Home       SAPA Project Test       Papers

Eysenck's PEN Model: Its Contribution to Personality Psychology

KwangMin Jang
Northwestern University

To explain individual differences in personality or temperament, Eysenck proposed the PEN model and Gray attempted to reformulate Eysenck's theory. This paper summarizes and evaluates the PEN model. Special attention is given to the contribution of the PEN model to an experimental approach to the study of personality. In the PEN model, personality is comprised of three major dimensions: extraversion, neuroticism, and psychoticism. These descriptive dimensions have psychophysiological roots in which cortical arousal causes extraversion, visceral brain activation causes neuroticism, and gonadal hormones and enzymes cause psychoticism.

In the search for a model of individual differences in personality, many theorists have suggested criteria for a good model. Among the criteria suggested are evidence of "temporal stability and cross-observer validity" (Costa & McCrae, 1992a, p. 653), universality, testability, replicability, and practicality (Eysenck, 1991; Gray, 1981). In particular, Eysenck (1990) distinctively suggests that an adequate model of personality must have two interlocking aspects: descriptive or taxonomic, and causal or biological. From this perspective, Eysenck (1991) further claims that the PEN model constitutes a paradigm in personality research. To support the theory, proponents of the PEN model (e.g., Eysenck & Eysenck, 1985) emphasize the use of not only correlational research methods such as factor analysis, but also experimental research methods. In the course of the research, some theorists (e.g., Gray, 1981; Revelle, Humphreys, Simon, & Gilliland, 1980; Zinbarg & Revelle, 1989) have attempted to modify Eysenck's original theory to better account for their empirical data.

Nevertheless, the PEN model deserves a good evaluation for its contribution to the development of personality psychology. Distinctively, the PEN model strongly advocates the scientific process for evaluating theories with experimental evidence. This paper will summarize and evaluate the PEN model, focusing especially on Eysenck's theory of individual differences in human temperament and Gray's reformulation of Eysenck's theory.

Descriptive Aspects

Hierarchical Taxonomy

Personality can be studied from either temperamental or cognitive aspects, or both (Eysenck & Eysenck, 1985). Eysenck (1991), however, focuses on the temperament aspect of personality in his PEN model. For better understanding of the PEN model, therefore, the study should begin with its description or taxonomy of personality or temperament. As Eysenck (1991) states, "In any science, taxonomy precedes causal analysis" (p. 774). In the course of taxonomy, any organisms can be organized into groups based on characters and their relationships. Eysenck describes in plain terms how taxonomy in the study of personality can be achieved using the correlational technique called factor analysis:
In the case of personality study the organisms concerned are human beings, preferably randomly chosen, or with sex, age and other restrictions; the characters are traits, measured by experiment, by rating, by self-rating, or in some other way (e.g. projective test, miniature situations, etc.). We can correlate traits over subjects, or subjects over traits, giving us groups of people showing similarity over traits, or groups of traits, cohering as factors over people. We can then look at the traits (or people) having the highest factor loadings in order to better identify the trait clusters. (Eysenck, 1991, p. 775)
Descriptively, individual differences in personality or temperament are analyzed in terms of traits, which can be defined as theoretical constructs based on "covariation of a number of behavioral acts" (Eysenck & Eysenck, 1985, p. 12). However, Eysenck (1991) further supposes that traits themselves intercorrelate and make up higher-order factors or superfactors, which Eysenck calls "types."

As a result, the PEN model proposes a hierarchical taxonomy of personality containing four levels (Eysenck, 1990). At the very bottom level of the hierarchy are behaviors such as talking with a friend on a single occasion. At the second level are habits such as talking with friends on multiple occasions, which are comprised of recurring behaviors. The third level of the hierarchy is that of traits or factors such as sociability, which are comprised of intercorrelated sets of habits. At the top of the hierarchy are superfactors or dimensions of personality such as extraversion, which are intercorrelated sets of traits or factors. Eysenck suggests three such superfactors: extraversion (E), neuroticism (N), and psychoticism (P). These three superfactors or dimensions of personality are orthogonal to each other, which means that they do not correlate with each other (Eysenck & Eysenck, 1985).

The PEN model is based on the principle of "aggregation," in which measures will have higher reliability if they are comprised of many items (Eysenck, 1990). That is, each superfactor in the PEN model is comprised of many different factors, habits, and behaviors, and thus reliability of measurement is increased. The PEN model is also based on the state-trait distinction. Traits are "semipermanent personality dispositions," whereas states are "transient internal conditions" that are produced when traits and situations interact with each other (Eysenck & Eysenck, 1985, p. 33). That is, the superfactors of extraversion, neuroticism, and psychoticism at the top level of the hierarchy are stable, whereas behaviors such as talking with a friend on a single occasion at the bottom of the hierarchy are changeable across time and situation. In this respect, the distinction between levels is very important for the analysis of personality in the PEN model.

Three Dimensions of Personality

There are vigorous debates regarding the number of dimensions that define personality (Costa & McCrae, 1992a, 1992b; Eysenck, 1991, 1992b, 1992c). In this respect, Eysenck strongly advocates that there are only three major dimensions or superfactors in the description of personality: extraversion-introversion; emotional stability versus instability, or neuroticism; and psychoticism versus impulse control (Eysenck & Eysenck, 1985). In the PEN model, these dimensions or superfactors are based on "constitutional, genetic, or inborn factors, which are to be discovered in the physiological, neurological, and biochemical structure of the individual" (Eysenck & Eysenck, 1985, pp. 42-43).

Researchers on the PEN model emphasize the dimensional aspect of personality, as opposed categorization (Eysenck, 1992a; Eysenck & Eysenck, 1985). That is, each person does not necessarily have either 100 percent or zero percent of extraversion, neuroticism, or psychoticism. An individual may show some degree of these superfactors on the continuum. A person may have high extraversion, moderate neuroticism, and low psychoticism. Eysenck (1992a), for example, provides empirical evidence to support this "dimensional or continutiy hypothesis" for psychoticism (p. 776). Following are three interesting points Eysenck (1992a) suggests after studying psychosis:

  1. Psychotic symptoms and illnesses do not form completely separate diagnostic entities, unrelated to each other, but are genetically related and form a general cluster with severity of illness the major distinguishing marker....
  2. Psychosis is not a separate diagnostic entity which is categorically separated from normality; it is merely an extreme along a continuum of abnormality shading into schizoid personality, 'spectrum' disorders, psychopathy and personality disorder, criminality and alcoholism, and average types of behaviour right to the other extreme of empathy, altruism and selflessness.
  3. This continuum is co-linear with the concept of psychoticism, embodied (however imperfectly) in the P scale of the EPQ, and also in a number of 'schizotypy' constructs and scales.... All the elements of this theory are empirically testable, and have been so tested on numerous occasions. (pp. 776-777)
On this continuum, a person with high extraversion is sociable, popular, optimistic, and rather unreliable, whereas a person with low extraversion is quiet, introspective, reserved, and reliable. A person with high neuroticism is anxious, worried, moody, and unstable, whereas a person with low neuroticism is calm, even-tempered, carefree, and emotionally stable. A person with high psychoticism is troublesome, uncooperative, hostile, and socially withdrawn, whereas a person with low psychoticism is altruistic, socialized, empathic, and conventional (Eysenck & Eysenck, 1985).

Furthermore, the superfactors of extraversion, neuroticism, and psychoticism appear to be universal. Such universality has been demonstrated in cross-cultural studies using the Eysenck Personality Questionnaire (EPQ). Evidently, the studies show that the same dimensions of personality emerge in many different nations and cultures other than Western countries (Eysenck, 1991; Eysenck & Eysenck, 1985).

Although the overall evidence supports the PEN model quite well, there are also many anomalies to be cleared up. For example, the trait of impulsivity was originally under the superfactor of extraversion in the Eysenck Personality Inventory (EPI), but later it was moved to psychoticism in the Eysenck Personality Questionnaire (EPQ) (Eysenck & Eysenck, 1985). What happened was that impulsivity correlated quite well with extraversion, "but even better with psychoticism" (Eysenck & Eysenck, 1985, p. 69). Some researchers, such as Gray (1981), disagree with this removal from extraversion and strongly believe that impulsivity, as well as anxiety, should be treated as uniquely important.

Causal Aspects

Based on a three-dimensional description of personality, the PEN model further attempts to provide causal explanation of personality. The PEN model looks for psychophysiological, hormonal, and other biological mechanisms responsible for the personality dimensions, so that the theory can be tested by scientific experiments. Eysenck and Eysenck (1985) clearly contend that "no theory would be considered valid that did not make testable and verified predictions" (p. 187). Consequently, Eysenck (1990) proposes the arousal theory, by modifying his inhibition theory to explain the causal roots of the three dimensions of personality.

Extraversion and Cortical Arousal

According to the arousal theory, Eysenck (1990) provides a biological explanation of extraversion in terms of cortical arousal via the ascending reticular activating system (ARAS). Activity in the ARAS stimulates the cerebral cortex, which, in turn, leads to higher cortical arousal. Cortical arousal can be measured by skin conductance, brain waves, or sweating (Eysenck, 1990). Because of the different levels of ARAS activity, "introverts are characterized by higher levels of activity than extraverts and so are chronically more cortically aroused than extraverts" (Eysenck & Eysenck, 1985, p. 197, emphasis added).

Based on the Yerkes-Dodson law, the arousal theory of the PEN model assumes that "some intermediate level of arousal is optimal for performance" (Eysenck & Eysenck, 1985, p. 199). The Yerkes-Dodson law suggests that arousal and performance have an inverted-U relationship. That is, task performance is impaired when motivation is either very low or very high, and performance is maximized at some intermediate level of "optimal" motivation.

Geen (1984) also supports Eysenck's arousal theory. In his experiments, Geen measures preferred stimulation levels in introverts and extraverts and effects on arousal and performance. The results show that introverts choose a lower level of noise than do the extraverts, and both introverts and extraverts show no difference in arousal and performance with preferred noise level. Geen therefore concludes that "the best performance for both introverts and extraverts...occurred when stimulation was given at the appropriate optimal level," and that "the data pertaining to the measure of performance are consistent with the Yerkes-Dodson Law" (p. 1311).

Neuroticism and Visceral Brain Activation

Eysenck (1990) also explains neuroticism in terms of activation thresholds in the sympathetic nervous system or visceral brain. The visceral brain is also referred to as the limbic system, which consists of the hippocampus, amygdala, septum, and hypothalamus, and regulates such emotional states as sex, fear, and aggression. It is responsible for the fight-or-flight response in the face of danger. Heart rate, blood pressure, skin conductance, sweating, breathing rate, and muscular tension in the forehead can measure activation levels of the visceral brain (Eysenck, 1990; Eysenck & Eysenck, 1985). Neurotic individuals have greater activation levels and lower thresholds within the visceral brain. They are easily upset in the face of very minor stresses. However, emotionally stable people are calm under such stresses because they have lesser activation levels and higher thresholds (Eysenck, 1990).

Empirically, Ormel and Wohlfarth (1991) report that neuroticism indeed has strong influence on psychological distress. They find that "temperamental dispositions seem more powerful than environmental factors" in predicting psychological distress, and that neuroticism is a "powerful determinant" of high psychological distress (p. 753).

Psychoticism and Gonadal Hormones

Eysenck (1990) also provides a biological explanation of psychoticism in terms of gonadal hormones such as testosterone and enzymes such as monoamine oxidase (MAO). Although there has not been a lot of research done on psychoticism in comparison with extraversion and neuroticism, the current research shows that people who show a psychotic episode have increased testosterone levels and low MAO levels.

Impulsivity and aggressiveness were negatively correlated with MAO, which plays a role in the degradation of the monoamines norepinephrine, dopamine, and serotonin (Eysenck, 1990, 1992a). Eysenck (1992a) reports that "low platelet monoamine oxydase (MAO) has been found in psychotic patients, and also in their relatives and inpatients who have recovered, suggesting that low MAO activity may be a marker for 'vulnerability'" (p. 774).

Gray's Reformulation

Gray (1981) agrees with Eysenck's notion of hierarchical description of personality, but does not agree with the idea that extraversion and neuroticism are the defining factors of personality. Gray thinks that Eysenck's removal of the impulsivity factor from the superfactor of extraversion is a bad idea, and strongly believes that impulsivity, as well as anxiety, should be treated as a major dimension in personality. To do that, Gray reformulates Eysenck's theory by rotating the dimensions of extraversion and neuroticism by 45 degrees. This results in two new dimensions: impulsivity (Imp-D), which is high on both neuroticism and extraversion, and anxiety (Anx-D), which is high on neuroticism, but low on extraversion. According to Eysenck and Eysenck (1985), however, "it would seem preferable to locate the anxiety dimension closer to the neuroticism dimension than to that of extraversion" to accord with questionnaire measures of anxiety. Then, "impulsivity would have to be moved closer to the dimension of extraversion so that it would remain orthogonal to the dimension of anxiety" (p. 210).


Gray's reformulation of Eysenck's theory also has neurological grounding and behavioral consequences. People with high impulsivity are highly sensitive to reward and non-punishment, whereas people with high anxiety are highly sensitive to signals of punishment, non-reward, and novelty. The underlying system for impulsivity is the behavioral activation system (BAS), which involves "the medial forebrain bundle and the lateral hypothalamus" (Eysenck & Eysenk, 1985, p. 211). In contrast, the underlying neurological system for anxiety is the behavioral inhibition system (BIS). The BIS consists of "an interacting set of structures comprising the septo-hippocampal system (SHS), its monoaminergic afferents from the brain stem and its neocortical projection in the frontal lobe" (Gray, 1981, p. 261).

According to Gray (1981), impulsivity and anxiety are the most important dimensions of personality. Accordingly, extraversion and neuroticism are "secondary consequences of the interactions between impulsivity and anxiety systems" (p. 261). That is, a person in whom the behavioral inhibition system (BIS) of anxiety is more powerful than the behavioral activation system (BAS) of impulsivity becomes introverted, and a person in whom the BAS is relatively more powerful than the BIS becomes extraverted. Introverts, for example, are more sensitive to signals of punishment, non-reward, and novelty than they are to signals of reward and non-punishment. Thus, in Gray's perspective, the superfactor of extraversion reflects the "relative" strength of both impulsivity and anxiety, whereas neuroticism reflects their "joint" strength, in which increments in the sensitivity of either the impulsivity or anxiety system provide increments to neuroticism (p. 261). Gray's theory also predicts that extraverts will exhibit superior conditioning with a rewarding unconditioned stimulus, opposed to Eysenck's prediction of introverts' superiority in conditioning (Zinbarg & Revelle, 1989).

Some research results (e.g., Anderson & Revelle, 1994; Revelle et al., 1980; Zinbarg & Revelle, 1989) indicate that impulsivity and anxiety are more consistently and strongly associated with individual differences in performance than extraversion and neuroticism. Revelle et al. (1980) reported that the administration of moderate doses of caffeine hindered the performance of introverts and helped the performance of extraverts on a cognitive task similar to the verbal test of the GRE. However, these phenomena were affected by time of day and impulsivity. In the morning, low impulsives were hindered and high impulsives helped by caffeine. But this pattern reversed in the evening. The researchers concluded that the results from the test require a revision of Eysenck's theory. Instead of a stable difference in arousal between low and high impulsives, it appeared that these groups differed in the phase of their diurnal arousal rhythms. That is, low impulsives are more aroused in the morning and less aroused in the evening than are the high impulsives.

The data reported by Revelle et al. (1980) are of great importance in support of Gray's theory. Gray (1981) even called these data "a dagger that goes to the heart of Eysenckian theory" (p. 258). This finding challenges Eysenck's arousal theory, and indicates the complexity of the interaction between personality traits and situations.

Impulsivity and Anxiety Versus Extraversion and Neuroticism

Although Gray's theory is important to know for understanding impulsivity and anxiety, Eysenck's theory is undeniably the basis of Gray's reformulation. Moreover, it is too early to draw the conclusion, as Gray (1981) suggests, that impulsivity and anxiety are more important dimensions of personality than extraversion and neuroticism. More research results have to be reported on BAS and BIS. As Gray (1981) admits, "little progress has been made in describing the structures" that go to make up BAS, in particular (p. 261). Moreover, other study results (Eysenck, 1990; Eysenck & Eysenck, 1985; Zinbarg & Revelle, 1989) require explanations different from Gray's reformulation. For example, the study on the effects of personality on salivation to lemon juice shows that the amount of salivation correlated "more highly with extraversion than with impulsivity" (Eysenck & Eysenck, 1985, p. 211).


All things considered, the PEN model has contributed to the study of personality in three distinctive ways. First of all, it combines both descriptive and causal aspects of personality in one theory (Eysenck, 1997; Stelmack, 1997). This characteristic clearly distinguishes the PEN model from most other trait theories such as the five-factor model (Costa & McCrae, 1992a, 1992b; Eysenck, 1991, 1992b, 1992c). By providing causal explanations in addition to the description of personality, the PEN model is supported by more credible evidence than purely descriptive models. The combination in one theory of two important aspects of personality makes it possible to understand personality as a whole.

Second, the PEN model is comprehensive in description by proposing a hierarchy of four levels and by making a clear distinction among those levels. This characteristic can play another critical role for the comparison with other trait theories. Even though Costa & McCrae's (1992a) five-factor model is also hierarchical, their model seems to mix up lower-level factors with higher-level superfactors (Eysenck & Eysenck, 1985; Eysenck, 1991, 1992b). That is, the big five dimensions of agreeableness and conscientiousness are traits at the third level that combine as part of a superfactor of psychoticism at the top level of the PEN model. For understanding the very nature of personality, fewer independent factors are better than many factors overlapping one another. Moreover, the five-factor model includes "intellect" or "openness" at the top level (Costa & McCrae, 1992a). But the PEN model draws a clear line between temperament and cognitive ability and treats intelligence differently. That does not mean the PEN model totally excludes intelligence from personality. Rather, advocates of the PEN model "adopt the more common view that intellectual processes can be discriminated from emotional ones" (Eysenck & Eysenck, 1985, p. 159).

Finally, the PEN model becomes most compelling because of its experimental approach to the study of personality, which makes the model more testable. Consequently, the PEN model is likely "to generate more specific predictions because knowledge about the functioning of the specified physiological structures is available" (Eysenck & Eysenck, 1985, p. 192). Other researchers can then go through the scientific process to evaluate the theory through various experiments. The experiments may not uniformly support the PEN model and may provoke some disagreement (e.g., Anderson & Revelle, 1994; Gray, 1981; Revelle et al., 1980; Stelmack, 1990; Zinbarg & Revelle, 1989). However, the experimental approach of the PEN model serves a good role model for other personality theories (Eysenck, 1991, 1992b, 1992c, 1997; Stelmack, 1997). Overall, the PEN model has contributed to the advancement of the study of personality by calling attention of the importance of a scientific approach to personality.

Peer Commentary

Elaboration and Clarification of the PEN Model's Contribution to Psychology

Neera Mehta
Northwestern University

In "The PEN Model: Its Contribution to Personality Psychology," Jang provides an excellent general overview of Eysenck's three-dimensional personality model. He begins by discussing Eysenck's use of a hierarchical taxonomy and the descriptive aspects of the three superfactors: psychoticism, extraversion, and neuroticism. He then goes on to further discuss the biological roots of these three dimensions of personality, presents Gray's model as an example of research that has modified Eysenck's model, and then further evaluates the PEN model. Although the main ideas of Eysenck's model are presented clearly in this paper, certain details need further clarification and elaboration.

Without providing much validation, Jang seems to conclude that Eysenck's personality dimensions are more empirically sound than Gray's dimensions. He states, "Although Gray's theory is important to know for understanding impulsivity and anxiety, Eysenck's theory is undeniably the basis of Gray's reformulation. Moreover, it is too early to draw the conclusion that impulsivity and anxiety are more important dimensions of personality than extraversion and neuroticism."

It is important to remember, however, that the PEN model's contribution to personality psychology does not only involve Eysenck's theories on personality dimensions, but also the modified personality models that have been inspired as a result of the PEN model. Indeed some of this research that has used Eysenck's model as a starting point, such as Gray's model, may be just as empirically significant, and in some aspects even more valid than the PEN model. Jang himself earlier states, "Some research results indicate that impulsivity and anxiety are more consistently and strongly associated with individual differences in performance than extraversion and neuroticism." This contradiction between statements indicates that in some aspects impulsivity and anxiety may indeed be "more important dimensions of personality." This commentary will attempt to clarify and elaborate on specific points concerning Eysenck's and Gray's models that were simplified by Jang, as well as present new research concerning these models.

The first issue that needs further clarification is the reason behind introverts' increased sensitivity to stimulation. Jang simply states, "Because of the different levels of the ARAS [ascending reticular activating system] activity, introverts are characterized by higher levels of activity than extraverts and so are chronically more cortically aroused than extraverts." He then supports this statement with research indicating that introverts choose lower levels of noise than extraverts. To a reader unfamiliar with this theory, however, the reasons why introverts choose lower levels of noise may not be clear. A more coherent description of the role of arousal in extraversion is necessary.

The ARAS is a primitive part of the brain that is not involved with higher cortical functions. It is responsible for general arousal and regulates patterns of wakefulness and attention. According to Eysenck's arousal model, introverts are more aroused than extraverts because the ARAS is regulated differently in introverts as opposed to extraverts. As Jang mentioned, introverts and extraverts seek optimal arousal in daily life. Jang does not explain, however, that because introverts are more aroused to begin with, they can tolerate less increase in arousal than extraverts. The introvert's cerebral cortex, therefore, must exert more inhibition over primitive lower brain centers, resulting in more inhibited behavior in introverts as opposed to extraverts. Introverts, therefore, perform better than extraverts under low levels of stimulation but perform worse at high levels of stimulation (Eysenck, 1967).

In regard to this finding, Jang did not note that this relation between extraversion and arousal was reported based on the Eysenck Personality Inventory extraversion (EPI-E) scale, not the Eysenck Personality Questionnaire extraversion (EPQ-E) scale, and specifically, the EPI-Imp (impulsivity) subscale and not the EPI-Soc (sociability) subscale. A reanalysis of the previous results, therefore, found that this arousal effect on extraversion was due to its impulsivity component (Revelle, 1997). This finding is very important in terms of the ongoing debate regarding impulsivity's role in personality. As noted by Jang, the findings in the study assessing performance and arousal using caffeine and a GRE-like test were eventually attributed to impulsivity and time of day. It is interesting to note that the ARAS regulates patterns of wakefulness and attention, which is, therefore, consistent with the time of day effects that were found in relation to impulsivity. These results, therefore, call into question the assumption that extraverts are always less aroused than introverts. They, therefore, question the basis of Eysenck's arousal model, and are indeed a major finding in support of Gray's model.

In terms of the causal aspects of neuroticism, Jang clearly explains the connection between neuroticism and visceral brain activation. What is not evident from this short paragraph on neuroticism, however, is that research on the biological roots of neuroticism has not been consistent, and even Eysenck admits that we know little about the organic underpinnings of this dimension (Eysenck, 1990). This shortcoming is important to note because admitting other possibilities when discussing theories such as these leaves room for improvement and reformulation by new research. For example, in terms of neuroticism, some later research indicates that brain asymmetry may result in increased neuroticism. Specifically, increased activation in the right hemisphere of the brain in anterior sites is associated with heightened negative affect (which is associated with neuroticism), decreased positive affect, or both. On the other hand, increased activation of the left hemisphere anterior regions results in heightened positive affect, decreased negative affect, or both (Davidson & Tomarken, 1989). Again, is it important when discussing the PEN model's contribution to personality psychology to emphasize other researchers that used this model's experimental approach as a springboard for their own research.

After discussing the causal aspects of neuroticism, Jang discusses psychoticism and its relation to hormones and enzymes. He reports the connection between decreased levels of monoamine oxidase (MAO) and increased psychoticism, and then further indicates that decreased levels of MAO are a marker for vulnerability to impulsivity and aggressiveness. He, therefore, links impulsivity to psychoticism here and then later addresses how impulsivity is considered a separate dimension in Gray's model. He does not, however, indicate why he chooses to support Eysenck's theory that impulsivity is a component of psychoticism, which he has supported by discussing the negative correlation between impulsivity and MAO. He also only briefly mentions the unstable role of impulsivity in Eysenck's model by saying, "What happened was that impulsivity correlated quite well with extraversion, but even better with psychoticism." As seen by the previously mentioned findings on impulsivity in relation to arousal and extraversion, understanding the history of impulsivity is crucial in understanding findings resulting from Eysenck's research.

In the early years of Eysenck's two-dimensional model, extraversion was considered a mixture of impulsivity and sociability. Impulsivity was considered an integral part of extraversion at that time, and sociability was given little importance. Through the years sociability gained importance, and eventually with the introduction of psychoticism to Eysenck's personality model, the EPI was replaced by the EPQ as a measure of personality traits. This change from the EPI to the EPQ is important to note because it meant a drastic difference in the definition of extraversion. It was originally reported that the EPQ measured extraversion and neuroticism with the same criteria as the EPI. The extraversion scale, however, had a significantly different item content. Specifically, the majority of the impulsivity items on the extraversion scale had been removed. Sociability now dominated extraversion, and impulsivity was considered to be part of psychoticism (Revelle, 1997). This difference in measures is important to consider in terms of research findings, such as the finding of introverts' preference for lower levels of noise as compared to extraverts'.

After coherently describing Gray's model, Jang reports that Eysenck's and Gray's theories predict opposite findings in relation to conditioning and extraversion. He states that Gray predicts that extraverts will have superior conditioning, whereas Eysenck predicts that introverts will have superior conditioning. Again, he simply states this difference between the two theorists without elaborating on the reason for this difference, which may not be clear to a reader unfamiliar with these theories.

According to Eysenck, introverts are more highly aroused than extraverts, and therefore, because arousal enhances certain types of learning, introverts should learn simple tasks more easily than extraverts. Gray, on the other hand, considers conditioning differences to be a result of the behavioral activation system (BAS) and the behavioral inhibition system (BIS). According to Gray, highly impulsive individuals should learn better from rewards, and highly anxious individuals should learn better from punishments. Their predictions are different, therefore, in that Eysenck thinks introverts in general are more conditionable than extraverts, whereas Gray thinks introverts are more conditionabe only in terms of punishments, whereas extraverts are more conditionable in terms of rewards (Zinbarg & Revelle, 1989). Again, although Jang reported this difference, he did not take a stand on which theory he felt was best supported by research, nor did he present any research relating to this issue. Research does indicate that impulsiveness and anxiety interact to provide a better fit to conditioning data than neuroticism and extraversion (Revelle, 1997).

In terms of Gray's model, unfortunately, agreement has not been reached about how to assess Gray's personality dimensions. The EPQ cannot be used because Gray views extraversion as a balance between impulsivity and anxiety, and the EPQ-E has a large component of sociability and little impulsivity. Sociability seems to play no role in Gray's model, and therefore, the two theories of extraversion differ in an essential way. Some researchers use measures of impulsivity, but even within Gray's theory there are doubts about whether impulsive behavior truly comes only from the BAS, as opposed to a combination of the BAS and the BIS (Carver & White, 1994). For instance, research indicates that a person with high BIS sensitivity is vulnerable to anxiety in certain situations; however, this person may also learn to avoid anxiety-provoking situations, and therefore may experience little day-to-day anxiety (Fowles, 1987). As seen from such findings, it is obvious that more research needs to be done on the source of impulsivity, and, as Jang claims, on the BAS and the BIS.

Although Jang simplifies many of the descriptive and causal aspects of this model, he also presents many positive aspects of the PEN model, especially emphasizing its important contribution to combining correlational and experimental psychology. He claims, however, that the PEN model is a good role model for other personality theories, but then he seems to belittle Gray's theory due to the fact that it is based on Eysenck's theory. He cites only one example, however, of a finding that refutes Gray's model (the salivation to lemon juice study), whereas many of the other cited examples throughout his paper support the dominance of impulsivity over extraversion. I, therefore, challenge Jang to justify his conclusion that Eysenck's dimensions of personality are "more important dimensions of personality" than Gray's dimensions of personality.

Peer Commentary

Why Did Eysenck Move Impulsivity?

Han S. Paik
Washington University

Jang has written a good review of Eysenck's PEN model and Gray's reformulation of that model. Both the descriptive and causal aspects of Eysenck's PEN model are adequately described. Each of the different levels of the hierarchy, in the taxonomic portion of the model, is drawn out well, and each of the biological causes for the three different dimensions is also explained carefully. Also, Gray's reformulation of the model is treated with sufficient representation. Overall, Jang has done a good job of covering the many different aspects involved with the PEN model and has explained them thoroughly. However, I hope to help in the clarification of some of the finer details of the model.

Jang states that, "Although the overall evidence supports the PEN model quite well, there are also many anomalies to be cleared up." He mentions that the trait of impulsivity was originally under the superfactor of extraversion but was later moved to psychoticism. Furthermore, the correlation of the factor of impulsivity with extraversion was quite good but was found later to be even better with psychoticism, which is the reason for the move of the factor. Unfortunately, there is no further explanation for this anomaly.

I feel that this is a very important detail. Although it may not overthrow the entire PEN model as a tool for trait psychology, it is an important detail that deserves attention. In Gray's reformulation of Eysenck's theory, Gray rotated the axes of extraversion and neuroticism by 45 degrees. These rotated axes were newly labeled impulsivity and anxiety. The impulsivity axis is the same factor that Eysenck moved from extraversion to psychoticism. If Gray felt that impulsivity is important enough to comprise a primary axis, then impulsivity must be very important indeed.

I will attempt to provide an explanation for Eysenck's movement of impulsivity. In 1957 Eysenck first developed the inhibition theory that originally underlay his PEN model. He had previously tested many different people and used factor analysis to arrive at the three different superfactors. In his initial development of his inhibition theory, he tested hysterics and found that impulsivity had the highest correlation with extraversion. Also, Eysenck developed the biological causes for each of the different superfactors in his 1957 model. Jang does a very good job of explaining all of these topics.

However, in 1967, when Eysenck developed the arousal theory, he changed his focus from hysterics to psychopaths. When testing psychopaths, Eysenck found that impulsivity had the highest correlation with psychoticism. Consequently, he moved the impulsivity factor from extraversion to psychoticism.

The reasons that impulsivity had different correlation values with the same superfactors are twofold. First, he tested different people. In his 1957 formulation, he tested hysterics, and in his 1967 formulation, he tested psychopaths. Second, he changed the test that he used to test his participants. In 1957 he used the Eysenck Personality Inventory, and in 1967 he used the Eysenck Personality Questionnaire. These two different tests consisted of different items. As a result, he changed the superfactor to which the factor of impulsivity belonged (Gray, 1981).

Jang obviously has a thorough understanding of Eysenck's PEN model and of Gray's subsequent reformulation of that model. However, he does not adequately explain the details of impulsivity and its migration from extraversion to psychoticism.

Peer Commentary

Following Eysenck's Experimental Approach

Jennifer A. Schroer
Northwestern University

Jang discusses Eysenck's (1967) contribution to personality study and the experimental approach to science, but he does not elaborate on what future researchers did with the information. Gray (1981) and Revelle, Humphreys, Simon, & Gilliland (1980) find fault with Eysenck's basic dimensions of the arousal model. Gray and Revelle et al. both use a laboratory setting to test predictions as to the behavior of extraverts and introverts and of high and low impulsives. Gray and Revelle et al. both follow the experimental approach recommended by Eysenck, but come up with a different conclusion as to structure.

Eysenck's PEN model is clearly described by Jang. However, Jang also needs to describe how Gray, and especially Revelle et al., follow Eysenck's use of experimentation to arrive at their results. Gray states that Eysenck's theory cannot perform the task that Eysenck set for it. But a theory is only ever killed by a better theory. Do we have one? This is an important question because as many problems as the PEN model has, it provided a basis for both Gray and Revelle et al. to expand upon. Eysenck gave future researchers a starting point that they previously did not have.

Gray expands upon Eysenck's theory by rotating the lines of causal influence 45 degrees. The new resulting dimensions are anxiety (Anx-D) and impulsivity (Imp-D). Gray uses neurological and behavioral data to reach the conclusion that Eysenck's PEN dimensions are only secondary consequences of the interactions between the anxiety and impulsivity systems.

Revelle et al. find through experimentation that introverts and extraverts have different diurnal rhythms instead of the chronic over- and under-arousal (respectively) suggested by Eysenck. Introverts are more aroused in the morning and extraverts are more aroused in the evening. The findings show that the effects of impulsivity and sociability depend on situational determinants of arousal. Revelle et al. built upon Blake's (1967) finding that introverts have a higher temperature in the morning (in agreement with Eysenck's postulate that they are more aroused than extraverts), but by late afternoon and evening extraverts have the higher body temperature. Revelle et al.'s conclusion, found through experimentation (an independent variable was heightened arousal due to caffeine intake, and the dependent variable was performance on a GRE-like test), is that time of day is a critical component of a personality model. Revelle et al. state, "It should be noted that our results suggest that the classic formulation of introversion/extraversion (Eysenck, 1967) needs to be reconsidered.... [C]omponents of introversion/extraversion function in such a different fashion as to make the analysis of the secondary factor of introversion/extraversion of dubious value. This emphasizes the importance of suggestions...to look at impulsivity and sociability separately" (p. 25). The experimental approach suggested by Eysenck is eventually the same approach that dethrones his theory.

Not enough attention is given by Jang to the expansion of Eysenck's arousal model based on new experimentation. Revelle et al. provide results that contradict Eysenck's arousal model. Jang concludes by agreeing with Eysenck's model even after he gives much information contrary to its structure. Many advances in Gray's reformulation of the PEN model are stated but then ignored in the paper's conclusion. The paper therefore is internally inconsistent. The conclusion should explain how much Gray and Revelle et al. have expanded Eysenck's scientific approach. They agree with Eysenck's basic premise of experimentation, but demand a reformulation of his theory based on their own experimental results.

Author Response

Who Is Right, Eysenck or Gray?

KwangMin Jang
Northwestern University

With many thanks to the peer commentators, I would like to clarify an important point that I raised in the conclusion of my paper. In the paper "The PEN model: Its Contribution to Personality Psychology," I concluded that it is through the experimental approach that the PEN model becomes most compelling for the study of personality. The experimental approach makes it possible for other researchers to go through their own experiments to evaluate the theory (Eysenck & Eysenck, 1985). Although other experiments may not support the PEN model and may even provoke some discrepancies (e.g., Anderson & Revelle, 1994; Gray, 1981; Revelle, Humphreys, Simon, & Gilliland, 1980; Stelmack, 1990; Zinbarg & Revelle, 1989), the experimental approach of the PEN model itself serves as a good role model for other personality theories (Eysenck, 1991). As the peer commentator Schroer correctly notes, the PEN model "gave future researchers a starting point that they previously did not have."

The peer commentaries on my paper mainly discuss the discrepancies regarding Eysenck's movement of impulsivity and Gray's reformulation of Eysenck's theory. In my paper, I briefly summarized those issues, but still needed further in-depth explanations and evaluation. The peer commentators gratefully raised many good points to clarify and elaborate on what I simplified about many aspects of the PEN model. They did not only add details for further clarification and elaboration, but also presented new research concerning Eysenck's and Gray's models.

In his peer commentary, Paik added a good explanation for Eysenck's movement of impulsivity from extraversion to psychoticism. As Paik described, Eysenck originally found that impulsivity was correlated with the personality superfactor extraversion when he tested hysterics with the Eysenck Personality Inventory (EPI) during the development of the inhibition theory. However, when Eysenck changed his testing to psychopaths with the Eysenck Personality Questionnaire (EPQ), he found that impulsivity correlated "even better with psychoticism" (Eysenck & Eysenck, 1985, p. 69). Consequently, Eysenck (1967) proposed the arousal theory to better accommodate the new findings.

Since this change has been made, many other researchers have questioned the validity of Eyseck's arousal theory. As mentioned in my paper and also in Schroer's commentary, Gray (1981) and Revelle et al. (1980), in particular, proposed the reformulation of Eysenck's arousal theory. In her peer commentary, Schroer correctly noted that, for Gray (1981), "Eysenck's PEN dimensions are only secondary consequences of the interactions between the anxiety and impulsivity systems." Schroer also precisely noted Revelle et al.'s (1980) findings that, "introverts and extraverts have different diurnal rhythms [depending on high or low impulsivity] instead of the chronic over- and under-arousal (respectively) suggested by Eysenck" (Schroer, above).

However, despite these disagreements with Eysenck's theory, we should not hasten to conclude that only Gray (1981) and Revelle et al. (1980) are correct and to "dethrone" Eysenck's theory as Schroer seemed to propose. We also should not deprecate Gray's theory because of the fact that it is based undeniably on Eysenck's theory. As Mehta accurately noted in her commentary, "the PEN model's contribution to personality psychology does not only involve Eysenck's theories on personality dimensions, but also the modified personality models that have been inspired as a result of the PEN model."

It is probably more worthwhile to emphasize the contributions and similarities of various theories to the PEN model, as described by Eysenck and Eysenck (1985). With a focus on the same two-dimensional personality space of extraversion-neuroticism, both Eysenck's and Gray's theories can explain many experimental findings through similar theoretical frameworks. "If, for example, it is found that high-anxiety subjects perform some task less efficiently than low-anxiety subjects, H. J. Eysenck can simply redescribe the findings as revealing the inferiority of neurotic introverts to stable extraverts" (Eysenck & Eysenck, 1985, p. 212). Similarly, Gray accounts for the unequal performances as differences of "susceptibility to punishment and non-reward: the greater the degree of introversion, the greater is this susceptibility" (Eysenck & Eysenck, 1985, p. 212). One finds that the subtle differences in their explanations are, in fact, derived from the similar theoretical frameworks of the two-dimensional personality space of extraversion-neuroticism.

There are differences between the two theories in making some predictions, of course, such as those pertaining to conditioning. As mentioned in my original paper and also in Mehta's commentary, Eysenck predicts introverts will be superior in conditioning, whereas Gray predicts extroverts will be more superior in conditioning. Some experiments, however, show equivocal evidence for both theories, which makes it hard to take a stand on either side. The study results (Eysenck, 1990; Eysenck & Eysenck, 1985; Zinbarg & Revelle, 1989), such as the study about salivation to lemon juice correlating more highly with extraversion than with impulsivity (Eysenck & Eysenck, 1985), demand further research, as I mentioned in my paper. Regarding Revelle et al.'s (1980) finding, the researchers' choice of a highly academic-type task (i.e., GRE) has been questioned for its methodological validity because "the psychological processes involved are largely unknown" (Eysenck & Eysenck, 1985, p. 286). Mehta also admitted in the conclusion of her commentary that more research on the source of impulsivity had to be conducted.

Unlike Mehta's challenge, therefore, my conclusion here, as well as in my original paper, is not to claim that, "Eysenck's dimensions of personality are 'more important dimensions of personality' than Gray's dimensions of personality." I, rather, agree with the claim that "the currently available evidence is insufficient to justify any change" of the PEN model (Eysenck & Eysenck, 1985, p. 286). Consequently, more research results on many experiments have to be reported for a more well-grounded conclusion.

Generally speaking, the peer commentators helped me to clarify and emphasize the most important point of my paper in terms of the contribution of the PEN model to the study of personality. It is through the experimental approach that the PEN model contributes to personality psychology. Although experiments have provoked current disagreement (Gray, 1981; Revelle et al., 1980), this is a natural process in developing any good theory. Apart from the argument of pros or cons, the PEN model deserves a fair evaluation simply because of its emphasis on an experimental approach.


Anderson, K. J., & Revelle, W. (1994). Impulsivity and time of day: Is rate of change in arousal a function of impulsivity? Journal of Personality and Social Psychology, 67, 334-344.

Blake, M. S. F. (1967). Relationship between circadian rhythm of body temperature and introversion-extraversion. Nature, 215, 896-897.

Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: The BIS/BAS Scales. Journal of Personality and Social Psychology, 67, 319-333.

Costa, P. T., & McCrae, R. R. (1992a). Four ways five factors are basic. Personality and Individual Differences, 13, 653-665.

Costa, P. T., & McCrae, R. R. (1992b). Reply to Eysenck. Personality and Individual Differences, 13, 861-865.

Davidson, R. J., & Tomarken, A. J. (1989). Laterality and emotion: An electrophysiological approach. In F. Boller & J. Grafman (Eds.), Handbook of neuropsychology (pp. 419-441). Amsterdam: Elsevier.

Eysenck, H. J. (1957). The dynamics of anxiety and hysteria: An experimental application of modern learning theory to psychiatry. London: Routledge & Kegan Paul.

Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL: Charles C. Thomas.

Eysenck, H. J. (1990). Biological dimensions of personality. In L. A. Pervin (Ed.), Handbook of personality: Theory and research (pp. 244-276). New York: Guilford.

Eysenck, H. J. (1991). Dimensions of personality: 16, 5, or 3?--Criteria for a taxonomic paradigm. Personality and Individual Differences, 12, 773-790.

Eysenck, H. J. (1992a). The definition and measurement of psychoticism. Personality and Individual Differences, 13, 757-785.

Eysenck, H. J. (1992b). Four ways five factors are not basic. Personality and Individual Differences, 13, 667-673.

Eysenck, H. J. (1992c). A reply to Costa and McCrae. P or A and C--The role of theory. Personality and Individual Differences, 13, 867-868.

Eysenck, H. J. (1997). Personality and experimental psychology: The unification of psychology and the possibility of a paradigm. Journal of Personality and Social Psychology, 73, 1224-1237.

Eysenck, H. J., & Eysenck, M. W. (1985). Personality and individual differences: A natural science approach. New York: Plenum.

Fowles, D. C. (1987). Application of a behavioral theory of motivation to the concepts of anxiety and impulsivity. Journal of Research in Personality, 21, 417-435.

Geen, R. G. (1984). Preferred stimulation levels in introverts and extraverts: Effects on arousal and performance. Journal of Personality and Social Psychology, 46, 1303-1312.

Gray, J. A. (1981). A critique of Eysenck's theory of personality. In H. J. Eysenck (Ed.), A model for personality (pp. 246-277). Berlin: Springer.

Ormel, J., & Wohlfarth, T. (1991). How neuroticism, long-term difficulties, and life situation change influence psychological distress: A longitudinal model. Journal of Personality and Social Psychology, 60, 744-755.

Revelle, W. (1997). Extraversion and impulsivity: The lost dimension? In H. Nyborg (Ed.), The scientific study of human nature: Tribute to Hans J. Eysenck at eighty (pp. 189-212). New York: Elsevier.

Revelle, W., Humphreys, M. S., Simon, L., & Gilliland, K. (1980). The interactive effect of personality, time of day, and caffeine: A test of the arousal model. Journal of Experimental Psychology: General, 109, 1-31.

Stelmack, R. M. (1990). Biological bases of extraversion: Psychophysiological evidence. Journal of Personality, 58, 293-311.

Stelmack, R. M. (1997). Toward a paradigm in personality: Comment on Eysenck's (1997) view. Journal of Personality and Social Psychology, 73, 1238-1241.

Zinbarg, R., & Revelle, W. (1989). Personality and conditioning: A test of four models. Journal of Personality and Social Psychology, 57, 301-314.

Last modified August 1998
Visited times since July 2001

Home to Personality Papers

Home to Great Ideas in Personality