Abstract
Negotiating the transition from dependence on parents to relative independence is not a unique demand for today's youth but has a long evolutionary history (transmission) and is shared across mammalian species (translation). However, behavioral changes observed during this period are often described as delinquent. This review examines changes in explorative and emotive behaviors during the transition into and out of adolescence and the underlying neurobiological bases in the context of adaptive and maladaptive functions.
Introduction
Adolescence is a transitional period from childhood to adulthood, with an onset that includes pubertal maturation and an offset that is marked by independence from the parent. The paradox observed for human adolescents is that while they are stronger, faster, and more resistant to disease, and have better reasoning and decision-making skills than children, mortality increases 200% for them during this time (Dahl, 2001). This increase in deaths is not attributable to disease, but rather to preventable forms of death including accidental fatalities, suicide, and homicide (Centers for Disease Control, 2007) and is related to difficulties in the regulation of behavior and emotion (Steinberg, 2008).
To understand this paradox, this review highlights central themes of the NIH Blueprint for neuroscience research on neurodevelopment (National Institutes of Health, 2006). This report underscores the importance of “periods of rapid developmental transition, such as the transition into and out of puberty” and of using “translational developmental neuroscience” to understand how behavior is transmitted from one generation to the next and translated across species. By examining the development of adolescent behavior through this lens, what may appear to be aberrant behavior at first glance may simply be adaptation of evolutionarily conserved mechanisms to the current environment.
More and more, scientists are attempting to explain aberrant behavior in the context of gene and environment interactions, with the environment accounting for much of the variance (Caspi et al., 2010, Risch et al., 2009). In contrast to the argument of the environment being variable and the genome being relatively stable, the environment can be an extreme source of stability for nervous systems and behaviors (Finlay, 2007). A long evolutionary history can lead to expected experiences and specialized mechanisms for adapting to these experiences. Greenough (Greenough et al., 1987) coined the terms experience-expectant and experience-dependent to differentiate processes that appear to have evolved as a neural preparation for incorporating specific experiences versus specific processes to incorporate experiences that are unique to the individual, such as learning about one's specific environment. When an evolutionarily expected environment is altered, then evolutionarily conserved mechanisms may be maladaptive for the current environment, leading to altered or aberrant behavior for that context.
One potential example of an altered environmental experience is the prolongation of adolescence with the postponement of adulthood (Cline, 1941). Prolonged adolescence may be lengthened further by earlier pubertal onset, as suggested by recent studies showing a declining age of secondary sex characteristics (e.g., breast development among girls) in the United States (Biro et al., 2010) and in Copenhagen (Aksglaede et al., 2009). The resulting prolongation of adolescence may result in a mismatch in the expected environment—based on a long evolutionary history—and the actual one, which must be considered in understanding how these neural systems develop and how behaviors emerge and are maintained.
Figure 1 depicts how prolongation of adolescence may result in increases in the duration and the degree (magnitude) of an imbalance between hormonally sensitive limbic processes relative to more age-dependent (experience) driven cognitive processes across development during the transition from dependence to independence (Casey et al., 2008a). We (Casey et al., 2008a) and others (Ernst et al., 2006, Ernst et al., 2009, Geier and Luna, 2009, Steinberg, 2008) have suggested that adolescence is a period of functional activation of basic motivational and emotional systems at a time when prefrontal cortical systems involving rational decisions and actions are not fully mature. We provide empirical evidence for this model, framing the review in the context of key questions asked by ethologists to understand animal behavior (Tinbergen, 1974): (1) what may cause or control adolescent behavior; (2) how does the behavior change across development; and (3) how may the behavior be adaptive?
Figure 1. Model of Adolescent Development Prolongation of adolescence may result in increases in the duration and the degree (magnitude) of an imbalance between hormonally driven limbic processes relative to age-dependent (experience) driven cortical processes, In the illustration above, blue represents cortical control, pink represents subcortical reactivity, and red represents a further shift with earlier pubertal onset. Increasing the duration and degree of this imbalance may lead to greater difficulty as the individual moves from dependence on parents to relative independence.
What Causes or Controls Adolescent Behavior?
Parents often attribute their adolescent's mood swings and provocative behavior to surging hormones. However, animal and human studies suggest that both puberty and age influence changes in explorative and emotive behaviors in adolescents. To address causal influences on behavior during adolescence, we begin by defining adolescence. Adolescence is the transition between pubertal onset and parental independence where puberty refers to changes in reproductive maturation (Graber and Brooks-Gunn, 1996) and adolescence refers to the transition from dependence on parents to relative independence. The latter involves emotional, psychological, social, mental, and physical changes and growth (Schulz et al., 2009, Spear, 2010). Pubertal onset occurs with the release of gonadal hormones—testosterone released from the testes in males and estrogen and progesterone released by the ovaries and uterus of females. These gonadal hormones contribute to the development of secondary sexual characteristics influencing the physical body appearance and function that the adolescent must adjust to, in addition to influencing neural function via binding to testosterone and estrogen receptors in brain. The hormonal and brain changes coincide with increased sexual activity and interest (Sisk and Zehr, 2005) and with changes in arousal and the salience of motivational stimuli (Friemel et al., 2010). In contrast, chronological age (experience) has been postulated to be associated with cognitive maturation and control (Spear, 2010, Steinberg, 2005). Thus pubertal onset and chronological age appear to be two causal forces shaping adolescent development.
Disentangling age and pubertal effects is more challenging, however (Blakemore et al., 2010). Many factors influence pubertal onset including both genetic and environmental ones (Chumlea et al., 2003, Herman-Giddens et al., 1997, Kaplowitz et al., 2001, Lee et al., 2001, Tanner, 1989). Specifically, pubertal onset has been suggested to be accelerated in obese individuals (Karlberg, 2002) but delayed in the malnourished (Argente, 1999), the anorexic (Muñoz and Argente, 2002), or those engaged in activities where low body fat is reinforced (e.g., ballet, wrestling) (Roemmich et al., 2001). Importantly, early puberty has been associated with poor outcomes in both sexes. These outcomes include earlier use of alcohol and illegal substances, earlier sexual behavior, higher risk for mental health problems, and increased risk for delinquency (Bratberg et al., 2007, Deardorff et al., 2005, Waylen and Wolke, 2004, Graber et al., 1997, Kaltiala-Heino et al., 2003).
Recent animal and human imaging studies support separate and interactive roles of puberty and age in controlling behavior and altering brain development. Sisk and colleagues showed that the presence or absence of testosterone during adolescence in male hamsters could program adult sexual behavior (Schulz et al., 2009), but preadolescent exposure did not. However, preadolescent testosterone exposure impacted brain development by altering limbic circuitry (e.g., increases amygdala volume), important in social development (Schulz et al., 2009, Sisk et al., 2003). Human imaging studies of individuals with endocrine disruptions provide converging evidence for hormonal influences differing as a function of age of exposure. In an imaging study of adolescents with congenital adrenal hyperplasia, which is associated with elevated testosterone in utero, Ernst et al. (2005) showed enhanced amygdala activity to cues of empty threat (fearful faces) relative to controls; this activity was specific to females and similar to male controls. Using a similar behavioral probe in adolescent males with familial hyperandrogenism, which causes elevated testosterone postnatally, Mueller et al. (2009) showed elevated limbic activity and faster behavioral responses to fearful faces relative to controls. Together, these animal and human findings provide evidence of hormonal influences on brain systems that may play a role in processing affective and social stimuli varying as a function of age.
Disentangling age and pubertal driven changes is difficult in typically developing organisms given the strong associations between the two and the wide variability in timing of puberty (Spear, 2010). Nonetheless, recent human behavioral and imaging studies have attempted to disentangle these by examining adolescents of the same age but in different stages of sexual maturation (Forbes et al., 2010) or by examining the effects of sexual maturation using a wider age range but then statistically controlling for age (Silk et al., 2009). Sexual maturation is typically assessed by physical exam or Tanner staging by the individual or parent, but more and more hormonal measures are being assessed, which present challenges given the large fluctuations in early stages of puberty and during the menstrual cycle yet move the field forward beyond the use of age as a proxy for pubertal maturation.
To date, these studies suggest that mid- to late puberty is positively associated with elevated physiological reactivity to emotional cues (Silk et al., 2009) and that testosterone level is positively associated with anticipation of reward in males (Forbes et al., 2010). These findings suggest an important role of pubertal hormones in social and emotional processes during adolescence. How these processes change as the individual transitions into and out of adolescence is less clear from studies that limit the age range of the sample or statistically control for age effects, especially when these samples are younger than the age at which inflections in risky behavior have been shown to peak (13 to 17 years, see Steinberg et al., 2009). To address this question, we review recent behavioral, imaging, and animal studies of this transitional period, highlighting changes specific to adolescents relative to children and adults.
How Does Behavior Change during Adolescence?
Changes in explorative and emotive behaviors have been observed across species as the organism moves from parental dependence to independence during the transition into and out of adolescence. These behavioral findings are reviewed and then discussed in the context of changes in the underlying neurobiology with age.
Human Behavioral Studies
A cornerstone of behavioral development is the ability to resist temptation in favor of long-term goals. Lapses in this ability have been suggested to lie at the very core of adolescent behavior (Steinberg et al., 2009). Resistance to temptation or delay of immediate gratification has been studied in the context of social, developmental, and cognitive psychology. In preschoolers, this ability has been measured by assessing how long they can resist an immediate reward (e.g., a marshmallow) in favor of a larger reward later (e.g., two marshmallows) (Mischel et al., 1989). Although individuals vary in this ability even as adults, developmental studies suggest periods, such as adolescence, when an individual may be particularly susceptible to temptations (e.g., Eigsti et al., 2006).
There is a wealth of behavioral evidence from experimental paradigms in controlled laboratory settings that show a steady improvement in the ability to suppress an inappropriate response in favor of an appropriate one from infancy to adulthood, termed cognitive control (Casey, 2005, Casey et al., 2005, Davidson et al., 2006). However, when it is advantageous to suppress a response to incentive-related cues, cognitive control suffers (Somerville et al., 2010a). This reduced control is especially evident during the period of adolescence, when suboptimal choices in sexual and drug-related behaviors have been suggested to peak (Casey et al., 2008b, Eaton et al., 2008, Spear, 2000, Windle et al., 2008). These observations imply that developmental trajectories in cognitive control can be modulated by emotionally charged or reinforcing contexts (e.g., social and sexual interactions), in which cognitive control demands interact with motivational drives. Moreover, these data suggest the importance of distinguishing between cognitive and motivational processes in understanding how they may interact across development.
Motivation has been shown to modulate cognitive control in at least two ways. First, being rewarded for performance on a given task can improve performance more than when not rewarded (Geier et al., 2010). Second, the capacity to exert control can be challenged when required to suppress thoughts and actions toward appetitive cues (Somerville et al., 2010a). Recent studies of adolescent development have begun to compare cognitive control capacity in relatively neutral versus motivational contexts. These studies suggest a change in sensitivity to environmental cues, especially reward-based ones at different points in development, and suggest a unique influence of motivation on cognition during the adolescent years (Somerville and Casey, 2010).
Ernst and colleagues (Hardin et al., 2009, Jazbec et al., 2006) first showed that adolescent behavior is differentially biased in motivationally charged contexts relative to adults. Using an antisaccade task with a promise of financial reward for accurate performance on some trials but not others, they showed that promise of a reward facilitated adolescent performance more than adult performance. Recently, this finding has been replicated (Geier et al., 2010).
In contrast to enhancing performance, rewards can diminish performance when suppressing responses to stimuli that lead to high gain. For example, using a gambling task in which reward feedback was provided immediately during decision-making (“hot” trials, which heightened task-elicited arousal) or withheld until after the decision (“cold” deliberate decision-making trials), Figner et al. (2009) showed that adolescents made disproportionately more risky gambles compared to adults but only in the “hot” condition. Using a similar gambling task, Cauffman et al. (2010) have shown that this sensitivity to rewards and incentives actually peaks during adolescence, as demonstrated by a steady increase from late childhood to adolescence in tendency to play with more advantageous decks of cards, followed by a subsequent decline from late adolescence to adulthood. These findings illustrate a curvilinear function, peaking roughly between 13 and 17 and subsequently declining (Steinberg et al., 2009).
More real-world experiments have begun to examine how peers, as possible secondary reinforcers, influence adolescent behavior more than adult behavior. Using a simulated driving task, Gardner and Steinberg (2005) showed that adolescents make riskier decisions in the presence of peers than when alone. The number of risky decisions decreases by young adulthood (Chassin et al., 2004, Steinberg, 2008). Taken together, these studies suggest that during adolescence, motivational cues of potential reward are particularly salient and can lead to improved performance when provided as a reinforcer or rewarded outcome but to riskier or suboptimal choices when provided as a cue. In the latter case, the motivational cue can diminish effective goal-oriented behavior.
Finally, there is evidence to suggest that motivational processes related to sensitivity to rewards and sensation-seeking behavior are distinct from impulsivity with very different developmental patterns (curvilinear versus a linear function, respectively). This distinction is further supported in a recent study by Cauffman et al. (2010) using self-report measures of sensation seeking and impulsivity. They showed that the often-conflated constructs of sensation seeking and impulsivity developed along different timetables in over 900 individuals between the ages of 10 and 30. Specifically, differences in sensation seeking across age followed a curvilinear pattern, peaking between 10 and 15 years and declining or remaining stable thereafter. In contrast, impulsivity followed a linear pattern, decreasing with age. These findings suggest that heightened vulnerability to risk taking in adolescence may be due to a complex interaction of sensation seeking in the context of relatively immature behavioral control abilities typical of this period of development (Cauffman et al., 2010).
Human Imaging Studies
Over the past decade, developmental scientists have begun to use neuroimaging to understand inflections in behavior during the developmental window of adolescence relative to those preceding or following it (Galvan et al., 2007, Hare et al., 2008, Somerville and Casey, 2010). Depiction of how the brain changes during adolescence relative to both childhood and adulthood is needed to explain the previously described inflections in behavior. Moreover, this approach provides the opportunity to link neural processes with cognitive and motivational ones. For example, neural correlates of cognitive control would presumably develop linearly from childhood to adulthood, whereas neural correlates of incentive and emotive processes related to desire, fight, and flight may have a different developmental pattern.
One of the first studies to examine incentive-related processes across the full spectrum of development from childhood to adulthood was completed by Galvan et al. (2006) in 7- to 29-year-olds. Specifically, she focused on dopamine-rich circuitry and manipulated the magnitude of reward outcome, borrowing heavily from nonhuman primate studies showing dopaminergic neuronal firing to rewards. Animal work has shown that behavior can be altered by anticipated reward outcome (Pavlov, 1927), and electrophysiological studies show that striatal neurons play an important role in rewarded behavior, given their sensitivity to changes in reward magnitude and frequency (Cromwell and Schultz, 2003, Fiorillo et al., 2003). Galvan thus examined whether dopamine-related neural circuitry showed differing responses to reward manipulation as a function of age. She found that the ventral striatum, a dopamine-rich area shown previously in adult imaging to be associated with addiction and reward (Delgado et al., 2000, Elliott et al., 2000, Volkow et al., 1997), was sensitive to varying magnitudes of monetary reward (Galvan et al., 2005). Moreover, this response was exaggerated in adolescents, relative to both children and adults, indicative of increased signal (Galvan et al., 2006) or more sustained activation (Delgado et al., 2000). In contrast to the pattern in the ventral striatum, orbital prefrontal regions related to optimizing gains in goal-oriented behavior showed protracted development across these ages in a more linear fashion.
How does this enhancement of signaling in the ventral striatum in adolescents relate to real-world behavior? In a follow-up study, Galvan et al. (2007) examined the association between activity in the ventral striatum to large monetary reward and anonymous self-report ratings of risk taking (Fromme et al., 1997, Katz et al., 2000) and impulsivity (Conners et al., 1999) from a sample of 7- to 29-year-olds. There was a positive association between ventral striatal activity to monetary reward and the likelihood of engaging in risky behavior, but no association between activity in this region and impulsivity measures. These findings further support the dissociation of impulsivity and exploratory processes of sensation seeking and risk taking described earlier (Steinberg et al., 2009).
Although several laboratories (Ernst et al., 2005, Galvan et al., 2006, Geier et al., 2010, Van Leijenhorst et al., 2010a) have shown exaggerated ventral striatal responses to cues signaling reward outcome in adolescents, at least one laboratory has shown diminished activity of this region in adolescents in similar reward paradigms (Bjork et al., 2008, Bjork et al., 2010). This diminished response has been suggested to be similar to reward-deficiency syndrome, a condition often linked with addiction in adults. Since dopamine helps to link actions to sensations of pleasure, its redistribution during development may raise the threshold for stimulation (see Comparitive Studies). Accordingly, activities that once caused excitement can cease to provide such thrills. However, understanding a change in activities by adolescents may be informed by both considering the influence of the developmental context and the role of dopamine in learning. For example, dopamine is essential in learning to optimize gains and in detecting novelty or violations in the expected environment as shown by Schultz et al. (Cromwell and Schultz, 2003, Fiorillo et al., 2003). Second, while behavior is largely shaped by parents in early childhood, peers may motivate behavior more in adolescence (e.g., Gardner and Steinberg, 2005). Accordingly, one can imagine alternative explanations for why activities that once caused excitement can cease to provide such thrills. For example, if activities previously approved or reinforced by parents or teachers are not reinforced by peers, then these behaviors may be extinguished. New activities shared with peers or that gain peer approval or reinforcement may then be the new activities that the adolescent will seek. This later interpretation of adolescent behavior based on learning theory would not be consistent with a reward- or dopamine-deficient hypothesis of adolescence.
Further support of the role for an elevation in dopamine-rich circuitry being related to risky behavior and sensitivity to reward comes from Van Leijenhorst et al. (2010a) in adolescents showing upregulation of this circuitry with high-risk relative to low-risk choices when gambling. In adults, Christopoulos et al. (2009) have shown that a risky choice is more probable when striatal activity is higher, whereas increased prefrontal activity correlates with increased probability of a safe choice and with higher risk aversion. Thus, elevated ventral striatal activity would seem more consistent with the observed increase in risky behavior during this age (Figner et al., 2009). Moreover, diminished activity of dopamine-rich circuitry has been associated with clinical populations that are characterized by impulsivity problems like ADHD (Durston et al., 2003, Epstein et al., 2007, Vaidya et al., 1998) and do not show heightened ventral striatal responses to incentives (Scheres et al., 2007).
A scientific area that has received less attention is determining how cognitive control and motivational systems interact over the course of development. As mentioned earlier, Ernst and colleagues (Hardin et al., 2009, Jazbec et al., 2006) showed that promise of a monetary reward facilitated adolescent cognitive control behavior more than for adults. Geier et al. (2010) recently identified the neural substrates of this cognitive upregulation using a variant of an antisaccade task during functional brain imaging. In adolescents and adults, trials for which money was at stake speeded performance and facilitated accuracy, but this effect was larger in adolescents. Following a cue that the next trial would be rewarded, adolescents showed exaggerated activation in the ventral striatum and supplementary frontal eye fields while preparing for and subsequently executing the antisaccade, suggesting a reward-related upregulation in control of goal-directed eye movements.
Rewards can enhance or diminish goal-directed behavior. The capacity to exert control over one's actions is especially challenged when required to suppress an action toward positive or appetitive cues. The observation that adolescents take more risks when appetitive cues are present versus absent during gambling tasks makes this point (e.g., Figner et al., 2009). Recently, Somerville et al. (2010a) provided empirical evidence for a specific reduction of impulse control in adolescents when faced with cues signaling appetitive value. Using a task that contained social appetitive cues (e.g., happy faces) that facilitated approach responses, she showed the developmental trajectory of subjects' abilities to flexibly approach or avoid positive or neutral stimuli. Specifically, adolescents showed a unique pattern of errors relative to both children and adults, characterized by a reduction in the capacity to suppress an approach response toward a positive, appetitive social cue (see Figure 2).
Figure 2. Behavioral Performance to Appetitive Relative to Nonappetitive Social Cues across Development Gray line represents missing a target go trial (misses); black line represents making a commission error (false alarm) by inappropriately responding to a nontarget when instructed to withhold a response. Behavioral performance is biased toward appetitive social cues (happy faces) relative to nonappetitive social cues (calm faces) during adolescence relative childhood or adulthood. Data are represented as mean ± standard error of the mean (SEM). Adapted with permission from Figure 2 of Somerville et al. (2010a).
The decrement in performance during adolescence was paralleled by enhanced activity in the ventral striatum. Conversely, activation in the inferior frontal gyrus was associated with overall accuracy and showed a linear pattern of change with age for the no-go versus go trials (see Figure 3).
Figure 3. Development of Ventral Frontostriatal Function(A) Ventral striatal region showing differential activity as a function of age.(B) Plot of ventral striatal activity to appetitive cues as a function of age.(C) Differential activity in the inferior frontal gyrus as a function of task (nogo > go).(D) Plot of inferior frontal gyrus activity to nogo relative to go trials as a function of age. Recruitment decreases linearly with age. Data are represented as mean ± SEM. Adapted with permission from Figures 3 and 4 of Somerville et al. (2010a).
Recently, developmental studies have begun to provide support for strengthening in the connections of dopamine-rich frontostriatal circuitry. Using diffusion tensor imaging and functional magnetic resonance imaging (fMRI), Casey and colleagues (Casey et al., 2007, Liston et al., 2006) and others (Asato et al., 2010) have shown greater strength in distal connections within these circuits across development and have linked connection strength between prefrontal and striatal regions with the capacity to effectively engage cognitive control, in typically and atypically developing individuals (Casey et al., 2007, Liston et al., 2006). These studies illustrate the importance of signaling within corticostriatal circuitry, which supports the capacity to effectively engage in cognitive control and underscore the importance of specific brain circuitry over individual regions.
Collectively, the developmental imaging literature suggests that the prefrontal cortex, thought to support cognitive control (Astle and Scerif, 2009, Bitan et al., 2006, Bunge et al., 2002, Luna et al., 2001, Luna et al., 2010, Stevens et al., 2009, Tamm et al., 2002), undergoes protracted maturation (Giedd et al., 1999, Huttenlocher, 1990, Sowell et al., 2003, Shaw et al., 2008) while striatal regions sensitive to novelty and reward manipulations may develop earlier (Casey et al., 2002, Luna et al., 2001) or be more functionally active with the onset of puberty. As noted, although several groups have shown heightened activation of the ventral striatum in adolescents in anticipation and/or receipt of rewards compared to adults (Ernst et al., 2005, Galvan et al., 2006, Geier et al., 2010, Van Leijenhorst et al., 2010a), others have shown suppressed activity (Bjork et al., 2004, Bjork et al., 2010). To understand and attempt to constrain interpretations of these imaging findings, we turn to the recent animal literature on adolescence.
Comparative Studies
Adolescence is not special to humans. Rather, a variety of species must acquire the basic skills of transitioning from dependence to relative independence from parental care (Spear, 2010). Nonhuman adolescents show age-typical ways of responding to their environment. These behaviors include increases in peer interactions, novelty seeking, and elevations in consummatory behaviors (Spear, 2000) thought to serve a number of adaptive functions despite the effect risky behaviors may have on adolescent mortality rates (Crockett and Pope, 1993, Irwin and Millstein, 1986, Spear, 2010). Adaptive functions include increasing the probability of reproductive success among males across species, improving life circumstances, and enabling procurement of additional resources, as well as exploring adult liberties and greater ability to face and surmount challenges (Belsky et al., 1991, Csikszentmihalyi and Larson, 1987, Daly and Wilson, 1987, Meschke and Silbereisen, 1997). Risk taking also encourages individuals within a species to emigrate away from the home territory around the time of sexual maturation, thereby reducing inbreeding within the population and avoiding the lower viability of inbred offspring due to greater expression of recessive genes (Bixler, 1992, Moore, 1992). Within mammalian species, sexual maturation marks the time when males emigrate away from the home territory (Pereira and Altmann, 1985, Schlegel and Barry, 1991). So what are the neuroanatomical correlates of this behavior?
Seminal animal work has delineated dopamine-rich frontostriatal circuitry in motivated behavior. For example, using single-unit recordings in monkeys, Pasupathy and Miller (2005) showed that when flexibly learning a set of reward contingencies, very early activity in the striatum provides the foundation for reward-based associations, whereas later, more deliberative prefrontal mechanisms are engaged to maintain the behavioral outputs that can optimize the greatest gains; these findings have been replicated in lesion studies (Gill et al., 2010, Hauber and Sommer, 2009). A role for the striatum in early temporal coding of reward contingencies prior to the onset of activation in prefrontal regions has been extended to humans (Galvan et al., 2005). These findings suggest that understanding the interactions between regions within frontostriatal circuitry is critical for developing a model of cognitive and motivational control in adolescence.
Frontostriatal circuits undergo considerable elaboration during adolescence (Benes et al., 2000, Brenhouse et al., 2008, Cunningham et al., 2008, Tseng and O'Donnell, 2007) that is particularly dramatic in the dopamine system. Animal studies suggest that peaks in the density of dopamine receptors D1 and D2 in the striatum occur early in adolescence, followed by a loss of these receptors by young adulthood (Seeman et al., 1987, Tarazi and Baldessarini, 2000, Teicher et al., 2003). In contrast, the prefrontal cortex does not show peaks in D1 and D2 receptor density until late adolescence and young adulthood (Andersen et al., 2000, Weickert et al., 2007). It remains unclear how changes in dopamine systems may relate to motivated behavior, as controversy remains as to whether it is a result of less active or hypersensitive dopamine systems (e.g., Robinson and Berridge, 2003, Volkow and Swanson, 2003). This controversy is consistent with the mixed adolescent imaging findings of diminished or elevated ventral striatal activity in anticipation of appetitive outcomes (e.g., Bjork et al., 2010, Somerville et al., 2010a). However, given the dramatic changes in dopamine-rich circuitry during adolescence, it is likely to be related to changes in sensitivity to rewards distinct from pre- and postpuberty (Brenhouse et al., 2008, Spear, 2010). Nonetheless, equating the relative appetitive quality of rewards when investigating reward-related behavior across ages and species is an area that requires more investigation.
How Is Adolescent Behavior Adaptive?
Because information about gender and social status are essential for reproduction and survival of a species, it seems plausible that specialized mechanisms have evolved to process socioemotional cues when needed (Fernald, 2004, Finlay, 2007, Insel and Fernald, 2004). Changes in behavior during adolescence appear to occur in parallel with the increase in pubertal hormones, resulting in more exploratory behavior and seeking sexual partners. In conjunction with a “push” mechanism, however, there would need to be a mechanism for “pulling” back when detecting cues that signal threat or danger. In other words, if the organism leaves the nest but is immediately consumed by a predator, it has no more increased the chances of reproduction and survival than it would have had it stayed.
Over the past decade, a number of imaging studies have begun to examine the sensitivity of adolescents to emotional cues and information (Ernst et al., 2005, Guyer et al., 2008a, Guyer et al., 2008b, Guyer et al., 2009, Monk et al., 2003, Rich et al., 2006, Williams et al., 2006). One of the most developmentally comprehensive of these studies was completed by Hare et al. (2008) based on an initial sample of 80 subjects between 7 and 32 years. Hare went beyond simple examination of the limbic activity, which has been shown by several groups to be higher in adolescents than in adults, in order to: (1) show specific changes in the brain and in behavior in adolescents relative to both adults and children; (2) examine not only transient patterns of brain activity, but also changes in activity over time with repeated exposure to empty threat; and (3) assess whether these changes related to self-report ratings of everyday anxiety (Hare et al., 2008).
Hare showed that adolescents have an initial exaggerated amygdala response to cues that signal threat (fearful faces) relative to children and adults (see Figures 4A and 4B). The initial heightened response in amygdala activity is age dependent and specific to adolescents relative to children and adults. This developmental curvilinear pattern is strikingly similar to previous findings of elevated accumbens activity to appetitive social and monetary cues (Ernst et al., 2005, Galvan et al., 2006, Guyer et al., 2008a, Guyer et al., 2008b, Guyer et al., 2009, Monk et al., 2003, Rich et al., 2006, Williams et al., 2006).
Figure 4. Developmental and Individual Differences in Amygdala Responses to Empty Threat(A) Greater engagement of the amygdala to empty threat (fearful faces) in adolescents relative to children or adults.(B) Localization of amygdala activation on a coronal image.(C) Association of habituation of amygdala response with repeated exposure to empty threat and everyday ratings of anxiety. Adapted with permission from Figures 2 and 4 of Hare et al. (2008).
More in-depth examination of the blood oxygen level-dependent signal in the amygdala with repeated presentation of the fearful face (i.e., repeated exposure to empty threat) across experimental trials showed attenuation over time. The extent to which activation of this region diminished with repeated trials was correlated with the anonymous self-report ratings of everyday anxiety (Figures 4B and 4C). The failure of the amygdala response to return to baseline over time was associated with coupling of the amygdala and ventromedial prefrontal cortex. Specifically, inverse functional coupling of these regions, consistent with greater top-down regulation of the amygdala from prefrontal projections, was correlated with a more diminished signal in the amygdala over time.
These findings suggest that initial emotional reactivity to potential threat, as indexed by elevated amygdala activity, is typical of, or normal for adolescence, but that failure of this response to subside over time with no impending threat is atypical or maladaptive and may be indicative of risk for anxiety. Consistent with this suggestion is clinical imaging data showing elevated amygdala activity to fearful faces using similar paradigms with children and adolescents diagnosed with anxiety and depression (e.g., Thomas et al., 2001). Future studies of populations at risk for anxiety will need to examine carefully not only what triggers a heightened threat response in the amygdala, but also the brain processes that support anxiety responses that are sustained over time (Somerville et al., 2010b).
The observation of imbalanced activity in the amygdala-ventromedial prefrontal network, as shown by elevated amygdala and less prefrontal activity in anxious individuals, is consistent with a variety of work in animals (Baxter et al., 2000, Milad and Quirk, 2002), in humans (Delgado et al., 2006, Etkin et al., 2006, Haas et al., 2007, Johnstone et al., 2007, Urry et al., 2006), and in childhood and adolescent mood and anxiety disorders (Guyer et al., 2008a, Monk et al., 2008), implicating an inverse association between these structures that governs affective output. In particular, increased ventromedial prefrontal activity is inversely correlated with responding in the amygdala and predicts behavioral outcomes such as fear extinction, downregulation of autonomic responses (Phelps et al., 2004), and more positive interpretations of emotionally ambiguous information (Kim et al., 2004).
Recent translational studies using genetically altered mice and human genetic imaging suggest one pathway to the uncoupling within this circuitry (Soliman et al., 2010) and resulting variability across individuals. During adolescence, when the amygdala response is heightened relative to that observed in children and adults (imbalance), more top-down control is needed. Environmental or genetic factors that result in less coupling between these regions with development, to help provide that downregulation, may lead to heightened emotion and, ultimately, risk for anxiety-related disorders.
Our model depicted in Figure 1 shows an imbalance between hormonally driven limbic and more age-dependent cognitive systems during adolescence. We suggest that this imbalance may be lengthened as a result of the mismatch between the expected experiences of mating with sexual maturation and the delay in this behavior with the prolongation of adolescence (refer back to Figure 1). Consistent with this view is how the brain differs in sensitivity to testosterone levels across development. Earlier onset of testosterone secretion engendered either experimentally in animals or by altered dietary or behavioral patterns in humans results in experience-dependent alterations in social and sexual development of the adolescent (Schulz et al., 2009). The degree of this imbalance between hormone-sensitive brain regions and cortical ones would be lengthened and accentuated for individuals with earlier pubertal onset when the cognitive systems are less fully developed, resulting in less adaptive behavior for the current environment and greater risk for mental health problems (Graber et al., 1997, Kaltiala-Heino et al., 2003).
Conclusions
Negotiating the transition from dependence on parents to relative independence is not a unique demand for today's youth, but has a long evolutionary history and is shared across mammalian species. This period of transition has been prolonged in Western civilization for humans, with the increasingly occurring postponement of parental independence in many individuals. In this review, we highlighted recent neurobiological studies of adolescence to address the questions of potential mechanisms of behavioral change, how behavior changes during the transitions from childhood to adolescence to adulthood, and how these behavioral changes are adaptive from a comparative and evolutionary framework. To theoretically ground the empirical findings, we provided a plausible neurobiological model for understanding adolescence. The intention is to move away from psychopathologizing adolescence in order to explain why some teens—but not others—are more vulnerable to poor outcomes and why there is a 200% increase in mortality during this developmental period. As such, we identified potential markers of risk for mental health problems that go beyond adaptive and typical exploratory and emotive behaviors of adolescence.
The model builds on animal work (Laviola et al., 1999, Spear, 2000) and human behavioral (e.g., Figner et al., 2009, Gardner and Steinberg, 2005) and imaging studies of adolescents (Ernst et al., 2005, Galvan et al., 2006, Galvan et al., 2007, Geier et al., 2010, Hare et al., 2008, Van Leijenhorst et al., 2010a, Van Leijenhorst et al., 2010b) to illustrate how hormonally sensitive subcortical limbic regions and age-dependent prefrontal cortical regions must be considered together in understanding adolescent behavior. The model is consistent with different developmental trajectories for signaling in these regions, with limbic projections developing earlier than prefrontal control regions. Accordingly, the adolescent is biased by elevated subcortical limbic responses to motivational and emotional cues relative to less mature cortical recruitment (i.e., imbalance theory), compared to children, for whom this frontolimbic circuitry is still developing, and compared to adults, for whom these systems are fully mature. With development and experience, the functional connectivity between these regions is strengthened and provides a mechanism for top-down modulation of the subcortical systems (Hare et al., 2008). Thus it is the limbic corticosubcortical circuitry, along with functional strengthening of connections within this circuitry (which develop with experience), that may provide a mechanism to explain changes in both impulsivity and risk taking observed across development. Atypical early experiences can lead to failure in the typical development of functional circuitry, resulting in a heightened imbalance.
This view of adolescence is consistent with previous ones (Ernst et al., 2006, Ernst et al., 2009, Geier and Luna, 2009, Steinberg, 2008) in that it provides a basis for nonlinear inflections observed in behavior from childhood to adulthood, as a result of earlier maturation of subcortical projections relative to less mature top-down prefrontal ones. Specifically, the triadic model (Ernst et al., 2006) proposes that motivated behavior has three distinct neural circuits (approach, avoidance, and regulatory). The approach system is largely controlled by the ventral striatum, the avoidance system by the amygdale, and, lastly, the regulatory system by the prefrontal cortex (Hare et al., 2005). The current model differs from others in that it is grounded in empirical evidence for brain changes not only in the transition from adolescence to adulthood, but also in the transition into adolescence from childhood and later out of adolescence into adulthood. Moreover, the model does not suggest that the ventral striatum and amygdala are specific to approach and avoidance behavior given recent studies showing valence independence of these structures (Levita et al., 2009), but rather are systems that are important in detecting and learning about appetitive and emotive information in the environment for the development of adaptive behavior.
In sum, we suggest that adolescence, as a transitional period from parental dependence to adult independence, requires explorative and emotive mechanisms. These mechanisms have evolved over generations, for successfully adapting to new information and environments. From this perspective, aberrant behavior is the result of a change in an evolutionally stable environment (experience-expectant) (Greenough et al., 1987). When there is a violation between the expected and actual environment during development, mechanisms that have evolved to survive in that environment will be altered. The result is a disruption of a structured collaboration between the information in the organism and the environment (Finlay, 2007) leaving the organism less prepared and functionally adaptive. Accordingly, “development can no longer be viewed as a simple passage from the embryo to the mature organism directed by the information encoded in the genes, but rather a structured collaboration between the information in the organism and the environment” that occurs throughout development (Finlay, 2007, p.34) and evolution.