Elsevier

Biological Psychiatry

Volume 49, Issue 12, 15 June 2001, Pages 973-979
Biological Psychiatry

The unmet needs in diagnosis and treatment of mood disorders in children and adolescents
Emotional development and psychiatry

https://doi.org/10.1016/S0006-3223(01)01115-5Get rights and content

Abstract

This paper discusses the current practice of using single word labels for emotional states that ignore the nature of the incentive, cognitive representations, physiologic profile, and especially the developmental stage of the agent. The universal cognitive advances that are derivatives of brain maturation over the first dozen years are accompanied by significant emotional states that infants and young children cannot experience. The cognitive transitions at the end of the first year, the middle of the second year, 5–7 years and early adolescence have special implications for affect states.

Introduction

Aword can name a phenomenon with different antecedents and future actualizations, as well as events with no referent in nature. The popular terms for varied emotions belong to one or the other of these categories. A fair proportion of scientists and clinicians write about the generic concept emotion, or any specific emotion, as if the semantic term referred to a unitary natural phenomenon and the intellectual challenge was to achieve consensus on the correct definition and discover the most valid measurement.

Most definitions of emotion permit almost any change in physiology or consciousness to be categorized as an emotion. Rolls (1999), for example, defines emotion as the state created by a reward or the sign of a reward. Keltner and Gross (1999) say emotions are brief episodic patterns of perception, experience, physiology, action and communication that occur to specific challenges. Buck (1999) regards emotion as a subjective experience of a feeling or desire. I do not suggest that these definitions are incorrect, only that they are so general it is difficult to determine which human states are not emotions. Further, these permissive definitions contrast with the empirical emphasis on the states called fear, anxiety, sadness, anger and surprise, which comprise a small proportion of the affect states that the Rolls, Keltner and Gross, and Buck definitions permit.

A more serious problem is the continued reliance on familiar words, like fear and sad, which presumes that the human communities that invented these contextually unconstrained terms eons ago possessed a profound insight into the fundamental nature of human emotions. These words were intended to communicate the fact that a human agent experienced consciously a change in body tone that could not be attributed to an organic need for food, water, warmth or relief from pain. Members of the community also understood that the agent accommodated to the incentive and her accompanying thoughts when she selected a word or phrase to describe her emotional state.

Two historical events altered the original meanings of emotional words as used in daily life. First, behavioral biologists and neuroscientists borrowed some of these words to describe the states of animals that displayed particular behaviors. For example, they attributed fear to a rat who avoided a brightly lit alley in an elevated T-maze or froze to a tone that had been paired with shock, and assumed that the word fear did not change its popular meaning (Belzung and Le Pape 1992; Davis 1992, Le Doux 1993, Le Doux 1994). Further, many investigators wrote as if the term fear referred to a single brain state that was fundamentally similar in humans and animals across different incentives, rather than a family of states, each mediated by a different pattern of brain activity.

Support for this particularistic claim is the fact that the neural circuit activated when a rat freezes to a tone that had been associated with electric shock is different from the circuit activated when a rat avoids a brightly lit alley in an elevated maze (Treit et al 1993). Further, a rat who experiences electric shock to the paws while in a distinctive chamber shows a number of conditioned reactions when placed in the chamber at a later time, including freezing, a rise in heart rate and defecation. Rats without an amygdala do not acquire the conditioned reactions of freezing or heart rate increase, but do acquire conditioned defecation (Antoniadis and McDonald 2000). This result means that the rats with a lesioned amygdala who defecated in the chamber could not be in the same brain, or emotional state as the rats with an intact amygdala who froze in the same context.

Moreover, it is not obvious that fear is the best name for the family of states produced by classical conditioning of freezing with shock as the unconditioned stimulus. Pavlov did not attribute a state of hunger to the dogs that were conditioned to salivate to an auditory stimulus that had been followed by the delivery of food powder. A biologist who called a female rhesus unfaithful because she copulated with four male rhesus in an hour would be a target of derision. But few scientists smile when an author writes that a rat that shows increased activity in the lateral and central nuclei of the amygdala as it freezes to a tone that signals shock is fearful. Francis Bacon called these semantic errors, “Idols of the Marketplace.” A human fear state, like the perception of pain, combines activity in brain, peripheral body targets and thought; it is not a particular brain state.

A second historical event was the decision to use answers to interviews and questionnaires as evidence for human emotional states. If a respondent in an epidemiologic study told a telephone interviewer that she usually avoided crowds because she felt uncomfortable, the latter assumed that the person experienced the emotion of anxiety or fear, even though there were no measurements of the person’s physiology when she either thought about crowds or joined them.

Verbal report, observed behavior, and physiology are used as evidence for inferring emotional states in humans; however, because these three classes of information are not always highly correlated the meaning of an emotional term inferred from one source is not equivalent to the meaning inferred from another. That is, a verbal report of depression need not be highly related to the biological profile presumed to be the basis for this emotion; a period of sleeplessness need not be related either to a report of depression or a biological profile. Further, the three sources of evidence form a pyramid with respect to number and variety of human emotional states. Self-reports yield the smallest number of affective categories because of the limitation on the number of words available to name feeling states. Behaviors are the foundation of a larger number of affects. Physiologic profiles to particular incentives provide the largest number of categories; hence, different physiologic profiles can accompany a particular behavior or self-reported emotion. The implication of this last claim, which Newton would have applauded, is that scientists should begin their inquiry with particular phenomena rather than popular words. That is, investigators should begin with the fact that the amygdala is necessary for acquiring a conditioned freezing reaction to a compartment but is not required for acquiring a conditioned defecation reaction, and they should probe the reason for these phenomena rather than begin inquiry by assuming, a priori, that “fear” is a state that rats experience.

One problem with the current constructs of emotion is that investigators are prone to award the same emotional label to four distinctly different phenomena. I shall use the emotion of anxiety as an illustration, but the following discussion applies to all currently popular emotional concepts.

A verbal judgment without the presence of any accompanying change in bodily state or physiology is one basis for attributing anxiety to an individual. It is common for individuals to say that they are anxious over a possible storm or their child’s high fever. However, if scientists had measured their physiology at the time they made the statement, or even when the feared event was imminent, they would not have detected any particular change in brain or bodily activity. Let us call this meaning judged anxiety.

A second category is a verbal report of anxiety accompanied by a physiologic profile, but not the one scientists assume is theoretically appropriate. For example, an individual with a viral infection might feel tense, and in an attempt to understand the unexpected sensations might decide that she is worried about her job. Let us call this meaning constructed anxiety.

A third meaning is used most often by investigators who work with animals. A particular stimulus provokes activity in the brain structures believed to be relevant to fear (e.g., activation of the amygdala and its projections), but the person does not consciously detect any consequences of the changed physiology. Let us call this meaning physiologic anxiety.

The fourth referent is the one scientists would like to believe occurs when the term anxious is attributed to a person. An event, either a thought or external provocation, has generated the physiologic state believed to accompany anxiety, and the person detects the sensory consequences of the altered physiology and interprets it as reflecting an anxious state. Let us call this meaning anxiety(Kagan 1998).

The distinction between physiologic anxiety and anxiety, as defined above, has theoretical importance because the affect that follows a change in physiology to an incentive is influenced by whether the agent does or does not consciously detect the change. Most adults react to an unexpected stimulus with a time-locked series of event-related potentials over the first 300–600 msec, as well as a decrease in heart rate that reaches its trough about 3 sec after the onset of the event. Most individuals are unaware of a changed feeling state when the event-related potential and the heart rate decrease occur. It is necessary, therefore, to invent different names for the states characterized by undetected, contrasted with detected, changes in physiology. I call the former internal tone and the latter a feeling state. Damasio (1999) calls the former an emotion and the latter a feeling state that has pierced consciousness.

Imagine three women, each of whom has just realized that, in a month, her last child is leaving home for college. Over the next few days, the first woman detects a loss of appetite and energy, thinks about these changes, decides they are due to the imminent loss of the adolescent, and concludes she must be sad. The second woman detects the same bodily changes, but decides that she has been working too hard. The same bodily changes in the third mother go undetected, even though she seems to her husband to be less affectionate and more quarrelsome. The imminent loss of the child generated similar bodily changes in internal tone in all three women, but we must use different terms for their feeling states. It is an error to call all three women sad because both their conscious states and behaviors are different. At a minimum, we should call the states sadness A, sadness B, and sadness C. Because evaluation of consciously detected physiologic changes probably does not occur in animals, investigators should distinguish between the emotional states of animals and humans and resist the pleasing assumption that the same affect terms can be applied to rats, monkeys, and humans.

Finally, investigators must differentiate between transient, acute emotions provoked by specific incentives, on the one hand, and chronic moods that are less clearly linked to external events, on the other. An acute state of anxiety following a physician’s diagnosis of cancer should be distinguished from the chronic uncertainty of melancholics like Sylvia Plath. Four different emotional constructs are created by crossing the distinction between consciously detected, compared with undetected, changes in physiology with the distinction between acute emotions and chronic moods. Each of these four states requires a different name. Adults who report that they worry a great deal but do not detect any changes in internal tone show EEG patterns different from those who worry and, in addition, are conscious of physiologic changes (Nitschke et al 1999). Unpublished work from my laboratory reveals that among 10-year-old children who report being shy and timid, only those who had been classified at 4 months as temperamentally high reactive showed right frontal activation in the EEG, whereas those who had been classified as low reactive did not show this physiologic property (Kagan and Snidman 1999).

The agent’s interpretations of changes in internal tone in a context are central components of every human emotional state, but these interpretations are influenced by maturing cognitive competences. Hence, the universal changes in cognitive abilities that accompany brain maturation over the first 12–15 years, together with the knowledge that is acquired with experience, should be accompanied by new families of emotions. This assumption means that the emotional states of infants or very young children have to be different, perhaps qualitatively, from those of adolescents and adults. Some investigators prefer the parsimonious premise that each affect has a core physiologic (or brain) state that does not change with age. Although encountering a large upright bear on a trail might elicit a similar state in 2-year-olds and adults during the first 200 m/sec, the cognitive evaluations that occur during the next few seconds, including the recognition that one might be able to cope with this dangerous object, will be very different in young children and adults. These evaluations are not simply epiphenomena, for they alter the brain state created during the first 200 m/sec and, in so doing, generate a specific emotion.

Although a detailed understanding of the complex relations among brain growth, cognitive processes and emotional states is incomplete; nonetheless, some reasonable hypotheses are possible. The evidence suggests that important cognitive transitions occur in children between 6 and 12 months, 1 and 4 years, 5 and 8 years, and at puberty. The following sections describe the implications of these transitions in cognitive ability for emotional states.

Table 1contains a selected list of suggested emotional states, along with the relevant incentives, that accompany particular developmental stages. This list is not intended to be exhaustive, and excludes states produced primarily by states of hunger, pain, cold, intense visual and auditory stimulation, and bitter or sour tastes.

The infant during the first 6 months is alerted by, and attends to, changes in the sensory surround and events that are discrepant from acquired schemata. This state might be called alert attention or surprise. If the infant is able to relate the new event to an acquired schema—called assimilation—it may smile. This reaction is common in 4-month-olds when an adult gazes at the infant. The smile reflects successful assimilation of the person’s face and is a state one might call the pleasure of assimilation.

A cry to loss of an object, for example when a pacifier is removed, reflects a third state one might call distress to loss. Some observers have called this state anger because of the shape of the infant’s mouth and other facial features when the infant is crying. But I believe that loss of an object by a child older than 2 years is always accompanied by a representation of an external target. Because this representation is absent in young infants, it is useful to award different names to the reactions of infants and children.

The growth of connectivity between medial temporal sites and frontal cortex, which began earlier, attains a critical level during the last half of the first year to permit a new cognitive competence and a new emotion (Diamond and Goldman-Rakic 1989). The infant is now able to retrieve schematic representations of past events, even if they are not in the perceptual field, and to hold in a working memory circuit both the retrieved schemata and the representations created by the immediate context, while attempting to relate the latter to the former. If assimilation does not occur in a few seconds, a new affect state is created and the infant may cry. Some theorists call this state fear to unfamiliarity. The universal display of crying to a stranger and temporary separation from a caretaker, which occur between 6 and 12 months, are classic examples of this affect state.

Four important cognitive competences that emerge in most children between 1 and 4 years make several new families of emotions possible. The first is an initial set of schemata and semantic representations for the categories good and bad. Two-year-olds are aware of some of the actions and events that adults treat as a reason for punishment because they violate the moral standards of the family. The second is the initial ability to infer some of the thoughts and feelings of another (Kagan 1981). This second competence is evident in the disposition, first observed in the second year, to track the direction of gaze of another adult. If a parent who has been playing with the infant suddenly directs her gaze to a corner of the room the child is likely to direct her gaze in the same direction (Tomasello 1999). This behavior implies that the child infers that the adult has relevant knowledge about the environment.

The combination of an appreciation of good and bad actions—that is, praiseworthy and punishable responses—and the capacity to infer the mind of another renders the 2-year-old vulnerable to a new affect when he behaves in ways that are inconsistent with his understanding of good and bad. The child who has committed a punishable act will lower his head or blush when he spills food or breaks a vase. Most observers call this state shame or embarrassment(Lewis 1992).

The third competence is the child’s initial, conscious awareness of her intentions and of changes in internal tone. The 4-month-old infant feels the pain of a pin and the sweetness of sugar, but is not aware consciously of these sensory states. That is, the infant does not reflect on these sensations or retrieve representations of past similar states. The 2-year-old, by contrast, consciously recognizes her feelings. The combination of self-awareness and inferring the mind of another makes the affect state of empathy possible. One sign of this state is that 2-year-olds will show behavioral signs of concern if a parent seems to be in pain or distress Radke-Yarrow and Zahn-Waxler 1983, Zahn-Waxler 1991.

The awareness of self’s intentions also permits an affect some might call pride when the child successfully completes a difficult task following effort. One sign of this state is that children often smile after building a difficult block construction. This smile may resemble the smile of assimilation of the 4-month-old, but the affect state of pride is a different member of this family because the incentive and cognitive representations are distinct.

The appreciation of good and bad acts and inference of another’s intentions permit a state one might call moral anger. One sign of this state is the fact that 3-year-olds protest if an older sibling teases them because the young child appreciates that the sibling’s action violates a moral standard. This affect state is a member of the anger family, but should be differentiated from the state of the infant who has just lost a toy.

Finally, children older than 3 years are able to integrate past with present and to anticipate future events by reflecting on retrieved schemata and semantic representations. This state is enhanced between the third and fourth birthday (Loken 2000). This ability renders the child vulnerable to a state some might call sadness when the child reflects on the loss of a state of affairs that had been desirable, and to a state some might call anxiety over possible harm, loss or adult disapproval. Both states require the ability to integrate past, present and future. Infants are incapable of this integration.

The cognitive talents that emerge between 5 and 7 years in children from industrialized societies, between 7 and 11 years in those from societies without schools, permit additional affect states. These talents are part of the developmental stage Piaget (1952) called concrete operations. Although there is controversy surrounding the mechanisms Piaget posited to explain cognitive development, there is no disagreement over the empirical facts that represent universal cognitive advances in the childhood years. One significant ability is the capacity to re-run mentally a past behavioral sequence. This phenomenon, which Piaget called reversibility, allows children to experience guilt when they have violated a personal standard, especially one that caused distress to another. The child reflects on the sequence of actions that led to a violation and realizes that he could have suppressed the asocial act. The insight that he is responsible for an action that harmed another or destroyed property is often followed by a feeling of guilt. Infants feel neither shame nor guilt when they throw food on a clean tablecloth; 2-year-olds will feel shame; 6-year-olds will feel guilt. Freud recognized that guilt emerged at this age, but his interpretation involved the oedipal conflict, not reversibility.

A second feature of this stage is the ability to compare a diverse set of events on the same dimension or property. This competence, which Piaget called seriation, is revealed when children given a set of six sticks of varying length arrange them in a pattern from shortest to longest. The ability to rank a set of objects on some property permits children to compare self with others on attributes of concern to the child like strength, size, attractiveness, popularity and skills. The results of that comparison create new emotional states. If the child judges that she possesses less of a desired characteristic than a friend does, she may experience an emotional state for which there is no consensual name. This state could be called self-doubt, envy, sadness, or low self-esteem. I prefer the first term, but the important point is that 3-year-olds do not experience this emotion.

The ability to compare self with others leads not only to judgments of difference, but also to detection of properties that the child shares with others. This judgment is accompanied by the assignment of self and others to a common symbolic category. If the shared features are distinctive (e.g., family name, ethnic group, uncommon physical feature) the child is liable to experience a vicarious emotion appropriate to the experience of the other. This combination of processes is called an identification. The girl who is identified with her mother experiences one class of vicarious affect when the mother is praised but a different class of emotion if the parent is criticized or behaves in ways that violate the child’s moral standards. We might call the first state vicarious pride and the second vicarious shame. These states belong to the same family as the pride and shame of the 3-year-old following an action, but I believe it is theoretically useful to regard the vicarious states as different because the incentives and accompanying thoughts are distinct. A person who harms another is vulnerable to guilt; a person identified with a parent will experience shame if the parent harms another.

The passage into puberty is accompanied by three new cognitive abilities and, therefore, new emotional families. These competences belong to the stage Piaget called formal operations. One ability is the detection of semantically based, logical inconsistencies among one’s beliefs. Recognition of these inconsistencies renders the adolescent vulnerable to an emotional state one might call cognitive uncertainty. For example, an adolescent recognizes the inconsistency between his belief that his father is a wonderful man and the fact that the father is a vocational failure. Arthur Miller awarded this state to Biff in “Death of a Salesman.” This state differs from both shame and guilt because the agent has not acted in ways that violate community or personal standards.

The ability to analyze the consistency among a set of beliefs is accompanied by the capacity to arrange the beliefs in a pattern that defines an ideal. When mathematicians say that a proof is beautiful they mean that a set of equations has elegance, simplicity, and the proper level of difficulty in being comprehended. When adolescents combine sexual desire toward another with the belief that the other possesses ideal traits, which often includes physical attractiveness, the state we call love emerges.

A third intellectual competence is the capacity to be convinced, using logic, that one has exhausted all possible solutions to a problem. Younger children are not capable of this conviction because they cannot perform a logical analysis of a set of premises and a conclusion. If the problem involves the adolescent’s security, safety or acceptability to others, the conviction that no coping action is possible can lead to an emotion some might call the depression of hopelessness. This state can motivate a suicide attempt. Because young children are incapable of this conviction, the state of a 7-year-old who tried to kill himself with a knife is not to be compared with the state of a pregnant adolescent who tried to slash her wrists. The affects of the two agents are distinct.

These suggestions have implications for the current practices of psychiatrists, clinical psychologists, and epidemiologists. The most important is that information on cognitive competences, behavior, and physiology should be added to interview evidence in arriving at a psychiatric diagnosis. Currently, the assignment of a person to the nosological category depressive disorder, for example, is based primarily on the patient’s verbal description of their symptoms, and especially their reports of their conscious feelings, because these feelings are salient to the individual. This evidence leads the investigator or clinician to focus on the person’s emotional state and to be indifferent to the patient’s beliefs, living conditions, and cognitive abilities. Most patients who report feeling depressed usually have other characteristics, including substance abuse, a chaotic home, problems in sustaining attention, the perception of being a victim of prejudice, high standards for achievement, identification with a person or group that generates a feeling of doubt, academic failure, lack of friends, or low expectations of gaining desired goals.

These features are potentially as significant for the diagnosis as the verbal description of depressed feelings. However, because the feelings are salient to the patient, and drugs are better able to alter feelings than to change beliefs or living conditions, it is understandable that psychiatrists have made the felt emotion the defining feature of the clinical syndrome. Hippocrates and Galen were wiser for they assumed that the defining feature of melancholia was an imbalance in body humors. An analogy to rashes is useful. The itch that accompanies a body rash is also salient, but the itching is a sign, not the cause, of an ailment that could be the result of an allergy, shingles, measles or other conditions. Selection of the best therapy for the rash requires understanding its etiology.

The central features of depressive disorder are verbal descriptions of sadness, apathy, low energy level, sleeplessness, and poor appetite. If these symptoms occurred in a middle-class 8-year-old who had failed to meet a personal standard on academic achievement promoted by the family, the profile is more properly called a guilt reaction. If the same symptoms occurred in an adolescent from a poor family who, in addition, was failing in school and was convinced of the futility of improving her life, a more correct label is a depression of hopelessness. Finally, some youths who report the same symptoms may have inherited a biological diathesis for depression, a category that used to be called endogenous depression. These three patients may belong to different categories, even though all report sadness and apathy. Although future research may reveal that the names I have given to these three types are not the most accurate, the main point is that the affect state the patient describes can have different origins. The diagnosis should include not only the reported feeling state but also thoughts, cognitive abilities, behaviors, and physiology.

Future studies on emotional development in children, as well as diagnostic classifications, should add assessments of cognitive function, direct behavioral observations to relevant incentives and, if possible, appropriate physiologic measurements. Consider groups of 6- and 14-year-olds described by their parents as apathetic and subject to bouts of depression. It would be useful to gather information on EEG asymmetry, sympathetic tone in the cardiovascular system and behavioral reactions to failure on age-appropriate tasks. Should the two groups differ on these measurements, investigators might question assigning them to the same diagnostic category. Finally, some investigators should initiate a longitudinal study of children at risk for depression, from infancy to adolescence, to see if 1-year-olds who display extreme levels of distress or apathy to separation, neglect or unpredictability are more likely than others to become depressed adolescents. If the predictive relation is affirmed, those who believe that a temperamental diathesis to depressive affect is preserved across development would be supported; however, the view advocated in this paper would be supported if the predictive relation was not realized.

Section snippets

Summary

This article has tried to make two main points. First, the current terms for affect states are flawed because they are unconstrained by class of incentive, cognitive structures and physiology, and each should be replaced with constructs for families of affects that specify the developmental stage of the child and the incentives for the emotion. The emotional states are not Platonic essences defined by particular profiles of brain activity that transcend species, age and local context. The

Acknowledgements

Preparation of this paper was supported in part by grants from the W.T. Grant Foundation and the Bial Foundation.

References (21)

There are more references available in the full text version of this article.

Cited by (0)

View full text