Example: Attachment system(s)

This point may be illustrated by the study of social attachment. Attachment is typically assessed through patterns of responses to a behavioral paradigm wherein the infant is separated and reunited with his or her parent. The ingenuity of this approach is the focus on how the infant organizes a broad range of complex behaviors (e.g., affect expression, social interaction, exploration, attributions, comfort, wariness) in response to changing circumstances. Because this approach relies exclusively on the measurement and categorization of overt behaviors, little attention has been directed toward isolating the constituent developmental processes underling the behaviors. For example, security and distress behaviors might reflect distinct neural systems (cf., Bischof, 1975). There are other examples of psychological constructs that, like security and distress, appeared to be opposite poles of the same system, but turned out to have distinct neural circuitries. For example, hunger–satiety and positive–negative affect (Critchley & Rolls, 1996) first seemed to be opposite poles of the same continua, but are based on distinct neural systems.

What might such constituent attachment systems look like? Ancillary evidence suggests that infants may have a system of neural circuitry that becomes activated in response to fear (of being left alone, of separation; Caldji et al., 2004; Kalynchuk & Meaney, 2003). Such a fear system might protect the infant from the environment, motivate proximity to caregivers to deal with stress, or generate cries that alert the caregiver to perceived threat over the course of development (LeBar & LeDoux, 2003). Such a fear system might become coordinated with a distinct affiliation system that motivates social contact including shared affect between infant and caregiver, elicitation of joint attention, and social responsivity.

A number of studies have implicated neurohypophyseal peptides, such as the oxytocin (OT) system, as a neurobiological mechanism responsible for the regulation of complex social behaviors such as affiliation and parental care (Carter, Braver, Barch, Botvinick, & Noll, 1998; Insel, 1992, 1997). OT is a peptide hormone that is produced in the hypothalamus and released centrally and peripherally into the blood stream via axon terminals in the posterior pituitary (Kendrick, Keverne, Baldwin, & Sharman, 1986). Quite distinct from hormonal markers of the fear system, OT receptors appear to be part of the neural system of reward circuitry that includes the nucleus accumbens (NAC; Lovic & Fleming, 2004). In the prairie vole, a monogamous rodent, higher levels of OT are associated with decreases in stress hormones and increases in positive social interactions and attachment behaviors (Witt, Carter, & Walton, 1990). In additional, OT appears to facilitate the consolidation of social memories, a key component of selective social attachments (Popik, Vetulani, & Van Ree, 1992). In contrast, hormonal markers of fear response systems appear to impair memory functions (Erickson, Drevets, & Schulkin, 2003).

Most research on the functions of the OT system has been conducted with nonhuman animals, and evidence for a neurobiological marker of attachment-related behavior in humans remains compelling, but inconclusive (Wismer Fries, Ziegler, Kurian, Jacoris, & Pollak, in press). However, these animal studies suggest that the neurobiological circuitry involving affiliative behaviors may be one subcomponent of social attachment in humans. Moreover, a central question is how reward circuitry becomes linked with the circuitry of the fear system. Multiple systems (including OT, dopamine [DA], estradiol, arginine vasopressin, serotonin [5-HT], and progesterone) are implicated in feedback loops involving reward circuitry and the regulation of social behaviors in the brain (Insel & Fernald, 2004). Thus, the interesting question is not which neuropeptides are activated during attachment behaviors, but how a social bond develops and becomes tied to a response system (Insel, 2003). One possibility is that because social attachment requires that stimuli become linked to major information streams, early adversity might affect attachment-related behaviors through corticolimbic pathways.

Both DA and 5-HT pathways within the mesolimbic system and prefrontal cortex (PFC) are particularly sensitive to adversity (e.g., Braun, Lange, Metzger, & Poeggel, 1999; Ziabreva, Schnabel, Poeggel, & Braun, 2003). It has been demonstrated that alterations of monoaminergic receptor densities in the medial PFC, amygdala, hippocampus, and NAC may induce lasting effects on behavioral flexibility and regulation (Jinks & McGregor, 1997; Ragozzino, Detrick, & Kesner, 1999). However, there are multiple ways in which breakdowns within corticolimbic circuitry could affect an individual's ability to use contextual cues to guide behavior (Dawson, Hessl, & Frey, 1994; Heidbreder, Weiss, Domeney, Pryce, Homberg, Hedou, Feldon, Moran, & Nelson, 2000). For example, the ventromedial cortex (including the orbitofrontal cortex [OFC]) is critical for associating incoming stimuli with existing response–reinforcement contingencies (Adolphs, 2002; Bechara, Damasio, & Damasio, 2000). The medial PFC provides goal-directed motor plans selected within the NAC on the basis of contextual and emotional associations from both the hippocampus and amygdala (Grace, 2000). Therefore, a focus only on one particular brain region or only on the complex behavioral output of the attachment system is unlikely to reveal how various aspects of information become integrated into a coordinated response system or the means by which such a system may be changed by experience.

Influences of “associated” processes

In addition to issues of fear and affiliation, other affective, perceptual, cognitive, and motor processes are activated in situations in which attachment-related behaviors are observed and measured. A focus on the development of these brain–behavior processes can clarify which systems are coordinated with attachment behaviors.

One example of associated processes is the communication, signaling, and decoding of emotional signals that occurs in situations in which attachment is observed. Selective impairments following brain damage suggest that different neural systems may support how individuals communicate what they are feeling to others, generate their emotional states, and recognize and encode emotional signals from others (Heilman, 1994). An early stage in the processing of social information involves the child's perception of salient signals from the environment, including facial expressions, tone of voice, and body language. Most research on emotion perception has been directed at the network of cortical circuitry involved in face perception (De Renzi, 1997).

Elegant work by Kanwisher and colleagues (e.g., Kanwisher, McDermott, & Chun, 1997) has demonstrated that regions in the fusiform gyrus respond more to faces than other animate regions such as human hands. Similar brain regions emerge when subjects view dynamic (as opposed to static) facial images (Sato, Kochiyama, Yoshikawa, Naito, & Matsumura, 2004). This research raises two compelling issues that bear upon the development of psychopathology. The first issue is that studies of mature subjects do not address the developmental origins of neural architecture. Thus, the fusiform gyri might be prepared to processes faces specifically; or this brain region may not begin as face specific but instead may be particularly sensitive to experience and expertise with salient stimuli such as facial expressions (Tarr & Gauthier, 2000). A second question concerns specificity of the systems used to recognize faces (face vs. hands), emotions (angry vs. sad), and identities (mother vs. stranger). Rather than difficulty processing faces per se, children who have experienced early adversity show atypical patterns in decoding specific emotional signals (e.g., anger vs. sadness, or mild vs. intense anger) in the human face (Pollak, 2004).

Less frequently studied, but also central to aspects of emotion communication, is children's understanding of tactile, gestural, olfactory (e.g., pheromones), and auditory affective cues. The human voice contains in its acoustic structure a wealth of information on the speaker's identity and emotional state, yet little is known about the neural basis of vocal emotion communication. A recent study with adult subjects found that voice-selective regions along the upper bank of the superior temporal sulcus showed greater neuronal activity in response to vocal sounds, whether speech or nonspeech, than to nonvocal environmental sounds. These regions selectivity responded more to human voices than to matched control stimuli (Belin, Zatorre, Lafaille, Ahad, & Pike, 2000). These voice-selective areas in the superior temporal sulcus may represent the counterpart of the face-selective areas in human visual cortex and may be critical for understanding children's perception of auditory emotional signals. Indeed, recent electrophysiological studies of severely abused children revealed that auditory anger cues from their own mothers engaged more attentional resources in these children relative to controls (Shackman & Pollak, 2005). Thus, questions remain about whether and how the type and amount of emotional experience children receive affect the neural systems that subserve emotion processing.

A second example of processes that are inextricably tied to social attachment is attention. Within a social encounter there are countless cues that the developing child must attend to including interoceptive information (e.g., subjective feeling states: distress, pain, comfort, anxiety, pleasure), contextual features (familiar vs. unfamiliar settings or people), and the social signals being expressed by others. To successfully select and utilize these cues, the child will rely upon attentional processes that include orienting to relevant stimuli, filtering of irrelevant stimuli, shifting attention between key features in the environment, and sustaining attention over time. Through processes of selective attention, some sensory inputs will become more readily available to influence memory and behavior (Egeth & Yantis, 1997; Posner, 1994).

Regions of the parietal and frontal lobes appear to be involved in mediating various aspects of attentional control (for review, see Goldberg, Bisley, Powell, Gottlieb, & Kusunoki, 2002). For instance, the parietal cortex may represent a general-purpose mechanism for directing visual attention (Wojciulik & Kanwisher, 1999) including the abilities to orient attention toward sensory stimuli while ignoring other stimuli (Mirsky, 1996). Areas of the PFC also play a role in selective attention and filtering of irrelevant information. For example, patients with lesions of the lateral PFC have difficulty focusing attention while filtering out irrelevant information (Gehring & Knight, 2002). Dopaminergic systems linking the frontal lobes with midbrain regions appear to play a role in how attentional responses are related to behavioral regulation (King, Tenney, Rossi, Colamussi, & Burdick, 2003; Pruessner, Champagne, Meaney, & Dagher, 2004), but little is known about the development of these systems in humans.

Less is currently known about how attentional networks become differentially activated in response to emotion-related information. Using Stroop-type tasks, Compton, Banich, Mohanty, Milham, Herrington, Miller, Scalf, Webb, and Heller (2003), found that the left dorsolateral PFC and medial PFC are involved in maintaining attentional set for both neutral and emotional stimuli, suggesting a common selective attention system. However, this same study demonstrated that areas of the right occipitotemporal region are recruited when emotional (compared with neutral) stimuli were to be ignored, suggesting a specialized emotion filtering system. In addition, anterior regions of the anterior cingulate cortex (ACC) appear more active in response to attended emotional stimuli (Fichtenholtz, Dean, Dillon, Yamasaki, McCarthy, & LaBar, 2004) and when subjects attend to their own internal emotional responses (Lane, Fink, Chau, & Dolan, 1997). The roles of these processes in forms of affective regulation, such as attachment, require further examination.

Summary

A challenge in attempting to understand the neural processes underlying children's socioemotional behaviors is that we often lack clarity or specificity about the particular affective processes in question. Most research methods lack the specificity necessary to determine links to discrete neural systems. This point is illustrated by the construct of social attachment. This complex phenomenon is likely to emerge from multiple processes including (but not limited to) biobehavioral systems involved in affiliation and fear regulation, affective processes such as the expression and perception of multimodal emotion cues, and attentional processes. To advance understanding of the developmental basis of attachment problems in children, it is necessary to discern how these various systems come to function in synchrony. Through which mechanisms do various emotional processes become relatively activated with reference to each other? An affective neuroscience approach may be useful in excavating the neural foci of activation, the network of functional connectivity, and the interactions of various networks associated with both separation distress and attachment/security. Rather than viewing such an approach as biological reductionism of complex psychological processes, such a perspective, in tandem with traditional approaches, allows us to address a new generation of compelling developmental questions.

Linking biology with experience

Two kinds of information appear critical for understanding processes of change: the constraints of the developing nervous system and the sources of plasticity within that same system. Even the most powerful young learners are not blank slates. Ultimately, affective plasticity or learning must be explained by a combination of biological predisposition and experiential input.

To begin to unpack how affective systems can change in response to different kinds of experiences, measurement of the effects of early adversity on child development must now become more nuanced. In the past, it was sufficient to select study populations based on whether children did or did not experience some form of adversity, such as child maltreatment. However, now it is necessary to parse the effects of variations in adverse experiences on developmental outcomes. For example, factors such as the intensity/severity, chronicity/duration, and developmental timing of experience may be critically related to the development of different forms of psychopathology (cf., Manly, Kim, Rogosch, & Cicchetti, 2001). It is not trivial to measure a child's experience, nor is it clear that there are currently solid standards for evaluating how adverse an experience was for a given child or precisely how long the experience lasted. However, with the exception of some work done on rats, little is presently known about sensitive periods for the impact of experience on the stress and emotion–regulatory systems. Moreover, even in the existing rodent models, there is little specificity about what constitutes the underlying mechanisms.

The value of an affective neuroscience approach is the possibility of linking a child's experience, physiology, behavior, and genes. Indeed, the kinds of questions that need to be answered to address public health questions are unlikely to be addressed from only one level of analysis. For example, more precise models are necessary to determine what kind of events lead to long-term alternations in neural systems and how to define individual differences in the thresholds after which environmental influences begin to dysregulate affective systems. To advance research in these areas, we need not only to identify gene polymorphisms in clinical populations but also to address what the gene is doing: how it causes or maintains disorders, how it directs or constrains an organism's interaction with the environment, whether it has the same function early in development as in adulthood, and whether the gene functions similarly across species. However, genetic analyses in isolation are unlikely to advance our understanding of psychopathology. Whatever the sequential particulars of the three billion DNA subunits that make up a child's genotype, they are meaningful only in the context of phenotype, of a child's life history and experiences. Developmental approaches will need to consider how experiences turn on and off certain genes and through which mechanisms genetic polymorphisms act in combination with “stressful” experiences to produce psychopathological syndromes (see, e.g., Caspi et al., 2003).

Candidate mechanisms of plasticity

Social stimuli are diverse and complex. Therefore, the developing child's behavioral responses need to be guided by some representations based upon past experiences. Networks involving the ventromedial cortex and OFC are critical for processes such as associating incoming stimuli with existing response contingencies, linking stimuli with motivational significance, and for holding these representations in working memory as behavioral responses are selected (Adolphs, 2002; Gottfried, O'Doherty, & Dolan, 2003; O'Doherty, Kringelbach, Rolls, Hornak, & Andrews, 2001). Animals with lesions to these systems demonstrate impaired object–reward association memory (Gaffan & Parker, 2000) and impaired contextual fear conditioning (Li, Inoue, Nakagawa, & Koyama, 2004). Which systems might explain the associations between early adversity and the emergence of socioemotional behaviors?

One candidate is the network involving the medial PFC. In particular, the ACC has extensive reciprocal connections with the limbic system via the hippocampal formation, the shell region of the NAC and the amygdala. Part of this network is likely to involve dopaminergic and serotonergic modulation of cortical activities related to socioemotional behavior (Ziabreva et al., 2003). These cognitive–affective interactions are best understood through work with nonhuman primates. The amygdala is implicated in the processing of social information (Adolphs, Tranel, Hamann, Young, Calder, Anderson, Phelps, Lee, & Damasio, 1999; see Aggleton & Young, 2000, for review). In primates, the amygdala projects directly to the medial thalamic nucleus, which in turn projects to the ventromedial PFC. This circuit appears to participate in learning and memory-related processes (Mishkin, Malamut, & Bachevalier, 1984). Early reports of amygdala lesions in monkeys described profound social and emotional changes, including problems appreciating the emotional significance of stimuli (Kluver & Bucy, 1939). Similarly, it has been reported that human patients who have suffered bilateral amygdala damage display deficits in the recognition of facial expressions of emotion (Adolphs et al., 1999; Broks et al., 1998; Calder, Young, & Rowland, 1996). These subjects show normal performance on tests of visuoperceptual processing of facial features, suggesting that the impairment is specific to the reading of emotional information (Adolphs, Baron–Cohen, & Tranel, 2002).

One brain region is unlikely to be responsible for the range of complex emotional behaviors. Yet, there are possible mechanisms through which the amygdala contributes to deficits in social behavior. One hypothesis is that the amygdala plays a critical role in stimulus–reinforcement conditioning, whereby neutral cues acquire positive and negative incentive status and emotional meaning (Aggleton & Young, 2000; Bechara, Tranel, Damasio, Adolphs, Rockland, & Damasio, 1995; Emery & Amaral, 2000; Gottfried et al., 2003; LeDoux, 1996). Consistent with such a view, individuals with lesions to the amygdala show poor performance on associative learning tasks involving emotional expressions (Boucsein, Weniger, Mursch, Steinhoff, & Irle, 2001). In addition, monkeys with lesions of the amygdala do not show the usual pattern of initial wariness and situation evaluation in dyadic interactions (Emery, Capitanio, Mason, Machado, Mendoza, & Amaral, 2001).

It is likely that this circuitry also informs issues of learning and development. As a case in point, a failure to perceive potential threat in the environment was found in infant monkeys who had received amygdala lesions as neonates (Bauman, Lavenex, Mason, Capitanio, & Amaral, 2004). In contrast to control and hippocampus-lesioned monkeys, amygdala-lesioned monkeys did not maintain proximity to their mothers. Interestingly, the mother–infant interactions appeared normal for the first few months of the infants' lives. A compelling aspect of these data is that the amygdala deficits were not apparent until the infant had opportunities for situational and contingency learning to occur. Thus, the amygdala may allow for proper evaluation of the situation through utilization of current sensory information about the partner, current contextual information, and information about past experiences.

Repeated stressful experiences appear to alter the balance of both excitatory and inhibitory inputs in the limbic system (Poeggel, Helmeke, Abraham, Schwabe, Friedrich, & Braun, 2003). Corticolimbic afferents from areas such as the amygdala, hippocampus, and PFC carry contextual information that may cause release of DA and facilitate approach to rewarding objects. However, the basolateral amygdala system may also receive stress signals from the ventral tegmental area (VTA), facilitating learning through DA neurons within VTA, NAC, and other limbic and cortical areas (Datla, Ahier, Young, Gray, & Joseph, 2002; Stevenson & Gratton, 2003). Another possibility is that DA in the medial PFC allows animals to attend to behaviorally relevant stimuli (Passetti, Dalley, & Robbins, 2003). Impairments of DA within the basolateral amygdala would offset an organism's ability to appraise situations, thereby leading to inappropriate behavioral responses in situations of uncertainty or risk. Such a view is consistent with data suggesting that dopaminergic mechanisms may impair PFC functioning during stress, resulting in behavioral responses that are mediated more by subcortical than by prefrontal systems (Arnsten & Goldman–Rakic, 1998). On this view, the medial PFC would provide goal-directed motor plans generated from both the hippocampus (context) and amygdala (affective state). Dysfunction within this circuit might render the individual unable to combine environmental cues efficiently, leading to inappropriate behavioral responses. If correct, this theory would link chronic stress to problems such as attention deficit and hyperactivity (Heidbreder et al., 2000) and account for how DA may trigger autonomic and neuroendocrine responses to stress (Abercrombie, Keefe, DiFrischia, & Zigmond, 1989).

Although the DA system acts as a learning signal for behavioral reinforcement (Schultz, 1998), the numerous reports of DA release in response to aversive stimuli in animals are difficult to fit into theories of reward. These seemingly contradictory findings have lead to the suggestion that DA is involved in motivation and attention underlying behavioral responses to important events (whether aversive or appetitive; Berridge & Robinson, 1998; Redgrave, Prescott, & Gurney, 1999; Salamone, Cousins, & Snyder, 1997). A recent study suggested that repeated social stress affected DA binding in the NAC (which projects to the VTA) and also affected animals hypothalamic–pituitary–adrenocortical (HPA) responses (Lucas, Celen, Tamashiro, Blanchard, Blanchard, Markham, Sakai, & Mcewen, 2004). Indeed, the effects of early stress appear to be specific to the ascending DA systems (e.g., DA transporter) and nonspecific in terms of alterations in CRF expression in the PVN of the hypothalamus, differences in pituitary ACTH, and release of glucocorticoids (Meaney, Brake, & Gratton, 2002). Early adversity appears to alter the development of the HPA axis and extrahypothalamic corticotropin-releasing hormone (CRH; Shea, Walsh, MacMillan, & Steiner, 2004). Behaviorally, such changes are reflected as an individual's diminished capacity to maintain homeostasis if challenged by adverse events. However, critical unanswered questions remain such as whether such atypical responsivity increases the risk of psychopathology, whether such disorganization following adversity occurs in genetically vulnerable individuals, whether hypo- versus hyperresponsivity may change over development (DeKloet & Oitzl, 2003), and the effects of different kinds of adversity on these processes.

One possibility is that early stress exaggerates a typical developmental process such that chronically elevated glucocorticoids speed the overall reduction of neuron densities in core regions of the limbic system (Poeggel, Lange, Hase, Metzger, Gulyaeva, & Braun, 1999). However, the direction of causality in these findings is not clear. Stress-induced DA release is partly influenced by circulating cortisol levels (Marinelli & Piazza, 2002), which also increase during stress. Pruessner et al. (2004) have demonstrated that psychological stress causes DA release in humans. A lab-induced psychological stressor caused a significant release of DA in the ventral striatum as measured by positron emission tomography in adults who reported that they received poor parental care in childhood (Pruessner et al., 2004). This model may be particularly helpful in understanding long-term behavioral responses to stress such as depression and posttraumatic stress disorder.

Summary

The neurophysiological models presented here are certainly underspecified and overly simplistic. However, these recent findings are useful in thinking about the mechanisms through which experience may influence individual developmental organization. Such models will become substantially more complex and sophisticated as further studies on mechanisms are completed. The key is that in defining a mechanism of development, whatever the level of analysis, the focus must be on processes or mechanisms that are responsible for or driving developmental change. Those developmental changes may be experience-independent maturational processes or may be dependent upon experience/environmental input. Understanding the ways in which the environment affects developmental organization is central to a developmental approach to addressing etiological and treatment issues in psychopathology.

Summary

At present, there is little empirical data upon which to define commonly used regulatory terms or to disambiguate between various regulatory processes. Undoubtedly, multiple physiological systems are involved in mediating the developmental impact of early adversity on the individual's ability to regulate their emotional behavior. Although the CRH/HPA system has been studied the most, other systems such as thyroid, 5-HT, growth hormone, and the neurobiological processes regulating sleep and diurnal rhythms have received less attention from developmental psychopathologists and will need to be integrated into models of risk (cf., Dahl, 1996). Thus, an affective neuroscience perspective may aid in specifying the processes that are most central for understanding the risks associated with early adversity. An affective neuroscience approach is likely to clarify the processes through which early adverse experiences result in developmental changes in emotion regulatory abilities. Nonhuman animal data indicate that many brain systems undergo protracted periods of postnatal development. These periods of time in which neural systems are undergoing organization may allow windows of vulnerability or experience-dependent fine-tuning of key regulatory systems including attention, learning, emotion, and memory (e.g., Black, Jones, Nelson, & Greenough, 1998). Experience-dependent changes in regulatory signaling between brain regions, through dopaminergic and serotonergic systems (Goldman–Rakic, Bourgeois, & Rakic, 1997), appears important for understanding plasticity of the neural circuitry involved in emotion regulatory processes.