Openness to Experience: Expanding the boundaries of Factor V

The fifth factor in lexical studies of trait adjectives is commonly interpreted as Intellect, whereas the corresponding factor derived from questionnaire studies is typically identified as Openness to Experience. Intellect as a construct is problematic because it erroneously suggests an equivalence of Factor V with intelligence, describes aspects of Factor III (Conscientiousness) as well as of Factor V, and fails to suggest the diverse psychological correlates that Factor V is known to have. By contrast, Openness to Experience is a broader construct that implies both receptivity to many varieties of experience and a fluid and permeable structure of consciousness. Data from analyses of adjectives, established personality questionnaires, and Hartmann's (1991) new Boundary Questionnaire support these interpretations. The construct of Openness can be transported across geographical and cultural boundaries to function as a universal dimension of personality structure.


INTRODUCTION
Contemporary consensus on the five major factors of personality originated in the convergence of two lines of research: the lexical tradition, in which natural language trait adjectives are taken to define the personality sphere (Goldberg, 1990;John, Angleitner and Ostendorf, 1988), and the questionnaire tradition, in which scales are developed to measure psychological constructs suggested by personality theories (e.g. Eysenck and Eysenck, 1975;Jackson, 1984). When a broad range of either adjectives or scales is sampled, their intercorrelations can usually be summarized by five recurrent factors (Digman and Inouye, 1986). Four of these-Extraversion, Agreeableness, Conscientiousness, and Neuroticism versus Emotional Stability-are very similar across the two traditions. The fifth factor, however, appears to be best described as Intellect when it is based solely on trait adjectives, but as Openness to Experience when it is derived from psychological constructs. A comparison of these two constructs and labels will be a major focus of this entire Special Issue.
For readers not already acquainted with the constructs, illustrative cases might be a useful starting point, and as personifications of Intellect and Openness one could hardly improve upon the protagonists of Hermann Hesse's (1930~968) medieval novel, Narcissus and Goldmund. Both are recognized by their peers as extraordinary individuals, both share aspects of what Narcissus called 'the creative mind ', (p. 272), yet they are in some respects very different. Narcissus is a monk; a brilliant scholar and thinker with penetrating insight into human character, he is also ascetic and self-disciplined, prepared to restrict his life to the library and monastery. Goldmund, the novel's real hero, is a sculptor, a lover of many women, and a wanderer, committed to nothing but seeking. As the epigraph shows, he is also distinguished by a particular style of consciousness that I hope to show is an important aspect of Openness. I will argue that it is the ability to accommodate such wide-ranging and non-obvious correlates that recommends the construct and label of Openness to Experience.

FACTOR V IN THE LEXICAL TRADITION
Research in the lexical tradition began with the hypothesis that traits important in human interactions will have been encoded into natural languages, principally as trait-descriptive adjectives (Norman, 1963). An analysis of trait adjectives in the dictionary ought therefore to lead to an understanding of trait structure in human populations. Lexical analyses led to the Five-Factor Model (FFM; Goldberg, 1993), and the success of the FFM in mapping personality constructs (Ozer and Reise, 1994) might be seen as powerful support for the lexical hypothesis.
However, the lexical hypothesis has many forms. There are legitimate reasons to doubt that lay conceptions form a scientifically adequate basis for the study of personality (Tellegen and Waller, in press), so a weak form, in which lay trait language is held to provide merely a rough guide to the range of personality traits, is most plausible. In this view, lexical studies are only one of many sources of evidence on personality structure. Researchers adopting this weak version of the hypothesis might wish to consult the larger psychological literature in selecting adjectives to factor-a strategy used by Cattell (cf. John et al., 1988) and Costa (1985b, 1987).
Lexical purists argue, however, that deviations from a strong form of the lexical hypothesis lead to subjective biases in the selection of adjectives: 'To avoid arbitrary preferences . . . one should select personality variables to be representative of trait adjectives in the natural language' (Peabody, 1987, p. 59). Unfortunately, it is not self-evident what constitutes the body of trait terms in a language. Ought one to include every relevant entry in an unabridged dictionary, including terms like acaroid, halituous, and raptril, which are 'unknown even to most well-educated and intelligent persons' (Norman, 1967, p. 13)? Should the list be restricted to commonly used trait adjectives (Goldberg, 1990), or to terms spontaneously generated in self-descriptions (John, 1990)? And what precisely is a trait-descriptive adjective? Does this category include state terms (angry), abilities (musical), physical descriptors (lean)? Should both evaluative and descriptive terms be examined (Waller, in press), or should purely evaluative terms be discarded?

Alternative lexical versions of Factor V
Different investigators have adopted different positions on these questions, with resulting differences in the structure of personality they report. It is a testimony to the robustness of the FFM that despite these differences most lexical studies have reported five rather similar factors (e.g. Goldberg, 1990;Ostendorf, 1990). But the differences are not trivial, and they are particularly consequential for the interpretation of Factor V. In creating the Dutch taxonomy, adjectives were rated by panels of judges to determine their relevance to personality; in one of these screenings, 'the intention was to select stable traits of personality only, rather than stable traits of capacity, ability, intelligence, or skill' (Brokken, 1978, p. 16). Presumably as a result, De Raad, Hendriks and Hofstee (1992) reported a fifth Dutch factor defined by critical, mutinous,$erce against virtuous, meek, docile that bears little resemblance to either Intellect or Openness. In the German taxonomy, where ability traits were explicitly included, Factor V was interpreted as Intelligence or General Ability, with smaller loadings for terms like creative and imaginative (Angleitner and Ostendorf, 1989).
The German results thus parallel most English language studies, on which the Intellect interpretation was originally based. In 1983, Goldberg suggested using such bipolar terms as stupid-intelligent, ignorant-knowledgeable, and unanalyticalanalytical to measure this factor; in 1990, he recommended a set of unipolar terms that included intellectual and bright versus simple and unreflective. John (1990) asked a panel of judges to review the literature on the FFM and select items from Gough and Heilbrun's (1983) Adjective Check List (ACL) to measure each of the five factors; at least 90 per cent of them agreed that Factor V was defined by such terms as clever, ingenious, logical, and joresighted. Given such a collection of adjectives, the label Intellect is certainly understandable.
But very different results emerge when a different strategy for sampling the lexicon is used. Tellegen and Waller (1 987) varied the usual lexical approach in three ways: (i) they systematically sampled pages of the dictionary rather than attempting an exhaustive enumeration; (ii) they used phrases rather than single words to convey the meaning of the traits; and (iii) they made no attempt to eliminate mood terms or adjectives that might be considered primarily evaluative, such as good and evil. Factor analysis of the resulting items lead to a seven-factor solution, in which Positive Valence and Negative Valence factors supplemented five substantive factors. These latter five showed a one-to-one correspondence with the dimensions of the FFM, but with notable alterations to Factor V. That factor, which they labelled Conventionality, contrasted conservative; conventional; and thought of as old fashioned with curious, inquisitive; odd, peculiar; and progressive, favor social reform (Waller, 1994). It is hard to recognize Intellect in either pole of this factor, but the low pole is recognizably a form of Openness.
This seven-factor model has been interpreted as a challenge to the comprehensiveness of the FFM (Waller, in press), but there are other ways to interpret the data. In a joint factor analysis with the Revised NEO Personality Inventory (NEO-PI-R; Costa and McCrae, 1992b) only five factors were found, the two evaluative factors splitting their loadings across several of the substantive factors (Costa and McCrae, 1994). The seven factors appear to tap interesting aspects of personality, but it cannot be concluded that they represent the basic structure of personality (Costa and McCrae, 1992a). The seven-factor solution does show, however, that not all lexical approaches emphasize the construct of Intellect.

The construct validity of Intellect
For lexicologists, the criterion for naming adjective factors might well be semantic: what label best captures the meaning of the defining terms? By this criterion, Zntellect is certainly a reasonable choice for many versions of Factor V. But for personality psychologists who seek to understand dimensions of individual differences an entirely different criterion is appropriate, namely, construct validity. Here rational and semantic interpretations of factors must yield to evidence on the network of their empirical correlates. Does the term Intellect suggest the range of correlates that Factor V is known to have (McCrae and Costa, in press), or does it instead suggest attributes that are not strongly related to Factor V? What, in short, is the convergent and discriminant validity of the Intellect construct? From this perspective there are serious problems with the choice of Intellect as a label for Factor V.
First, and most obviously, the term suggests that high scorers on the factor should be intelligent. Goldberg (1981) argued that the social utility of trait terms related to Factor V was to convey whether 'X is smart or dumb (How easy will it be for me to teach X?)' (p. 161). If this claim were true, it would certainly establish Factor V as a major dimension of individual differences, because intelligence has always been a central concept in psychology.
But in fact high scorers on Factor V are only slightly more intelligent than low scorers when ability tests are used as measures of intelligence. For example, Goldberg's (1 983) Intellect scale correlated only 0.27 with Vocabulary scores from the Wechsler Adult Intelligence Scales (WAIS; Wechsler, 1958), and a joint factor analysis of 80 bipolar trait adjectives with eight cognitive tests showed that Factor V and general intelligence formed separate factors (McCrae and Costa, 1985b). Although intellectual interests is a key definer of Factor V, intellectual ability is only a modest correlate.
Second, the construct of Intellect lacks discriminant validity with respect to Factor 111, Conscientiousness. John's (1 990) judges assumed that terms like logical and foresighted would be definers of Factor V-surely they are part of what laypersons mean by 'smart'-but these adjectives did not define that factor; instead, they loaded on the Conscientiousness factor (McCrae, 1990). Indeed, the term intelligent itself often shows as high or higher loadings on Factor I11 as on Factor V (McCrae and Costa, in press). In an Abridged Big Five Circumplex (ABSC) analysis, terms like intelligent and intellectual were classified as as V+III+ rather than as V + V + (Johnson and Ostendorf, 1993).
When laypersons use terms related to Intellect, they appear to mean something akin to applied intelligence, which would commonly be seen in academic achievement, a correlate of Conscientiousness (Digman and Inouye, 1986). Hesse's scholar Narcissus is prototypically intelligent, and he would have scored as high on Factor I11 as on Factor V (and perhaps highest of all on the sixth factor, g . ) Third and finally, Intellect fails to suggest the full range of attributes that are related to this factor: intellectual curiosity is an important part of Factor V, but only a part. With the construct of Intellect in mind, who would imagine that high scorers on Factor V would be easily hypnotized, or seek thrills and adventures, or be empathic, or cope by using regression in service of the ego, or believe in paranormal phenomena (McCrae and Costa, in press)? Such correlates are, however, predictable from the construct of Openness. The relations among Openness, Intellect, Conscientiousness, and measured intelligence are represented in Figure I. A schematic representation of relations among Openness to Experience, Intellect, Adopting a broader interpretation of Factor V leads to the selection of rather different adjectives to measure it, and a series of analyses stimulated by personal communications with L. R. Goldberg and 0. P. John (July 13 to September 15, 1989) examined the empirical consequences of different item selections. Two sevenitem scales were constructed from bipolar adjectives that defined Factor V in selfreports (McCrae and Costa, 1985b) and peer ratings (McCrae and Costa, 1987). The first scale was composed of items that appeared to represent Goldberg's construct of Intellect: analytical, intelligent versus stupid, perceptive versus imperceptive, cur-ious, imaginative versus down-to-earth, creative, and cultured. The second scale (Non-Intellect) consisted of items related to Factor V but less obviously related to Intellect: original versus conventional, broad versus narrow interests, complex versus simple, daring versus unadventurous, liberal versus conservative, independent versus conforming, and prefer variety versus routine. Coefficient alphas for the two scales were similar: 0.68 and 0.66.
"Correlation with Intellect significantly higher than correlation with Non-Intellect. bCorrelation with Non-Intellect significantly higher than correlation with Intellect. 'Factor from McCrae, Costa and Busch, (1986).
These data merit several comments. First, both brief adjective scales show many significant correlations despite the fact that data administrations are separated by as much as 20 years for some subjects. The findings are unlikely to be due to method variance, because peer ratings predict self-reports, and both peer ratings and selfreports predict performance measures (education and divergent thinking). Supplemental analyses showed that these associations were essentially unchanged when measures of the other four factors were partialled, and were only slightly reduced when WAIS Vocabulary scores were controlled. Measures of Factor V-including semantically narrow Intellect scales-thus have diverse and important psychological correlates, to whose interpretation we will turn in the next section.
The second noteworthy aspect of Table 1 is the comparison of the Intellect and Non-Intellect scales. In a few cases there are significant differences (Meng, Rosenthal, and Rubin, 1992), but the general conclusion is that these two scales (which themselves correlate from 0.54 to 0.65 in the different samples) are essentially interchangeable as predictors of this set of criteria. At least at a global level, it appears to be possible to measure Factor V quite adequately without the use of any Intellect-related terms (mean correlations actually show a slight advantage for the Non-Intellect scale). Some construct broader than Intellect is thus needed to encompass the full scope of the factor.

OPENNESS IN THE PSYCHOLOGICAL LITERATURE
How might one characterize individuals who combine intellectual curiosity with broad interests, liberal views, adventurous tendencies, and a need for variety? We have argued that Openness to Experience is an apt label, because it suggests a preference for the new and different in many different aspects of life (McCrae, 1993-1 994;McCrae and Costa, 1985a;. The concept of Openness to Experience was not invented to describe the adjective definers of Factor V; it has long had currency among personality psychologists. Rogers (1961) wrote about openness to feelings, Rokeach (1960) about the open mind, Tellegen and Atkinson (1974) about openness to absorbing experience, and MacKinnon (1960) about openness to experience in general as a common feature of creative individuals. Kaplan and Singer (1963) conducted an experimental study of sensory acuity and concluded that 'openness to sense impressions apparently runs parallel to openness to ideas' (p. 490).' Paul Costa and I borrowed the term most directly from the work of Coan (1974), whose Experience Inventory was the starting point for the development of the Openness scales in the NEO-PI-R. As Coan noted, 'people vary considerably in the range and types of experience to which they are open ' (1972, p. 346), so our strategy, like his, was to measure Openness in many different areas. In the NEO-PI-R, facet scales measuring Openness to Fantasy, Aesthetics, Feelings, Actions, Ideas, and Values are summed to measure total Openness. Research using these scales has shown the extraordinary richness of the Openness construct. Here is a trait dimension that affects nearly every aspect of the individual's life, from political attitudes (Riemann, Grubich, Hempel, Mergl and Richter, 1993;Trapnell, 1994) to musical preferences (Dollinger, 1993). Openness is seen as much in affective and sensory systems (McCrae, 1993(McCrae, -1994 as in cognitive systems. It is thus not surprising that sexuality-so central a part of the human experience-is pervasively influenced by Openness: open men and women have more information about sex, wider sexual experience, stronger sexual drives, and more liberal sexual attitudes (Costa, Fagan, Piedmont, Ponticas and Wise, 1992;Meston, Trapnell and Gorzalka, 1993). When Hesse depicts Goldmund's absorption in erotic experience he illustrates an important manifestation of Openness.
Reflection on the many correlates of Openness has suggested two complementary aspects of the dimension, motivational and structural. The motivational aspect is perhaps most easily conveyed. Although the word openness may suggest to some readers a kind of passive tolerance of new experience, in fact open people are characterized by an active pursuit of novelty, a quest to 'clarify, intensify, or otherwise enlarge our experience' (Canaday, 1980, p. 5). The need for novelty in experience is universal, ranked as such by anthropologist Ralph Linton (1945) alongside needs for emotional bonds and long-term security, and anticipated evolutionarily by the exploratory behavior of animals (Berlyne, 1955). Individual differences in so basic a need understandably have powerful consequences.
The structural aspect of Openness is less easily grasped; it refers not to the contents of consciousness so much as to the organization of the contents in a particularly fluid and permeable structure. As Rokeach (1960) argued, closed individuals have beliefs that are tightly compartmentalized, not easily affected by contradictory beliefs or by corrective information (Davies, 1993). Open individuals, by contrast, have more flexible attitudes. The same open style of consciousness can be seen in divergent thinking, in which remote associations are easily made, and in synesthesia, in which the distinctions between different sensory modalities are blurred (Rader and .
Earlier, the label Intellect was criticized because it erroneously suggested that high scorers on Factor V would necessarily be high in cognitive ability. The label Openness is liable to a parallel criticism, because Openness is often mistakenly interpreted in interpersonal terms to refer to self-disclosure, a characteristic that is in fact more closely related to Extraversion and Agreeableness. Trusting and talkative people reveal much about themselves, they open themselves out to the world. Whether they are equally receptive to information and experience coming in from the world is something quite different, and depends on their standing on Factor V.

Jung and his interpreters
The question then arose: 'Am I the one who is sitting on the stone, or am I the stone on which he is sitting?' This question always perplexed me, and I would stand up, wondering who was what now, The answer remained totally unclear, and my uncertainty was accompanied by a feeling of curious and fascinating darkness.
C . G. Jung, Memories, Dreams, Reflections (1961, p. 20) It is difficult to determine who should be credited with the first identification of Openness as a major dimension of personality, but one name recurs throughout the history of the idea: C. G. Jung. In Psychological Types (Jung, 1923(Jung, /1971) he suggested distinctions between introversion and extraversion, thinking and feeling, and sensation and intuition that have influenced researchers in this area for decades. As Coan (1974) wrote, 'for those who wish to extend our explorations to a broader range of human experience, [the Jungian system] is a fund of insights that we cannot afford to overlook' (p. 58). Guilford began the tradition of factor analytic explorations of personality in the 1930s by showing the multidimensionality of purported measures of introversionextraversion (e.g. Guilford and Guilford, 1934). Eventually he stressed the distinction between social introversion and what he called thinking introversion (Guilford, 1977)-a factor reflected in the Thoughtfulness scale of the GZTS. Thoughtfulness is a correlate of Openness (McCrae, 1993(McCrae, -1994; see also Table l), and the interpretation of Factor V as thinking introversion has recently been revived by Wolfe (1993), who proposed the label Bookishness. At least a part of what Jung meant by introversion seems to have been related to Openness, but it was confounded with the social introversion that is more commonly assessed in personality questionnaires.
The distinction between thinking and feeling functions has also been important for researchers on Openness. Jung believed that thinking and feeling were mutually exclusive approaches to rational decision-making, characterizing entirely different types of individual. Hesse (1930Hesse ( /1968, who was deeply influenced by Jung (and was even briefly his patient), set out more or less consciously to portray these types in the characters of Narcissus and Goldmund; the result, however, appears instead to have been the illustration of two different aspects of the single dimension of Openness. Costa and McCrae (1 976) found a dimension in the Sixteen Personality Factor Questionnaire that appeared to show an integration of openness to both ideas and feelings, at least in the oldest age group; they interpreted this synthesis as the result of Jungian processes of individuation. However, when they subsequently created scales to measure explicitly these two facets of Openness, they showed that at all ages, individuals who were more open to ideas tended also to be more open to feelings McCrae, 1978, 1980). In that research, as in Hesse's characterizations, openness to ideas and to feelings appear to be not polar opposites, but different facets of a single common factor.
Psychometrically, thinking and feeling can be placed at opposite poles of a dimension only if the former is interpreted as cold rationality, the latter as warm sentimentality. As warmth versus coldness, this contrast is closely related to Factor 11, Agreeableness (McCrae and Costa, 1989).
Most important of the Jungian distinctions is that between sensation and intuition. Jung defined intuition as 'perception via the unconscious' (192311971, p. 538), and argued that it is incompatible with sensation, which is conscious perception: 'When I try to assure myself with my eyes and ears of what is actually happening, I cannot at the same time give way to dreams and fantasies about what lies around the corner' (p. 539). When MacKinnon began his studies of creativity in 1949 at the Institute of Personality Assessment and Research, he introduced an early version of the Myers-Briggs Type Indicator (MBTI; Myers and McCaulley, 1985), which measures Sensation versus Intuition and found a strong association of Intuition with creativity: 'We would expect creative persons not to be stimulus-and object-bound in their perceptions but ever alert to the as-yet-not-realized, and this is precisely what we find to be true of all our creative groups ' (1960, pp. 377-378).
Yet Jung's dichotomy between sensation and intuition is in some ways flawed. Just as both breadth of thought and depth of feeling characterize the open individual, so too do both sensory acuity (Kaplan and Singer, 1963) and imagination (Saucier, 1992). The MBTI Sensation versus Intuition scale does not really contrast sense perceptions with hunches; in fact, none of its items deals directly with sensory input. Instead, this scale contrasts a preference for the factual, simple, and conventional with a preference for the possible, complex, and original. As such, it is a good measure of Openness (McCrae, 1993(McCrae, -1994. Jung himself never wrote about Openness-that term does not appear in the index to his collected works2-and as we have seen, his conceptualizations as often confused as clarified individual differences in this area. Why, then, has he had such influence on students of Openness? Perhaps because Jung himself was almost archetypally open to experience. How else can we characterize someone who first embraced the radical tenets of psychoanalysis and then rejected its dogmatic orthodoxy? Who travelled to India, Uganda, and New Mexico in search of spiritual insights? Who wrote volumes on the interpretation of alchemy and proposed that flying saucers were a modern myth? The spirit of Openness was beautifully illustrated in his 'Foreword to the I Ching', when Jung explained his decision to consult the Oracle itself on the wisdom of an English translation: 'Not even the strangeness of insane delusions or of primitive superstition has ever shocked me. I have always tried to remain unbiased and curious-rerum novarum cupidus. Why not venture a dialog with an ancient book that purports to be animated?' (Jung, 1958, p. 594).
In Memories, Dreams, Reflections Jung (1961) recounted a number of peculiar experiences that testify to a particular kind of mind. His dreams were vivid and real to him, so much that the boundary between dream world and reality was often tenuous. Similarly, his sense of identity was weak; he felt himself to have multiple identities, or to merge his own identity with his surroundings. Anyone who has read his works knows that his cognitive style is often tortuous, following a sweep of associations rather than a logical course of development. All of these features suggest a particular structure of consciousness, in which the rigid dichotomies between reality and fantasy, self and other, cause and effect are softened. In some individuals this may represent a form of psychosis; in others it is only the modus vivendi of an extremely open mind.

OPENNESS AND BOUNDARIES IN THE MIND
The concept of permeability or fluidity in consciousness is certainly not new. In addition to research on dogmatism in attitudes and ideology, there have been studies of a number of phenomena related to the structure of mental contents. Cognitive psychologists have examined remote associates as predictors of creativity (Guilford, 1967). Rader and Tellegen (1987) developed a measure of synesthesia, which represents a kind of permeability across sensory modalities. Chapman, Chapman and Raulin (1978) assessed perceptual aberration, and Haan (1965) created a measure of regression in service of the ego; these two variables might be considered maladaptive and adaptive variants, respectively, of the tendency of primary process thinking to penetrate into consciousness.3 But perhaps the most extensive and integrated consideration of the structure of consciousness in recent years has been in the work of Ernest Hartmann.
Hartmann is a psychoanalytically oriented psychiatrist with a long-standing interest in sleep, dreams, and nightmares (Hartmann, 1984). His studies of chronic nightmare sufferers (Hartmann, Russ, Oldfield, Sivan and Cooper, 1987) led him to some * However, he did refer to misoneism, 'a deep and superstitious fear of novelty' (Jung, 1964, p. 31) that in his view characterized many of his contemporaries' reactions to the idea of the unconscious.
Both these scales are related to NEO-PI-R Openness (McCrae and Costa, in press), and the Perceptual Aberration scale is related to what might be viewed as permeability in handedness: high scorers are twice as likely as controls to be ambilateral rather than either right-or left-handed (Chapman and Chapman, 1987). surprising observations. Individuals who frequently had nightmares did not show a high incidence of psychiatric disorder, nor were they able to identify traumatic incidents in their childhood that might have been responsible for the recurrent nightmares. They did, however, seem to share a set of personality characteristics: they could be described as ' "unguarded", "undefended", "fluid", "artistic", "vulnerable", "open" ' (Hartmann, 1991, p. 16).
Hartmann interpreted these characteristics as signs of a particular structure of mind. Just as they were unable to screen out terrifying images and affects from their dreams, so these people appeared unable or unwilling to place barriers between their own identity and that of others, or between their established beliefs and unconventional ideas, or between one affect and another. The mental boundaries that separate the contents of consciousness were exceptionally thin. This is precisely the phenomenon that Hesse describes in his characterization of the student Goldmund, for whom the real world and the dream world were separated only by 'a quivering film', a 'thin skin'. Mundane reality was inundated by images and associations: A Latin initial changed to his mother's perfumed face, a long note in the Ave became the gate to Paradise, a Greek letter a galloping horse, a rearing serpent that quickly slithered off through the flowers, leaving the rigid page of grammar in its place (Hesse, 19304968, p. 62).
Hartmann (1 99 1) came to regard thickness or thinness of boundaries as 'a broad dimension of personality and an aspect of the overall organization of the mind' (p. 49), and believed it to be 'a new dimension of personality' (p. 3). He traced its intellectual roots to such constructs as James's tender-mindedness, Freud's Reizschutz or stimulus shield, Blatt and Ritzler's (1974) permeable ego boundaries, and Gardner, Holzman, Klein, Linton andSpence's (1959) cognitive controls. (Rokeach (1960) is a striking omission from this list.) Whether or not thickness versus thinness of boundaries is really a new dimension of personality, certainly Hartmann's approach to measuring it is unique. In designing a questionnaire, he applied the metaphor of boundaries in a wide array of areas or categories, creating scales to measure thinness of boundaries with regard to sleep and waking, thoughts and feelings, persons, places, values. 'I like heavy solid clothing' would be a subtle item in most scales; here it is a literal operationalization of thick boundaries.
The Boundary Questionnaire (BQ; Hartmann, 1991) was administered to 124 men and women in the BLSA in 1993, together with Tellegen, Grove and Waller's (1991) Inventory of Personal Characteristics # 7 (Costa and Mccrde, 1994). These subjects had previously completed the NEO-PI (Costa and McCrae, 1985) and MBTJ in 1986, and the ACL (from which John's (1990) Big Five prototypes could be scored) in 1987. Complete data on the BQ were available for 53 men aged 32-91, and 71 women aged 2690. Table 2 provides some descriptive information on the category scales of the BQ. For each of the 12 categories, the best item is given as an example. Most of the scales show adequate internal consistency, although Childhood/adolescence/ adulthood and Interpersonal scales have psychometric problems. The category scores were correlated with the 18 facets of NEO-PI N, E, and 0, and with the A and C domain scales; the single largest correlate is reported in the Table. In most cases some facet of 0 shows the strongest association, although the BQ also has content related to N, E, and low C .  (14) Unusual experiences (19) Thoughts, feelings, moods (16) Childhood, adolescence, adulthood (6) Interpersonal (1 5 ) Sensitivity ( 5 ) Neat, exact, precise ( 1 1) 113. I awake from one dream into another.  (20) Opinions about peoples, nations, groups (14) Opinions about beauty, truth (7) 0 There is considerable heterogeneity in BQ item content, and it is not immediately clear what the sum of these scales measures. (Note in particular that the Interpersonal category, typified by the item 'I am a very open person' is strongly related to El: Warmth; it is unrelated to total Openness, r = 0.14, n.s. Self-disclosure should not be confused with Openness to Experience.) There is, however, a unifying theme across most of the items; the first general factor correlates 0.95 with the simple sum of items. To get a better sense of what that theme is, Table 3 presents the 10 items that correlate most highly with the total. A reading of these items suggests that Hartmann has succeeded quite well in measuring his concept of thin versus thick boundaries, from the preference for blurred edges in paintings to the categorical separation of the sexes, as cut-and-dried and rule-bound as 'a rigid page of grammar'.

136.
1 like paintings or drawings with soft and blurred edges. I have had deyd vu experiences. My thoughts blend into one another. At times I feel happy and sad all at once. At times I have felt as if I were coming apart. I had a difficult and complicated adolescence. I can easily imagine myself to be someone of the opposite sex. versus There is a place for everything, and everything should be in its place. A man is a man and a woman is a woman; it is very important to maintain that distinction. A good relationship is one in which everything is clearly defined and spelled out.

58.
Nore: All correlations are greater than 0.46 in absolute magnitude.
The BQ total score is somewhat higher in women than in men (Y = 0.31), and lower in older respondents than younger (r = -0.30). It is significantly related to WAIS Vocabulary scores, Y = 0.18, p < 0.05, but not to years of education, r = 0.14, n.s. But the data in Table 2 suggest that its chief correlates should be with personality dimensions, and correlations with NEO-PI N, E, 0, A, and C domains are 0.32, 0.27, 0.66, 0.01, and -0.05, respectively; the first three are statistically significant. The correlation of 0.66 between Openness and BQ total Score is reniarkable given the independent origins of the instruments and the lapse of seven years between their administrations. BQ Total Score is significantly related to all six facets of Openness, rs = 0.37-0.58, but especially to Openness to Fantasy, Aesthetics, and Feelings. These are the facets that are most strongly related to Tellegen and Atkinson's (1974) Absorption (Glisky, Tataryn, Tobias, Kihlstrom and McConkey, 1991), so these data replicate Hartmann's (1991) own report of a moderately strong correlation (0.54) between BQ Total Score and Absorption.
Given Hartmann's background as a psychiatrist and his interest in nightmares, it is hardly surprising that many of the items in the BQ have a slightly psychopathological flavor, accounting for the modest correlation with NEO-PI N. Similar item content could be found in Coan's (1974) Experience Inventory, the direct precursor to the NEO-PI Openness scales. In creating the NEO-PI we chose to exclude some of the more eccentric and bizarre aspects of the domain that Coan had identified. Ironically, this has led some to suggest that Openness may not be relevant to personality psychopathology (Hyler and Lyons, 1988). Openness per se is not pathological, but some forms of psychopathology can be expressed in open or closed ways (Costa and Widiger, 1994).

The five-(or six-) factor synthesis
Thus far I have discussed several approaches to personality measurement: the lexical approach, with its five-factor and seven-factor variants operationalized, respectively, by John's (1990) and Tellegen, Grove and Waller's (1991) instruments; the tradition of personality theory and research from which the NEO-PI-R and the concept of Openness were derived; the Jungian typology assessed by the MBTI; and the 'new' dimension of personality offered by Hartmann. How can these approaches be integrated, and how should the integration by characterized?
A straightforward approach is a joint factor analysis. Complete data on each of these instruments, plus WAIS Vocabulary and years of education, were available for 85 participants, a rather small sample for the analysis of 24 variables. However, the results were clear. Six eigenvalues exceeded 1.0, with a break after the sixth. Table 4 reports factor loadings, which have been reordered and reflected into the standard lexical order. The identity of the factors is beyond doubt: they are the Big Five plus general intelligence. Note that all the subjects in this analysis joined the BLSA Stress and Coping project in 1986; thus, these data constitute an independent replication of the factorial separation of Openness from intelligence reported in McCrae and Costa (1985b). Both studies support Brand's (1984Brand's ( , 1994 view that intelligence is an independent sixth factor. The fifth factor is defined chiefly by NEO-PI Openness and MBTI Intuition, but also by thin boundaries, unconventionality, and John's Intellect-saturated version of Factor V. Perhaps the only surprise is the high loading of Positive Valence. In this context, Positive Valence seems to measure not social desirability but distinctiveness, an extraordinary, remarkable quality that contrasts with the ordinary normality of average, conventional people. Among the phrases that high scorers used to describe themselves are elegant, rejined; gijted, talented; and exceptional, special. Compare Hesse's description of Narcissus and Goldmund: 'Both were refined, both were different from the others because of obvious gifts and signs; both bore the special mark of fate.. . two exceptional human beings ' (193011968, pp. 17,36). MacKinnon (1960) also noted that 'the truly creative individual has . . . a sense of destiny . . . and almost inevitably a measure of egotism ' (p. 375). Whether open individuals are indeed remarkable in some objective sense is debatable; it seems clear, however, that they regard themselves as special.
What label should be given to this factor? In particular, is it reasonable to regard it as a measure of Intellect? It would seem strange to have both Intellect and Intelligence factors in the same analysis, and surely the sixth factor is best interpreted as Intelligence. In addition, Intellect seems a poor choice for a factor that is defined by unconventionality, thin mental boundaries, and intuition.
Is Imagination, Saucier's (1992) candidate, a better label? Doubtless high scorers on this Factor V are imaginative, drawn to the 'dream-filled, super-real world of images', yet Imagination, too, seems too narrow a label. It does not convey the blurring of affects or identities that Hartmann (1991) noted, nor the liberal ideologies that Trapnell(l994) pointed out. Openness seems a better metaphor, conveying both the idea of welcoming new input-whether sensory, cognitive, or affective--and the notion of permeability that characterizes the structure and functioning of open minds (Rokeach, 1960).

A JOURNEY TO THE EAST
From the wanderings of Goldmund to the Oriental Tour of 19th-century gentlemen, travel has always been regarded as a broadening experience, useful for putting one's accustomed ideas and values in perspective. Personality psychology can benefit in much the same way by pushing its constructs across geographical and cultural boundaries. Is there a Factor V in non-Indo-European cultures, and if so, is it best characterized by Intellect, or Openness, or some other term?
Surely there is some equivalent of Intellect in the Far East. As most Westerners know, the Confucian tradition placed great emphasis on scholarship, and academic achievement is still highly valued among Chinese and Japanese cultures at home and abroad (Bond, 1991). Yet China also has a mystical tradition, Taoism, that regards mere intellect as a distinctly limited approach to the world. Of Taoist Master Chuang Tzu's teachings it was said: Thoughts such as his that can cross the Dark Streams of death, mount to the Royal Empyrean, that know neither east nor west, north nor south, but plunge into the bottomless chasm; thoughts from which all boundaries have loosened and dropped away.. . -how can you hope to reach them by the striving of a petty intelligence? (Waley, 1939, pp. 35-36).
The elements of Factor V are thus clearly to be found in Asian cultures, but do they covary in the same way? In early studies using Norman's (1963) measure of the FFM, the status of Factor V was problematic. Bond, Nakazato and Shiraishi (1975) clearly replicated four of the five factors, with coefficients of congruence ranging from 0.87 to 0.93. Factor V, however, showed a very marginal congruence coefficient of 0.72. In this Japanese sample, the factor was defined by artistically sensitive and polished, re3ned, but not by intellectual and imaginative, which instead loaded chiefly on Factors 111 and 11, respectively. In a sample of Chinese university students given the Norman instrument in English, Bond (1979) also replicated the FFM, but once again found that intellectual loaded on Factor 111, Conscientiousness, instead of Factor V. Citing the Chinese saying, 'if people do not pursue their studies, they cannot know what is morally right', Bond concluded that for the Chinese, intellect is inextricably tried to 'moral integrity and social responsibility' (p. 53)-aspects of Conscientiousness.
Yet in other Chinese studies, intellectual does join other definers of Factor V.4 Yang and Bond (1990) translated Norman's instrument into Chinese and administered it to large samples. In ratings of father, friend, and self, Factor V was defined by artistically sensitive, intellectual, and imaginative. Polished, re$ned loaded more consistently on Factor I, Extraversion.
One problem with all these studies is the instrument used. More contemporary measures, derived either from the lexical or the questionnaire tradition, might give more consistent results. Yik and Bond (1993) combined imported and indigenous lexical approaches to personality assessment in creating the Sino-American Person Perception Scales (SAPPSs). Goldberg provided adjective markets for each of the five factors, including Intellect, and I suggested adjectives to measure non-Intellectrelated aspects of Openness. When translated into Chinese and administered to 389 Hong Kong Chinese students, six, rather than five, factors emerged, with separate Intellect and (non-Intellect) Openness factors. Further, when another set of adjectives were selected to represent distinctly Chinese aspects of personality, a joint factor analysis in a new sample showed eight factors: the Western Intellect, Openness, Neuroticism, and Extraversion factors were joined by factors labelled Application, Restraint, Assertiveness, and Helpfulness. (It should be noted that in this instance, Assertiveness does not refer to social leadership, but rather to assertion of the self against social pressure; its best items are determined, independent, strong in opinions, and individualistic.) These data might be interpreted to mean that the FFM is a strictly Western phenomenon, one that can be recovered in Chinese samples only when imported instruments impose their view of personality. However, it is also possible that some SAPPS scales may represent lower-level traits. For example, Application and Restraint appear to correspond to what McCrae and Costa (1987) called the 'proactive' and 'inhibitive' aspects of the broader factor of Conscientiousness.
It is possible to test this hypothesis through a joint factor analysis of the SAPPS scales with data from a Chinese translation of the NEO-PI-R. A small group of Chinese students completed both instruments, and the intercorrelations were reported C. H. Hwang (1982, personal communication, April 4, 1994 pointed out that 'hollow-mindedness', an intellectual humility that promotes openness to new ideas, was an important virtue for scholars in the Confucian tradition. by Cheng, Cheng, Ng and Yip (1991) under the supervision of M. H. Bond. Table  5 reports an analysis of these intercorrelations. This is a clear replication of the FFM in Chinese, with Application and Restraint, as hypothesized, the chief definers of a Conscientiousness factor. And how is Factor V defined? Chiefly by NEO-PI-R Openness and by the non-Intellect Openness SAPPS scale, and secondarily by Assertiveness and Intellect. High scorers on Factor V in Chinese prefer variety and are eager to change, but they are also individualistic and analytical. It seems clear from the body of Chinese and Japanese research that Intellect in the East, even more clearly than in the West, combines Fiske's (1949) Inquiring Intellect with an applied Intellect that is chiefly related to Conscientiousness; within AB5C systems that combination would probably be classified as 1II-tV-t.
Once again Intellect appears to be a dubious label for Factor V.
When the facet scales of the NEO-PI-R are themselves analyzed in Chinese (McCrae, Costa and Yik, in press) and in Japanese (Gondo, Shimonaka, Nakazato, Ishihara and Imuta, 1993), a clear Openness factor emerges (although in both cases 0 4 : Openness to Actions is a weak definer of the f a~t o r ) .~ These data support the view that the FFM is a universal model of personality structure, and that in the East as in the West, Factor V is best construed as a broad dimension in which intellectual interests and imagination are joined by aesthetic sensitivity, affective responsiveness, and liberal attitudes.
Would high scorers on Factor V in China and Japan also show the permeability of consciousness that Hartmann's Boundary Questionnaire measures? No data have yet been collected on that question, nor have any Taoist sages or Zen masters com-All six 0 facets clearly define that factor in a Korean version of the NEO-PI-R (K. I. Lee, personal communication, March 31, 1994).

R. R. M c C r a e
pleted the NEO-PI-R. But the essential message of Eastern mysticism is the loss of the bounded individual ego in an unbounded whole, a transcendental experience to which individuals with thin boundaries are especially likely to be open.

CONCLUSION
Openness to Experience seems to be poorly represented in natural languages (McCrae, 1990). Aspects of Openness are so seldom expressed in Hungarian adjectives that no clear Factor V emerges in that language at all Szirmak and De Raad, 1994). In English, such familiar traits as 'prefers variety', 'has broad interests', and 'is aesthetically sensitive' are represented by no single adjective. This problem is probably even more acute for less observable traits that characterize the structure of consciousness-tendencies to experience synesthesia, isolation of affect, or fluid ego boundaries.
In an earlier article in this journal I urged those interested in personality structure to move 'beyond the confines of the dictionary' (McCrae, 1990, p. 127) to examine natural language in speech and literature. Speech in everyday life may be more rhetorical than descriptive (De Raad and Caljt, 1990), but literature can certainly be illuminating. Hesse's descriptions of Goldmund, for example, point to traits that would probably never be identified in a list of adjectives, yet are crucial to an understanding of individual differences in experiencing the world.
Perhaps it is unreasonable to ask researchers to review systematically the vast corpus of world literature in search of new traits. But it is not unreasonable to ask that they survey the psychological literature. If we wish to understand personality traits and their structure, it is time to move beyond the comfortable boundaries of the lexicon to consider also the constructs of such thinkers as Jung, Tellegen, and Hartmann. Research so far suggests that the FFM encompasses all these systems, and by encompassing them it is clarified and enriched.