Internalizing problems of childhood and adolescence: Prospects, pitfalls, and progress in understanding the development of anxiety and depression

The focus of this article is on internalizing problems that are experienced by children and adolescents. We provide an historical perspective, selectively examine the current state of knowledge, consider advances and gaps in what is known, and identify new research directions. Diagnosis, epidemiology, theory, and research first are considered separately for anxiety and depressive disorders. These internalizing problems, however, whether clinical or subclinical, share many common features and show high comorbidity rates. We emphasize the importance of systematic analysis of comorbid anxiety and depression, including their comorbidity with externalizing problems. This could lead to more valid classification of subtypes of internalizing problems and further an understanding of the diverse conditions that constitute internalized distress. We highlight the need to study anxiety and depression within a developmental psychopathology framework, as well as to include both categorical and dimensional assessments of these problems in the same research designs. This will be essential for understanding the complex interplay of biological and environmental processes that contribute to the emergence, progression, and amelioration of internalizing problems over time.

We highlight the need to consider anxiety seen in normal and subclinical problems? Or instead, do different etiologic factors operate and depressive disorders within a developmental psychopathology framework, in order to produce different levels of severity of problems? In what ways might early subsyn-to address a number of core questions. How and why do these serious emotional distur-dromal forms of internalized distress increase the later likelihood of diagnosed anxiety or bances emerge over the course of childhood and adolescence? Are there developmental depressive disorders? Different definitions of problems, heterogeneity of measures, varia-precursors of these disorders in adolescents and adults? How are anxiety and depression tions in samples, and severity of problems studied across disciplines have made general-conceptualized during the first 2 decades of life? What are some of the major biological izations difficult. The current article reflects both the tensions and opportunities inherent in and environmental factors that contribute to their emergence? To their co-occurrence? To attempts to integrate psychiatric and psychological approaches. This integration is one of continuity and change over time? How can the study of normative biological, develop-the major tasks in the new century.
The study of internalizing disorders in chil-mental, and socialization processes advance understanding of the etiology and develop-dren and adolescents originally was based on application of adult models to earlier periods ment of internalizing problems?
Psychiatric diagnoses of anxiety and mood of development. This did not encourage, and sometimes impeded, an understanding of the disorders are based on discrete classification systems. A clinical disorder is a distinct en-origins and development of these disorders.
However, with the advent of a developmental tity, construed as a disease; it is either present or absent, depending upon whether specific psychopathology perspective, there has been an increase in the use of longitudinal and age diagnostic criteria are met. Disorders have traditionally been viewed as qualitatively differ-cross-sectional research designs to provide information about adaptive and maladaptive ent from the lesser problems of the "walking wounded" or "worried well." In psychology, functioning in childhood and adolescence.
This makes it increasingly possible to develop dimensional classification systems typically are used to characterize internalizing prob-explanatory models and treatments of internalizing disorders that are grounded in a firm lems (i.e., there is a gradation or continuum of severity of disturbance). This approach understanding of the phenomenology of these problems in childhood and adolescence. does not yield psychiatric diagnoses, though scores typically are normed, in an effort to distinguish normal, subclinical, and clinical Anxiety Disorders problems. There also tends to be less differentiation among subtypes of internalizing prob-Diagnosis and classification lems in dimensional than categorical systems. Factor analytic work has provided a rationale The diagnosis and classification of anxiety disorders in children and adolescents is rela-for combining symptoms of anxiety and depressed mood into one scale (Achenbach, tively new. The first Diagnostic and Statistical Manual of Mental Disorders used to clas-1991). At the same time, this precludes comparisons of the two most prominent types of sify these problems (DSM-II; American Psychological Association [APA], 1968) rec-internalizing problems.
We consider both categorical and dimen-ognized the existence of two disorders: withdrawing reaction and overanxious reaction. sional perspectives in this article. The different approaches have led to a series of ques-The DSM-III (APA, 1980) marked a historic shift to empirically based, behaviorally de-tions that have just begun to surface. Do risk and protective factors known to influence the scriptive diagnostic criteria. A separate diagnostic section on "anxiety disorders of child-development of internalizing disorders also function on a continuum? In other words, do hood and adolescence" included separation anxiety disorder, overanxious disorder, and lower levels lead to less severe disturbance Internalizing problems of childhood and adolescence 445 avoidant disorder of childhood or adoles-terized by pervasively excessive worry and is seen most often in older children and adoles-cence. The DSM-IV (APA, 1994) identifies separation anxiety disorder as the only anxiety cents. Social phobia (i.e., fear and avoidance of social and performance situations) becomes disorder unique to childhood. Overanxious disorder and avoidant disorder now are sub-more common in middle childhood and adolescence. Panic disorder is rare in childhood sumed under generalized anxiety disorder and social phobia, respectively. There are several but more frequent in adolescence. The emergence of panic disorder may be associated other anxiety disorders in DSM-IV that apply to both adults and children (e.g., panic disor-with puberty (Hayward, Killen, Hammer, Litt, & Wilson, 1992). Obsessive-compulsive der, agoraphobia, specific phobia, social phobia, obsessive-compulsive disorder, general-disorder consists of repetitive intrusive thoughts (obsessions) or behaviors (compul-ized anxiety disorder, and posttraumatic stress disorder). sions). Onset occurs during early and middle childhood (e.g., Swedo, Rapoport, Leonard, Lenane, & Cheslow, 1989;Riddle, 1998). Epidemiology Advancements in identification and diagnosis Theory and research of anxiety disorders in children have contributed to improved epidemiological studies (and Our contemporary understanding of biological, environmental, and developmental factors vice versa). As recently as the mid-1980s, in a review of the extant research (Orvaschel & in the etiology of anxiety disorders reflects a preceding era characterized by periods of crit-Weissman, 1986) no data were available regarding anxiety disorders. A decade later in ical reappraisal and integration. Shifts from general to specific, singular to integrative, and a comprehensive review (Costello & Angold, 1995), prevalence rates of 5.7-17.7% for any conceptual to quantitative can be seen within and between major orientations of study. A anxiety disorder in children were reported. Other studies have confirmed similar rates. historical review of representative theories and research exemplifies some of these transi-Differences in prevalence rates may reflect several factors. Differentiation of anxiety tions. symptoms from anxiety disorders has largely depended upon the presence or absence of Psychoanalytic, psychodynamic, and relational theories. Freud (1936) attributed anxi-functional impairment and subjective distress criteria. The inclusion of functional impair-ety to the presence of unconscious infantile libidinal or aggressive wishes towards paren-ment criteria has been shown to decrease rates of anxiety disorders. The informant may also tal figures, a conception that lost favor over time. However, the focus on childhood expe-influence rates of reported cases (e.g., children often report more anxiety symptoms than riences and the quality of the parent-child relationship in shaping children's sense of secu-are seen in their parents' reports).
Age-related appearance of specific anxiety rity as well as their anxieties and fears has remained central to several of the psychody-disorders in children has been well documented (e.g., Costello & Angold, 1995). Sep-namic and relational theories that followed.
The collaboration of a psychoanalyst, John aration anxiety disorder, the fear of separation from primary caretakers, presents most fre-Bowlby, and a developmental psychologist, Mary Ainsworth, led to theory-driven re-quently during early and middle childhood. Specific phobias (i.e., fear and avoidance of search on the early formation of anxiety problems that develop in the context of mother-identifiable objects or situations) have been described in children of all ages. Elevated infant interactions and depend on the nature of the attachment relationship. Caregiver sen-rates of animal phobias may appear in early childhood in comparison to social-related sitivity and consistency contribute to secure attachment, presumably by heightening the in-phobias in adolescence (Marks & Gelder, 1966). Generalized anxiety disorder is charac-fant's comfort and sense of safety (Ainsworth, 1978;Bowlby, 1973Bowlby, , 1988. Attachment theo-be drawn from early experimental studies, generalizations from naturalistic research are rists proposed a relationship between insecure attachment (anxious-resistant) and the devel-less straightforward. Parents who "model" anxiety and their children who "reproduce" it opment of anxiety in children. Anxious-resistant attachment is, in fact, associated with in-may also indicate shared genetic susceptibility either to anxiety problems per se or to biolog-creased fearfulness and inhibited behaviors in children (e.g., Cassidy & Berlin, 1994). It also ically based, predisposing characteristics (e.g., low thresholds for physiological arousal). predicts later anxiety disorders in children and adolescents (Warren, Huston, Egeland, & The role of cognition in theories about anxiety has been emphasized in recent dec- Sroufe, 1997).
The theoretical models and empirical re-ades. Research on anxiety disorders in adults describes a bias toward increased attention to search generated by Bowlby and Ainsworth reflect one of the first efforts to investigate threatening or dangerous situations (e.g., Beck & Clark, 1997). Research on anxiety in the etiology of (internalizing) disorders within a developmental psychopathology framework. children suggests similar cognitive processes.
Anxious children have been found to interpret Their pioneering work led to research not only on the beneficial effects of sensitive ambiguous information as threatening more often than nonanxious children (e.g., Hadwin, caregiving but also on the adverse consequences of negative parenting for children 's Frost, French, & Richards, 1997). Biased attention to signals of threat or danger are anxiety. Examples of the latter include parental restriction and negative feedback (e.g., thought to create cognitive distortions through a process of overactivation. Biased attention Krohne & Hock, 1991), parental high control leading to perception of low personal control could occur for a number of reasons. In addition to socialization experiences that sensitize (e.g., Chorpita & Barlow, 1998), and maternal intrusiveness and overprotectiveness (e.g., children to dangerous, anxiety-provoking conditions, temperament may render some chil- Bowen, Vitalo, Kerr, & Pelletier, 1995).
dren particularly vigilant and vulnerable to overactivation. Behavioral and cognitive learning theories. Pavlov's classical conditioning model of anxiety, first used to study the learning of fear in Biological models Genetic influence. Twin and adoption stud-animals, soon was extended to research with humans. Its early application to children was ies support the theory that heritable factors play some role in the expression of anxiety seen in the case of "Little Albert," a young child who was conditioned to develop a spe-problems and disorders (Kendler, Neale, Kessler, Heath, & Eaves, 1992). There is a famil-cific phobia (Watson & Rayner, 1920). Several theorists (e.g., Rachman, 1977;Seligman, ial pattern seen in the pathogenesis of childhood anxiety disorders. For example, there is 1971) challenged the overencompassing characteristics of the original behavioral model, an increased prevalence in first-degree relatives of children with anxiety disorder com-noting that some stimuli are more biologically prone to become feared objects. Also, several pared with relatives of children with attentiondeficit hyperactivity disorder (ADHD) or additional learning processes have been identified over time. Anxiety could be learned never psychiatrically ill children (e.g., Last, Hersen, Kazdin, Orvaschel, & Perrin, 1991). from caregivers or others in the child's environment (e.g., through modeling and conta-While twin and family studies provide indirect evidence of heritability, specific genetic gion of anxious, fearful behavior; explanatory styles that emphasize frightening and danger-mechanisms at the molecular level have not been identified and environmental influences ous aspects of life; or specific child-rearing and discipline practices likely to induce fear are also likely to be operative. Temperament. Particular temperament in children through threats or actual infliction of harm). While causal inferences about the traits may be early indicators of fearfulness or anxiety. Behaviorally withdrawn and inhib-environment in the learning of anxiety could ited behavior when encountering unfamiliar have lower baseline cortisol than control groups (see review by Stansbury & Gunnar, persons and situations is one such trait (Kagan, Reznick, Snidman, Gibbons & Johnson, 1994). Youths with clinical levels of anxiety have shown high cortisol reactivity relative to 1988). It is postulated to reflect a lowered threshold of activity in limbic and hypothala-controls, in response to a social challenge (Granger, Weisz, & Kauneckis, 1994). mic structures and has been associated with increased sympathetic activity (Kagan, Rez-Neurobiological processes and brain circuitry. Advances in science and technology nick, & Snidman, 1987). Family and twin studies indicate a relationship between behav-over the past few decades have contributed greatly to our understanding of the biological ioral inhibition and increased risk of anxiety disorders (see review by Turner, Beidel, & substrates of adult anxiety disorders. Studies indicate the likely involvement of multiple Wolf, 1996). Generalized social anxiety at adolescence (but not specific fears, separation neurotransmitters and brain structures. Abnormalities in the serotonin, noradrenergic, and anxiety, or performance anxiety) can be predicted from behavioral inhibition in the 2nd GABAergic systems have been implicated (Johnson & Lydiard, 1995;Longo, 1998). year of life, though only for girls, suggesting the moderating effects of gender (Schwartz, There has been very little research with children. Snidman, & Kagan, 1999).
Physiological regulatory processes. Re-Gray's neurophysiological model of anxiety focuses on approach and withdrawal pat-search on the psychophysiology of internalizing (and externalizing) problems has pro-terns (Gray, 1982(Gray, , 1991. They are part of the fight-flight system that responds to protect duced models wherein both classes of psychopathology have been associated with the organism. Research on neural regulation suggests differentiating roles of left and right distinct physiological profiles. The focus has been mainly on the autonomic nervous system anterior frontal lobe brain regions in controlling approach and withdrawal behaviors (Da-(ANS) and the hypothalamic-pituitary-adrenal (HPA) axis systems, both of which are vidson & Fox, 1982;Fox, 1991). The left frontal brain region is implicated in approach involved in the body's stress response. Anxiety symptoms and disorders have been associ-behaviors and positive emotions, while the right frontal region is associated with with-ated with high heart rates (e.g., Gerardi, Keane, Calhoon, & Klauminzer, 1994) and drawal behavior and negative emotions. Infants observed to be high on motor activity delayed electrodermal habituation (Birket-Smith, Hasle, & Jensen, 1993). Children hos-and negative affect have shown greater right frontal activation at 9 months of age and more pitalized with separation anxiety disorder were found to have higher heart rates than inhibited behavior at age 14 months . In another study other patients (Rogeness, Cepeda, Macedo, & Fischer, 1990). (Fox, Rubin, Calkins, Marshall, Coplan, Porges, Long, & Stewart, 1996), increased The HPA axis system is important in the regulation of arousal (Gold, Goodwin, & right frontal EEG activation was observed in socially inhibited 4-year-old children. In con-Chrousos, 1988; Stansbury & Gunner, 1994). Healthy adaptation (indeed survival) relies on trast, left frontal activation was correlated with outgoing behavior. These studies provide the ability to produce increased levels of cortisol under stress and to reduce production additional evidence for biological manifestations of inhibited behavior thought to play a once stress has abated. However, chronic exposure to stress may predispose individuals to role in the development of anxiety disorders.
Functional neuroimaging research also anxiety and affective disorders, by causing long-term overproduction of hormones such provides insights into the neurobiology of anxiety disorders (Rauch, Savage, Alpert, as cortisol in the HPA system. Children and adolescents with internalizing problems tend Fischman, & Jenike, 1997). Anatomical brain circuits connecting specific brain regions may to have higher baseline cortisol, whereas those with externalizing problems tend to be involved, and disruption of components at varying points in the circuit may produce sim-where parental influences on children's emotional states have been found (Ge, Best, Con-ilar symptoms. Here too, research has been based primarily on adult and animal models. ger, & Simons, 1996;Ge, Conger, Lorenz, Shanahan, & Elder, 1995). This may be an-Pine and Grun (1999) review neuroanatomical models of childhood anxiety correlates of other "critical period" in which to study the interaction of nature and nurture. adult phobias, generalized anxiety, and panic disorder. These disorders are postulated to reflect abnormalities in limbic-based amygdala, Depressive Disorders septohippocampal, and brain stem-hypothalamic circuits. A potentiated startle reflex via Diagnosis and classification enhanced amygdala activation may identify children at risk for phobic disorders (Davis, Our focus is on the two most common mood disorders, major depressive disorder (MDD) 1997; Grillon, Dierker, & Merikangas, 1997). Septohippocampal dysfunction is believed to and dysthymic disorder (DD; APA, 1994).
Manic-depressive disorder is rare in child-contribute to generalized anxiety disorder in adults and children (Daleiden & Vasey, hood and adolescence and is not considered here. There is relatively little systematic data 1997). Neurocircuits connecting brain stem periaquaductal gray and the medial hypothala-available on its phenomenology and rates of occurrence. This is changing, however, and mus may play a role in the etiology of separation anxiety and panic (Panksepp, 1998; there is now heightened interest in the study of bipolar disorder within a developmental Shekhar & Keim, 1997). Caution is warranted, as little is yet known of the biological framework (Akiskal, 1998;Carlson, Bromet, & Sievers, 2000). By the 1970s, similari-development of specific brain circuits and how this relates to development of anxiety.
ties were noted in childhood and adult unipolar depression, and they came to be Bridging research efforts are needed to further our understanding of the complex inter-considered isomorphic disorders (e.g., Malmquist, 1971). Depression in children and ado-play of biological and environmental factors that lead children toward or away from having lescents began to be diagnosed using adult criteria. Major depression in childhood is anxiety problems. Davidson and Rickman (in press) report associations between right fron-characterized by a period of disturbance in mood that may include depressed affect, an-tal brain activation and behavioral inhibition, in a sample of children studied at both 3 and hedonia, or irritability. Cognitive or vegetative symptoms must also be present. Dysthy-9-10 years of age. Inhibition was stable over time, only for a small subgroup that also had mic disorder is a persistent, milder, but more chronically depressed mood (or irritability for shown right frontal asymmetry at both time points. In keeping with a diathesis-stress children). Cognitive or vegetative symptoms also may be present. With few exceptions model, the authors speculate that behavioral inhibition may not remain stable for most (e.g., irritability manifest instead of depressed mood, duration criteria for DD), DSM-IV cri-children with a biological predisposition, due to the impact of early training and experience teria for child and adult mood disorders are nearly identical. on neural plasticity. Many individuals are saturated with particular emotional climates from early in life. These different emotional Epidemiology environments could contribute either to stability or change in inhibition over time, through Overall the prevalence rate of MDD children ages 4-18 years is between 2 and 8% (Poz-differential, experiential shaping of the emotional circuitry of the brain. While Davidson nanski & Mokros, 1994). Rates are low in childhood. Less than 1% of preschoolers and and Rickman emphasize the importance of early experience, plasticity (and hence shap-about 2% of school-aged children meet criteria for MDD (Kashani & Orvaschel, 1988), ing) is possible at other points in development. Adolescence is a period of rapid change, but by adolescence the rates are between 2 and 8%. There is an increase in prevalence ternalized processes. It was assumed that children did not have the ego-and cognitive-between ages 15 and 18 years. Adolescents report lifetime prevalence rates of MDD as developmental capabilities to experience sustained guilt, misery, and despair (i.e., those high as 15-20%, comparable to adult rates (e.g., Lewinsohn, Clarke, Seeley, & Rohde, symptoms of internalized distress that are part of the depressive experience). Later it was 1994). The prevalence of DD is less than 2% in childhood and up to 8% in adolescence proposed that depression in children might be "masked" or manifest as "depressive equiva- (Lewinsohn, Hops, Roberts, Seeley, & Andrews, 1993). Children whose first affective lents," seen in somatic complaints or behavioral disturbances (e.g., Cytryn & McKnew, episode is DD are about 2 years younger at time of onset than children whose first epi-1972; Sperling, 1959). It is now clear that young children are capable of the experiences sode is MDD (Kovacs, Akiskal, Gatsonis, & Parrone, 1994).
implicated in the phenomenology of depression, such as excessive guilt (see review by There are also age changes in the phenomenology of depression. In a comparison of de-Zahn-Waxler & Kochanska, 1990) and other intropunitive emotions. Spitz and Wolf's pressive symptoms in children and adolescents with MDD (Ryan, Puig-Antich, (1946) seminal work on anaclitic depression in infants in orphanages indicated the pres-Ambrosini, Robinovitch, Robinson, Nelson, Iyengar, & Tworney, 1987), many similarities ence of depressive syndromes very early in life. Acknowledgment that young children are were found (e.g., severity, depressed mood, insomnia, irritability, suicidal ideation). How-capable of grief reactions (Bowlby, 1960) also helped pave the way for theory and research ever, prepubertal children had more somatic complaints and psychomotor agitation (as on the development of depression in childhood. well as separation anxiety and phobias), while adolescents had more anhedonia, hopeless-Psychodynamic views of family relations influenced later research by developmental ness or helplessness, and hypersomnia. Relative to adults, adolescents are more often psychologists on parenting practices and family processes associated with child and adoles-changed by an episode of major depression, in ways suggesting that early-onset depression cent depression. Based on an attachment perspective, healthy development was thought to is a more pernicious, severe form of the disorder .
be promoted by a family environment characterized both by close relationships and the fostering of autonomy. Low levels of parental Theory and research warmth and support, as well as parental rejection, hostility, and family conflict, are known Psychoanalytic, psychodynamic, and relational theories. Beginning with Freud, psy-to be associated with depression in children and adolescents (McCauley, Pavidis, & Ken-choanalytic theories have emphasized the importance of unresolved childhood experiences dall, in press). Longitudinal research designs demonstrate the predictive power of observed that result in adult depression, in particular the loss of a real or imagined love object. One parental characteristics such as hostility and low warmth on adolescents' depressive symp-common theme is that repeated disappointments and parental failures to meet the child's toms over time, controlling for initial symptoms levels (Ge et al., 1996). Other observa-psychological needs may give rise to depression. Unconscious reactions to early child-tional research also has identified adverse conditions (i.e., less supportive, more conflic-hood circumstances, which also included sexual and aggressive impulses toward parents, tual family environments associated with greater depressive symptomatology in adoles-were thought to result over time in harsh superego development, as extreme guilt, high cents, both concurrently and prospectively; Sheeber, Hops, Alpert, Davis, & Andrews, standards, and self-blame became internalized. Adult depression, but not childhood de-1997). Longitudinal research has tended to focus on samples with low to moderate levels pression, was believed to result from these in-of depressive symptoms. The impact of the Abramson, 1984). Greater integration of this literature with developmental research on environment on depressive disorders is unknown and remains an important topic for in-self-concept and emotion regulation is warranted, given the low self-regard and dysregu-vestigation. An increasingly prominent view attributes youth depression to a combination lated emotions that characterize depression. of psychological stress and biological vulnerability that impact the individual (i.e., a diath-Biological models. Genetic influence. Twin and adoption stud-esis-stress model; Post, Weiss, Leverich, George, Frye, & Ketter, 1996).
ies in adult populations suggest that genetic factors account for at least 50% of the variance in the transmission of mood disorders. Behavioral and cognitive learning theories. Behavioral theories of depression emphasize Although genetic factors may increase risk for depression, environmental factors, particu-the importance of learning and environmental contingencies. The initial demonstration of larly nonshared experiences within and outside the family, including differential treat-"experimental neurosis" in Pavlov's laboratory led to a host of studies using similar para-ment of children by parents (Plomin, 1994) also have an impact. The interaction of ge-digms to create symptoms of anxiety and depression (reviewed by Mineka & Kihlstrom, netic and environmental factors has been hypothesized in the development of more severe 1978). Overmier and Seligman's (1967) study was the first, however, to have specific impli-forms of depression (Rende, Plomin, Reiss, & Hetherington, 1993). Increased prevalence of cations for depressive symptoms. Dogs initially exposed to uncontrollable shock became depressive disorder has been observed in relatives of depressed youth. Children of a de-impaired in subsequent efforts to escape from or control shock. The model has been applied pressed parent are at substantial risk for depression, and risk increases further if both to other species (Maier & Seligman, 1976;Mineka & Henderson, 1985), culminating in parents are affected (Birmaher, Ryan, Williamson, Brent, Kaufman, Dahl, Perel, & Nel-the "learned helplessness" model of depression where the organism learns to give up in son, 1996;McCauley et al., in press).
The concept of depression stemming from contrast to research on anxiety, there has been little emphasis on the role of temperament in learned helplessness was expanded to include cognitive distortions characterized by certain depression. Physiological processes, however, have been studied and many factors associ-attributional styles in adults (e.g., failure attributed to stable, global, internal factors; Se-ated with anxiety are also part of a depressive profile in adults. High resting heart rates and ligman, Abramson, Semmel, & von Baeyer, 1979). A crosscutting theme of several cog-low electrodermal activity and reactivity typically have been found for adult depressed pa-nitive-behavioral theories is that repeated, uncontrollable aversive environmental experi-tients (Lahmeyer & Bellur, 1987;Thorell, Kjellman, & D'Elia, 1987), a pattern reported ences create distortions (e.g., negative schemas, biased attributions, negative self-con-earlier for anxiety. Little research has been done with depressed children and adolescents. cept) or behavior deficits (e.g., deficient social skills). Distortions thus can create risk Disturbances in HPA axis function also are associated with depression. Many adult pa-for or become part of the depressive experience. While most cognitive-behavioral theo-tients tend to have elevated basal cortisol levels, show a flattening of the circadian rhythm, ries originally focused on adult depression, there have been downward age extensions to and fail to suppress cortisol production in response to the DST test (Akil, Haskett, Young, childhood and adolescence. Several studies have shown associations between depression Grunhaus, Kotun, Weinberg, Greden, & Watson, 1993). This line of research has been ex-and cognitive style in children that are consistent with adult models (e.g., Seligman, tended to the study of depressed children and adolescents. Evidence of HPA axis dysregula-Peterson, Kaslow, Tanenbaum, Alloy, & tion has not been observed in children and ad-mothers show reduced left frontal EEG activity (e.g., see Dawson, Frey, Self, Panagio-olescents with the frequency and consistency seen in studies of depressed adults. Several tides, Hessl, Yamada, & Rinaldi, 1999). This pattern has been replicated many times, and studies of baseline plasma cortisol secretion comparing depressed child and adolescent recent work suggests the role of environmental processes in patterns of brain asymmetry outpatients with controls revealed no significant differences. Some other studies have related to emotionality. Reduced left frontal EEG activity can be predicted from observed found results similar to those with depressed adults who fail to show cortisol suppression maternal behaviors toward their infants during play even after accounting for maternal de-following the Dexamethasone Suppression Test (DST; see review by Birmaher et al., pression (Dawson, Frey, Panagiotides, Self, Hessl, & Yamada, 1997). These mothers were 1996).
Neurobiological processes and brain cir-less affectionate, less engaged during play, and more intrusive. cuitry. Consistent with research on anxiety disorders in adults, several neurotransmitters Post et al. (1996) have presented a model that places affective illness in the context of have been identified in the pathophysiology of mood disorders, including abnormalities in an evolving developmental neurobiological framework and a series of molecular neurobi-the serotonergic, noradrenergic, and GABAergic systems. Information about brain re-ological adaptations. Environmental experiences are postulated to mediate effects on gions and circuitry associated with depression in adults has been derived from accumulated gene expression, including psychosocial stressors and the neurobiology of episode recur-research based on brain lesions, neuroimaging, and electroencephalographic (EEG) ac-rence. In this view, social support may be capable of generating an inhibitory effect on tivity. As previously discussed, the left frontal brain region is associated with approach and illness progression by decreasing the perception and neurobiological impact of stressors positive emotions; the right frontal lobe is involved in internalizing patterns of withdrawal (even at the level of gene expression). and negative emotions that include sadness and depressed mood. Studies that investigate Anxiety and Depression: A Developmental mood have implicated both frontal and pari-Psychopathology Perspective etal regions of both left and right hemispheres (Davidson, 1994;Heller, 1990). Individuals Developmental psychopathology has been defined as "the study of the origins and course with depressed moods show frontal lobe asymmetry, reflected mainly in low activation of individual patterns of behavioral maladaptation" (Sroufe & Rutter, 1984, p. 18), with of the left hemisphere relative to controls. Depression also has been associated with re-an emphasis as well on factors that contribute to resilience and adaptive functioning (Cic-duced activity in the right parietal region. Relations between frontal and parietal lobe chetti, Ganiban, & Barnett, 1991;Cicchetti & Schneider-Rosen, 1984; Cicchetti & Toth, activity remain to be studied, as do interconnections of cortical and limbic systems impli-1995, 1998). Because many psychological problems generally do not arise de novo, but cated in the regulation of moods, including depression. Neurochemical transmission be-emerge over time from prodromal or subclinical symptoms, a developmental perspective tween limbic and cortical structures that affect arousal and activation of various brain struc-permits study of the evolution of a disorder.
It is assumed that the organism and the envi-tures may be a crucial link in the corticalsubcortical interaction (Heller, 1990). ronment are mutually and interactively involved in determining developmental course. While research with children and adolescents is quite limited, some studies indicate Even with an established history of problems, the particular form of psychopathology may patterns of brain activity in children at risk that parallel findings with depressed adults. still be in formative stages (Newman, Moffitt, Caspi, Magdol, Silva, & Stanton, 1996). What Infants and young children of depressed factors contribute to stability or change in in-and curious. Within a functionalist framework, emotions are viewed as organizing, ternalizing problems over time? To improvement or deterioration in functioning? To shifts adaptive, and having regulatory functions for internal dynamic processes and interpersonal in types of disorders (e.g., from anxiety to depression)? To different patterns of comor-interaction (Campos, Barrett, Lamb, Goldsmith, & Stenberg, 1983). No emotion is bidity?
Recent conceptual, methodological, and intrinsically more adaptive or dysfunctional than any other emotion. Emotions per se are analytic advances make it increasingly possible to incorporate key factors into transac-not viewed as dysfunctional, but negative emotions often have maladaptive qualities of tional, longitudinal research, designed to test ways in which nature and nurture interact to high intensity, long duration, and situational inappropriateness (Watson & Clarke, 1992). affect developmental outcomes. We now know more about putative biological and en-Among the prominent dysfunctions are disconnections between the experiential and ex-vironmental factors that confer risk or protection. However, we lack data that would help pressive components of emotion (Plutchik, 1993), as well as regulatory problems (Cole, to distinguish atypical from normative patterns of development. For example, how do Michel, & Teti, 1994). Internalizing problems not only involve prolonged, intense expres-frontal lobe activity (EEG) and reactivity to stress (ANS and HPA axis activity) differ as sions of anxiety and sadness but also efforts to control or suppress negative emotions. a function of age? What constitutes nonnormative reactions at any given age, as well as Young female adults asked to suppress negative emotion show greater physiological reac-across age? Similarly, how do environmental processes likely to include observational tivity, which may be costly in the long term (Richards & Gross, 1999). There is a need for methods (e.g., affective climate, parental warmth or rejection, expression of authority, research on processes by which dysregulation and disconnection develop in childhood, and instructional and explanatory style) differ as a function of age? We also lack descriptive data how they relate to psychopathology.
One of the earliest developmental tasks in on normative and nonnormative aspects of the experience, expression, and regulation of childhood involves the ability to regulate behaviors and emotions in an appropriate man-emotion, as well as many aspects of social and interpersonal development. How do these ner. Most children show varying degrees of mastery of this process. They are able to cope processes differ at a specific point in time, or across time, including developmental varia-with the challenges in their everyday lives in ways that allow them to become increasingly tions that may result from different socialization experiences? Considerable research effort independent and socially skilled. A number of children show deficits in the first years of life, will need to be devoted to this issue in the 21st century. There are several other issues and may be diagnosed with regulatory disorders. Problems in these early years are amor-central to an understanding of the phenomenology and etiology of internalizing problems phous and difficult to characterize, compared with later development, where there is in-from a developmental perspective that have not received adequate attention. We consider creased differentiation and specificity. The ways in which very young children's fearful-them next. They include the roles of emotion, comorbidity, gender, and culture. ness, anger, and failures to regulate emotional distress relate to internalizing (and comorbid externalizing) problems by late preschool and Emotions and Internalizing Problems early elementary school remains unclear. Negative emotions become more manage-Emotion regulation and dysregulation able with time for most children. Nervous systems mature, children come to understand The lack of emphasis on analysis of the roles and functions of emotions in the development the meaning, complexities, and consequences of their feelings, and they develop coping of anxiety and mood disorders is both notable strategies. Socialization also is thought to play From a developmental perspective, early affective biases that may precede or be part of a role, including the learning of display rules (i.e., what to conceal, what to reveal, what are the depressive experience are likely also to involve worry and anxiety; guilt, shame, and the appropriate forms of expression). This literature, however, is based mainly on age self-reproach; lack of pleasure (anhedonia); suppression of anger and hostility; and em-cross-sectional comparisons of different age groups. We still know little about individual pathic overarousal where the self becomes submerged in the problems of others. Recall differences in regulatory processes, how atrisk samples compare with normative groups, that the moral emotions such as guilt were first implicated in adult depression by Freud, and how this relates to the development of specific emotional and behavioral problems. who emphasized childhood precursors of this disorder, seen in a harsh superego and inter-We do know that regulatory problems continue throughout childhood, adolescence, and nalization of criticism and blame. Different configurations of emotional profiles may be adulthood, the implications of which merit further attention.
markers of different subtypes of anxiety and depression.

Emotions and the development of psychopathology Comorbid anxiety and depression
Comorbidity rates for anxiety and depressive Some emotion theorists have focused on functional continuities between emotions as epi-disorders can range as high as 70%, with rates from 20 through 50% most likely to be re-sodic states and psychopathology (Izard, 1977;Malatesta & Wilson, 1988). The repeti-ported (see reviews by Angold, Costello, & Erkanli, 1999;Brady & Kendall, 1992). In ad-tive nature and the emotional salience of everyday social interactions become the basis olescents, there is actually greater comorbidity between anxiety and depressive disorders for affective biases. These biases are thought to become the central organizing axes for per-than among anxiety disorders, and comorbidity is much more common than the "pure" sonality over time. When they become consolidated through repetitions of discrete emo-types. Comorbidity rates are likely to be underestimated (e.g., major depression accom-tions, and organized into rigid patterns, this may lead to specific forms of psychopathol-panied by subclinical anxiety would not qualify as comorbidity). Yet the co-occurring ogy. Examples include anger biases or traits in antisocial personality, sadness in depres-anxiety could have significant clinical implications later in development. Links between sion, and fear in some anxiety disorders. Models based on these views remain to be depression and anxiety have been construed in different ways: comorbidity may indicate tested in longitudinal research, focusing both on child characteristics (e.g., stress reactivity the presence of a single underlying dimension; separate, distinct disorders; or concep-seen in ANS and HPA arousal) and socialization experiences (e.g., how parents and peers tual and measurement problems signaling the inability of categorical systems to reliably respond to negative emotion displays) that shape particular affective biases implicated in characterize the relevant phenomena. There is now increased recognition of ways in which particular disorders.
Advances in research on emotions and in-the study of comorbidity can inform our understanding of the development of the differ-ternalizing disorders will require expanded conceptions of the structure and organization ent forms of psychopathology (Angold et al., 1999). of affects implicated in these problems. Both anxiety and depressive disorders clearly are Anxiety disorders in childhood and adolescence often precede and predict later depres-characterized by more than one prevailing emotion. For example, while sadness is a cen-sive disorders. Initial evidence was based on retrospective accounts of lifetime histories ob-tral emotion-trait-bias linked to depression, other emotions are arguably as important. tained from diagnostic interviews. For exam-ple, in one study of children with comorbid tively rare in childhood. Or it may be that continuity is present but takes a less direct anxiety and depression, two thirds became anxious before they became depressed (Ko-form (i.e., from early anxiety to later depression). Anxiety does show stability over time vacs, Gatsonis, Paulauskas, & Richards, 1989). Several prospective, longitudinal studies have (e.g., first-grade children's self-reported anxiety symptoms had significant prognostic found that anxiety temporally precedes depression in children, adolescents, and even value in terms of their anxiety symptoms and adaptive functioning in fifth grade, based on young adults (e.g., Breslau, Schultz, & Peterson, 1995; Cole, Peeke, Martin, Truglio, & an epidemiologically defined sample; Ialongo, Edelsohn, Werthamer-Larrson, Crockett, & Seroczynski, 1998;. Also, prepubertal-onset anxiety Kellam, 1995).
Because anxiety involves dysregulation of disorder precedes later recurrent major depressive disorder across several generations of limbic, vegetative, and autonomic systems, the heightened, sustained arousal eventually families at high risk for depression (Warner, Weissman, Mufson, & Wickramaratne, 1999). taxes these systems in ways that cause the organism to shut down and withdraw from envi-A common theme surfaces across theories, regarding the causal role anxiety may play in ronmental stimulation-in short, to become depressed. Not all forms of depression are the emergence of some forms of depression. The precipitating role of anxiety in the devel-preceded by anxiety, and anxiety does not invariably lead to depression. But the pattern opment of depression is consistent with an attachment perspective (Bowlby, 1960): the ini-occurs with sufficient regularity to postulate a subtype of depression that differs in its devel-tial response to object loss consists of anxiety and agitation, depleting the organism and re-opmental history from depression that has not been preceded by anxiety. It is noteworthy, sulting in despair. Over time, the anxiety becomes intolerable and the organism gives up, too, that the opposite pattern of depression leading to later anxiety has not emerged as a resulting in depression. Cognitive theories postulate causal models in which anxiety has developmental phenomenon.
The question of whether anxiety and de-a direct influence on liability for major depression. In the learned helplessness model of pressive disorders are meaningfully distinct or part of a common constellation has been con-depression, the initial response to an uncontrollable situation is to become highly anx-sidered extensively in work on "negative affectivity" or "internalizing syndrome" in both ious. One developmentally relevant model focuses on the role of rumination in the children and adults (e.g., Brady & Kendall, 1992;Clark & Watson, 1991;Kendler, Neale, prediction of current and subsequent depression (Nolen-Hoeksema, 1998;Nolen-Hoek-Kessler, Heath, & Eaves, 1992;King, Ollendick, & Gullone, 1991). Overlapping and dis-sema & Girgus, 1994). Rumination is the tendency to focus passively and repetitively on tinct features of anxiety and depression have been explained using a tripartite model one's symptoms of depression without taking action to relieve them. It involves worry, per-(Clarke & Watson, 1991). The model postulates that anxiety and depression share a gen-severation, and even obsession about one's inner state. Anxiety and worry have childhood eral distress factor referred to as "negative affectivity." Low levels of positive affectivity origins; hence, ruminative qualities may emerge at point in time prior to signs of de-are relatively unique to depression, and somatic tension and arousal are relatively pression. The ruminative process can feed upon itself so as to magnify the problems, unique to anxiety. This three-factor model has been challenged by data suggesting other fac-overwhelm the child, and eventually create a state of depression. Unlike disruptive behav-tor solutions (Burns & Eidelson, 1998) and variations in factor solutions for children as a ior problems, depression does not show a great deal of stability from childhood to ado-function of age (Cole, Truglio, & Peeke, 1997). However, it provides an important lescence. It may be a discontinuous phenomenon, reflecting the fact that depression is rela-starting point for further research on the phe-nomenology of internalizing problems, by and externalizing problems clearly emerge as separate factors, they are also significantly in-focusing on different affective cores that underlie different types and combinations of in-tercorrelated, again indicating their often comorbid nature. ternalizing problems.
Anxiety and depression also have been as-There is growing evidence that diverse constellations of behavior may reflect diverse sociated with different forms of cognitive processing (see review by Mineka, Watson, & etiologies, having different correlates and developmental pathways. For example, Harring-Clark, 1998). An attentional bias for threatening information is primarily associated with ton, Fudge, Rutter, Pickles, and Hill (1990) found that children with depressive disorder anxiety, memory biases are primarily associated with depression, and judgmental or inter-were 4 times more likely to suffer from some depressive disorder in adulthood. However, pretive biases are associated with both anxiety and depression. This is consistent with Oatley the children with depression and conduct disorder had lower rates of depression in adult-and Johnson-Laird's (1987) proposal that there may be unique modes of cognitive oper-hood than did children with pure depression.
The presence of anxiety in antisocial youth, ation associated with different basic emotions. In this view, depression and anxiety evolved assessed in terms of autonomic reactivity, also appears to reduce later criminal behavior in response to different environmental circumstances. Anxiety, like fear, is associated (e.g., Raine, Venables, & Williams, 1995).
Depression or anxiety may function as a pro-with scanning the environment in anticipation of potential threat. Depression involves reflec-tective factor, reducing conduct problems over time. Children with conduct disorder tive consideration of events that have led to failure and loss. Research on childhood ante-also have shown higher levels of cortisol when a comorbid anxiety disorder was pres-cedents of different biases is warranted (Zahn-Waxler, 2000). ent (McBurnett, Lahey, Frick, Risch, Loeber, Hart, Christ, & Hanson, 1991), providing corroborative neuroendocrinological evidence for Comorbid internalizing and the anxiety component. Other research has externalizing problems found the opposite pattern. For example, the link between aggression early and late in the The incidence of disruptive behavior disorders (conduct, oppositional, ADHD) is ele-school year in first-grade males was strengthened by the presence of comorbid anxious vated in youth with either mood or anxiety disorders (see reviews by Compas & Ham-symptoms, rather than attenuated (Ialongo, Edelsohn, Werthamer-Larrson, Crockett, & men, 1994;Nottelmann & Jensen, 1995). There is also substantial comorbidity between Kellam, 1996). Considerable conceptual and empirical work is needed to unravel the com-internalizing and externalizing syndromes, seen in Achenbach scores in the .50s and .60s plex and diverse ways in which having one type of problem influences the developmental (e.g., Hinden, Compas, Howell, & Achenbach, 1997). The common and distinct fea-course of another comorbid condition.
Calculations of comorbidity rates for inter-tures of internalizing and externalizing problems have been studied in a birth cohort at nalizing and externalizing problems are based on the research sample as a whole. While sub-ages 18 and 21 years (Krueger, Caspi, Moffitt, & Silva, 1998). Latent structure and sta-samples can be identified based on different constellations of overlapping disorders, there bility of 10 common DSM-III-R (APA, 1987) mental disorders were examined. A two-factor have been few efforts to develop distinctive profiles based on these subgroups. Such an model in which some disorders reflected internalizing problems and others reflected ex-approach would permit greater in-depth analysis of possible different etiologies. Both cat-ternalizing problems provided the most optimal fit to the data. Individuals retained their egorical and dimensional approaches could be used in concert to identify and investigate relative positions to a significant degree, over the 3-year period. Even though internalizing subtypes of internalizing problems that be-come lumped together in calculations of co-associated with problems (e.g., anger, fear, sorrow). morbidity. This would facilitate examination Future research would benefit from longiof the implications of symptoms versus a ditudinal research designs that also include agnosis of depression (Gotlib, Lewinsohn, & variables based on direct assessments of bio-Seeley, 1995). This complementary approach logical (e.g., DNA, dopaminergic and serocould help to establish more veridical profiles tonergic neurotransmitters) and environmental for types of internalizing problems (Klimes-(e.g., specific types of childrearing and disci-Dougan, Kendziora, Zahn-Waxler, Hastings, pline practices) processes. The influence of Putnam, Fox, Suomi, & Weissbrod, 1997).
these factors could then be estimated in mod-Detailed analysis of different constellations of els based on direct assessments of these facsymptoms within a diagnostic category could tors rather than on inferences derived solely help further to refine the classification profrom the degree of genetic relatedness. With cess.
regard to environmental processes, other lon-A number of broad classes of dysfunction, gitudinal research on the co-occurrence of adexpected to create a range of adverse environolescent depressive symptoms and conduct mental conditions for children, have been asproblems in adolescents has measured parentsociated with both internalizing and externaling effects directly (Ge et al., 1996). Different izing problems (e.g., Gotlib & Avison, observed parenting behaviors predicted both 1993)-for example, psychosocial stress, the separate occurrence and the co-occurrence poverty, parental marital discord, parental of these internalizing and externalizing probpsychopathology, maltreatment, and parental lems at later time points, controlling on initial emotional unavailability. A concentration of problems. Research that incorporates the multiple risk factors is typically found in strengths of both types of designs (i.e., genetimore severe and varied problems in children cally informed samples and actual measure- (Sameroff, Seifer, & Bartko, 1997). Selection ment of environmental, and biological, variof well-defined samples will be important for ables) will be central to scientific advances in identifying environmental factors specific to this area. the development of internalizing, externaliz-Finally, we consider the implications of ing, and comorbid problems. comorbidity of emotions for understanding Depressive symptoms and antisocial bethe phenomenology of internalizing problems. havior appear to share a common genetic Basic or discrete negative emotions such as diathesis. They may co-occur because of a fear, sadness, and anger co-occur as blends in common genetic liability that increases vulnormal individuals (e.g., Watson & Clark, nerability to both types of problems (O' Con-1992) and in those with clinical conditions ner, McGuire, Reiss, Hetherington, & Plomin, (Biederman, Farone, Mick, & Lelon, 1995;1998). In this work, genetic influences were Kovacs, Feinberg, Crouse-Novak, Paulausidentified separately for depressive symptoms kas, & Finkelstein, 1984). Terms such as "suland antisocial behavior, and for their co-oclen," "moody," or "irritable" are used to decurrence, in adolescents studied using a twinscribe emotional states that become more sibling design. O'Conner et al. (1998) hypothtraitlike blends of different negative emotions esize that depressive and antisocial symptoms seen in both internalizing and externalizing may be correlated because genes influence a problems. Future research could focus on proneurotransmitter system associated with both cesses by which certain emotions within the behavioral syndromes. Such a view would be comorbid constellation come to predominate consistent with the demonstrated efficacy of (e.g., the child who initially is both angry and some psychopharmacological agents in treatfearful, but who eventually is characterized ing both internalizing and externalizing synprimarily by fearfulness). Negative emotions dromes. The medications may help to regulate thus could become differentially channeled emotionality and, hence, act more generally (e.g., as one comes to predominate over time to influence the ultimate form of disorder). upon the wide a range of negative emotions Symptoms of irritability, in particular, twice as likely (sometimes more) as males to become anxious and depressed, a pattern that characterize both internalizing and externalizing disorders. Features of major depression continues throughout adulthood. Comorbidity of anxiety and depressive disorders is much among clinically referred children and adolescents include a high frequency of dysphoria more common in girls than boys (Lewinsohn, Rohde, & Seeley, 1995); moreover, having and irritable or angry mood (Biederman et al., 1995;Kovacs et al., 1984). Irritability in de-more than one anxiety disorder during childhood or adolescence is virtually exclusive to pressed children is commonly described by parents and teachers, and is often what leads females (Lewinsohn, Zinbarg, Seeley, Lewinsohn, & Sack, 1997). to a diagnosis of depression (Poznanski, 1982). And it is not uncommon for anxious Until recently, most of the research on childhood disorders has focused on external-children to show heightened irritability. This further underscores the importance of re-izing problems known to be more prevalent in males throughout the life span, and beginning search on the etiology of the dysregulated emotions central to internalizing disorders. As in early childhood. Because the nature of problems more common to girls are also less noted, children who meet criteria for major depressive disorder frequently meet criteria disruptive to others and surface later, research on etiology and development of internalizing for ADHD, conduct disorder, and anxiety disorders (Biederman et al., 1995). These rela-problems has lagged. The presence of this very strong sex difference, suggests the need tionships would be consistent either with processes whereby depression reflects hostile to better understand both constitutional and environmental factors that place females at impulses turned inward or antisocial, disruptive behavior reflects masked depression. Nei-such high risk for anxiety and mood disorders. Most explanations have focused on bio-ther process is likely to be completely "successful," hence the manifest comorbidity. logical processes and experiential factors associated with adolescence and pubertal de-Affective disturbances associated with disruptive behavior disorders and depressive disor-velopment (see Zahn-Waxler, 2000). Here we emphasize the significance of processes ders thus are not as distinctly different as they might first appear to be (Cole & Zahn-Wax-set in motion much earlier in development that are more difficult to detect. ler, 1992). Careful analysis of patterns of emotion dysregulation associated with internalizing and externalizing problems might Dispositional characteristics. Early constitutional advantages for girls may provide a help to elucidate their common and separate pathways. This will require research designs number of qualities that decrease risk for early behavior problems. Young girls, on av-that incorporate genetic, psychiatric, and developmental research designs. The resources erage, show greater language skills, more rapid physical maturation, and advanced regu-and cooperation necessary make this a daunting task. latory capabilities. This may contribute to their known advantage in terms of frustration tolerance, ego and impulse control, internal-Gender ized standards of conduct, empathic sensitivity and ability to interpret others' emotions, During childhood girls exhibit far fewer externalizing problems (and less overall psycho-social maturity, and responsible interpersonal behavior (Zahn-Waxler, 2000). Why then do pathology) than boys, with no clear sex differences evident in depressive and anxiety girls, who appear resilient and relatively impervious to childhood mental disorders, later disorders. By adolescence, girls (but not boys) show a marked increase in anxiety and mood show so many internalizing problems? Continuities may present but in a more subtle form. disorders and symptoms (Lewinson et al., 1993;Nolen-Hoeksema & Girgus, 1994). Fe-Qualities that protect against antisocial behavior could also create risk for internalizing male sex becomes the strongest risk factor for internalizing problems. Females are at least problems.
We indicated earlier that anxiety could of-minate more than males, and rumination predicts future depression (Nolen-Hoeksema & ten be a precursor of depression, beginning in childhood. Girls are more prone to early Girgus, 1994). We propose that risk for some forms of later depression begin in childhood, fearfulness and worry than boys (Silverman, LaGreca, & Wasserstein, 1995), even norma-more often for girls because of their heightened reactivity and ruminative styles. tively, and they are somewhat more likely to be shy and inhibited. Dienstbier (1984) has proposed that different temperaments might Socialization experiences. The socialization data reviewed here is quite recent in origin lead to different emotion-attributional styles and levels of guilt. Proneness to emotional and often based on middle-class samples. Differential treatment of boys and girls may cre-tension should result in intense discomfort and distress following transgression. When ate conditions that predispose females more often than males to anxiety and depression. distress is internal, the child is more likely to experience the links between tension and Girls are more likely than boys to be socialized in ways that interfere with self-actualiza-transgression, and come to experience anticipatory anxiety. Temperamentally anxious tion (i.e., to be dependent, compliant, and unassertive;Hops, 1995). A number of different children may develop "affective maps" of their experiences where threat or stress-re-experiences are likely to create greater risk for females than males (reviewed in Zahn-Wax-lated information becomes particularly salient (Derryberry & Reed, 1994). Similarly, Da-ler, 2000). On average, sons are more valued than daughters. Even when there is serious masio, Tranel, and Damasio (1991) refer to an automatic guiding system that activates "so-conflict between parents, these couples show more mutual investment in their sons than matic markers" associated with one's past experience. These "maps" or "somatic markers" daughters. Socialization practices (particularly in fathers) that encourage sex-stereo-may facilitate the early, rapid development of mechanisms related to conscience, such as typed activities could lead to greater submissiveness and dependency on the opinions of guilt and restraint.
Because anxiety is more common in girls others in girls. Parents more often discourage exploration of the physical environment in than boys in childhood, and because it is associated with autonomic arousal (elevated heart girls than boys. And they more often place pressure on girls to anticipate the conse-rate and electrodermal activity), girls may become more physiologically aroused than boys quences of their negative acts. Girls are more likely to be reinforced for shyness and depen-under particular conditions of interpersonal conflict and distress (Zahn-Waxler, Cole, dency, their successes more often overlooked, and their physical aggression deemed more Welsh, . Girls may more readily develop somatic markers that facilitate inter-serious than for boys. Parents often have higher expectations for nalization of norms and standards of conduct. This proneness to experience internalized dis-mature interpersonal behavior in girls, are less tolerant of their anger and misbehavior, and tress also may lead to early rumination, creating a constellation of factors that could more likely to override or negate assertive behaviors in girls than boys. Many of the social-heighten risk for developing anxiety and mood disorders. Although this is a viable hy-ization practices directed more often toward girls contain messages that reflect pressures to pothesis, it remains to be tested. Because females are known to be more emotionally re-be prosocial, suppress anger, and curtail antisocial behavior. Suppression of anger, asser-sponsive than males to the problems of others, a wider range of interpersonal contexts tion, and other forms of self-expression may heighten internalized distress. Over time this may arouse them. Their automated, internal reactions that also include ruminative scripts may result in the indecision, self-criticism, self-blame, and low self-esteem that become may then contribute to entrenched, generalized patterns that are later seen as anxiety and part of the phenomenology of the depressive experience. This is one hypothetical example depression. Adolescent and adult females ru-of how disconnection between experience and than boys across cultures (Crijnin et al., 1997). While adult depressive disorders and expression of emotions (here, anger and hostility) could contribute to the development of syndromes are more prevalent in females than males in virtually all cultures, the ratios differ internalizing problems. It is noteworthy that females and males do not appear to differ in markedly (Kleinman & Cohen, 1997). In China, for example, female-to-male rates of their experiencing of anger and hostility but rather in their expression and acting out on depression and completed suicides are strikingly disproportionate, with suicide described these feelings.
as a "normative" response for povertystricken rural females. This indicates the need Culture and the development of to examine cultural characteristics that lead internalizing problems women from some parts of the world, such as China, to take their lives more often than men, Almost without exception, research considered here is based on samples of children and whereas in other places such as the United States the reverse is true. adolescents from the United States. It should not be assumed that etiology, prevalence, and phenomenology are consistent across cultures. Cultures vary both on some biological dimen-Reflections and Projections sions and in the kinds of behaviors they value, forbid, and condone. Both may shape the de-It has been illuminating to chart the remarkable progress made in our understanding and gree and direction of problems in childhood. There are differences in cultural norms re-treatment of early anxiety and depression over a relatively short period of time. Until quite garding displays and regulation of emotions, self-expression, and the role of the self in re-recently these disorders were not thought to be present in children and adolescents. Since lation to others (Markus & Kitayama, 1991;Han, Leichtman, & Wang, 1998). These vari-then, several internalizing disorders have been identified, assessed, classified, and treated ations, as well as different culturally defined ways of coping with distress, also may help with a variety of psychotherapeutic and pharmacological approaches. A number of biolog-to determine whether internalizing problems become predominant (McCarty, Weisz, Wani-ical and environmental factors implicated in the expression of these disorders have been tromanee, Eastman, Suwanlert, Chaiyasit, & Bond, 1999). This is a little-researched area, studied, as well as risk and protective factors that heighten or diminish the probability that one that is ripe for inquiry both into the universality and cultural specificity of processes such problems will surface. We have seen an increase in multivariate, longitudinal designs that underlie anxiety and depression.
Cross-cultural differences exist in the prev-used to more fully explicate the etiology of the internalizing symptoms and disorders that alence and patterning of problems. Internalizing problems in children occur more often in so commonly afflict children and adolescents.
Considerable attention has been devoted to some countries (e.g., Greece, Thailand, Puerto Rico), whereas externalizing problems are the development of methods to assess symptomatology and the factors implicated in their more common in other countries (e.g., United States, Germany, Sweden; Crijnin, Achen-expression. But it is just a beginning. As a field we are still in a very early developmental bach, & Verhulst, 1997; MacDonald, Tsiantis, Achenbach, Motti-Stefanidi, & Richardson, period with regard to our theoretical and applied knowledge base. While transactional, 1994). While comorbidity of internalizing and externalizing problems is common across developmental research is commonly recommended, we have not yet fully incorporated countries and cultures, the nature of these relationships will vary depending upon which this perspective into our study designs. Moreover, we are still some distance from ascer-type of problem is more prevalent in the different cultures.
taining the most salient constructs and, in turn, having the requisite armamentarium of Girls show more internalizing problems valid measures needed to test transactional of human suffering. To do so requires a profound understanding of the nature of children models.
The fluid nature of many affective and be-and adolescents across the time course. Symptoms of internalizing problems are known to havioral symptoms in children and adolescents continues to raise questions about the vary across different periods of development.
Historically, the use of top-down models (i.e., ultimate forms disorders will take over the course of time. There are indications of pro-studying anxiety and depression in adults and then working backward to understand similar gression from comorbidity to specificity, where comorbidity is seen as an earlier devel-problems in children and adolescents) created problems that continue to the present day. In opmental phenomenon (Nottlemann & Jensen, 1995). Initially, psychopathology is psychiatric research, participants who vary widely in age (e.g., from 6 to 16 years) often amorphous in form of expression, becoming more differentiated and clearly defined as per-are considered as one group for analytic purposes. This can limit generalizations and, sonalities begin to crystallize and the central nervous system reaches maturity, especially hence, the utility of the research findings. It is difficult to develop valid assessment sys-after puberty and midadolescence. For example, while current lifetime histories of depres-tems and programs for effective prevention, intervention, and treatment in a nondevelop-sion in adolescents are highly comorbid with other disorders, the rates are substantially mental context. Psychotherapeutic and pharmacological approaches to treatment also lower in adults (Rohde, Lewinson, & Seeley, 1991). Adolescent-onset depression may rep-originated from top-down models. While many treatment approaches are now highly resent a more serious form of the disorder. Alternatively, it may initially be more undif-sensitive to developmental level, some are not. Public attention recently has been drawn ferentiated, with greater specificity developing over time. Longitudinal research would to giving children in the first years of life the medications that had been developed and ap-help to address questions about determinants of these different developmental pathways.
proved for treatment of depressive disorders in adults and attentional problems in school-The literature on resilience is beginning to identify factors associated with a good out-age children. While developmentally sensitive research has become more common, it often come despite being reared under conditions of risk. We know relatively little about what fac-follows rather than precedes treatment decisions. tors are protective with regard to internalizing disorders as compared with other types of Psychologists have in-depth knowledge of developmental processes. They also have psychopathology. Many definitions of resilience center on achievement, social compe-methods that, if used in conjunction with those developed within a psychiatric tradition, tence, or finely honed interpersonal skills (Beardslee & Podorefsky, 1988; Luthar & would help to establish an integrated perception of the essential need for both categorical Zigler, 1991). Internalizing problems often coexist with competencies, and "high" func-and dimensional approaches to assess and classify internalizing problems. It is important tioning also can reflect aspects of disorder (Luthar & Zigler, 1991). Consequently, addi-that both psychiatry and psychology deal with a full range of symptomatology. There is a tional methods are needed to assess psychological health as a protective factor. reluctance to wrestle with questions about whether and when a quantitative difference The scope of this article did not include discussion of assessment and treatment, topics becomes a qualitative difference as well, or about dynamic processes that move the child that are less often considered within a developmental psychopathology framework. Scien-from normative, to subclinical, to clinical symptoms (or in the other direction). Psychol-tific knowledge of the origins, development, and phenomenology of internalizing problems ogists who rely solely on quantitative, dimensional measures often do not convey the de-is important in its own right. But ultimately this work is also in the service of valid assess-gree of psychopathology (or lack thereof) that is reflected in their findings. Psychiatrists who ment and successful treatment of these forms rely solely on qualitative, categorical mea-pear as temperamental qualities or personality characteristics. In contrast to the stability of sures based on diagnostic information often do not have sufficient appreciation of the sig-externalizing problems across time, there is little empirical evidence of continuity of de-nificance of subthreshold symptoms. Collaborative work would elucidate the wide range of pression. Lack of continuity may reflect the lack of developmental research on internaliz-significant symptoms.
A related issue concerns the models and ing problems, along with the fact that prodromal signs may be less obvious. Even less is statistical approaches utilized in different disciplines. It has been common in develop-known about the history of anxiety syndromes across these developmental periods, and prob-mental research for psychologists to test explanatory models using a variable-centered ably for the same reasons. Even with the increased awareness of internalizing disorders approach, attempting to understand developmental trajectories through analyses of the in children, only a minority receives treatment. When children evidence more "pure" group as a whole. Such a process does not lend itself well to understanding of subtypes forms of anxiety or depression, symptoms may go unnoticed as they suffer in silence. of anxiety and depression either in "pure" or comorbid form. It is easy for the individual to When they are accompanying externalizing problems, the disruptive behaviors may ac-become lost in what has come to be viewed as a "one-size-fits-all" model. A person-oriented tively divert attention away from the anxiety or depression. Some early expressions of anx-approach that focuses on attempting to define different clusters of individuals is also essen-iety could even be seen as positive as they may motivate children to behave in ways val-tial. The contributions of developmental psychology and psychopathology will become ued by adults.
In summary, anxiety and depressive disor-more salient to child and adolescent psychiatry (and vice versa) when the disciplines are ders are common in childhood and adolescence. These disorders are still understudied able to interface more on problems that would be expected to be of common interest.
relative to other disorders, highlighting the need to focus on biological and psychosocial A developmental psychopathology framework has much to offer for understanding the underpinnings of these illnesses. Longitudinal studies of the course and outcome of anxiety etiology and development of internalizing problems. There is evidence from prospective and depressive disorders in childhood and adolescence are critical-not only to further our research of continuity between adolescent and adult depression (Rao, Hammen, & Daley, knowledge of the impact of these disorders during the earlier stages of human develop-1999). Longitudinal research on links between childhood depression and adolescent or ment but also to increase our knowledge of their emergence in adulthood. The progress adult depression has not been conducted. Not all depression subsequent to childhood would that has been made and the foundations that have already been established for achieving be expected to have very early origins, but depressed adults often report symptoms dat-future goals should encourage us. We also should be prepared for sweeping changes in ing back to childhood. Clinical depression may emanate eventually from subsyndromal how internalizing problems are viewed and studied in this century. depressive styles that in childhood may ap-