Stress can be defined as any real or perceived threat to a person’s well-being or homeostasis. This is often divided into two broad categories of stressors: physiologic and psychogenic (e.g., Morilak et al., 2005). Physiologic stress follows from any real and imminent threat that requires physical adaptations in order to meet environmental demands and restore homeostasis (e.g., attack or injury). The bodily response to physiologic stress is present regardless of whether a person perceives or is consciously aware of the stimulus. In contrast, psychogenic stress relies upon a person’s perception and interpretation in order to attribute a stressful quality to the stimulus. There is considerable variability in response to psychogenic stressors, or whether an event is interpreted as stressful at all. Many physiologic stressors are also compounded by psychogenic interpretations of the event.

Historically, researchers viewed psychogenic stress as the primary source of individual differences in the stress response. It is now clear that personality and personal history can also influence the likelihood and frequency of exposure to physiologic stressors (Hammen, 1991, 2005). Furthermore, people vary widely in their exposure to stress, the magnitude of their physiological and emotional responses to adverse circumstances, the duration of their recovery following a stressor, the length of time needed to recover following negative life events, and whether there is adequate restoration during or between times of stress (Williams, Smith, Gunn, & Uchino, 2010). Examination of individual differences across these four component processes—exposure, reactivity, recovery, and restoration—may inform our understanding of the mechanisms by which adversity is associated with negative health consequences (Hawkley & Cacioppo, 2003; Uchino, Smith, Holt-Lunstead, Campo, & Reblin, 2007; Williams et al., 2010).

There are many mechanisms that appear to account for the lasting effects of stress on health and well-being. These mechanisms can be best understood within a conceptual framework that differentiates between those stressors that lead to temporary disruptions in homeostasis versus those that produce more lasting biological adaptations. This second process is termed allostasis—the process by which the body achieves stability through change (e.g., Sterling & Eyer, 1988). Whereas homeostatic processes promote stability via brief, temporary adjustments within the typical operating ranges of physiological systems, allostatic processes produce long-term alterations within these operating ranges and are invoked to maintain stability in the face of extreme or prolonged stress (Beauchaine, Neuhaus, Zalewski, Crowell, & Potapova, 2011; Lupien et al., 2006; McEwen, 1998).

In this review, we examine some of the biological adaptations (i.e., allostasis) that are associated with protracted distress. We also discuss how individual variability in the four primary stress processes: exposure, reactivity, recovery, and restoration could contribute to the emergence of adaptive or maladaptive emotion regulation strategies. There is strong evidence to suggest that adverse experiences, particularly in early development, can produce lasting alterations to the limbic-hypothalamic-pituitary-adrenal axis (LHPA), neurotransmitter and neuropeptide functioning, brain structure, and gene expression (e.g., de Kloet, Joels, & Holsboer, 2005; Mead, Beauchaine, & Shannon, 2010). In turn, these biological adaptations can affect emotion regulation and health.

Emotions are automated, rapid, and dynamic response tendencies that have been preserved across evolution to promote survival behaviors (Cole, Martin, & Dennis, 2004; Ekman, 1992; Ekman & Friesen, 1976; Gross, 1998b). Understanding emotional processes is important to the study of adolescent health, given that many behavioral problems are characterized by poor modulation of affect (e.g., Beauchaine, Gatzke-Kopp, & Mead, 2007). However, assessing the regulation of emotion presents conceptual and measurement challenges, because most regulatory efforts are internal and unobservable. In contrast, emotion dysregulation is often reflected in quantifiable behaviors. Accordingly, we give considerable attention to the construct of emotion dysregulation, which captures “patterns of emotion regulation that have acquired a maladaptive quality, such that emotions seem to interfere with functioning” (Cole & Hall, 2008, p. 266).

There is a growing literature on emotion dysregulation, psychopathology, and health behaviors (Davidson, 2000; Oshri, Rogosch, Burnette, & Cicchetti, 2011). However, there are inconsistencies across studies in terminology, measurement, and level of analysis (e.g., behavioral, physiological; Adrian, Zeman, & Veits, 2011; Cole et al., 2004; Goldsmith & Davidson, 2004; Gross, 2007). According to one conceptualization, dysregulated emotion can be distinguished from typical emotional responses by four key characteristics (Cole & Hall, 2008). First, dysregulated emotions are prolonged, and regulatory attempts are ineffective (e.g., anxiety, depressed mood, irritability, and anger can be difficult to modulate for adolescents with certain psychiatric diagnoses). Second, dysregulated emotions interfere with appropriate social and goal-directed behaviors (e.g., attending school). Third, dysregulated emotions are often expressed in inappropriate social contexts or when the emotion is unshared by others. Finally, the onset of dysregulated emotions can be too abrupt and/or recovery is too slow. Poor emotional recovery (e.g., dysphoria that does not respond to situational changes) and emotional lability (e.g., unpredictable and quick changes of mood) are common among adolescents with psychological problems.

Consistent with this framework, emotion regulation and dysregulation have almost always been defined as individual processes. Regulation is thought to occur internally, when a person influences which emotion is felt, when the emotion occurs, or how it is experienced and expressed behaviorally (Gross, 1998a). Similarly, emotion dysregulation is viewed as a person’s failure to modify emotions in any or all of these ways, often leading to context-inappropriate affect and/or behavior (Gratz & Roemer, 2004). Thus, most researchers operate on the assumption that emotion regulation and dysregulation occur within, rather than between individuals.

We suggest that emotion regulation is better understood as an individual and interpersonal process (Hughes, Crowell, Uyeji, & Coan, 2012). Because we are a social species, humans are adapted to use social proximity as the default strategy for all metabolically intensive activities, including emotion regulation. This hypothesis has been articulated within social baseline theory (SBT), which suggests that all people are “hard-wired to assume close proximity to conspecifics, and to utilize social proximity as a baseline affect regulation strategy” (Coan, 2008; Coan, 2010, p. 213). From this perspective, healthy and resilient outcomes may also be due to effective co–regulation across the life span, rather than individual regulatory efforts alone. This may be particularly true for adolescents who are transitioning from adult to mostly peer sources of co-regulation (Hughes et al., 2012).

Resilience has also been defined and measured variably across studies. However, nearly all definitions emphasize adaptive or competent functioning across one or more domains, despite significant adversity or trauma (Cicchetti & Blender, 2006; Curtis & Cicchetti, 2003; Luthar et al., 2000; Shannon, Beauchaine, Brenner, Neuhaus, & Gatzke-Kopp, 2007). This has been operationalized as a dynamic developmental process where a person achieves “desirable outcomes in spite of significant challenges to adaptation or development” (Masten & Coatsworth, 1995, p. 737). Such challenges are often operationalized by experiences of neglect, poverty, or exposure to severe or prolonged abuse (Cicchetti & Blender, 2006).

Positive adaptation and competence are often measured at the individual level and can include positive developmental outcomes in spite of stressful life circumstances (e.g., academic success, strong social skills, effective emotion regulation strategies), or a surprising absence of problems (e.g., lack of psychopathology despite a strong familial loading; Shannon et al., 2007). Because such outcomes are often assessed within the child or adolescent, it is tempting to view resilience as constellation of personality traits or innate abilities. However, resilient functioning is a dynamic, collaborative, and fluctuating product of biology–environment interactions across development (Masten, 2004).

A challenge of resilience research is that adaptive functioning is often discussed in relation to an expected—but hypothetical—negative outcome. For this reason, resilience can only be understood properly when there is significant risk exposure. Without such adversity, resilience is difficult to distinguish from a wide range of typical outcomes and the construct loses much of its meaning (see Shannon et al., 2007). Although resilience is often defined broadly, our focus here is on the development of emotion regulation skills or the absence of psychopathology, when clinical problems characterized by emotion dysregulation could be expected.

The literatures on stress, resilience, and emotion regulation have partially independent traditions. Yet there are clear advantages to conceptualizing these phenomena within a unified framework. The term resilience often implies that a person has acquired at least some emotion regulation skills in spite of significant stress exposure (Buckner et al., 2003). More importantly, resilience and emotion regulation are both complex developmental processes that can be altered subtly or dramatically through stress exposure (Charney, 2004). There are many common themes across these three constructs. However, we find an interpersonal understanding of stress, emotion regulation, and resilience to be well suited for understanding how these three processes interact and develop across time. By adolescence, interpersonal stressors are key precipitants of depression, anxiety, aggression, and suicidality (Clarke, 2006; Hilt, Cha, & Nolen-Hoeksema, 2008; Johnson et al., 2002). Similarly, both resilience and emotion regulation can be conceptualized more effectively as transactional, rather than solely individual processes (Masten, 2001).

An interpersonal perspective has advantages over a strictly individual-differences approach and is more consistent with the current understanding of resilience (Cicchetti & Blender, 2006). Early work on resilience implied that some children had an innate capacity to withstand adversity (see Masten, 2001 for a review). This innate capacity was hypothesized to emerge through a combination of high intelligence and an easy disposition. Criticisms of this perspective are twofold. First, there are no public heath interventions that can readily affect these highly heritable aspects of early infant temperament. Second, personality characteristics are neither inherently problematic nor beneficial. Rather, the same trait could confer vulnerability or resilience depending upon contextual risk and/or protective factors. More recent definitions of resilience assume that interactive processes shape adaptive outcomes and, therefore, neither risk nor resilience can be attributed to purely heritable or environmental sources of transmission (Shannon et al., 2007).

As alluded to above, a similar conceptual shift can be identified in the literature on emotion dysregulation. Over the past two decades research on emotion regulation/dysregulation has increased exponentially (e.g., Cole et al., 2004; Cole, Michel, & Teti, 1994). However, there are inconsistencies across studies in the assessment and interpretation of emotion regulation/dysregulation. By and large, developmental researchers have examined emotional processes within dyadic or other complex family systems (e.g., Crowell et al., 2008; Deater-Deckard & Petrill, 2004; Fosco & Grych, 2008; Sroufe, Duggal, Weinfield, & Carlson, 2000). Researchers studying older adolescents or adults have traditionally pursued an individual-differences approach to evaluating regulatory successes and failures, often paying minimal tribute to the interpersonal context in which regulation or dysregulation occur (e.g., John & Gross, 2007).

From a life span development perspective, emotion regulation and resilience can be viewed as transactional, interpersonal processes that emerge initially within early attachments and continue through peer relationships and pair bonding. Stress or threat within key childhood attachment relationships can produce lasting alterations to the biological systems involved in stress reactivity, recovery, and restoration (Schore, 2001). Across development these biological adaptations may also affect internal working models of key relationships, alter peer bonding behavior, and may lead to insecure expectations of co-regulation by adolescence or adulthood (Diamond, 2001; Hughes et al., 2012). Even though attachments are often viewed as noncentral to the daily functioning of adults, it is well established that healthy relationships in adulthood can reduce cardiovascular arousal, lower glucocorticoid levels, attenuate threat-related brain activation, and promote health and longevity (see Coan, 2008 for a review). In contrast, interpersonal stressors may shape and maintain problems with emotion regulation and health from infancy through adulthood.

Stress reactivity is the rapid cascade of physiological, cognitive, emotional, and/or behavioral responses that follow a stressor and shift an organism into a heightened state of biological and behavioral preparedness (Williams et al., 2010). A common assumption is that heightened stress reactivity represents a maladaptive “holdover” from a period of our evolutionary past when stressors were more frequent and lethal (see Boyce & Ellis, 2005, for a review). In our present context this vulnerability is often viewed as problematic, leading to psychopathology and health problems for highly reactive individuals.

Yet, accumulating evidence reveals that, among children, heightened reactivity alone is not associated consistently with poor health. Rather, highly reactive children may experience greater or fewer health problems depending on their developmental context. In one study, high-reactive children reared in stressful home or childcare environments also experienced the highest levels of respiratory illness. However, the high-reactive children who were in protective environments were the healthiest of all. Across both high- and low-risk settings, the low-reactive youth showed no effect of context on their health (Boyce et al., 1995). Such findings suggest that high reactivity in children can confer risk or benefit depending upon environmental factors.

Further complicating matters, the relation between reactivity and psychological outcomes may depend upon the timing and number of biological assessments. For example, children with high salivary cortisol early in the school year (when peer groups are forming) but normal levels later in the year (when peer groups were established) were rated as outgoing, well liked, and socially competent; those children who maintained high cortisol reactivity or who changed from low/normal to high levels were rated as solitary and had greater negative affectivity (Gunnar, Tout, de Haan, Pierce, & Stanbury, 1997). These findings imply that heightened reactivity may be beneficial initially, but problematic if sustained—especially when the environmental context is no longer novel or threatening. Stressors, such as starting school, likely elicit strong homeostatic responses. However, if sustained, allostatic processes may be invoked, leading to long-term biological adaptations in stress response systems.

There are also differences in stress reactivity that emerge in studies that include repeated assessments across early development. A common pattern is one in which neurobiological systems are initially hyper-reactive to extreme stressors (e.g., maternal separation, chronic exposure to a dominant male) then, following prolonged stress, shift to a chronically hypo-reactive baseline state (Mead et al., 2010; Miller, Chen, & Zhou, 2007; Susman, 2006). Abused monkey infants, for example, display elevated levels of cortisol during the first month of life, the period when infants receive the harshest treatment from mothers. Over time, the abused infants have low levels of cortisol, particularly in the morning (see Loman & Gunnar, 2010). A similar down-regulation has been found in studies of rodents who were separated repeatedly from their mothers (Meaney, Brake, & Gratton, 2002) or who lost in an aggressive interaction with a dominant male (Covington & Miczek, 2005). In both studies, decreased dopamine transporter binding led to biobehavioral differences in the stress response and enhanced behavioral sensitivity to drugs of abuse (i.e., the stress-exposed rodents also self-administered greater amounts of cocaine). Taken together, these studies reveal a pattern of hypo-reactivity to prolonged stress exposure. Such dampened responses may reflect allostatic adaptations to chronic adversity.

Unfortunately, the direction of biological adaptations to stress is not consistent across studies, and chronic hyperactivation is also a common pattern (e.g., Sanchez et al., 2010). This may be because measurements are affected by the time of day the assessments are administered, the length of time post-stressor, the nature of the stressor, the developmental timing of adversity, the genetic vulnerabilities of the stress-exposed individual, the contextual factors that either exacerbate or buffer against adversity, and the measure of “reactivity” itself (see Cicchetti, Rogosch, Gunnar, & Toth, 2010). Furthermore, child temperament can also affect the direction of biological responses to adversity. In one study, toddlers were coded as either inhibited/vigilant (i.e., “doves”) or bold/aggressive (i.e., “hawks”; Davies, Sturge-Apple, & Cicchetti, 2011). For dove children, interparental aggression predicted greater cortisol reactivity over a 1-year period. Conversely, hawk children showed diminished cortisol reactivity over the year. Dove children also showed increased internalizing symptoms but decreases in attention/hyperactivity problems whereas hawk children were more likely to develop problems with attention/hyperactivity.

Thus, the research linking stress exposure to stress reactivity and then to subsequent health problems paints a complicated developmental picture. Although several unifying theories are beginning to emerge (e.g., Miller et al., 2007), a life span theory is premature due to inconsistencies in the definition and measurement of key constructs, differences in samples, and a paucity of life span longitudinal studies. The research reviewed here suggests a developmental trajectory in which high reactivity can confer risk or benefit depending upon both infant temperament and the environmental context (Boyce & Ellis, 2005; Davies et al., 2011). At typical levels of adversity, there appears to be a linear relation with health outcomes—higher reactivity often predicts greater health risk. At either end of the distribution, however, the findings are less consistent. Youth who have been exposed to extreme adversity often show a down-regulation across biological measures of reactivity, which may be due to hypervigilance or orienting (which lowers heart rate; Porges, 1995), or it could be due to the nature of laboratory stress paradigms, which are often mild when compared with the stressors of daily life. In contrast, youth raised in exceptionally protective environments may show high reactivity to laboratory stressors. However, this reactivity could show no relation or a negative association with health problems (Boyce et al., 1995). It is clear that researchers will need to examine the moderating effects of temperament on reactivity for youth raised in both high- and low-risk environments.