Psychology, Plans and the structure of behavior. New York: 1965, 2, 625-633. Holt, Rinehart & Winston, 1960. MISCHEL, W., ZEISS, R., & ZEISS, A. R. Internal- MISCHEL, W. Theory and research on the antece- external control and persistence: Validation and dents of self-imposed delay of reward. In B. A. implications of the Stanford Preschool I-E Scale. Maher (Ed.), Progress in experimental personal- Journal of Personality and Social Psychology, 1973, ity research. Vol. 3. New York: Academic Press, in press. 1966. MOORE, B. S., UNDERWOOD, B., & ROSENHAN, D. L. MISCHEL, W. Personality and assessment. New Affect and altruism. Developmental Psychology, York: \Viley, 1968. 1973, 8, 99-104. RECONCEPTUALIZATION OF PERSONALITY 283

Moos, R. H. Situational analysis of a therapeutic tive behaviorism versus cognitive theory in verbal community milieu. Journal of Abnormal Psy- operant conditioning. Psychological Review, 1966, chology, 1968, 73, 49-61. 73, 306-326. Moos, R. H. Sources of variance in response to STRONG, E. K., JR. Vocational interests 18 years questionnaires and in behavior. Journal of after college. Minneapolis: University of Min- Abnormal Psychology, 1969, 74, 405-412. nesota Press, 1955. NEISSER, U. Cognitive psychology. New York: TVERSKY, A., & KAHNEMAN, D. Belief in the law Appleton-Century-Crofts, 1967. of small numbers. Psychological Bulletin, 1971, NORMAN, D. A. Memory and attention. New York: 76, 105-110. Wiley, 1969. VERNON, P. E. Personality assessment: A critical PATTERSON, G. R., & COBB, J. A. Stimulus control survey. New York: Wiley, 1964. for classes of noxious behaviors. In J. F. Knutson WACHTEL, P. L. Psychodynamics, behavior ther- (Ed.)i The control of aggression: Implications from apy, and the implacable experimenter: An in- basic research. Chicago: Aldine, 1971. quiry into the consistency of personality. Journal PETERSON, D. R. The clinical study of social be- of Abnormal Psychology, 1973, in press. havior. New York: Appleton-Century-Crofts, WALLACE, J. An abilities conception of personality: 1968. Some implications for personality measurement. PIAGET, J. The construction of reality in the child. American Psychologist, 1966, 21, 132-138. New York: Basic Books, 1954. WALLACH, M. A. Commentary: Active-analytical PREMACK, D. Reinforcement theory. In D. Levine versus passive-global cognitive functioning. In (Ed.), Nebraska Symposium on Motivation: 1965. S. Messick & J. Ross (Eds.), Measurement in Lincoln: University of Nebraska Press, 1965. personality and cognition. New York: Wiley, RAUSH, H. L. Interaction sequences. Journal of 1962. Personality and Social Psychology, 1965, 2, 487- WALLACH, M. A., & LEGGETT, M. I. Testing the 499. hypothesis that a person will be consistent: RAUSH, H. L., DITTMAN, A. L., & TAYLOR, T. J. Stylistic consistency versus situational specificity The interpersonal behavior of children in resi- in size of children's drawings. Journal of Person- dential treatment. Journal of Abnormal and ality, 1972, 40, 309-330. Social Psychology, 1959, 58, 9-26. WKINER, B., FRIEZE, I., KUKLA, A., REED, L., ROBBINS, L. N. Dissecting the "broken home" as a REST, S., & ROSENBAUM, R. M. Perceiving the predictor of deviance. Paper presented at the causes of success and failure. New York: General NIMH Conference on Developmental Aspects of Learning Press, 1971. Self-Regulation, La Jolla, California, February WEINER, B., & KUKLA, A. An attributional analysis 19, 1972. of achievement motivation. Journal of Person- ROTTER, J. B. Social learning and clinical psychol- ality and Social Psychology, 1970, IS, 1-20. ogy. Englewood Cliffs, N. J.: Prentice-Hall, WHITE, R. W. Motivation reconsidered: The con- 1954. cept of competence. Psychological Review, 1959, RUMMELHART, D. E., LlNDSEY, P. H., & NORMAN, 66, 297-333. D. A. A process model for long-term memory. WHITE, R. W. The abnormal personality. New (Tech. Rep. No. 17). La Jolla: University of York: Ronald Press, 1964. California at San Diego, Center for Human WITKIN, H. A. Psychological differentiation and Information Processing, 1971. forms of pathology, Journal of Abnormal Psy- SCHACHTER, S. The interaction of cognitive and chology, 1965, 70, 317-336. physiological determinants of emotional state. In WOLF, R. The measurement of environments. I n L. Berkowitz (Ed.), Advances in experimental A. Anastasi (Ed.), Testing problems in perspective. social psychology. Vol. 1. New York: Academic Washington, D. C.: American Council on Educa- Press, 1964. tion, 1966. SCHACHTER, S., & SINGER, J. E. Cognitive, social, WOLPE, J. The systematic desensitization treat- and physiological determinants of emotional ment of neuroses. Journal of Nervous and Mental state. Psychological Review, 1962, 69, 379-399. Disease, 1961, 132, 189-203. SCHNEIDER, D. J. Implicit personality theory: A ZIGLER, E., & PHILLIPS, L. Social competence and review. Psychological Bulletin, 1973, 79, 294-309. outcome in psychiatric disorder. Journal of SHWEDER, R. A. Is a culture a situation? Unpub- Abnormal and Social Psychology, 1961, 63, 264- lished manuscript, Harvard University, Depart- 271. ment of Social Relations, 1971. ZIGLER, E., & PHILLIPS, L. Social competence and SHWEDER, R. A. Semantic structures and person- the process-reactive distinction in psychopathol- ality assessment. Unpublished doctoral disserta- ogy. Journal of Abnormal and Social Psychology, tion Harvard University, Department of Social 1962, 65, 215-222. Relations, March 1972. SPIELBERGER, D. C., & DsNiKE, L. D. Descrip- (Received December 12, 1972)