88 Biological Bases of Personality
disorder and is recognized as a primary personality trait only
in the Zuckerman-Kuhlman model.
EXTRAVERSION/SOCIABILITY
All models of basic personality, with the exception of
Cloninger’s, recognize extraversion (E) as a primary and
basic personality factor, but different models have defined it
differently. In his earlier model Eysenck regarded E as a com-
bination of two narrower traits: sociability and impulsivity.
This amalgam was questioned by Carrigan (1960) and
Guilford (1975), who claimed that sociability and impulsivity
were independent traits. Sybil Eysenck and Hans Eysenck
(1963) initially defended the dual nature of extraversion.
However, the introduction of psychoticism (P) into a new ver-
sion of their questionnaire resulted in a drift of impulsivity-
type items to the P dimension, leaving E defined primarily by
sociability and activity types of items. Hans and Michael
Eysenck (1985) finally defined E in terms of the subtraits: so-
ciable, lively, active, assertive, sensation seeking, carefree,
dominant, surgent, and venturesome.
Costa and McCrae (1992a) defined their E superfactor in
terms of subscale facets: warmth, gregariousness (sociabil-
ity), activity, excitement seeking (sensation seeking), and
positive emotions. Neither Eysenck nor Costa and McCrae
now include impulsivity in the E factor; Eysenck now
includes it in the N superfactor, and Costa and McCrae place
it in their neuroticism factor. Both Eysenck and Costa and
McCrae include activity and sensation seeking as compo-
nents of their E factors.
Zuckerman et al. (1993) include only sociability and iso-
lation intolerance in their sociability superfactor. In the alter-
native five, impulsivity and sensation seeking form another
primary factor instead of being subsumed under E, and activ-
ity comprises another major factor. In spite of these differ-
ences in the content of the E factor in the three models, the
questionnaire measures of the factors intercorrelate highly
and have high loadings on a common factor (Zuckerman
et al., 1993).
Cortical Arousal
Eysenck’s (1967) theory of extraversion has shaped much of
the psychobiological research on this trait even to the end
of the century (Strelau & Eysenck, 1987). The model suggests
that introversion-extraversion is based on arousal characteris-
tics of the cerebral cortex as regulated by the reticulocortical
activating system. The extravert’s cortex in waking, nonstim-
ulating conditions is underaroused relative to his or her
optimal level of arousal. In these conditions the extravert is
prone to seek out exciting stimulation in order to increase the
level of arousal to a level that makes him or her feel and func-
tion better. The introvert is usually closer to an optimal level
of arousal in low stimulation conditions and has less need to
seek additional stimulation to feel better. The introvert may be
overstimulated at a level of stimulation that is positive for the
extravert.
The theory was initially tested with measures of brain ac-
tivity from the electroencephalogram (EEG). Spectrum
analyses break the raw EEG into bands characteristic of dif-
ferent degrees of arousal: sleep (delta), drowsiness (theta),
relaxed wakefulness (alpha), and alert excitement (beta).
Alpha has often been regarded as inversely related to arousal
on the assumption that any interruption of this regular wave
means an increase in arousal. However, some have used the
frequency of alpha within the usual band (8–13 Hz) as a
measure of relative arousal or alpha amplitude as an inverse
measure of arousal. EEG spectrum characteristics are highly
if not completely heritable (Lykken, 1982).
The findings relating extraversion to EEG criteria of
arousal in various conditions from nonstimulating to mentally
engaged have been summarized by Gale (1983), O’Gorman
(1984), and Zuckerman (1991). Gale tried to reconcile the
wide variety of results with the hypothesis that differences be-
tween introverts and extraverts appear only in moderately ac-
tive conditions and not in either low stimulation (eyes closed,
no stimulation) or activating conditions. Both O’Gorman and
Zuckerman concluded that neither Eysenck’s broad hypothe-
sis nor Gale’s narrow hypothesis, limiting the prediction to
specific experimental conditions, were consistently sup-
ported by studies. Zuckerman noted that among the best stud-
ies, those confirming Eysenck’s hypothesis used samples
with either all female or equal male and female participants,
whereas those with all male or a preponderance of male par-
ticipants did not support the hypothesis.
A large study utilizing the full spectrum range of EEG,
three levels of activating conditions, measures of impulsivity
as well as E, and a test of the interaction of personality,
arousal level, and performance, found only weak evidence
supporting Eysenck’s hypothesis (Matthews & Amelang,
1993). Correlations of .16 (about 3% of the variance) were
found between activation in the low arousal bands (delta and
theta) and E and one of its components, impulsivity. These
correlations controlled for the influence of the other two
Eysenck factors, neuroticism and psychoticism. The sociabil-
ity component of E was not related to any index of cortical
arousal. The significant results linking E to low arousal bands
were found only in the least stimulating condition (reclining,
eyes closed). The fact that the differences were not found in