PUBERTY AND BEHAVIOR, EMOTIONS, AND COGNITION

Until the last two decades, studies of pubertal processes considered primarily physical
morphological characteristics or menarche in assessing the relationship between biological
changes and psychological development. These studies now are enriched by assessment
of biological substances—specifically, hormones—that are essential for pubertal
development.
Pubertal Status and Pubertal Timing
The literature linking puberty and psychological development includes assessment that
considers both pubertal status and timing of puberty. Pubertal status refers to the degree
of physical maturation on indicators that include breast, genital, and pubic hair
Puberty and Behavior, Emotions, and Cognition 23
development and hormone levels. Timing of puberty refers to pubertal status relative to
same-age peers.
Pubertal Status: Hormones
The use of hormones to assess degree of pubertal development is a relatively new undertaking.
In the following section, hormone levels are examined in relation to their
contribution to behavior, emotions and cognition.
Testosterone and Estrogen: Antisocial Behavior “A focus on either social or biological
factors can yield only part of the story of aggressive and violent behaviors: integrative
investigations are essential to complete the picture” (Cairns & Stoff, 1996, p. 338). Relative
to other domains, the integration of biological and psychological processes in
relation to antisocial behavior has been evident for some time. Increases in antisocial
behavior (physical aggression, relational aggression, conduct disorder symptoms, behavior
problems, delinquent and violent behavior, and early and risky sexual activity
and arrests) have been attributed to changes in hormones, physical maturation, altered
reactivity to life stressors, and, most recently, brain changes during puberty. An unresolved
issue is whether antisocial behavior existed prior to puberty but became transformed
to more annoying or more serious problems at puberty or whether problems
emerge de novo at puberty.
The links between the biology of puberty and problem behavior have been extensively
reviewed elsewhere (Brain & Susman, 1997; Buchanan, Eccles, & Becker, 1992).
Major advances in relating pubertal status and psychological parameters partially result
from advances in hormone assay technologies, whereby hormones can reliably and
sensitively be measured in small quantities of blood and saliva.
The steroid hormone testosterone (T) is implicated in physical aggression in animals
and antisocial behavior in humans (Brain & Susman, 1997; Mazur & Booth, 1998). The
relationship between T and problem behavior is hypothesized to derive from pre- and
early postnatal organizational effects of hormones on brain development and later activational
influences when T begins to rise at puberty. The argument is that because
males are exposed to higher concentrations of androgens than females during pre- and
postnatal development and onward, and because males tend to express more physical
aggression than females, androgens must be implicated in aggressive behavior and
dominance in males (Mazur & Booth, 1998). Because T rises at puberty and externalizing
behavior problems also rise at puberty, it follows that T is hypothesized to influence
antisocial behavior.
Evidence for the relationship between T and aggressive behavior is derived from a
few correlational and experimental studies. In boys in the later stages of pubertal development,
Olweus, Mattson, Schalling, and Low (1988) examined the causal pathway
between T and provoked and unprovoked aggression and reported that T exerted a direct
causal path on provoked aggressive behavior. T appeared to lower the boys’ frustration
tolerance. For unprovoked aggressive behavior (starting fights and verbal aggression),
the findings were somewhat different. T had no direct effects on unprovoked
aggressive behavior, but there was an indirect effect of T with low frustration tolerance
as a mediator of aggressive behavior. The authors concluded that higher levels of T
24 Puberty and Psychological Development
made the boys more impatient and irritable, in turn increasing readiness to engage in
unprovoked aggressive behavior.
Higher levels of T appear to affect observable behavior in subtle ways. Adolescent
boys’ perceptions of dominance were reflected in peers’ T concentrations (Schaal,
Tremblay, Soussignan, & Susman, 1996). T was significantly higher in peer-perceived
leaders than in nonleaders. In addition, T levels and body mass additively predicted social
dominance (Tremblay et al., 1998). It is not yet evident whether social dominance
leads to higher T levels or whether higher T and greater body mass lead to higher social
dominance.
The relationship between T and antisocial behavior is far from consistent across
studies. T was positively related to substance use (Bauman, Foshee, Koch, Haley, &
Downton, 1989; Martin et al., 2001), coitus in females (Halpern, Udry, & Suchindran,
1997), and sexual activity in males (Halpern, Udry, & Suchindran, 1998), but there was
a negative relationship between T and behavior problems in healthy young boys (Susman
et al., 1987). Similarly, there was no relationship between a diagnosis of conduct
disorder problems and T in 4- to 10-year-old children (Constantino et al., 1993). In
girls, T was not related to aggressive or dominance behaviors in three studies that included
early-puberty girls (Brooks-Gunn & Warren, 1989; Inoff-Germain et al., 1988;
Susman et al., 1987). It is noteworthy that the associations between antisocial behavior
and T are less apparent in girls and in male children and younger adolescents (Brooks-
Gunn & Warren, 1989; Constantino et al., 1993; Nottelmann et al., 1987; Susman et al.,
1987) than in older adolescents (Olweus et al., 1988) and adults (Mazur & Booth, 1998).
These developmental inconsistencies should be expected given the different constructs
assessed across studies. Questionnaires that assess the molar aspects of aggressive behavior
may not have adequate sensitivity for capturing subtle differences in the behavior
of adolescents that covary with T levels. The relatively inconsistent links between T
and antisocial behavior in adolescents compared to adults indicates that elevated T and
antisocial behavior may be a consequence of aggressive behavior (Constantino et al.,
1993). However, a causal influence for T in behavior should be considered. Boys who
consistently displayed disruptive behavior problems and were anxious across six years
were significantly lower on T than were boys who were not disruptive and anxious
(Schaal et al., 1996). These disruptive and anxious boys were also later in their pubertal
development. The effect of antisocial behavior on suppression of gonadal steroids
may be mediated by stressors and related hormones.
Experimental studies are the preferred approach for establishing the cause-effect relationship
between hormones and antisocial behavior. To examine this cause-effect relationship,
T or estrogen was administered to delayed-puberty boys and girls in a
placebo-controlled, randomized, double-blind, crossover design study. The boys and
girls were being treated with physiological doses of T (boys) or conjugated estrogens
(girls) (Finkelstein et al., 1997; Liben et al., 2002; Schwab et al., 2001; Susman et al.,
1998). Each 3-month treatment period was preceded and followed by a 3-month placebo
period. The doses of gonadal steroids were calculated to simulate concentrations in
blood in normal early (low dose), middle (middle dose), and late (high dose) pubertal
adolescents. Significant increases in aggressive impulses and physical aggression against
peers and adults were seen in boys but only at the middle dose. In contrast, significant
Puberty and Behavior, Emotions, and Cognition 25
increases in self-reported aggressive impulses and in physical aggression against both
peers and adults were seen in girls at the low and middle dose but not at the high dose of
estrogen (Finkelstein et al., 1997). In brief, experimental treatment with T and estradiol
resulted in changes in aggressive behavior, suggesting a causal role for these hormones.
The effects of estrogen are less frequently examined in relation to antisocial behavior
than are those of T. In the few studies that have examined estrogen, the relationship
between T and aggressive behavior in girls mimics the relationship between T and aggressive
behavior in boys (Inoff-Germain et al., 1988). The lack of progression of research
on estrogen and antisocial behavior reflects two issues. First, only males were included
in the majority of studies on hormones and antisocial behavior, as physically
aggressive behavior and violence occur more frequently in men and aggressive and violent
behavior in girls rarely comes to the attention of the judicial system. Second, the
logic of the arguments regarding T, aggressive behavior, and brain development does
not take into account the empirical findings on the different forms of aggressive behavior
in males and females. Females tend to show higher levels of relational aggression
than do boys. Therefore, the relationship between female phenotypic aggressive
behavior and hormones requires reconsideration. In human adolescents, dominance
may be a productive area to consider for establishing the dynamic integration between
hormones and social hierarchies as these hierarchies are established and dissolve rapidly
in adolescent peer groups.
Emotions The role of pubertal status in emotions is of sustained interest given that
depression symptoms begin to rise during the middle to late pubertal years (Lewinsohn,
Clarke, Seeley, & Rohde, 1994; Nolen-Hoeksema & Girgus, 1994). The stressful
change hypothesis suggests that girls experiencing the pubertal transition will manifest
higher levels of distress than will pre- or postpubertal girls (Ge, Conger, & Elder,
2001b). This hypothesis is based on the assumption that puberty is a stressful transition
that requires reorganization of adaptive coping strategies (Caspi & Moffitt, 1991).
Some researchers have argued that it is the novelty of a situation, rather than the magnitude
of the stressor, that leads to emotional distress in adolescents (Susman, Dorn,
Inoff-Germain, Nottelmann, & Chrousos, 1997). This perspective leads to the prediction
that given the novel nature of the hormonal and physical growth changes, puberty
constitutes a stressful adolescent transition.
In those studies that examine hormones and emotion, there are group differences for
depression in girls at different stages of puberty but a larger effect for T than for stage
of pubertal development (Angold, Costello, Erkanli, & Worthman, 1999). When Tanner
stage and hormones were entered simultaneously into a statistical model, the effect
for Tanner stage became nonsignificant, but the effect for T and estradiol remained unchanged.
In a parallel study, girls were grouped by pubertal breast stages and four stages
of estradiol secretion (Warren & Brooks-Gunn, 1989). The hormonal stages revealed a
significant curvilinear trend for depressive affect (increase, then decrease), impulse control
(decrease, then increase), and psychopathology (increase, then decrease), indicating
significant differences in these indexes during times of rapid increases in hormone
levels. In the clinical trial study just described, adolescents with delayed puberty treated
with physiological doses of T or estrogen showed few changes in emotions as a result
of treatment, with the exception of increased withdrawn behavior in girls (Susman et al.,
26 Puberty and Psychological Development
1998). Collectively, the studies with adolescents who are progressing normally through
puberty demonstrate an association between pubertal status and depression.
Adrenal Androgens and Cognition The role of hormones in cognition has not received
much attention in the last decade. In a recent study, maturational status as indexed by
T was related to better spatial abilities (Davison & Susman, 2001). As T increased, there
was a corresponding increase in spatial ability in boys, and to a lesser extent in girls. In
an experimental sex hormone treatment study, spatial performance showed traditional
sex differences but did not vary with levels of actively circulating sex steroids (Liben et
al., 2002). Further longitudinal research is required to resolve the question of how pubertal
hormones and cognition change in an integrated fashion.
Adrenal Androgens
Adrenal androgens represent a class of hormones that traditionally received little attention
in relation to behavior. The adrenal androgens DHEA, DHEAS, and Δ4-A are
secreted by the adrenal glands and begin to rise during adrenarche.
Adrenal Androgens and Problem Behavior In the last two decades reports began to appear
showing relationships between adrenal androgens and antisocial behavior in adolescents.
In healthy pubertal-age girls, DHEAS correlated negatively with aggressive
affect (Brooks-Gunn & Warren, 1989), and the interaction between negative life events
and DHEAS and aggressive affect also was significant. Girls with lower concentrations
of DHEAS who experienced negative life events had more aggressive affect than did
girls with fewer negative life events. In a study that included healthy 9- to 14-year-old
boys and girls, there was a relatively consistent pattern of higher DHEA and Δ4-A and
lower DHEAS and problem behaviors (Nottelmann et al., 1987; Nottelmann, Inoff-
Germain, Susman, & Chrousos, 1990; Susman, Dorn, & Chrousos, 1991; Susman et al.,
1987). Adrenal androgens also were correlated with dominance in girls while interacting
with their parents (Inoff-Germain et al., 1988). The links between adrenal androgens
and problem behavior extended to sexual behavior as well. Higher levels of adrenal
androgens were related to sexual behavior and activities during adolescence (Udry
et al., 1985; Udry & Talbert, 1988). In addition, youth with conduct disorder (CD) had
significantly higher levels of DHEA and DHEAS than did non-CD youth and reported
higher intensity of aggression and delinquency (van Goozen, Matthys, Cohen-Kettenis,
Thijssen, & van Engeland, 1998). Support for the links between adrenal androgens and
behavior problems was reported in girls with early adrenarche. These girls had higher
levels of adrenal androgens and more behavior problems than did the on-time girls
(Dorn, Susman, Nottelmann, Inoff-Germain, & Chrousos, 1999). Higher levels of adrenal
androgens may be an etiological process in the higher incidence of behavior problems
in premature-adrenarche girls.
Adrenal Androgens and Emotions There is a relatively consistent pattern of high adrenal
androgens associated with negative affect (Nottelmann et al., 1990; Nottelmann et
al., 1987; Susman et al., 1991; Susman et al., 1987). DHEAS levels interact with timing
of puberty and depression in girls. Girls with high levels of DHEAS and early maturation
had the highest emotional arousal and depressive affect scores (Graber, Brooks-
Puberty and Behavior, Emotions, and Cognition 27
Gunn, & Warren, in press). Girls with premature adrenarche, who also have higher
levels of adrenal androgens, were more anxious than were their peers with on-time
adrenarche (Dorn et al., 1999). In addition, higher levels of DHEA predicted the onset
of the first episode of major depression during adolescence (Goodyer, Herbert, Tamplin,
& Altham, 2000). As is the case for sex steroids, the described associations do not
imply causality. Variations in levels of hormones may arise from more distal genetic or
experiential origins than recent life events and behavior.
Pubertal Status: Physical Maturation
The relationship between pubertal status (i.e., degree of physical development) and
psychological development is hypothesized to result from changing social roles, including
increasing family and peer pressures, and cultural expectations for mature behavior.
The assumption is that problem behavior and negative emotions will increase
with advancing physical development independent of age.
Antisocial Behavior Indeed, more advanced pubertal status contributed significantly to
predicting female delinquency (Flannery, Rowe, & Gulley, 1993), symptoms of attentiondeficit/
hyperactive disorder (ADHD), oppositional disorder (ODD), and conduct disorder
(CD; Ge, Brody, Conger, Simons, & Murry, 2002). More advanced pubertal development
provides adolescents with opportunities to become involved with older and
deviant peers.
Emotions Tanner stage had a larger effect on depression than did age, suggesting that
biological change rather than a specific age is related to depression in both boys and
girls (Angold, Costello, & Worthman, 1998; Ge et al., 2001b; Ge et al., 2002). Additionally,
the association between pubertal status and internalizing distress is stable over
time. Pubertal status in 7th grade was significantly related to both internalized distress
and hostility assessed at Grades 8 and 10, respectively, and to externalized hostility at
Grade 9 (Ge, Conger, & Elder, 2001a). Overall, more advanced stages of physical development
favored more negative emotions.
Family The classic study by Steinberg and Hill (1978) showed that family-adolescent
interactions are moderated by pubertal development. Conflict between mothers and
sons was highest during midpuberty. The same findings emerged in African American
adolescents whose parents reported having more verbal aggression with sons during
midpuberty than during early or late puberty (Sagrestano, McCormick, Paikoff, &
Holmbeck, 1999). The sons reported more “hot” issues and having hot discussions
when they were more physically developed. However, the importance of midpubertal
status in family conflict is not consistent across studies. Later as opposed to midpubertal
status also was associated with more conflict and diminished closeness with parents
(Crockett & Petersen, 1987; Savin-Williams & Small, 1986). The lack of consensus regarding
pubertal status, conflict, and family interactions is not surprising given the
variation in methods for assessing family interaction.
Cognition Cognitive maturation normatively advances with chronological age rather
than pubertal maturation. For instance, spatial ability was associated with chronolog-
28 Puberty and Psychological Development
ical age but not with pubertal stage in a sample of healthy young adolescents (Davison
& Susman, 2001). As discussed earlier, pubertal status in delayed-puberty adolescents
was not associated with spatial ability (Liben et al., 2002) and thus denotes consistency
across studies.
Psychopathology Pubertal status was associated with a sharp increase in rates of
unipolar depression (Angold & Worthman, 1993). Girls begin to have higher rates of
both diagnosed depressive disorders and subclinical levels of depressive symptoms
than do boys in early to middle adolescence (Angold et al., 1998; Ge, Conger, Lorenz,
& Simons, 1994; Laitinen-Krispijn, van der Ende, & Verhulst, 1999). In prepubertal
children (Tanner stage 1) disorders with hypomanic-manic symptomatology (Schraufnagel,
Brumback, Harper, & Weinberg, 2001) and seasonal affective disorders (Swedo
et al., 1995) were more common, whereas in pubertal-age adolescents (Tanner stages
3-5) there was a predominance of depressive symptomatology. Other studies found no
association between prepubertal and early-adolescent bipolar disorder (Geller et al.,
2000). Overall, results are inconclusive regarding the connections between puberty and
psychopathology. A major deficit in the literature is that few longitudinal studies of
adolescent psychopathology employ state-of-the-art indexes of pubertal development,
such as Tanner stage and hormone concentrations.