. 45
( 51 .)


0.8 5


Inhibited Uninhibited
(a) 4-year-old group

% BOLD signal change

0.5 4
(c) Uninhibited Inhibited
Figure 8.9 (a) Levels of cortisol in children that were determined to be shy and wary. Adapted
from Schmidt et al. (1997), (b) Colorized group statistical map superimposed on coronal
group averaged T1 structural image in Talairach space. Signi¬cant fMRI signal changes
10 5; Talairach coordinates x, y,
(arrows) are shown in the right (peak P value 2.5
10 4; x, y, z
21, 6.5, 14) and left (P 4.2 21.5, 6.7, 18) amygdalae
(Amy) and occipito-temporal cortex (OTC), and (c) Percent (%) blood oxygenation level-
dependent (BOLD) signal change (versus ¬xation) in Amy to novel versus familiar faces
in adult subjects who were inhibited and uninhibited in the second year of life. One
standard error of the mean is indicated. Adapted from Schwartz et al. (2003a)

behavioral and physiological assessments of these children at 51„2 years of age sug-
gested that cortisol levels were discriminative between the two temperamental
extremes. Additionally, at 71„2 years of age, those described as shy and timid in the
initial assessment were quiet and socially avoidant in novel social situations sug-
gesting that this temperamental category is stable over time (Kagan et al., 1988).
This has important health implications, as shy children with high levels of cortisol
are vulnerable to allergic symptoms (Bell et al., 1990; Kagan et al., 1991), vascular
disease (Bell et al., 1993) and anxiety disorders (Van Ameringen et al., 1998; Kagan
252 J. Schulkin et al.

and Snidman, 1999), perhaps because of the chronic worry that they experience in
social contexts or in unfamiliar environments.
Linking behavioral, physiological, and endocrine measures in humans to neural
activation in hypothesized regions of interest, such as the amygdala, is currently
underway. One should note that the amygdala is involved in a broad array of
behavioral regulation, including the response to novelty. Novel events are poten-
tially dangerous, and the amygdala is involved in the perception of what is novel
and what is familiar. More recent evidence, using functional magnetic resonance
imaging (fMRI) in humans, has expanded on some of the earlier insights into the
diverse functions of the amygdala with regard to perception of novel stimuli such
as unfamiliar faces (Schwartz et al., 2003a; Figure 8.9(b)). Theoretically, these same
inhibited children who were described at 2 years of age as shy and socially wary
should reveal greater activation of the amygdala when shown novel faces. A longi-
tudinal study of these shy and fearful children when they reached adulthood found
that amygdala activation was greater when viewing novel faces than viewing famil-
iar faces, compared to those categorized as uninhibited during early childhood
(Schwartz et al., 2003b) (Figure 8.9(c)). The amygdala activation to familiar faces
by the inhibited and uninhibited adults did not differ. The increased amygdala
activation to unfamiliar faces suggests that increased glucocorticoids paired with
inhibited temperament during early childhood may have lifelong effects on pro-
cessing of emotional stimuli, even though at 71„2 years cortisol levels were not as
discriminating of the inhibited and uninhibited children as they had been during
earlier assessments (Kagan et al., 1988).

Prefrontal cortex and temperament, cortisol, and CRH
The prefrontal cortex is tied to temperamentally fearful and distressed infants and
behaviorally inhibited toddlers; these children exhibit a pattern of greater relative
right frontal electroencephalography (EEG) activation at rest and heightened star-
tle responses to a stranger approach at 9 months of age (Schmidt et al., 1997; 1999a).
These temperamentally fearful and distressed infants who develop shyness later in
childhood are also characterized by elevated basal and reactive salivary cortisol
(Gunnar et al., 1989; 1996).
The pattern of frontal brain activity and salivary cortisol responses in tempera-
mentally shy children is preserved up through the school age years. For example, we
have noted that children who were classi¬ed as extremely shy and socially wary at age
4 exhibited elevated morning salivary cortisol (Schmidt et al., 1997) and greater rela-
tive right frontal EEG activation at rest (Schmidt et al., 1997; Davidson and Rickman,
1999) compared with their socially outgoing counterparts. Temperamentally shy chil-
dren also exhibit a greater increase in right, but not left, frontal EEG activity and heart
rate in response to social challenge compared with their non-shy counterparts at age 7
253 Glucocorticoid facilitation of CRH in the placenta and the brain

(Schmidt et al., 1999a) and they display a relatively lower decrease from baseline levels
on salivary cortisol reactivity measures (Schmidt et al., 1999a). Six-month-old human
infants who show withdrawal behaviors displayed the same right-greater-than-left
frontal activation and higher basal and reactive cortisol concentrations found in non-
human primates (see below Buss et al., 2003).
Turning to nonhuman primates, a subset of young rhesus monkeys can be char-
acterized as anxious and fearful by observing their behavioral reactions to stressful
situations. These monkeys freeze for longer periods of time than other rhesus mon-
keys not characterized by fearful behavioral responses and have high levels of cor-
tisol (Champoux et al., 1989). These characteristics can be induced in macaques by
manipulating rearing conditions. When macaques are raised in an arti¬cial envi-
ronment with their peers, instead of in a more naturalistic environment with their
parents and extended family, alterations in behaviors and in hormones like corti-
sol and CRH are observed. In adult rhesus monkeys, high levels of cortisol and
high levels of CRH from the CSF are associated with behavioral inhibition (Habib
et al., 2000; Kalin et al., 2000).
A subset of these macaques not only have higher levels of CRH and cortisol than
other monkeys, but they also demonstrate greater fearful temperament and greater
activation of the right hemisphere. Increased relative right hemisphere activation
has been linked to withdrawal and negative perception of events (Davidson and
Rickman, 1999; Kalin et al., 2000). Differences in temperamental expression to a
number of unconditioned fear-related stimuli may re¬‚ect frontal neocortical acti-
vation (Kalin et al., 2001). Ibotenic acid lesions (cell body destroyed and ¬bers left
intact) of the macaque amygdala left a number of unconditioned behavioral trait-
like responses intact (Kalin et al., 2001), in addition to the normal asymmetry asso-
ciated with trait-like dispositions (Figure 8.10).
Much of the research on the role of prefrontal cortical regions in emotional behav-
ior, affective experience and temperament, characterizes emotion into an approach
and withdrawal dichotomy. EEG studies indicated that during reward and punish-
ment paradigms, reward trials were associated with greater left hemisphere acti-
vation while punishment trials were associated with greater right hemisphere
activation (Davidson, 2000). Extending these EEG ¬ndings, individual differences in
baseline frontal lobe asymmetry suggested that greater right frontal activation might
be associated with a more negative affective style, and vice versa with greater baseline
left-than-right frontal activation. Frontal cortex responses to reward and punish-
ment interact with underlying baseline asymmetries in activation (Davidson et al.,
1990). When these EEG recordings were performed with children, the generalization
was upheld, showing that those children who displayed social competence had
greater relative left frontal activation, while those who were characterized as with-
drawn had greater right frontal activation (e.g. Schmidt et al., 1997; 1999a). Greater
254 J. Schulkin et al.

70 100

CSF, CRH (pg/ml)
Cortisol ( g/dl)
0 0
Left Right Right
(b) Frontal asymmetry groups
(a) Frontal asymmetry groups

Figure 8.10 (a) Levels of systemic cortisol, and (b) CSF, CRH in left and right frontal brain activation
in macaques who were more fearful, and demonstrated greater right prefrontal activation.
Adapted from Kalin et al. (1998a; 2000)

right frontal electrical activity is stable over time in nonhuman primates, and is cor-
related with more defensive responses and elevated cortisol concentrations (Kalin
et al., 1998b). The CSF CRH concentrations are also elevated and stable over time in
monkeys with extreme right frontal activation (Kalin et al., 2000).
Interestingly, the medial prefrontal cortex (anterior cingulate) also plays a part
in the glucocorticoid response to stress. When the cingulate is lesioned in rats,
restraint stress leads to increased plasma ACTH and corticosterone levels, and
corticosterone implants to the cingulate signi¬cantly decrease plasma cortico-
sterone in response to stress (Diorio et al., 1993; see also Sullivan and Gratton, 2002).
However, these manipulations do not affect glucocorticoid levels in the absence
of stress.
The prefrontal cortex is divisible into several functional areas, and the medial and
orbital prefrontal (OMPFC) regions appear to be particularly important for emo-
tional regulation (Davidson, 2000). The OMPFC is reciprocally connected to the
amygdala (Amaral and Price, 1984), and in primates glucocorticoid receptors are
distributed in the prefrontal cortex to a much greater extent than they are in rodents
(Sanchez et al., 2000). CRH receptors are expressed in this region of the brain, and
postmortem studies in suicide patients who were diagnosed with severe depression
indicated decreases in CRH receptor distribution (Nemeroff et al., 1988). The fol-
lowing section discusses the implications of altered cortisol and CRH to increased
vulnerability to psychiatric disorders.

Elevated cortisol, CRH, and vulnerability to affective disorders
Increased exposure to stress or uncertainty during early life may produce a vul-
nerability to developing affective disorders. Research on affective disorders such
255 Glucocorticoid facilitation of CRH in the placenta and the brain

as depression and anxiety indicate neural activation and neuroendocrine pat-
terns similar to those observed in those exposed to suboptimal conditions pre- or
postnatal. However, when reviewing this literature, keep in mind that there is also
a signi¬cant genetic contribution to the vulnerability to mood disorders. Therefore,
those with genetic vulnerability may require little to no early exposure to environ-
mental factors to develop psychiatric disorders, while others who are exceptionally
resilient may experience extreme traumas early in life and emerge relatively
Many investigators have found increased functional activity in the amygdala
of patients with depression (Drevets et al., 2002). This increased amygdala activation
correlated with negative affect in a sample of medication-free depressives
(Abercrombie et al., 1998) and was also seen in patients suffering from a number of
anxiety disorders (see Davis and Whalen, 2001). Prefrontal cortex activity is also cor-
related with anxiety and depression (Davidson, 2000). These effects are largely later-
alized in both amygdala and prefrontal cortex (Davidson, 2000; Drevets et al., 2002).
Often in depression, particularly in those with co-morbid anxiety (Gold et al., 1988),
hypercortisolemia, hyperactivity in the HPA axis, and high levels of CRH in CSF are
found (Nemeroff et al., 1984; Arborelius et al., 1999; Holsboer, 2000). Melancholic
depressives (those with hyperarousal, fear, and anhedonia symptoms) reportedly
show a positive correlation between abnormally high levels of cortisol and high but
normal levels of CSF, CRH (Wong et al., 2000) indicating a lack of negative feedback
control of CRH by cortisol. In addition to this apparent dysfunction of negative feed-
back in the HPA axis, elevated levels of cortisol may involve sustained hyperactivity
in the amygdala via feed-forward processes. One study has found a signi¬cant posi-
tive correlation (r 0.69) between glucose metabolism in the amygdala measured
by [F-18]2-deoxy-2-¬‚uoro-D-glucose (FDG) positron emission tomography (PET)
and plasma cortisol levels in both unipolar and bipolar depressives (Drevets et al.,
2002). There is now some evidence that cortisol infusions increase glucose metabo-
lism in the amygdala (Erickson et al., unpublished). It is intriguing to speculate that
the cause of ¬rst depressive episode in patients who also have enlarged amygdala
(Frodl et al., 2002) may be increased chronic levels of glucocorticoids and blood ¬‚ow
in the amygdala (Figure 8.11).
Although the research has developed along two separate paths, activity in the
amygdala in a number of different anxiety disorders has been shown to be highly
reactive to triggers that evoke anxious reactions (Davis and Whalen, 2001), and the
HPA axis is hyper-responsive in anxiety disorders, particularly post-traumatic
stress disorder (PTSD) (Mason et al., 1988; Yehuda et al., 1991; Yehuda, 2002).
PTSD patients tend to have lower basal hypocortisolemia than normals (Mason
et al., 1988; Yehuda, 2002), though not always, but increased reactivity of the HPA
axis to cortisol, suggesting that CRH- and ACTH-secreting cells are sensitized to
256 J. Schulkin et al.


Plasma cortisol ( g/dl)




0.85 0.90 0.95 1.00 1.05
(b) Regional/global metabolism: left amygdala
Figure 8.11 Amygdala and prefrontal cortex activation in depression. (a) Areas of abnormally
increased CBF in familial major depressive disorder (MDD). Analyses show areas of
increased CBF in depressed patients relative to controls in the amygdala and medial


. 45
( 51 .)