Early Childhood Predictors of Adult Anxiety Disorders
Jerome Kagan and Nancy Snidman
This paper considers the influence of temperamental
factors on the development of anxious symptoms in chil-
dren and adolescents.
About 20 percent of healthy children are born with a
temperamental bias that predisposes them to be highly
reactive to unfamiliar stimulation as infants and to be
fearful of or avoidant to unfamiliar events and people as
young children. Experiences act on this initial tempera-
mental bias and, by adolescence, about one-third of this
group is likely to show signs of serious social anxiety.
These children are also likely to have one or more
biological features, including a sympathetically more
reactive cardiovascular system, asymmetry of cortical
activation in EEG favoring a more active right frontal
area, more power in the EEG in the higher frequency
range, and a narrower facial skeleton. The data imply that
this temperamental bias should be conceptualized as
constraining the probability of developing a consistently
fearless and spontaneous profile rather than as determin-
ing an anxious or introverted phenotype.
try 1999;46:1536–1541 © 1999 Society of Biological
Key Words: Anxiety, inhibition, temperament
most of this century with the views of physicians and
philosophers in ancient Athens, Rome, and Alexandria is
the stigma the former group attributes to individuals who
are excessively apprehensive over physical harm, task
failure, or unfamiliar social settings. Citizens in ancient
societies regarded anger, not anxiety, as the more harmful
emotion that individuals should try to control. The con-
temporary judgment that chronic anxiety is a greater
obstacle to adjustment than chronic anger, boredom, or
sexual frustration is the product of historical factors.
Conformity to authority and to the values of the local
community were adaptive in early societies; worry over
ne feature that distinguishes European-American
conceptualizations of psychological variation during
the evaluative judgments of others is easier if one is
vulnerable to apprehension. These anxiety states are a
burden in modern society because assuming risk, defend-
ing an unpopular opinion, engaging strangers, and meeting
difficult obligations serve effective adaptation. Thus, the
current concern with, and intensive study of, anxiety and
its symptoms are understandable.
Current hypotheses regarding the causes of variation in
susceptibility to this family of feelings have cycled back to
the views of Hippocrates and Galen who claimed that
some individuals inherited a constitutional vulnerability to
anxious states, a quality called “temperament” by modern
The suggestion by Thomas and Chess (1977) that
temperamental factors render some children especially
susceptible to fear and anxiety was followed by elegant
discoveries in neuroscience laboratories that made it pos-
sible for scientists to speculate on the possible biological
substrates for vulnerabilities to fear and anxiety. It took
less than 50 years after the first Chess and Thomas
publication to persuade the psychiatric and psychological
community that some children inherited a biology that
made it easier for them to feel uncertain, tense, or
apprehensive to unfamiliarity and challenge and, as a
result, to be more vulnerable to one or more of the anxiety
This paper summarizes what scholars have learned
about these temperamental biases. My colleagues and
I—especially Nancy Snidman, Doreen Arcus, J. Steven
Reznick, and Mark McManis—have been studying this
interesting problem for almost 20 years and much of the
evidence to be summarized comes from my laboratory.
Before presenting the data, however, it is necessary to
make a methodological point.
The primary information in the work to be described
came from direct observations rather than parental de-
scriptions of children. Investigators who rely only on
parental descriptions assume that these verbal reports
correspond closely to what would be detected by direct
behavioral observations. Unfortunately, the degree of cor-
respondence between the two sources of information is
modest at best; therefore, generalizations based only on
parental descriptions have a special meaning and validity
(Spiker et al 1992; Seifer et al 1994; Rosicky 1993; Klein
1991; Perrin and Last 1992; Kagan 1998).
From Harvard University, Cambridge, Massachusetts.
Address reprint requests to: Jerome Kagan, Department of Psychology, Harvard
University, 33 Kirkland Street, Cambridge, MA 02138.
Received February 5, 1999; revised May 3, 1999; accepted May 4, 1999.
© 1999 Society of Biological Psychiatry0006-3223/99/$20.00
Reactivity in Infancy
Research with animals reveals that the amygdala is re-
sponsive to unfamiliar events and, in addition, is a neces-
sary structure for the acquisition of conditioned reactions
that imply a fear state when aversive events, like shock,
are the unconditioned stimuli (Davis et al 1995; Blanchard
and Blanchard 1988; LeDoux 1996). The amygdala con-
tains receptors for a large number of neurotransmitters and
neuromodulators that are relevant to the display of fearful
behavior, including GABA, opioids, norepinephrine, and
corticotropin releasing hormone (Amaral et al 1992). The
basolateral area of the amygdala sends projections to the
ventral striatum and excitation of this structure induces
limb movement in animals (Rolls 1992). Second, the
amygdala is the origin of the amygdalofugal pathway
whose projections to the central gray and anterior cingu-
late modulate distress cries. Thus, infants born with a low
threshold in the amygdala and its projections should
display vigorous limb movements and become easily
distressed to unfamiliar stimulation. Further, as young
children, these infants should be avoidant of, or fearful to,
unfamiliar events. These children are called inhibited. In
contrast, infants born with an amygdaloid neurochemistry
that raised the threshold of these structures should display
minimal motor activity and minimal crying to unfamiliar
stimulation and should be minimally fearful to, or avoidant
of, unfamiliar events. These children are called uninhibited.
The Consequences of Variation in Infant
A sample of 462 healthy, Caucasian, middle-class, four-
month-old infants were administered a battery of visual,
auditory, and olfactory stimuli (Kagan 1994). About 20%
of the sample showed a combination of frequent vigorous
motor activity, including arching of the back, combined
with fretting and crying. These infants were called high
reactive. About 40% of the sample showed the opposite
profile of low motor activity and minimal distress to the
same battery and these children were classified as low
reactive. The remaining infants belonged to other temper-
Almost 80% of the sample of children returned to the
laboratory when they were 14- and 21- months old and
encountered a variety of unfamiliar social and nonsocial
events. The unfamiliar events included: interacting with an
unfamiliar examiner, placement of heart rate electrodes
and a blood pressure cuff, an unfamiliar liquid being
placed on the child’s tongue, and the appearance, in
different episodes, of a stranger, a clown, and an odd-
looking robot made of metal combined with requests by an
adult to approach each of these discrepant objects. Each
episode was scored for the display of an unambiguous fear
reaction, where fear was defined strictly as a display of
fretting or crying to any of the unfamiliar incentives or
failure to approach the stranger, clown, or robot despite a
friendly invitation to do so. The mean number of fears was
2.6 at 14 months and 2.7 at 21 months and about 30% of
the sample showed 4 or more fears at both 14 and 21
months. The high reactive infants displayed significantly
more fears than the low reactives and were more likely to
show four or more fears at both ages (Table 1).
A majority of the high and low reactive children were
evaluated again when they were 4.5 years of age. We
assumed that inhibition of speech in an unfamiliar social
setting is analogous to an animal’s freezing in an unfamil-
Table 1. Differences between High Reactive and Low Reactive Infants from 14 Months to 7.5
Mean fear at 14 months4.2
t(286) ? 11.1, p ? .001
Mean fear at 21 months
t(243) ? 4.67, p ? .001
Mean spontaneous comments at 4.5 years
t(191) ? 1.96, p ? .05
Mean smiles at 4.5 years
t(191) ? 3.69, p ? .01
Percent inhibited with peers at 4.5 years
Percent uninhibited with peers at 4.5 years
Mean spontaneous comments at 7.5 years
?2(1) ? 35.5, p ? .001
t(109) ? 2.37, p ? .01
Mean smiles at 7.5 years
t(109) ? 1.97, p ? .05
Percent with anxious symptoms at 7.5 years
?2(2) ? 12.8, p ? .01
aNumbers in parentheses are standard deviations.
Temperament and Anxiety1537
iar context and, therefore, expected that the high reactive
infants would be far less talkative than low reactives while
interacting with an unfamiliar adult. Further, because the
high reactive infants smiled less often during the evalua-
tions at 14 and 21 months and spontaneous smiling in a
social interaction requires a low level of uncertainty, we
expected that high reactives should smile less often with
the examiner than low reactives. The videotapes of a 60
min battery involving the child and an unfamiliar female
examiner were coded for the number of spontaneous
comments and smiles. In addition, between 3 and 6 weeks
after the laboratory session, the child and parent returned
for a play session with two other unfamiliar children of the
same age and gender. The three parents sat on a couch in
a large playroom while the three children played with
The 4.5-year-olds who had been high reactive infants
displayed significantly fewer comments and smiles with the
examiner and, in addition, were more likely than the low
reactives to be classified as shy/inhibited when playing with
the unfamiliar peers (Table 1). Only 4% of the entire sample
(9 children) displayed a profile at 4.5 years that was incon-
sistent with their original 4 month temperament. Three high
reactive infants were spontaneous and sociable and 6 low
reactive infants were shy. The majority of the children, both
high and low reactive, were neither extremely inhibited nor
uninhibited at 4.5 years (Kagan et al 1998).
Development of Signs of Anxiety
A group of 164 of the original sample of 462 children
were evaluated again when they were 7.5 years old.
Initially, all mothers were sent a questionnaire and
asked to rate their child on a 3-point scale for descrip-
tions of age-appropriate behavior. Twelve of the ques-
tions dealt with anxious symptoms (for example, “My
child becomes quiet and subdued in unfamiliar places,”
“My child is afraid of thunder and lightning,” “My child
is afraid of animals,” “My child has nightmares”). A
score of 1 was assigned to the child if the mother said
that the anxious behavior was sometimes true of the
child, a score of 2 was assigned if the mother said the
behavior was often true of her child. Children with a
total score of 9 or more across the 12 questions were
regarded as potential members of a category of anxious
children. The mothers of these potentially anxious
children were interviewed on the telephone and asked to
provide specific examples to support their descriptions.
These interviews often revealed that some mothers had
exaggerated the seriousness of their child’s behavior
and these children were eliminated from the potentially
anxious group. The teachers of the remaining children
were interviewed on the telephone. Each teacher, who
had no knowledge of the child’s prior behavior or the
purpose of the interview, described and then ranked the
child with respect to all children of the same gender in that
classroom for the qualities of shyness and fearfulness.
The maternal questionnaire and the subsequent inter-
a trio of investigators. If all three agreed that the evidence
indicated the child met criteria for anxious symptoms, the
child was categorized as anxious (42 children or 26% of the
group met that criterion). We selected 107 control children
from the rest of the sample who did not meet criteria for any
symptom and brought all 164 children to the laboratory
where each was administered a variety of procedures. As at
was coded from videotapes.
More high than low reactive infants had acquired
anxious symptoms—45% of high reactives but only 15%
of low reactives and 21% of all remaining children were
classified as possessing anxious symptoms. Further, the
high reactives displayed fewer spontaneous comments and
smiles than the low reactives while interacting with an
unfamiliar female examiner (Table 1). Only a modest
proportion of children were consistently inhibited or con-
sistently uninhibited on all assessments from 14 months to
7.5 years. Only 18% of the high reactives showed a
combination of high fear scores at 14 and 21 months,
inhibition with the examiner and inhibited behavior with
same gender peers at 4.5 years, and, in addition, display of
anxious symptoms at 7.5 years. Not one high reactive
infant developed the complementary profile of low fear at
14 and 21 months, uninhibited behavior at 4.5 years, and
no anxious symptoms at 7.5 years. Thus, the classification
high reactive constrained the likelihood that a child would
develop a consistently uninhibited phenotype. Most high
reactive infants developed a profile in the average range
(Kagan et al 1999).
The 23 high reactives who had anxious symptoms
differed from the 27 high reactives who did not show signs
of anxiety. More members of the former group had a
narrow facial skeleton, higher sitting diastolic blood pres-
sure, and a greater magnitude of cooling of the tempera-
ture of the fingertips while listening to a series of digits
they were asked to remember. The latter 2 variables imply
that these children had greater sympathetic influence on
the cardiovascular system.
The predictive value of the facial skeleton was not a
surprise. We had discovered earlier that the 14- and
21-month-old children who had a narrow facial skeleton
were more inhibited than those with broad faces (Arcus
and Kagan 1995). We interpret these results as implying
that the genes that control the growth of the maxilla, which
is a derivative of the neural crest and, therefore, ectoder-
mal in origin, are correlated with the genetic factors that
1538J. Kagan and N. Snidman
contribute to inhibited behavior. It is relevant, therefore,
that an allelomorph of the agouti gene, located on chro-
mosome 2 in mice, is linked to craniofacial anomalies
(Asher et al 1996). Both melanocytes, whose activity is
modulated by the agouti gene, and facial bone are deriv-
atives of the neural crest. Mice and rats who are homozygous
for the recessive form of the agouti gene (called non-agouti)
have all black fur and are tamer and less fearful than animals
with the agouti gene (Cottle and Price 1987; Hayssen 1997).
Other data affirm an association between high reactivity as
targets at 7.5 years. For example, an asymmetry in skin
temperature between the fingertips of the index fingers of the
left and right hand is a result of differential constriction of
arteriovenous anastomoses. The index fingers typically have
larger asymmetries than the middle or ring fingers (average
asymmetry of ?0.3°C favoring a cooler left hand). When the
distribution of asymmetries was divided into terciles, signif-
icantly more high than low reactives were either in the top or
the bottom tercile (80% vs. 58%), and girls with large
asymmetries were more likely to have anxious symptoms
than girls with smaller asymmetries.
Davidson (1992) and Fox and Davidson (1988) have
suggested that children who are avoidant of or fearful to
unfamiliar events show greater desynchronization of alpha
frequencies over the right frontal compared with the left
frontal area under resting conditions (Davidson 1995). In
addition, high reactives showed greater activation of the
right frontal area when they were 9- and 24-months-old,
whereas low reactives showed greater activation over the
left frontal area (Fox et al 1994). Because neural activity
in the amygdala is transmitted to the frontal lobes via
cholinergic fibers projecting from the basal nucleus of
Meynert, it is possible that greater desynchronization of
alpha frequencies in the right frontal area reflects greater
activity in the right amygdala (Kapp et al 1994; Lloyd and
Kling 1991). We have begun to evaluate the children from
this longitudinal sample at 10 years of age. At the time of
this writing, 28 high reactives and 24 low reactives have
been measured and significantly more high than low
reactives are showing greater EEG activation under resting
conditions over the right frontal area (30% vs. 8%)
whereas more low reactives are showing greater activation
over the left frontal area (55% vs. 25%) (chi square ? 4.0,
p ? .05). This replication of Fox and Davidson findings
lends robustness to this empirical association. In addition,
more high than low reactives had greater power in the 14-
to 30-Hz range than in the 8- to 13-Hz range at frontal sites
(eyes opened and relaxed) (chi square ? 4.1, p ? .05).
This result suggests greater cortical arousal in the children
who had been high reactive infants.
Inhibition in Children of Panic Disorder
An early study on a small sample suggested that children
with a panic parent were more likely than controls to show
inhibited behavior in a laboratory setting (Rosenbaum et al
1988). A subsequent collaboration involved an evaluation
of two larger samples of Caucasian, middle-class children
(65 4.5-year-olds and 83 6.5-year-olds), who had a parent
with panic disorder or panic combined with depression.
We compared the behavior and physiology of these
children with that of 42 4.5-year-old and 36 6.5- year-old
Caucasian children from middle-class families with no
psychiatric symptoms. Each child was administered a
battery by an examiner who was blind to the diagnostic
status of the parent.
More panic than control children were emotionally
subdued as they interacted with the unfamiliar female
examiner. Nineteen percent of children with a panic
parent, but only 5% of control children, had values in the
lowest quartile of the distributions for both spontaneous
smiles and comments while interacting with the female
examiner. These behavioral differences between the panic
and control children were clearer if the parent reported
onset of panic disorder before 21 years of age.
The autonomic variable that best differentiated the
panic from control children was a large temperature
asymmetry (?1°C) between the right and left index
fingers while the child was watching film clips. It will be
recalled that this sympathetically mediated measurement
also differentiated high from low reactives in the longitu-
Thus, about 1 in 5 middle-class, Caucasian children
living with a parent who had panic disorder was likely to
become an inhibited child, compared with a probability of
1 in 20 for matched control children. If one adds the large
temperature asymmetry as a diagnostic marker, then 1 in
10 children with a panic parent, but only 1 in 100 control
children combined a subdued style of interaction with this
sign of sympathetic lability.
A group of 10 girls living with a panic parent (15% of
all girls) were extremely subdued—they displayed fewer
than 10 spontaneous comments and 10 smiles across the
battery. These extremely subdued girls were qualitatively
different from the remaining girls living with a panic
parent who were emotionally more spontaneous with the
examiner. The 10 extremely subdued girls had higher
resting heart rates, were more likely to have a very cool
right compared with left index finger, and more likely to
have very light blue eyes. These 3 features are more
Temperament and Anxiety1539
characteristic of extremely inhibited children in the larger
longitudinal sample (Kagan 1994).
The major implication of this work is that high reactive
infants, many of whom show an inhibited profile to
unfamiliar events and situations in the second year, are at
slightly higher than normal risk for the later development
of some form of anxious symptomatology. This suggestion
is affirmed by longitudinal study of an independent sample
of 79 13-year-olds, who had been classified as inhibited or
uninhibited in the second year of life (Kagan et al 1988b).
These adolescents were interviewed by Carl Schwartz, a
child psychiatrist, who had no knowledge of their initial
temperamental classification or later laboratory behavior.
More of the adolescents who had been classified as
inhibited rather than uninhibited in the second year had
symptoms of social anxiety (61% vs. 27%) (Schwartz et
al, in press). These inhibited children were not more likely
to have developed specific target phobias or separation
anxiety, implying that inhibited children might be at special
risk for the development of social phobia during the adoles-
cent and adult years. College students who reported high
levels of social anxiety remembered being very shy when
they were young children (Mick and Telch 1998; van
Ameringen et al 1998).
It is possible that inhibited children are especially
susceptible to anxiety or PTSD after threatening events.
Only 10 children from a large group of school children
who were kidnaped and terrorized for 2 days developed
post-traumatic stress disorder (Terr 1979); these 10 might
have been temperamentally inhibited children. In 1984, a
sniper in a building across the street from a Los Angeles
elementary school fired at the children on the playground
killing 1 and injuring 13 children. One month later a group
of psychologists and psychiatrists who interviewed the
victims of this event judged 38% of the group to be
anxious. A primary quality that differentiated the anxious
from the nonanxious children was a prior avoidant per-
sonality among the former group (Pynoos et al 1987).
It is important to appreciate, however, that the majority
of inhibited children probably will not become anxiety
disorder patients in later life. Given the frequency of
anxiety disorder in the population, a variable that correctly
predicted 90% of individuals with adult anxiety disorder
and eliminated 90% of those who would not, would be
correct in only one-third of the cases.
The Notion of Constraint
It will be recalled that only 18% of the high reactive
infants were consistently inhibited at every evaluation, but
not one member of the group was consistently uninhibited
from 1–7 years of age. This fact suggests that the relation
between infant reactivity and the later development of a
consistently inhibited style is real, but modest. When the
probability of one event following another is low, say the
correlation is less than 0.4, it is likely that the relevant
antecedent event is affecting the consequent either indi-
rectly, only at extreme values, or in combination with
other factors. Under these conditions, it is more accurate to
use the verb “constrain” rather than “determine” to de-
scribe the relation between the earlier and later event. For
example, consistent nurturance of young children during
the first five years of life does not predict, with a high level
of confidence, the quality of one’s marriage, amount of
education, or degree of professional accomplishment.
Early nurturant care probably constrains, in a significant
way, the likelihood that children raised in such families
will become criminals. Because no high reactive infant
became a consistently uninhibited child, it is more accu-
rate to write that a high reactive temperament constrains
the probability of the child becoming consistently unin-
hibited, rather than to claim that high reactivity determines
the development of an inhibited or anxious profile. The
replacement of the word “determine” with “constrain” is
not idle word play for the connotations surrounding the
two words are different.
The evidence and ideas presented in this paper support the
usefulness of combining biological and behavioral evi-
dence. During the first half of this century, psychiatrists
and psychologists ignored the modest but significant
contribution of temperament to personality profiles. The
child’s social history determines the meaning he or she
will impose on an event; the combination of temperament
and history determines the ease with which the event
activates limbic structures and the subsequent emotional
response. The fact that only a small proportion of high
reactive infants became consistently fearful, anxious chil-
dren implicates an important role for the environment. We
must now search for the integrated profiles that emerge
from particular temperamental dispositions encountering
varied life histories and resist the desire to simplify the
problem by proceeding as if we could eventually have a
quantitative estimate of the separate contributions of
temperament and life events.
The preparation of this article was supported by grants from the W. T.
Grant Foundation, Foundation Bial, and NIMH Grant 47077. We thank
Mark McManis, Donna Steinberg, Jenny Mongkolcheep, Melissa Lewis,
Eric Peterson, and Vali Kahn for their contribution to this research.
1540 J. Kagan and N. Snidman
This work was presented at the scientific satellite conference, “The
Role of Biological and Psychological Factors on Early Development and
Their Impact on Adult Life,” that preceded the Anxiety Disorders
Association of America (ADAA) annual meeting, San Diego, March
1999. The conference was jointly sponsored by the ADAA and the
National Institute of Mental Health through an unrestricted educational
grant provided by Wyeth-Ayerst Laboratories.
Amaral DG, Price JL, Pitkanen A, Carmichael ST (1992):
Anatomical organization of the primate amygdaloid complex.
In: Aggleton JP, editor. The Amygdala. New York: Wiley-
Arcus DM, Kagan J (1995): Temperament and craniofacial
skeleton in childhood. Child Dev 66:1529–1540.
Asher JH, Harrison RW, Morell R, Carey MC, Friedman TB
(1996): Effects of Pax3 modifier genes on craniofacial mor-
phology, pigmentation, and viability. Genomics 34:285–291.
Blanchard DC, Blanchard RJ (1988): Etho-experimental ap-
proaches to the biology of emotion. In: Rosenzweig MR,
Porter LW, editors. Annual Review of Psychology. Palo Alto,
California: Annual Reviews, 43–68.
Cottle CA, Price EO (1987): Effects of the nonagouti pelage-
color allele on the behavior of captive wild Norway rats.
J Comp Psychol 101:390–394.
Davidson RJ (1992): Anterior cerebral asymmetry and the nature
of emotion. Brain Cognit 20:125–151.
Davidson RJ (1995): Cerebral asymmetry, emotion, and affective
style. In: Davidson RJ, Hugdahl K, editors. Brain Asymmetry.
Cambridge: MIT Press, 361–388.
Davis M, Gewirtz JC, McNish KA, Kim M (1995): The roles of
amygdala and the bed nucleus of the stria terminalis in the
acquisition of fear-potentiated startle using both explicit and
contextual cues. Presented at the 25th meeting of the Society
for Neuroscience, San Diego.
Fox NA, Davidson RJ (1988): Pattern of brain electrical activity
during facial signs of emotion in ten month old infants. Dev
Fox NA, Calkins SD, Bell MA (1994): Neural plasticity and
development in the first two years of life. Dev Psychopathol
Hayssen V (1997): Effects of the non-agouti coat-color allele on
behavior of deer mice (Peromyscus maniculatus). J Comp
Kagan J (1994): Galen’s Prophecy. New York: Basic Books.
Kagan J (1998): Biology and the child. In: Eisenberg N, editor.
Handbook of Child Psychology, Vol. 3, 5th ed. New York:
Kagan J, Snidman N, Arcus D (1998a): Childhood derivatives of
high and low reactivity in infancy. Child Dev 69:1483–1493.
Kagan J, Reznick JS, Snidman N (1988b): Biological bases of
childhood shyness. Science 240:167–171.
Kagan J, Snidman N, Zentner M, Peterson E (1999): Infant
temperament and anxious symptoms in school age children.
Dev Psychopathol 11:209–224.
Kapp BS, Supple WF, Whalen PJ (1994): Effects of electrical
stimulation of the amygdaloid central nucleus on neurocorti-
cal arousal in the rabbit. Behav Neurosci 108:81–93.
Klein RG (1991): Parent-child agreement in clinical assessment
of anxiety and other psychopathology. J Anxiety Dis 5:182–
LeDoux JE (1996): The Emotional Brain. New York: Simon &
Lloyd RL, Kling AS (1991): Delta activity from amygdala in
squirrel monkeys (Saimiri sciureus): Influence of social and
environmental contexts. Behav Neurosci 105:223–229.
Mick MA, Telch MJ (1998): Social anxiety and the history of
behavioral inhibition in young adults. J Anxiety Dis 12:1–20.
Perrin S, Last CG (1992): Do childhood anxiety measures
measure anxiety? J Abnorm Child Psychol 25:567–578.
Pynoos RS, Frederick C, Neder K, Arroyo W, Steinberg A, Eth
F, et al (1987): Life threat and post-traumatic stress disorder
in school-age children. Arch Gen Psychiatry 44:1057–1063.
Rolls ET (1992): Neurophysiology and functions of the primate
amygdala. In: Aggleton JP, editor. The Amygdala. New York:
Rosenbaum JF, Biederman J, Gersten M, Hirschfeld-Becker DR,
Meninger SR, Herman JB, et al (1988): Behavioral inhibition
in children of parents with panic disorder and agoraphobia: A
control study. Arch Gen Psychiatry 45:463–470.
Rosicky J (1993, March): The assessment of temperamental
fearfulness in infancy. Presented at the meeting of the Society
for Research in Child Development, New Orleans.
Schwartz CE, Snidman N, Kagan J (1999): Adolescent social
anxiety as an outcome of inhibited temperament in childhood.
J Am Acad Child Adolesc Psychiatry 38:1008–1015.
Seifer R, Sameroff AJ, Barrett LC, Krafchuk E (1994): Infant
temperament measured by multiple observations and mother
report. Child Dev 65:1478–1490.
Spiker D, Kraemer HC, Constantine NA, Bryant D (1992):
Reliability and validity of behavior problem checklists as
measures of stable traits in low birth weight premature
preschoolers. Child Dev 63:1481–1496.
Terr LC (1979): Children of Chowchilla. Psychoanal Study Child
Thomas A, Chess S (1977): Temperament and Development.
New York: Brunner Mazel.
Van Ameringen M, Mancini C, Oakman JM ( 1998): The
relationship of behavioral inhibition and shyness to anxiety
disorder. J Nervous Ment Dis 186:425–431.
Temperament and Anxiety1541