A Remarkable Replication
Putting aside the clinical versus empirical-statistical arguments for the time being, it is now recognized that the emergence of five consistent factors from diverse psychological data was a remarkable breakthrough. Why should one be able to start with an assortment of statements, which broadly sample individual personal feelings, attitudes, interests, and beliefs, reduce them to five predictable themes, and then be able to repeat the process on a second independent set of statements and individuals, with comparable results? It is because there is something there of value, especially given that the five-factor “solution” has now been replicated many times in the world of psychometrics—the science of measuring human mental traits. The responses of a few hundred people on any broad-spectrum personality test, when factor analyzed, will generally reveal five main factors resembling the Big Five personality model, regardless of how many scales the personality test reported. This phenomenon is at its most compelling with the simplified analysis of descriptive adjectives that apply to people, and with minor hiccups has been shown to work in over a dozen languages (Saucier & Goldberg, 2003, 2006). Indeed, the principle holds whether people are describing themselves or being described by people who know them well (Connelly & Ones, 2011).
Remarkably (but perhaps not surprising to William McDougall), this is a consistent pattern whether the descriptive statements are about normal everyday lives or about lives in clinical populations experiencing seriously limiting pathological feelings (Livesley, Jackson, & Schroeder, 1992). Of course, a key unresolved issue is whether this truly reflects a deep and abiding nature of human personality structures or something about the languages we have devised/invented to talk about one another. In other words, the issue is whether these five dimensions encapsulate the dynamic energizing forces that underlie our personalities or merely the conceptual, descriptive categories of our culturally generated languages, used to describe our behaviors. Of course, this is a solid scientific beginning that can have multiple implications and interpretations, all of which can be empirically linked to neurobiological processes (in research mostly yet to be done), but which do not intrinsically tell us anything about those processes.
Our position is that factor analysis by itself does not illuminate how brain systems, genetic factors, and other physiological mechanisms (hormones, etc.) might be involved with each of the statistically revealed five factors. We believe that factor analysis is a purely descriptive tool. It is easy to get caught in the classic correlation trap of assuming that finding a strong correlation means that you have found the “cause” of something. For example, the fact that liking to talk means you are extraverted is “circular,” because liking to talk was part of the definition of extraversion in the first place. So the factor-analytic concept of extraversion cannot be seen as the cause of liking to engage in social conversion, or vice versa. We must look much further to clarify the underlying brain and psychological cause—one of our reasons for writing this book. In part, we created the Affective Neuroscience Personality Scales so we might better clarify the underlying brain processes (but more on that issue later).
The Big Five Model Is Robust
What makes the Big Five model so compelling is that it has been replicated many times across the years. The Big Five usually works regardless of the sample of individuals tested, as long as an adequate sample of personality test items (or scales) is analyzed. Start with a reasonably broad sample of personality descriptors and a population of subjects that is at least three times as large as the number of items you ask them to rate; for instance, with 70 items you would ideally have 210 subjects or more. Use a standard factor-analytic procedure to extract five factors, and the items loading on those five factors are very likely to resemble Extraversion, Emotional Stability, Conscientiousness, Agreeableness, and Openness to Experience. Given adequate samples, Extraversion, Emotional Stability, Conscientiousness, and Agreeableness are almost guaranteed. However, as many have found, identification of the most subtle of the variables, Openness to Experience, has often been more difficult, perhaps because this arises substantially from the most subtle of the emotional systems, the SEEKING system. This system, one of the oldest, in evolutionary terms (see Chapter 10), is so pervasive in everyday waking life that it tends to get molded into the opportunities that the social world provides to exercise one’s natural enthusiasm to know and find more and more. In any event, when scientists find a phenomenon as robust as the Big Five, they are academically bound to agree there is something there. But that something has to be analyzed, neuropsychologically, neurogenetically and neurobiologically for major progress to occur. We think the Affective Neuroscience Personality Scales provides one such tool—one that bridges past affective-emotional brain research with future understanding of our underlying mental nature. But, before heading further let us flesh out the history of the field.
EMERGENCE OF THE BIG FIVE PERSONALITY MODEL
Even though Cattell, more than anyone else, promoted factor analysis as an objective personality research tool, he certainly did not originate the idea that there might be five major personality traits that could encompass most of personality. That was left to many other psychologists, and the origins of the Big Five, or the Five-Factor Model (FFM), have been well described by John Digman (1990, 1996). As noted in Chapter 11, in 1933 Cattell (1933) published a correlational analysis of forty-six bipolar items that identified four of the Big Five dimensions, including a Conscientiousness factor, which Cattell labeled “Will,” and an Extraversion factor, which Cattell labeled “Surgency.” In addition, Cattell identified a dimension he called “Maturity” (or the Big Five Agreeableness dimension) and a “Well-adjusted” versus Maladjusted factor, which in retrospect align with Emotional Stability. This report was an historical stepping-stone on the journey to accepting what would come to be called the Big Five personality model.
Donald Fiske: The First Validation of the Big Five
As the historical saga played out, Donald Fiske (1917–2003) of the University of Chicago came close to a full FFM solution. Using twenty-two of Cattell’s rating scales, Fiske (1949) spent a year analyzing three different rating modes: self-ratings, peer ratings, and supervisor ratings. He identified five consistent dimensions in each data set. However, his five-factor solutions lacked a Conscientiousness factor. He used the labels of Social Adaptability (Extraversion), Conformity (Agreeableness), Emotional Control (Emotional Stability), and Inquiring Intellect (Openness to Experience). Unfortunately, Conscientiousness was not one of his factors, and his fifth factor was labeled Confident Self-Expression, which seemed more like an aspect of Extraversion.
Ernest Tupes: The Forgotten Validators of the Big Five
Tupes and Christal (1961, 1992), using thirty of thirty-five Cattell scales, were the first to identify a full five-factor personality solution that is still recognized as the Big Five. Parenthetically, it is not unusual in science to forget major originators of scientific perspectives that were critical stepping-stones to the heralded achievements of others. For instance, who remembers that Oswald Avery (1877–1955) was the first scientist to identify DNA as our hereditary material (with his colleagues at Rockefeller University Hospital, published in 1944), while most other investigators in the field thought that proteins (the products of DNA and RNA) were the sources (as opposed to products) of the genes. Avery never received the Nobel Prize; Francis Crick and John Watson, who decoded the molecular structure of DNA, did in 1954 (Reichard, 2002).
In 1961 two Air Force researchers, Ernest Tupes and Raymond Christal, not only reported a stable five-factor model in their own data but also found the same five factors when they reanalyzed Cattell’s and Fiske’s earlier data. However, their report appeared only as an Air Force technical report, which was not widely read by personality researchers. Eventually, their paper was considered so important that it was republished in 1992 in the widely circulated Journal of Personality. They labeled their five broad personality facto
rs Surgency, Agreeableness, Dependability, Emotional Stability, and Culture, which marked the beginning of the current Big Five personality model.
Tupes was working for the Air Force and trying to ensure his work would provide a sound foundation for subsequent Air Force personnel selection programs. Therefore, he used a subset of Cattell’s scales that had already been used to predict high-quality performance of Air Force senior officers, as well as new second lieutenants. To ensure his studies would generalize, he sampled a broad population, including data on more than just members of the Air Force. In the initial Air Force studies, he had used Air Force officer candidates, as well as senior Air Force officers, with some airmen having only a high school education. Subsequently, he added university students and graduate students to the mix, including a reanalysis of the Cattell and Fiske data, because like Fiske he had used a subset of Cattell’s scales.
His data from all eight studies were derived mostly from observer ratings rather than self-reports, and he incorporated a sampling of many different peer raters, including fellow Officer Candidate School mates, Air Force Command classmates, fellow fraternity members, female university students, clinical psychology graduate students, and “experts” such as staff clinical psychologists and psychiatrists. Clearly, his approach was a monumental effort that, in his words, sought “to isolate meaningful and relatively independent trait-rating factors which are universal enough to appear in a variety of samples, and which are not unduly sensitive to the rating conditions or situations” (Tupes & Christal, 1992, p. 227).
IBM Replication
The original Tupes and Christal (1961) study was also unique in terms of one important methodological point: after all of their analyses were completed with the electronic hand calculators available at the time, they were able to replicate one of their factor analyses on a computer. They had gained access to one of the first factor analysis software programs and an IBM 650 computer to serve up the results. Their new computer program produced results virtually identical to those they had previously generated by hand. Tupes and Christal noted that, in the future, this “will not only save many hours of labor, but will bring considerably more rigor to what has thus far been a rather loose area” (1992, p. 246). They further concluded that the consistency they found across their eight studies “has always been there, but it has been hidden by inconsistency of factorial techniques and philosophies” (p. 246).
More Factors?
Tupes and Christal (1961, 1992) found only five personality dimensions in their studies and reported that “nothing more of any consequence” (1992, p. 245) appeared in their reanalysis of Cattell’s data using all thirty-five trait clusters. However, like most others that preceded them, they believed there must be more factors and almost apologized by writing that “It is unlikely that the five factors identified are the only fundamental personality factors. There are quite likely other fundamental concepts involved among the Allport-Odbert adjectives on which the variables used in the present study were based” (p. 247).
How prescient was their prediction! But this was not a problem to be resolved by factor analysis working with tertiary-level (language) data. The personality field had to wait for neuroscience to blossom and especially on our ever-increasing understanding of the anatomical and neurochemical nature of mammalian emotions (Panksepp, 1982, 1998a; Panksepp & Biven, 2012). As previously discussed, from an affective neuroscience perspective, the Emotional Stability dimension of the Big Five may actually be composed of three separate and distinct brain systems that generate distinct negative feelings: RAGE/Anger, FEAR, and PANIC/Sadness, and, except for Conscientiousness, the other Big Five dimensions may correspond to other distinct emotional strengths and weaknesses. In any event, just with the recognition that there are at least three primal negative emotional networks in mammalian brains, affective neuroscience could potentially add two additional emotional personality dimensions, which might yield a Big Seven list, including Conscientiousness (more on this later).
That said, it seems that correlational analyses of our languages have issued generalizations, such as conflating distinct negative emotions into a single category of negative emotionality. In contrast, psychiatrists and psychotherapists need a much sharper vision, including realizing that the three negative primal emotions, RAGE/Anger, FEAR, and PANIC/Sadness, have distinct brain substrates, to guide development of distinct treatments for problems in living arising from these distinct systems (Panksepp et al., 2014; Panksepp, 2015, 2016; Panksepp & Yovell, 2014).
THE WIDER ACCEPTANCE OF THE FIVE-FACTOR MODEL
Somewhat like the young medical student and future neuroscience researcher Roger Bannister breaking the four-minute mile barrier in 1954, once broken, all great runners were up to the task. Unremarkably, everyone in human personality research now seemed able to find five major personality factors (often using slightly different semantic labels). Of course, this partly had to do with the emergence of ever more powerful computer processing, as well as more standardized factor-analytic techniques. In 1963 Warren Norman (1930–1998) of the University of Michigan, only six years out of graduate school, reported that peer ratings of twenty scales resulted in five factors, which he labeled Surgency, Agreeableness, Conscientiousness, Emotionality, and Culture (Digman, 1996). In 1964 Edgar Borgatta of the University of Washington obtained ratings of fraternity and sorority members using thirty-four scales he designed himself. Borgatta (1964) also reported five factors, which he named Assertiveness, Likeability, Responsibility, Emotionality, and Intelligence. In 1967 Gene Smith of the Harvard Medical School in Boston used forty-two bipolar scales to obtain ratings of three independent groups of first-year college students. The same five factors emerged from the factor analysis of each group, and Smith labeled the dimensions Extraversion, Agreeableness, Emotionality, Strength of Character, and Refinement. A clear, repetitive pattern was emerging. For easier comparisons, the factor names used in these studies are listed in Table 12.1
Yet it would still be many years before the Big Five would gain wide acceptance. Alternative models were still being proposed. Eysenck (1955), Peabody (1967), and McCrae and Costa (1983) all proposed three-factor models. Block proposed a two-factor model with three subtypes (Block, 1961), and Myers and McCaulley (1985) put forth a four-factor model with sixteen types. Also, new personality instruments such as the Personality Research Form (Jackson, 1984) were published with far more than five factors.
Lewis Goldberg
In addition to reanalyzing Cattell’s old data (Digman & Takemoto-Chock, 1981), John Digman and colleagues completed new studies supporting the Big Five (Digman & Inouye, 1986). Similarly, Paul Costa and Robert McCrae (1985) revised their three-factor Neuroticism Extraversion Openness Personality Inventory (NEO-PI) to incorporate five factors.
Elsewhere, Digman (1996) also describes how Lewis Goldberg, of the University of Oregon, played a central role in this Big Five renaissance. Goldberg (1990) published an influential article that was seen to definitively establish the generality of the Big Five as a robust phenomenon. He based his research on Norman’s (1967) 1,431 English language adjective synonyms clustered into seventy-five scales. This was the most comprehensive set of adjectives ever used in personality research, which eliminated the argument that the five-factor solution resulted from analyzing an incomplete personality domain. In one case, students rated themselves on 1,710 trait-descriptive adjectives working in one-hour segments until they had completed this Herculean task. Of course, none of Goldberg’s ambitious factor analyses would have been possible without ever more powerful modern computers.
To quash the argument that five factors would be found only with specific factor-analytic techniques, Goldberg (1990) used five mathematically different methods of extracting the factors, and also allowed the factors from each method to be either correlated with each other or completely orthogonal (a term that means the factor dimensions are mathematically “perpendicular” to each other, which basically means “indepe
ndent of each other”). With each of the ten procedures (five different factoring methods, each once using correlated and once orthogonal relationships), five factors were derived with virtually identical results. Goldberg tried extracting more than five factors and found that the basic five factors remained intact. with extra factors being fragments of the basic five.
Table 12.1. Factor Names Used in Early Five Factor Model Studies
Source Labels
Fiske, 1949 Social Adaptability Conformity (Missing) Emotional Control Inquiring Intellect
Tupes & Christal, 1961 Surgency Agreeableness Dependability Emotional Stability Culture
Norman, 1963 Surgency Agreeableness Conscientiousness Emotionality Culture
Borgatta, 1964 Assertiveness Likeability Responsibility Emotionality Intelligence
Smith, 1967 Extraversion Agreeableness Strength of Character Emotionality Refinement
One can see how several distinct negative emotions may fall under a general Emotionality factor. One function of language is to simplify complexities, to the point where psychologists often seem to believe that words are explanations for diverse preverbal psychological processes (e.g., the nominal fallacy, assuming that naming equals explaining; or the post hoc ergo propter hoc fallacy, assuming that correlation equals causation).
In any event, with his seminal paper, Goldberg demonstrated consistent replications of five factors regardless of whether people were describing themselves, rating peers they liked, or rating peers they disliked. He successfully challenged previous reservations about the generality and robustness of the Big Five personality model. Using a comprehensive collection of English language personality-descriptive adjectives, the Big Five structure had emerged regardless of samples and statistical variations. As if this weren’t sufficient, Goldberg also provided one hundred Big Five “markers” to add precision to the core definitions of these five basic personality dimensions (Goldberg, 1992).
The Emotional Foundations of Personality Page 25