Fifty Years of Personality

John F. Kihlstrom

University of Wisconsin

The year 1937, when Floyd Ruch's Psychology and Life appeared in its first edition, was also a signal year for the scientific study of personality. That year witnessed the appearance of Gordon Allport's Personality: A Psychological Interpretation, as well as the first edition of Ross Stagner's Psychology of Personality. Allport's book literally created the field of personality as a topic of academic study, just as his brother Floyd's text had done for social psychology 13 years earlier. There had been scientific work on personality before that time, of course -- else Ruch would not have been able to include in his first edition two chapters devoted to "Individual Differences" (Chapter 2) and "Personality and its Measurement" (Chapter 3) -- citing, among other items, a pioneering study of the measurement of ascendant and submissive traits conducted by the brothers Allport.

It is interesting to note that Ruch placed personality at a very early place in his first edition; later editions, and most other prominent texts, tend to put personality at the back of the book. Either placement is intellectually defensible. For example, individual personality might be seen as the culmination of the general mental processes described in the beginning of the course, and as a sort of prelude to the study of psychopathology. Still, it is a happy event to find personality treated at the very beginning of the book, as if it were the instigation for the study of general mental processes. Arguably, that is how most psychologists came to their field in the first place.

Ruch was not able, in his very first edition, to review much substantive empirical or theoretical work on the topic of personality. His discussion was centered on Galton and James McKeen Cattell, the concept of mental tests, Hull's work on aptitude testing, the beginnings of factor analysis, and the extension of this work by Woodworth and the Thurstones to the problem of personality measurement. It is instructive to compare the first edition with the latest, 11th edition (by Phil Zimbardo) in 1985. Now material relating to personality is located closer to tbe back of the book than the front. Many of the same themes come up in Chapter 12, "Assessing Individual Differences", which centers on a comparison of formal assessment methods with a lot of material on intelligence testing. But the real clue to the changes in the field is that this chapter is preceeded by another one, Chapter 11, that had no counterpart in the first edition: "Understanding Human Personality". Here we find a survey of not just the type and trait theories discussed by Ruch in 1937, but a large number of competing approaches including psychodynamic, humanistic, behavioristic, and cognitive theories of personality. The development of these theories of personality, and their heuristic use to guide empirical research, is the primary feature of the psychology of personality as it has developed over the ensuing 50 years.

Psychometric Approaches to Personality

The first scientific view of personality relied on the classical fourfold typology offered by Hippocrates, Galen, and Kant: melancholic, sanguine, choleric, and phlegmatic. Types were categories of people, and at the end of the 19th century categories were defined in terms of singly necessary and jointly sufficient defining features. But classical type theory encountered the problems of partial and combined expression that bedevil all attempts to construe natural categories in terms of proper sets: some people are much more representative of their type than others, and some people appear to combine the features of two or more types.

There are various ways to solve this problem: Ruch himself redefined types in terms of features distributed continuously but bimodally; more recent work on categorization by Rosch, Smith, Cantor, and Hampson, would suggest that types be construed in terms of fuzzy sets held together by family resemblance and represented by prototypes or exemplars. But the solution adopted at the time was that offered by Wundt: to reconstrue the classic fourfold typology in terms of two independent dimensions representing the speed and strength of emotional arousal. In this way, the theory of personality types was transformed into the theory of personality traits. With individual differences reconceptualized in terms of continuously distributed variables instead of discrete categories, the problems of partial and combined expression were solved. Personality measurement was put on a quantitative basis and the psychometric approach to personality had begun.

The psychometric approach to personality takes as its model the analysis of individual differences in intelligence and other human abilities. This viewpoint was clearly expressed in what might be called the "Doctrine of Traits". From this point of view, persons are viewed psychologically as collections of traits -- internal dispositions that cause individual differences in experience, thought, and action. Rather than being slotted into pigeonholes, individuals were located at points in multidimensional space, with the dimensions defined by the major personality traits. A large proportion of traditional personality research consists of the development of an instrument -- usually a paper-and-pencil questionnaire -- for assessing some trait, and demonstrating the validity of the questionnaire by relating it to some trait-relevant criterion behavior. The successful prediction of criterion behavior validates both the questionnaire and the investigator's theory of the trait -- a process known as construct validity. These efforts comprise the "meat and potatoes" of empirical research on such personalty traits as authoritarianism, achievement motivation, machiavellianism, and so on. Dimensions of Personality, a volume edited by London and Exner in 1978, provides good summaries of the most prominent of these efforts.

But how many traits are there? In 1936, Allport and Odbert had counted no less than 17,953 different trait terms in the English lexicon. Some of these were redundant, and others useless for personality description, but the problem remained of determining how many traits were needed to adequately conceptualize individual differences in personality. Thus began the search for a universally applicable structure of personality traits -- a highly economical, tightly organized set of trait terms that could be used to capture individual and group differences in any culture, in any temporal epoch. It should be noted that Allport himself, a devoted advocate of the uniqueness of each human personality, rejected the idea that individuals could be compared to each other. Nevertheless the search for a universal structure was initiated in the 1940s by Raymond B. Cattell, and continues today in what has come to resemble a quest for an personological Holy Grail -- a quest pursued mostly through factor analysis and other multivariate statistical techniques.

Now, almost fifty years after the quest began, the field has settled on two interrelated structures. One is Eysenck's "Holy Trinity" of extraversion-introversion, neuroticism, and psychoticism. The other is Norman's "Big Five" of extraversion, agreeableness, conscientiousness, emotional stability, and culturedness. Obviously, the two structures are related: Eysenck's E-I superfactor largely subsumes Norman's extraversion and agreeablness factors, and neuroticism is virtually synonymous with emotional stability; Eysenck's psychoticism factor is closely related to antisocial behavior, and thus conscientiousness; and Eysenck has proposed a fourth major dimension of individual differences, intelligence, which bears some relation to culturedness. The Big Five might be preferred on the grounds that Eysenck's psychoticism factor is not well established, and culturedness is considerably broader than mere intelligence. Thus, The Big Five appears to present an optimal balance between economy and richness of description. Further research will be required to determine whether it is applicable across a wide range of cultures, though preliminary work by Goldberg suggests an answer in the affirmative.

Just as Eysenck's E-I superfactor can be decomposed into Norman's extraversion and agreeableness, so these two traits can be further analyzed. In 1979, Wiggins showed that such interpersonal traits formed a regular circumplex -- a circular ordering in which the angular distances between traits indicates the correlation between them. In 1980 Russell proposed a similar circumplex ordering for the lexicon of affective state terms. The advent of these circumplexes is a new development in the search for a universally applicable structure of personality traits, but it does not contradict the earlier work leading up to The Big Five.

Where do traits come from? Allport himself was largely silent on this question, and other major trait theorists, such as Guilford and Cattell, have focused on descriptive and structural issues rather than explanatory and causal ones. A common theme that runs through the trait literature is that of heritibility: it is frequently supposed that individual differences in personality traits, like their counterparts in physical traits, have their origins in individual differences in genetic endowment. This position has been much debated in the literature on intelligence, and the arguments of Jensen and Eysenck, Kamin and Gould have their counterparts outside the domain of human abilities. Every so often a study is produced comparing identical and fraternal twins, or adopted children with their biological and adoptive parents, yielding evidence for a significant genetic component in one or more of the traits measured. Similarly, beginning with the classic study of Sears, Maccoby and Levin on patterns of child rearing, there are studies showing the influence of family environment on personality. For his part, Eysenck has moved beyond description to propose specific genetic-biochemical origins for his major personality traits: extraversion-introversion is rooted in levels of cortical arousal, neuroticism in autonomic arousal, psychoticism in testosterone; intelligence has its origins in speed of neural transmission.

Alternatively, Buss and Plomin have proposed that three basic temperaments of emotionality, activity, and sociability have their origins in inherited, constitutional factors. However, they also indicate how these innate individual differences interact with the social environment outside the child to produce individual differences in specific personality traits. A major challenge for personality trait theorists is to work out precisely how these interactions take place. One view is provided by Zajonc's confluence model, which holds that individuals form part of the environment that shapes them. Confluence theory, initially offered to account for the influence of birth order and family size on intelligence, may be extended to non-intellective features of personality as well.

One salutary result of the psychometric tradition has been the development of an increasingly sophisticated technology for the assessment of individual differences. Lanyon and Goodstein's Personality Assessment (1971) provide thorough coverage of this evolutionary history; more technically demanding coverage is provided in Wiggins' (1973)Personality and Prediction: Principles of Personality Assessment. The history of personality assessment begins with purely rational instruments such as Woodworth's Personal Data Sheet, introduced during World War I to identify draftees who were disposed to mental illness; it continues with multidimensional questionnaires developed through factor analysis, such as the Guilford-Zimmerman Personality Questionnaire and Cattell's 16 Personality Factor Questionnaire; and empirically derived inventories such as the Minnesota Multiphasic Personality Inventory and the California Psychological Inventory.

Each of these methods has its virtues: the rational method insures face validity, the factor-analytic method internal consistency, and the empirical method external validity. Although the empirical method would seem to be preferable, insofar as external validity is built into the test from the beginning, in fact studies by Goldberg and Jackson show that the three types of tests yield essentially equivalent validity coefficients. Because questionnaires devised by the rational method are arguably the least expensive, that method would seem to win over the others on grounds of utility. The most recent trend in the evolution of personality assessment has been the construction of new inventories by means of a combination of the three methods described. These instruments, such as the Jackson Personality Inventory and Tellegen's Multidimensional Personality Questionnaire, represent the state of the art in psychometry. However, it is too early to tell whether the increases in external validity are great enough to justify the enormous costs associated with their production.

The Challenge to the Psychometric Tradition

Although trait conceptions of personality dominated scientific research on personality from Allport's time on, the decade of the 1960s gave signs of increasing dissatisfaction with it. An early warning was given by the publication by Sarbin, Taft, and Bailey of Clinical Inference and Cognitive Theory (1960), and by Vernon of Personality Assessment: A Critical Survey; the gathering storm hit shore in 1968, with the near- simultaneous publicaton of Mischel's Personality and Assessment and Peterson's The Clinical Study of Social Behavior. Mischel's monograph was the signal event in the revolt against trait theory, mustering the evidence against traits (and their assessment), and offering an alternative conceptualization from the perspective of cognitive social learning theory.

When analyzed, the Doctrine of Traits appears to contain four distinct propositions. (a) Topographically different behaviors1 (such as smiling and talking) co-occur reliably, such that they represent different public manifestations of the same primary trait; and semantically different primary traits (such as ambitious and dominant) also tend to covary, such that they represent different facets of the same superordinate trait (such as extraversion). This gives traits the property of coherence, resulting in a hierarchical structure of individual differences with specific behaviors at the lowest level, and habitual behaviors, primary traits, and secondary and tertiary (etc.) traits at progressively higher levels. (b) Within each level of this hierarchical structure, there is appreciable stability over both long and short intervals of time. (c) And again, within each level there is consistency across a wide variety of different situations. (d) Finally, there is predictability in the sense that knowledge of individual differences at one level of the hierarchy permits reasonably accurate inferences about individual differences at another level. In particular, the Doctrine of Traits permits deductive inferences from the level of traits to the level of behavior.

Obviously no psychometrician expects to find perfect levels of coherence, stability, consistency, and predictability; at the very least there is measurement error to take into account. But from the point of view of the Doctrine of Traits, the more these qualities are displayed in the individual's experience, thought, and action, the better. The critical literature of the 1960s, however, cast profound doubt on the validity of all four propositions.

Consider the situation with respect to coherence. Comparing the classic factor-analytic approaches to the structure of personality, there was little consensus among Guilford, Cattell, and Eysenck concerning the names of the primary traits, or the relations between them. There is, of course, greater consensus on a structure like The Big Five. However, it is important to note that this structure is derived from memory-based ratings of personality, not from objective recordings of actual behavior, and that such data is highly vulnerable to contamination by the raters'implicit theories of personality -- i.e., their preconceived notions concerning te interrelations among behaviors and traits. Analyses by Richard Shweder, among others, strongly questions whether the coherence among personality traits represented by The Big Five (and similar structures) is to be found in the behavior of the observed or the mind of the observer.

With respect to stability, two types of evidence are available. Studies using self-report questionnaires typically reveal extremely high test-retest reliability coefficients, even over intervals of 20 years. But again, following Shweder, the accuracy of self-perceived stability must be questioned. What is needed, obviously, are observations or ratings made independently at two different time periods. A number of longitudinal studies of temperament summarized by Buss and Plomin reveal levels of stability that are rather modest, if statistically significant. Similar findings were obtained from the Fels Institute research of Kagan and Moss, and from the Berkeley study of Block. The bottom line is that there is at least as much change as stability across even short periods of time, and that stability decreases with increasing temporal intervals.

A similar outcome has been obtained in studies of cross-situational consistency. Beginning with the classic Hartshorne and May (1928) study of honesty in children (reported by Ruch in the first edition), through Mischel and Peake's study of friendliness and conscientiousness in college students, a number of investigators have contrived to make trait-relevant behavioral observations in widely different situational contexts. As in the study of stability, the evidence for consistency is quite modest. As a rule, consistency is highest two highly similar behaviors are measured in two highly similar situations (where the behaviors and situations are identical, the result is an estimate of test-retest reliability that bears on the question of stability); as either behaviors or test situations diverge, consistency diminishes progressively.

Finally, the degree to which specific behaviors (in specific situations) can be predicted from knowledge of generalized traits is rather poor. In 1968, Mischel coined the term "personality coefficient" to characterize the correlation of approximately 0.30 typically obtained between a questionnaire measurement of a trait and some the occurrence of trait-relevant behavior in some specific situation. The value is significantly greater than chance, but of course it leaves a large proportion of behavioral variance unaccounted for by the trait measure. Mischel's estimate was based on only a cursory survey of the relevant literature, but inspection of a number of different domains (such as the studies summarized in London & Exner, 1978) suggests it is not far off the mark. In summary, then, it appears that the evidence for coherence is ambiguous, and the evidence for stability, consistency, and predictability is modest. There is not to say that human experience, thought, and action is random and haphazard, and that persons are inherently unpredictable. Obviously they aren't. But evidence of the sort summarized by Mischel suggests that traits are relatively weak determinants of behavior, and that other factors must be considered in describing and explaining personality.

Given the dominance of the psychometric viewpoint within personality, it would be expected that Mischel's criticism would generate vigorous replies. Beginning in the early 1970s, and continuing into the present, there has been a continuing exchange of position papers, replies, and rejoinders in the pages of the Psychological Review, Psychological Bulletin, and American Psychologist. One common theme (announced by Block and Epstein, among others) has been that the instruments usually used to provide the predictor variables in such studies are far from perfect psychometrically; and that the single-item behavioral criteria usually targeted are inherently unreliable. Under these circumstances, a validity correlation of 0.30 might be very good indeed, and Mischel's criticism would be unfair. And studies of cross-situational consistency, which often correlate one single-item index of behavior with another, would be severly constrained and thus almost irrelevant. But the personality coefficient is not merely an artifact of the failure to employ psychometrically adequate predictors and aggregate criteria. For example, two of the "state-of-the- art" self-report inventories (which are, almost by definition, psychometrically adequate) described earlier were validated against peer ratings on the same trait dimensions (which aggregate information across behaviors, contexts, time, and observers).

Another tack has been to propose that there are individual differences in consistency itself, and further that consistency, far from being a generalized trait-like disposition, varies across trait domains within subjects. Thus, in the phrase of Bem and Allen, it may only be possible to predict "some of the people some of the time". Bem and Allen showed that self-reported consistency in friendliness was, in fact, a fairly good predictor of observed cross-situational consistency in friendly behavior. However, a similar analysis of conscientiousness was unsuccessful, and their findings were not confirmed by Goldberg and his associates. So, the sorts of critiques leveled against the Doctrine of Traits remain relatively intact. Traits may exist, they may be measurable by questionnaires and rating scales, and they may exert some palpable influence on experience, thought, and action, but there is more to personality than traits.

The Psychodynamic Alternative

Although it took some time before they were fully felt, the implications of the low levels of consistency obtained by Hartshorne and May, among others, were appreciated at the time of Ruch's first edition. Outside the domain of academic psychology, however, the clinic offered a rather different view of personality -- one that was not so troubled by evidence for incoherence, instability, inconsistency, and unpredictability. This was the psychodynamic (or depth-psychological) viewpoint, as examplified at the time of the first edition by Freud's classic psychoanalytic theory of personality. Recall that the evidence limiting the Doctrine of Traits comes largely from studies of surface behavior. These findings, so troubling for the psychometric view, presented little or no problem for the psychoanalytic view. After all, psychoanalysis posits a number of processes -- the defense mechanisms -- by which one behavior could be transformed into another. By virtue of reaction formation, for example, the child who deeply loves his mother may appear to hate her. Thus, psychoanalytic theory accepts surface inconsistencies as a given, and resolves the difficulty by analyzing the hidden, unconscious processes that govern the vicissitudes of conscious experience and public behavior.

By 1937 psychoanalysis had triumphed within clinical psychiatry, but it found the going rough within academic psychology. Allport himself had a particularly negative encounter with Freud; and an entire issue of the Journal of Abnormal Psychology was devoted to unenthusiastic accounts by a number of academics (including E.G. Boring) of their experiences on the couch. At that time there still may have been a clear collective memory of James' statement (in the Principles) that unconscious states are "the sovereign means of believing what one likes in psychology, and of turning what might become a science into a tumbling-ground for whimsies". (There is reason for thinking that when James told Freud that "the future of psychology is in your hands" he was not uttering a complement, but rather warning Freud not to mess things up.)

The difficulties encountered by classical Freudian psychoanalysis in gaining acceptance among academic psychologists are too familiar to require detailed exposition in this space. There was, first, the reliance on unobservable structures and processes in a time of rising behaviorism. Not even the database of psychoanalysis was available for public observation, having been gathered in the privacy of the consulting room and retrieved from the unverified memory of the analyst. Just as important, the interpretive rules linking unconscious and conscious mental life were poorly specified; and there were so many of them that virtually any inference could be sustained by the appropriate combination of primitive motives and defense mechanisms. It probably didn't help that psychometricians and psychoanalysts came from different cultures (London and Vienna), wrote in different languages (English and German), and had different training (statistical and medical). Nor did it help that the psychoanalysts were reluctant to subject their theories to formal tests, and that the few tests performed generally gave negative (or at best inconclusive) results.

Yet even in the 1930s some of the psychoanalytic viewpoint began to creep into academic psychology. Clark Hull had been recruited to Yale to lead the Institute for Human Relations, part of whose mission was to see how psychoanalysis could be integrated with academic social science. In this context, Dollard and Miller produced their rigorous analyses of frustration, aggression, conflict, and avoidance. Along with Sears, they tried (with some success) to translate a de-sexualized form of psychoanalytic theory into the principles of S-R learning theory. At roughly the same time, psychoanalysis clearly influenced the work of Murray and his colleagues at Harvard, as represented in the Explorations in Personality (1938). The Explorations was not a textbook, but one was produced by McClelland, trained at Yale and Murray's intellectual heir, in 1951: it paid a great deal of attention to psychoanalysis, and systematized Murray's emphasis on unconscious social motivation. (By introducing the concept of schema, derived from Piaget and the psychoanalyst Schachtel, to personality, McClelland also foreshadowed certain cognitive trends in the personality to be discussed below.)

After World War II, when clinical psychology emerged as a separate profession, psychologists began to have sustained working relations with psychoanalytically oriented psychiatrists. At this point psychoanalysis clearly entered the discipline as both a substantive theoretical point of view and as a source of hypotheses about mental life. One example of its influence is the "New Look" in perception initiated in the late 1940s and early 1950s. In this movement, Bruner, Klein, and others tried to demonstrate the influence of motives, attitudes, and other personality variables on perception, and were particularly concerned with unconscious influences on perception and memory. The emergence of psychoanalysis as a theory to be contended is clearly seen in the next great publishing event in the next great publishing event in personality. The appearance, in 1957, of Hall and Lindzey's Theories of Personality, modelled on Hilgard's classic Theories of Learning (1st Edition, 1948), literally remade the undergraduate course into a survey of comparative theoretical approaches. The very first substantive chapter in Hall and Lindzey was on Freud; the second on Jung; the third on Adler, Fromm, Horney, and Sullivan; the fourth on Murray. The factor-analytic theories that dominated the psychometric viewpoint got surprisingly little attention. Several generations of personality psychologists, including the present author, cut their teeth on this book and learned their Freud from Hall and Lindzey.

Certain developments within psychoanalysis eased the process of integration. These theoretical developments did not reflect a rejection of Freud, but rather a form of intellectual evolution: they all have, as their common ancestor, statements in Freud's latter work (e.g., Group Psychology and the Analysis of the Ego, 1921). After Freud's death in 1939, a serious movement began to de-sexualize and de-biologize Freud, while paying continual homage to the general principles of depth psychology. These principles of "metapsychology" were summarized in 1960 by Rapaport in terms of 10 "points of view", of which the most important are:organismic (all behavior is that of the integral and indivisible personality);genetic (all behavior is part of a genetic series, and... the temporal sequences which brought about the present form of the personality);topographic (the crucial determinants of behaviors are unconscious;dynamic (the ultimate determinants of all behavior are the drives);economic (all behavior disposes of and is regulated by psychological energy);structural (all behavior has structural determiners);adaptive (all behavior is determined by reality); and psychosocial (all behavior is socially determined).

One group of analysts began to explore the adaptive, reality- monitoring functions of ego processes, and fostered a movement known as psychoanalytic ego psychology (as opposed to the classic id-psychology, empahazing the drives and fantasy life). This movement was concerned with analyzing basic mental functions of perception, memory, and thought. These were, of course, the very things that academic experimental psychologists were also interested in. The ego psychologists sometimes did perform controlled laboratory experiments, and in other ways tried make contact with their colleagues in academic experimental psychology.

Sometimes their colleagues responded. The second edition (1956) of Hilgard's Theories of Learning contained an entire chapter on Freud, including an extensive review of the available empirical literature bearing on psychoanalytic theory. Hilgard explored some of the parallels between psychoanalytic theory and conventional learning theory with discussions of "The Pleasure Principle and the Law of Effect"; "The Reality Principle and Trial-and-Error Learning"; "Repetition-Compulsion in Relation to Theories of Habit Strength; but he also showed how psychoanalytic conerns had influenced analyses of learning processes with discussions of "Anxiety as a Drive", "Repression, Forgetting, and Recall"; and "Aggression and Its Displacement", and pointed out how integration between the two psychologies could be fostered studies of learning in relation to stages of development, the psychodynamics of thinking, and therapy as a learning process. The chapter was dropped in the 5th edition of 1981, but the simple fact of its appearance in 1956 consoldiates the impression that psychoanalysis, at least in its ego-psychological form, had begun to be integrated within academic departments of psychology.

The chief theoreticians of the ego-psychological movement were Hartmann, whose 1939 monograph Ego Psychology and the Problem of Adaptation gave the trend its name, and Rapaport, whose major books concerned Emotions and Memory (1944) and The Organization and Pathology of Thought (1951). In Diagnostic Psychological Testing (with Gill and Shafer, 1945-1946; revised by Holt, 1968), Rapaport and his colleagues attempted to put a psychoanalytic approach to personality assessment on a firm psychometric footing. The house organ of the ego-psychological movement was Psychological Issues, which in its earliest volumes published empirical studies of memory for prose, the structure and perception of events, cognitive styles, subliminal perception, and perceptual development. Many of these papers might have been written by an academic experimental psychologist; but, in the context of the sterile psychophysics and radical behaviorism characteristic of the time, their emphasis on motivation, emotion, pathology, and development mark these papers clearly as products of a distinctively psychoanalytic metapsychology. Some ego-psychological theory and research also appeared in mainstream psychological journals. A case in point is Robert White's 1959 paper in Psychological Review, reinterpreting Freud's psychosexual stages in terms of the development of competence and mastery.

A case could be made that psychoanalytic ego-psychology kept the study of higher mental processes alive during the peak years of behaviorism in America. But with the triumph of the cognitive revolution of 1956-1967, it no longer served this function. Still, the impulse to integrate psychoanalysis with cognitive psychology remains active, as evidenced by such books as Pribram and Gill's Freud's 'Project' Re-Assessed: Preface to Contemporary Cognitive Theory and Neuropsychology (1976), referring to Freud's unpublished Project for a Scientific Psychology, and Erdelyi's Psychoanalysis: Freud's Cognitive Psychology (1985). In addition, the MacArthur Foundation has established a Program on Conscious and Unconscious Mental Processes, whose aim is to promote interchange between psychoanalysts and cognitive scientists.

At the same time that the ego psychologists were attempting to merge psychoanalysis with classic experimental psychology, other psychoanalysts were establishing links to classic social psychology. Freud had proposed that each instinctual drive had an object, and for the sexual and aggressive drives that mattered most to psychoanalysis, the relevant objects were people. Soon after Freud's death a number of psychoanalysts, stimulated by theoretical developments in sociology and cultural anthropology, began to draw attention to the social (rather than instinctual) origins of conflict and anxiety: Freud's daughter Anna was part of this movement, but more conservative than Horney and Fromm in her continued adherence to the classic psychosexual theory. The difference between Freud and the neofreudians is perhaps best illustrated by Erikson's reinterpretation of the developmental stages in terms of the development of identity, trust, and other aspects of interpersonal relations. And by extending the concept of psychological development beyond adolescence to middle and old age, Erikson laid an important foundation for the current field of life-span developmental psychology.

The psychoanalytic object-relations movement emerged fully after the war, in the hands of Melanie Klein, Kohut, Fairbairn, and Winnicott. Good coverage of the work of this group may be found in Eagle's Recent Developments in Psychoanalysis (1984) and Greenberg and Mitchell's Object Relations and Psychoanalytic Theory (1985). As a group, the object- relations theorists are concerned with the same sorts of questions that concern social psychologists -- especially cognitive social psychologists: how people form internal, mental, representations of themselves and others; how these representations are structured; the relations between the information contained by these mental concepts and images and the properties of the external objects they ostensibly represent; and how mental representations of self and others influence interpersonal behavior. Cognitive social psychologists work out these problems in terms of stereotypes, person perception and memory, the self-concept, and expectancy confirmation effects; psychoanalytic object-relations theorists tend to focus their attention on the parent-child bond and the transference relationship between patient and client. Cognitive social psychologists strongly favor controlled laboratory experiments and objectively recorded behavior; psychoanalytic object-relations theorists strongly favor unstructured or semistructured clinical interviews and (most recently) the coding of recorded therapeutic encounters The methodological tools and investigatory contexts differ widely; but the basic problems are the same.

As contemporary psychoanalytic theory becomes further de-biologized and de-sexualized, cognitive psychologists have embraced motivation, affect, and nonconscious mental processes as proper topics for investigation, and social psychologists are increasingly concerned with the in vivo study of interpersonal relations, we may can expect renewed efforts, at least in certain quarters, at integrating psychoanalysis with mainstream academic psychology. Before these efforts can succeed, however, psychoanalysis must prove itself willing to accept the terms of modern scientific psychology, as opposed to a strictly hermeneutic approach grounded in clinical data: Gruenebaum's Foundations of Psychoanalysis: A Philosophical Critique (1984) is a step in this direction.

Flirting with Situationism

Types and traits, like motives and defenses, are hypothetical constructs that cannot be directly observed and whose existence must therefore be inferred. Beginning in the 1950s, increasing appreciation of the empirical difficulties of the psychometric approach, and the methodological difficulties faced by psychoanalysis, led some researchers the existence of these "invisible entities". Adopting a behaviorist stance with respect to personality, these critics abandoned intrapsychic variables and focused instead on variables that were external to the person: behavior and the situation in which it occurred. The situationist perspective on personality was stated most forcefully by B.F. Skinner in his Science and Human Behavior (1953), in which he argued that drives, traits, and states were conceptual baggage that should be discarded, and that most individual differences in behavior are due to individual differences in deprivation and reinforcement history.

In the 1960s, the tendency toward situationism was abetted by the success of classical experimental social psychology in demonstrating the powerful effect of the social situation on individual experience, thought, and action. More important than anything else, probably, was the growth of the behavior therapy movement within clinical psychology. Traditionally, the syndromes of psychopathology were treated as analogous to personality types; however, as Eysenck and others had shown, efforts of insight- oriented therapists to change their patients' underlying personalities were not notably successful. By contrast, the efforts of behavior therapists, applying the principles of classical and isntrumental conditioning, yielded considerably better outcomes. It became common for behavior therapists to equate mental illness with behavioral symptoms, to focus on the external conditions under which the symptoms occurred, and to change the behavior by altering the contingencies of reinforcement. It did not seem too much a generalization to equate normal personality with manifest behavior as well, and to apply a similar analysis.

In a 1973Psychological Review paper, Kenneth Bowers outlined the salient features of what might be called The Doctrine of Situationism. (a) The important causal factors in behavior, including social behavior in natural environments, reside in the environment rather than in the organism itself. (b) Behaviors are acquired, maintained in the repertoire, elicited or omitted on any particular occasion, and extinguished depending on the contingencies of stimulation and reinforcement. (c) Individual differences in behavior represent variations in reinforcement history. (d) A satisfactory explanation of the causes of behavior is provided by a description of the environmental conditions that are associated with it. (e) Finally, the cause of a particular response is identified with the stimuli and reinforcements to which it is functionally related.

As Bowers noted, the problems with the situationist perspective on personality are the same as the problems with the behaviorism in general. Research on preparedness, autoshaping, and species-specific defense reactions showed that there were important biological constraints on learning. And research on predictability and controllability in learning, observational and latent learning, and language acquisition and use (particularly Chomsky's critique of Skinner) revealed the important role played by internal cognitive structures and processes in the acquisition and maintenance of behavior. These lines of research undermined the behaviorist assumptions of the empty organism, arbitrariness of association, association by contiguity, and learning by reinforcement. This research, very little of which was conducted in the traditional domain of personality, simultaneously undermined the situationist emphasis on the external environment and refocused attention on the sorts of internal variables that the behaviorists had abjured.

In the final analysis, it is not clear how seriously situationism was ever taken by personality psychologists. Aside from Skinner, there are no theorists who have lent their names to the Doctrine of Situationism as strongly as Allport, Guilford, Cattell, and Eysenck lent theirs to the Doctrine of Traits, or as strongly as Freud, Murray, and McClelland advocated the psychodynamic point of view. An acknowledged target of Bowers' 1973 critique was Mischel, but earlier that year, and in the same journal, Mischel had made clear that his point of view was not, and never had been, that of a situationist. In 1987, it appears to be as difficult to find an avowed Skinnerian among personality psychologists as it is to find someone who will admit to voting for Richard Nixon in 1972.

That does not mean that Bowers had erected a straw man. Allport's 1937 text had critiqued situationism in no uncertain terms. In their first edition (1956) of Theories of Personality Hall and Lindzey had clearly identified a behaviorist perspective on personality exemplified by Dollard and Miller, Sears, and Mowrer; the second edition (1970) added Wolpe, Eysenck, and other advocates of behavior therapy, and devoted an entire chapter to Skinner. Behaviorism entered clinical psychology about 20 years after its triumph in experimental psychology, if that triumph can be marked by Skinner's The Behavior of Organisms: An experimental analysis (1938) and Hull's Principles of Behavior (1943). Arguably, the success of behavior therapy led some personality psychologists to take Skinner seriously.

But by the mid 1970s it was too late. The inadequacies of behaviorism, and thus of situationism generally, had become widely appreciated. Learning theorists, whose ostensible successes had led to the importation of behaviorism in personality and clinical psychology, were now forced to take account of the central role played by biological and cognitive structures and processes internal to the organism. Any personality psychologists who had been tempted to follow the path laid out by the behaviorists were forced to the draw the same conclusion, and the flirtation with situationism was over. Personality psychologists returned to focus on the internal structures and processes that had always been their domain.

The Evolution of Interactionism

However wrongheaded, the situationist critique did have a lasting and positive impact in that it reminded personality psychologists of the important role played by physical and social structures and processes in the outside world in determining the person's experience, thought, and action. In the 1970s the task of personality was explicitly redefined, so that investigators became explicitly concerned with understanding how structures and processes internal to the person -- the traditional domain of personality psychology -- interacted with those external to the person -- the traditional domain of social psychology.

This had long been an agenda item. In 1935, Lewin (pronounced Loo- win) had proposed a field theory view of personality in which experience, thought, and action were determined by the individual's life space, which in turn included both strictly instrinsic personal factors (e.g., motives, intentions, perceptions, memories, and emotions) and strictly extrinsic environmental factors (e.g., the physical environment, the presence and behavior of others, and roles and norms imposed by the culture). He expressed this view mathematically in a famous formula:

B =f(P, E).

Similarly, Murray's 1938 book contained a theory of personality in which behavior was determined by a combination of personal needs and environmental press (their combination in the individual's life was called the thema). But for approximately 40 years this agenda item remained at the level of rhetoric, without much attempt at systematic empirical analysis.

Beginning in the 1970s, a number of investigators began to articulate a Doctrine of Interactionism for personality, and put it into empirical practice. The Doctrine, as stated in Bowers' 1973Psychological Review piece, "denies the primacy of either traits or situations in the determination of behavior; instead, it fully recognizes that whatever... effects do emerge will depend entirely upon the sample of settings and persons under consideration.... More specifically, interactionism argues that situations are as much a function of the person as the person's behavior is a function of the situation" (emphasis in the original). Stated in the terms of Lewin's formula, B =f(P x E).

Interactionism in personality is commonly construed in statistical terms, modelled after the interaction of two main effects in the analysis of variance. By the mid-1970s, two separate lines of research had emerged clearly documenting these effects. One, conducted by Cronbach and his colleagues, concerned the aptitude-by-treatment interaction. Cronbach argued that, in at least some contexts, the payoff of personnel selection would be maximized if people with differing aptitudes were given different training programs. The other, initiated by Endler and Hunt, employed the S-R Inventory technique, measuring a variety of trait-relevant responses across a variety of trait-relevant situations. They, and many others, showed in a variety of domains (e.g., aggression, anxiety, and dominance) that individuals varied widely in the pattern of response across different situations. Bowers reviewed 11 different studies employing this technique, and concluded that the interaction of person and situation accounted for approximately 20% of the variance in behavior, while the main effects of these variables accounted for only about 10% each.

But statistical interaction is only one model for interactionism in personality. The classic prediction paradigm, as represented by Lewin's formula, is unidirectional: P and E factors combine somehow to produce B. More recently, however, Bandura has proposed that each of the elements in Lewin's equation exerts a causal influence on the others. Not only do the person and the environment jointly produce behavior, but the behavior itself feeds back on its determinants to shape both the person who emitted the behavior and the environment that elicited it. This is the essence of the Doctrine of Reciprocal Determinism, which represents a resurgence of holism (as opposed to reductionism) in psychology.

Recpirocal determinism may be illustrated by Gottman's studies of distressed couples, in which one partner's complaint elicits a second complaint from the other partner, setting off a cross-complaining "loop" that quickly degenerates into an exchange of negative affect; or by Patterson's studies of aggressive boys, in which the child's noxious behaviors elicit retalitory responses in kind from his parents, quickly developing into a vicious cycle. They may also be illustrated by Snyder's studies of expectancy effects (the self-fulfilling prophecy) in dyadic social interaction. Here, one person's (the actor's) beliefs leads him or her to behave in a particular way towards another person (the target); this tends to elicit behavior from the target that confirms the actor's beliefs. More recently, Swann has shown that when targets can act reciprocally to correct the actor's beliefs, when they do not match the target's self- concept (a process called self-verification). Expectancy confirmation is likely to emerge as a major vehicle for the study of reciprocal determinism.

Although the relevance of holistic concepts has been widely recognized in the natural and social sciences, only rarely has it been put in practice -- for the simple reason that while we possess a vast repertoire of statistical techniques for the analysis of unidirectional causation (e.g., analysis of variance and multiple regression), we do not yet have available means of comparable power for the analysis of reciprocal causal patterns. A major trend in contemporary statistics is devoted to the development of just such techniques by Thomas, Kenney, and others. The Working Group on Social Interaction, in a 1985 report to the Committee on Basic Research in the Behavioral and Social Sciences of the National Research Council, identified the study of reciprocal causal relations as a major agenda item for the next decade of research on personality and social psychology.

From Social Learning to Social Intelligence

A major problem for interactionism is to define more clearly the nature of the person-situation interaction. In his 1973 paper, Bowers asserted that persons affected situations in three ways: people may literally alter the situation that they enter by their mere presence, as when a black person walks through an all-white neighborhood, or a woman enters the drawing room of a men-only club; they may alter the situation by their behavior, as when an extravert tries to liven up a dull party, or a colicky child annoys its parents; or they may alter the situation by transforming their mental representation of it, as when a dissident construes prison as a means of scoring political points rather than as punishment. Research on this last mode, involving cognitive transformations of situations, marks an important stage in the development, over more than 40 years, of a cognitive perspective within the psychology of personality.

The first major personality theorist to adopt an explicitly cognitive perspective was Kelly, whose theory of personal constructs appeared in 1955. Kelly argued that individual differences in behavior could be understood in terms of individual differences in the categories -- personal constructs -- through which people, situations, and events were perceived and interpreted. The theory was elaborated in a Fundamental Postulate and 11 Corollaries, and resulted in the development of a new instrument -- the Role Construct Repertory Test -- for assessing the individual's repertoire of personal constructs. Kelly's viewpoint has been vigorously promoted in England by Bannister, Fransella, and Ryle; and computer-controlled modifications of the "Rep Test" have been developed in the United States by Rosenberg and Pervin.

Although Kelly can fairly be called the founding cognitive theorist in personality, even in the 1950s he was not the only theorist to adopt this stance. Lewin had emphasized the importance of the psychological over the physical environment, and Murray had stressed beta press. In 1941, Dollard and Miller had described personality as comprising habitual social behaviors acquired through learning, and sought to understand the social circumstances under which these habits were acquired. This "social learning theory" was constructed within the framework of Hullian S-R learning theory, with emphasis on the reduction of secondary, acquired drives, and imitation as a habit acquired through social reinforcement.

As noted earlier, however, in the 1950s a number of workers in learning theory began to be critical of both Hullian and Skinnerian approaches to behavior, and cognitive concepts began to creep into social learning theory as well. Thus, Rotter's 1954 monograph on Social Learning and Clinical Psychology, while adopting Miller and Dollard's term, was explicitly intended as a fusion of the drive-reduction, reinforcement theories of Thorndike and Hull with the cognitive learning theories of Tolman and Lewin. Although Rotter's version of social learning theory often used behaviorist vocabulary, and proposed a list of human needs serving drive functions, his emphasis on expectancies, values, choice, and locus of control gave it a clear cognitive twist.

In the final analysis, Rotter's approach was less a theory of learning, and more a theory of choice. He had little to say about how expectancies, values, needs, and behavioral options were acquired -- except to say that they were acquired through learning. It remained for Bandura to add an explicit theory of the social learning process. Like Miller and Dollard, Bandura stressed the role of imitation in social learning. However, his concept departs radically from theirs because imitation no longer functions as a secondary drive, and because reinforcement is given no role in learning per se. Although on the surface his 1963Social Learning and Personality Development, written with Walters, seems to draw heavily on Skinnerian analyses of instrumental conditioning and the contingencies of reinforcement, its true nature is revealed in its use of the emerging literature on language acquisition and use for arguments in favor of learning by precept and example rather than reinforcement. By emphasizing cognitive processes rather than reinforcement, observation over direct experience, and self-regulation over environmental control, Bandura took a giant step away from the behaviorist tradition and offered the first fully cognitive theory of social learning processes. His 1986 monograph, Social Foundatins of Thought and Action: A Social Cognitive Theory completes the break with behaviorist analyses begun more than two decades before by dropping the term "social learning" from the title.

Another step in the evolution of cognitive theories of personality was taken by Mischel, who received his doctoral training at Ohio State at the hands of Kelly and Rotter (and Rogers), and who for many years was Bandura's colleague. His 1968 critique of psychometric and psychodynamic approaches had embraced social learning as a viable alternative for personality psychology. Like the early theory of Bandura and Walters, Mischel's 1968 position was couched in behaviorist language, with many references to the specificity of stimulus-response relations, reinforcement, generalization, and discrimination, and the importance of assessing behavior in specific situations rather than in general. However, it also reflected a cognitive, phenomenological perspective wholly foreign to traditional learning theory. From Mischel's point of view behavior is controlled by the meaning of situations, and individuals have the power to transform those meanings through purely cognitive activities. More recently, he has proposed a new set of cognitive social learning person variables that determine how situations will be interpreted and how actions will be organized. These variables may be classified as cognitive and behavioral construction competencies, encoding strategies and personal constructs, expectancies and values, and self-regulatory systems and plans. They are construed as modifiable individual differences (in contrast to enduring traits), accessible to conscious introspection (in contrast to unconscious motives), and the products of cognitive development and social learning.

The next step in the evolution of the cognitive perspective on personality has been to take the vocabulary of cognitive social learning theory and translate it into the technical vocabulary of contemporary cognitive social psychology. One attempt to do this is represented in the a 1987 monograph by Cantor and Kihlstrom, Personality and Social Intelligence. These authors begin with the assumption that social behavior is intelligent behavior -- optional rather than obligatory, discriminatively flexible rather than rigidly stereotyped, and that individual differences in experience, thought, and action -- which are, after all, the private and public expressions of personality -- are due to individual differences in the intellectual resources that people bring to bear on the problems encountered in their life situations. The social intelligence repertoire is construed in terms familiar in cognitive psychology:declarative knowledge concerning ourselves, others, and the situations in which we encounter them; and procedural knowledge consisting of the rules, skills, and strategies by which we form impressions of ourselves and others, and plan goal-directed interactions with them.

The social intelligence perspective is not merely a translation exercise, like the attempts to recast psychoanalysis in the terms of S-R learning or information-processing theory. Rather, adoption of the conceptual apparatus of cognitive psychology entails adoption of its paradigms and methods as well. Rather than focusing on the measurement and expression of individual differences, personality psychologists of a cognitive persuasion are increasingly concerned with using procedures familiar in the laboratory study of cognition to reveal the general cognitive processes out of which individual differences are constructed.

Fifty Years Before, Fifty Years Hence

In 1937, personality psychology was almost totally divorced from the rest of mainstream academic psychology. Allport's idiographic psychology celebrated human uniqueness, while Allport's departmental colleagues were concerned with understanding processes of perception, memory, and learning that all humans shared in common. And as Cronbach noted in his 1957 address on "The Two Disciplines of Scientific Psychology: even those psychometricians who had nomothetic concerns used correlational techniques that were largely foreign to their colleagues who manipulated independent variables in controlled experiments. Psychoanalysis, with its concept of drive, its concerns for ego functioning and social relations, and its implicit theory of learning, had more potential for integration with the mainstream. But, with few exceptions, those who favored psychoanalysis were shut out of their academy by their emphasis on unconscious mental processes, symbolic interpretation, reliance on uncontrolled case histories and qualitative methods of analysis. Thus for the first 20 years of its existence as an academic subdiscipline, personality psychology was clearly on the fringes, if not beyond the pale, tolerated as a curiosity or as a concession to undergraduates who still wanted to "know how people work".

The process of integration began with social learning theory, and both Dollard and Miller and Rotter made their agenda clear. As personality psychologists made the effort to express their concerns within the lingua franca of general experimental psychology, and to perform controlled experiments, the two disciplines within scientific psychology became more closely linked. Integration was fostered by similar developments within clinical psychology, as practitioners construed both psychopathology and psychotherapy in terms of learning. The trend was not even set back by the cognitive revolution. In fact, an argument could be made that personality only flirted with situationism, not because it embraced behaviorist learning theory late, but in large part because early social learning theorists were cognitivists at heart.

It remains to be seen whether the cognitive revolution is the last one in psychology. Some think it is; others think that it is already battling against a biological revolution represented by sociobiology, behavior genetics, and beuroscience. If the latter, we may expect to see the nature of personality psychology shift again, to become less cognitive and more biological in nature. Such a shift, should it in fact occur, should not be taken as fickleness, or as evidence that personality has no inner core, or that personality is shaped by situational and cultural forces. Far from it.

If there is a moral to this story, it is that the burden of personality psychology is to express its concerns, and perform its research, within the conceptual and methodological framework provided by the general psychology of perception, memory, thought, and language. Only then will other psychologists take an interest in the activities of their personologist colleagues, and see the relevance of personality for their own work. And only then will students just entering the field see how the chapters on personality relate to the rest -- wherever in the textbook they happen to be placed.


In accordance with the publisher's instructions, references have been kept to an absolute minimum. In most cases, enough information has been given in the text to allow the interested reader to track down the relevant source materials, perhaps with the aid of the progressive treatments of the field to be found periodically in the Annual Review of Psychology. Three seminal papers in contemporary personality psychology are:

Bandura, A. (1978). The self-system in reciprocal determinism. American Psychologist,33, 344-358.

Bowers, K.S. (1973). Situationism in psychology: An analysis and a critique. Psychological Review,80, 307-336.

Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality. Psychological Review,80, 252-283.

The following textbooks provide excellent surveys of both theoretical and empirical issues:

Feshbach, S., & Weiner, B. (1986). Personality. 2nd Ed. Lexington, Mass.: Heath.

Hall, C.S., Lindzey, G., Loehlin, J.C., & Manosevitz, M. (1985). Introduction to theories of personality. New York: Wiley.

Mischel, W. (1986). Introduction to Personality: A new look. 4th Ed. New York: Holt, Rinehart, & Winston.

Monte, C.F. (1980). Beneath the mask: An introduction to theories of personality. New York: Holt, Rinehart, & Winston.

Pervin, L.A. (1984). Personality: Theory and Research. 4th Ed. New York: Wiley.

Singer, J.L. (1984). The human personality. San Diego: Harcourt Brace Jovanovich.

In addition, alternative histories of personality psychology can be found in the following volumes:

Hilgard, E.R. (1987). Psychology in America: A historical survey. San Diego: Harcourt Brace Jovanovich.

Koch, S., & Leary, D.E. (1985). A century of psychology as science. New York: McGraw-Hill.

Author Notes

Chapter prepared for a memorial volume, edited by E.R. Hilgard, honoring Floyd M. Ruch on the occasion of the 50th anniversary of the publication of his introductory textbook, Psychology and Life. The point of view represented herein is based on research supported by Grant MH-35856 from the National Institute of Mental Health, and an H.I. Romnes Faculty Fellowship from the University of Wisconsin. Preparation of this paper began while the author was Visiting Fellow (Research Scholar) at Macquarie University. Jacquelyn Cranney, Jacqueline J. Goodnow, Kevin M. McConkey, Judy A. Ungerer, John C. Turner, and Ian K. Waterhouse provided intellectual companionship in Australia; and Nancy Cantor, Judith M. Harackiewicz, Irene P. Hoyt, Stanley B. Klein, and Patricia A. Register made comments at various stages of the writing.


1In these statements, the term "behavior" should be read as shorthand for the totality of individual experience, thought, and action that is the subject-matter of psychology.Return to text.