Puberty is a critical period of development that largely coincides with early to mid-adolescence, where hormone-driven processes facilitate sexual maturation (Grumbach and Styne, 2003). Two broad hormonal processes drive pubertal development. Puberty begins with adrenarche, which is when the adrenal gland releases increased amounts of androgens, including dehydroepiandrosterone (DHEA), dehydroepiandrosterone sulfate (DHEAS), and androstenedione (Palmert et al., 2001; Biro et al., 2014). This androgen release leads to the development of some secondary sex characteristics, including pubic hair, acne, and body odor (Grumbach and Styne, 2003). Adrenarche typically occurs between 6 and 9 years (earlier in girls than boys) and levels of androgens increase until the early 20’s (Figure 1; Havelock et al., 2004). Gonadarche typically begins between 9 and 14 in females and 10 and 15 in males (Dorn et al., 2006) and involves increased production of testosterone from the testes in males and estradiol from the ovaries in females (Grumbach and Styne, 2003). Gonadal hormones are involved in the development of primary and secondary sex characteristics, including genital and (in females) breast development, body/facial hair growth, and the growth spurts associated with puberty (Grumbach and Styne, 2003). Both gonadal and adrenal hormones are involved in pubertal brain development (Herting and Sowell, 2017). Variation in the timing of the initiation of puberty is not well understood, but is thought to be influenced by levels of body fat, hypothalamic-pituitary-adrenal (HPA) axis activity, and genetics (Cizza et al., 2001). Notably, there is some indication that pubertal development varies by race/ethnicity (Keenan et al., 2014; Deardorff et al., 2021), though it is unclear whether these differences are actually race/ethnicity-based or due to factors correlated with, but not attributable to race/ethnicity (e.g., income, Deardorff et al., 2019).

Thorough reviews of adolescent brain development can be found elsewhere (e.g., Casey et al., 2008; Blakemore, 2012; Konrad et al., 2013). Briefly, major changes in brain structure take place between childhood and adulthood (Giedd et al., 1999). Structurally, the “pruning” of synapses (Huttenlocher and Dabholkar, 1997) and thinning of gray matter are observed (Gogtay et al., 2004), along with increases in white matter (Gogtay et al., 2004; Perrin et al., 2008). Moreover, substantial changes in neurotransmitter systems are also occurring (Murrin et al., 2007). These structural and organizational changes are thought to contribute to corresponding changes in brain function. Briefly, as adolescence progresses, the prefrontal cortex exerts more control over sensory and affective regions of the brain, temporally consistent with the progression of structural changes in the brain (Casey et al., 2010). These changes in brain structure and function are linked with behavioral changes seen during adolescence (Casey et al., 2010), including increased responses to reward (Galván, 2013) and greater risk-taking (Galvan et al., 2007). Critically, the majority of studies examining brain development have considered brain development primarily as a function of age. This is despite differential impacts of age and puberty on both brain and behavior (Crone and Elzinga, 2015) and the roughly 5-year range of normal pubertal onset (Dorn et al., 2006). It is perfectly normal to have one 14 year-old in Tanner stage one (pre-puberty) and another in Tanner stage 5 (“adult-like”); treating these youth as equivalent neurobiologically is unjustified.

While increases in hormones begin during puberty, both adrenal and gonadal hormones continue to increase into the late teens/early 20’s (Havelock et al., 2004; Kelsey et al., 2014), after most individuals are Tanner Stage 5 and “adult-like” (Dorn et al., 2006). These hormone increases appear to roughly parallel the temporal course of brain development (i.e., Gogtay et al., 2004), suggesting at least partially distinct roles in physical vs. neural development. There has been some work specifically looking at the role of puberty in adolescent brain development (for reviews see Blakemore et al., 2010; Herting and Sowell, 2017; Vijayakumar et al., 2018). Briefly, as puberty advances, there are decreases in gray matter volume and increases in white matter similar to what is seen in the age literature (Vijayakumar et al., 2018). When the brain development literature has considered puberty, it has tended to consider secondary sex characteristics (e.g., Tanner staging) as the marker of pubertal development (see Vijayakumar et al., 2018). However, this approach has limitations. As reviewed by Dorn et al. (2006), there are reliability and validity problems with both self-report and physician-exam measures of secondary sex characteristics. Furthermore, hormones are thought to drive both the development of secondary sex characteristics and brain development (Grumbach and Styne, 2003). While more work is needed, a number of studies have shown that changes in pubertal hormones are associated with structural changes in the brain in humans (e.g., Nguyen et al., 2013, 2017; Nguyen, 2018). Indeed, structural neuroimaging and hormone work has found that adrenal hormones may impact brain development more during earlier stages of puberty, while gonadal hormones may influence brain development primarily later in puberty (Herting et al., 2017). With respect to functional neuroimaging, far less work has been conducted and the results have been less consistent (see Vijayakumar et al., 2018). As noted by Vijayakumar et al. (2018), this is likely a function of the limited work that has been conducted, the variability in behavioral paradigms completed during scanning, and the differing methods of examining pubertal development.

There are two main biologically oriented theorical perspectives on why there is a dramatic increase in antisocial behavior in adolescence. The first perspective postulates that increased antisocial behavior can be attributed to uneven maturation of different brain regions (Cohen and Casey, 2014; Jones et al., 2014). The second perspective, largely drawn from the animal literature (e.g., Berthold and Quiring, 1944), implicates increases in testosterone with increases in aggression, particularly in males. Notably in a human neuroimaging study, testosterone was found to modulate the relationship between amygdala and prefrontal cortex development and this relationship between amygdala and prefrontal cortex development was, in turn, associated with aggression (Nguyen et al., 2016). Despite these findings, the link between testerosterone and aggression, particularly in the human literature, however, is not a straightforward one.

Abnormal response to reward information has been observed in antisocial youth (e.g., Frick and White, 2008; Frick et al., 2014; Blair et al., 2016). Failure to adequately encode reward information may increase the risk for antisocial behavior. One model focuses on the link between reward processing, frustration, and reactive aggression (Blair, 2010). The accurate encoding of reward information is necessary to accurately predict likely outcomes following actions in a given environment. Disrupted representation of reward information will result in poor prediction of likely outcomes in the environment leading to increased frustration. Frustration is associated with greater risk of reactive aggression (Blair, 2010). Alternatively, inaccurate representation of reward information might lead to increased risk- and sensation-seeking behavior, which is also associated with antisocial activities (Mann et al., 2017).

Neurobiological models of antisocial behavior implicate dysfunction within the striatum during reward processing (Matthys et al., 2013; Rosell and Siever, 2015; Blair et al., 2016). The striatum is a key region in the reward system and is responsive to reward cues and rewarding feedback (O’Doherty et al., 2006). Several studies have shown reduced activation within striatum to reward in antisocial youth (Finger et al., 2011; White et al., 2013; Cohn et al., 2015; Holz et al., 2017), though at least one study reported increased reward response (Bjork et al., 2010). Several studies have also reported reduced activation to reward in regions of vmPFC/orbitofrontal cortex (Rubia et al., 2009; Finger et al., 2011; White et al., 2013; Schwenck et al., 2017), which are implicated in reward value representation and/or in the regulation of striatum (see e.g., O’Doherty et al., 2017). Multiple theoretical positions associate striatal dysfunction during reward processing with the decision-making problems associated with increased risk for antisocial behavior (e.g., Matthys et al., 2013; Rosell and Siever, 2015; Blair et al., 2016). Supporting these theoretical positions, two Activation Likelihood Estimate (ALE) meta-analyses implicate dysfunction in striatum and interconnected regions of cortex in antisocial behavior in adolescents (Alegria et al., 2016; Noordermeer et al., 2016).

A substantial amount of change occurs in the striatum during pubertal development. Structurally, an accelerated longitudinal study found that striatum volume was negatively associated with pubertal development (Goddings et al., 2014). Increasing testosterone occurring as a result of puberty has been associated with increased striatal response to threat cues (Spielberg et al., 2014). Moreover, longitudinal studies have shown that striatal response to reward information rises from early adolescence to a peak in mid-adolescence before declining (Braams et al., 2015; Vijayakumar et al., 2019), which is consistent with cross-sectional data (Casey et al., 2008). Notably, level of reward response in striatum has been associated with substance abuse, an antisocial behavior (Braams et al., 2016). Interestingly, other evidence suggests a more complex relationship between pubertal hormones and reward. White et al. (2020) found that increased testosterone reactivity is associated with reduced striatum and vmPFC activation in rewarded relative to unrewarded trials. In the same sample, the gonadal hormone estradiol was also associated with reduced striatal response to rewarded relative to unrewarded trials in females (Ladouceur et al., 2019). Importantly, in the paradigm considered in both studies, reward information was irrelevant to successful task completion. White et al. (2020) argued that this finding is consistent with hormones being highly context-dependent, consistent with theoretical positions (e.g., Roney, 2016).

Poor decision-making is associated with antisocial behavior (for reviews see Matthys et al., 2013; Rosell and Siever, 2015; Blair et al., 2016). In recent years, the use of computational neuroimaging has begun to better illustrate the regions of the brain involved in specific computations involved in decision-making relative to earlier neuroimaging work (O’Doherty et al., 2007). One of the essential computations necessary for decision-making is the expected value that becomes associated with an action or stimulus based on previous learning (O’Doherty, 2004). Building on previous work showing increased activation in anterior insula to cues signaling risk or avoidance (Kuhnen and Knutson, 2005; Liu et al., 2007), anterior insula has been shown to represent expected value during avoidance behavior (Rigoli et al., 2016). It has been argued that this representation of expected value in anterior insula is involved in response selection and behavior change (Casey et al., 2001; Budhani et al., 2007; White et al., 2014). Youth with DBDs show reduced representation of expected value in anterior insula when avoiding sub-optimal choices (White et al., 2013, 2014). Notably, the degree of impairment in the representation of expected value during the avoidance of sub-optimal choice has been associated with increased levels of antisocial behavior (White et al., 2016). Failure to appropriately represent the computations needed for successful behavior change might account for the punishment insensitivity observed in antisocial youth (Frick and White, 2008; Matthys et al., 2013; Frick et al., 2014; Rosell and Siever, 2015; Blair et al., 2016; White et al., 2016).

To our knowledge, no developmental neuroscience work has specifically considered pubertally driven changes in avoidance learning. However, the anterior insula has been considered in broader developmental neuroscience work on reward response. For example, a meta-analysis of reward processing found that adolescents show greater anterior insula activation during reward processing than adults (Silverman et al., 2015), suggesting adolescent changes in anterior insula functioning. Increasing DHEA during late childhood is associated with reduced functional connectivity between insula and amygdala (Barendse et al., 2018). An additional study found that increasing levels of estradiol in adolescent females were associated with anterior insula activation during social, but not monetary, feedback (Op de Macks et al., 2017). Estradiol is a major driver of pubertal development in females and DHEA is largely responsible for adrenarcheal maturation. These findings point to the possibility that puberty is altering functioning in anterior insula and highlight this as a potentially fruitful target for future work in this area.

With respect to amygdala functioning, there is heterogeneity between antisocial youth in terms of neural correlates. A lack of emotional empathy is specifically associated with the presence of DBDs and callous-unemotional (CU) traits (Viding et al., 2014; Blair et al., 2016). CU traits include a lack of guilt or remorse, a lack of concern about poor performance, callousness, and shallow affect (Frick and White, 2008; Frick et al., 2014) and are included as the “Limited Prosocial Emotions” specifier for Conduct Disorder in DSM-5 (American Psychiatric Association, 2013). The lack of empathy shown by these youth is thought to increase the likelihood of an individual selecting antisocial, particularly aggressive, social tactics (Pardini, 2006; Frick and White, 2008; Blair et al., 2014; Frick et al., 2014). The lack of empathy seen in antisocial youth displaying high levels of CU traits is thought to stem from reduced amygdala response to fear/distress cues in others (Blair et al., 2014, 2016; Viding et al., 2014). Notably, the relationship between CU traits and proactive aggression appears to be mediated by the level of amygdala response to fear/distress cues (Lozier et al., 2014). Furthermore, there is some evidence that CU traits may be related to differences in the relative coupling of the diurnal rhythms of testosterone and cortisol (Johnson et al., 2014).

In contrast, many youth with DBDs not exhibiting CU traits show increased amygdala response to fear/threat cues (Viding et al., 2012; Blair et al., 2014, 2016), similar to youth with exposure to traumatic events and maltreatment (McCrory et al., 2017). Indeed, there is a high comorbidity between trauma exposure and antisocial behavior (Lewis et al., 2019). Increased activation of the basic threat system, of which the amygdala is a part (Nelson and Trainor, 2007), is thought to increase the likelihood of increased reactive aggression and other reactive antisocial behaviors (Blair et al., 2016). Notably, even in youth with high levels of CU traits, high levels of trauma are associated with increased, rather than decreased, amygdala response to fearful faces (Meffert et al., 2018). It has been argued that the tendency among antisocial youth to misinterpret ambiguous stimuli as threatening (Crick and Dodge, 1996) can be accounted for by this increased response to fear/threat stimuli (Crowe and Blair, 2008).

Both elevated and diminished amygdala responsiveness is associated with increased risk of antisocial behavior. The amygdala increases in volume over the course of adolescence, both as a function of age and puberty (Goddings et al., 2014). Longitudinal increases in testosterone as a result of pubertal development are associated with increased amygdala responses to threat cues (Spielberg et al., 2014). In a different longitudinal sample, amygdala responsiveness to faces increased from age 10 to 13 (Moore et al., 2012). Moreover, earlier pubertal timing was associated with increased amygdala reactivity at age 10, with a significant increase in magnitude at age 13 (Moore et al., 2012). Adding data from additional waves of collection, Vijayakumar et al. (2019) found that amygdala responsiveness to emotional expressions increased with pubertal development, peaking at Tanner Stage 3 and then declining over Tanner Stages 4 and 5. Interestingly, in the same study, amygdala responsiveness, at least in females, was found to have an inverse U-shaped relationship with testosterone levels (Vijayakumar et al., 2019). In short, it seems clear that amygdala responsiveness changes occur as a result of puberty, potentially moderated by sex, though further work is needed.

Based on the animal literature, a link between hormones, particularly testosterone, and antisocial behavior, especially aggression, has been hypothesized. Indeed, higher levels of testosterone have been found in individuals meeting criteria for DBDs (Aromäki et al., 1999; Chance et al., 2000; Pajer et al., 2006) and in incarcerated individuals (Kreuz and Rose, 1972; Bain et al., 1987; Dabbs et al., 1987; Stålenheim et al., 1998; Maras et al., 2003; van Bokhoven et al., 2006). Importantly, however, evidence suggests that the relationship between testosterone and antisocial behavior in youth with DBDs is moderated by levels of CU traits (Blanchard and Centifanti, 2017), social dominance (Rowe et al., 2004), or by harsh parenting (Chen et al., 2018). Moreover, there is a report of altered coupling between HPG and HPA axes in youth with Conduct Disorder, though only in those who also had high levels of CU traits (Johnson et al., 2014). In sum, while testosterone does appear to be elevated in individuals at high-risk for antisocial behavior, the mechanism by which this association is conveyed remains unclear and is likely moderated by psychosocial context, other hormone systems, and other risk factors. Moreover, as discussed above, the relationship between hormones and antisocial behavior is likely to be mediated by other complex social processes and contextual factors.

As noted above, increases in hormonal output occur during puberty (Grumbach and Styne, 2003). If increased levels of hormones are associated with antisocial behavior in individuals with DBDs, it would make sense that levels of antisocial behavior would increase during puberty. There is some evidence that there is a large spike in antisocial behavior during adolescence in some youth (Moffitt, 2003, 2018; Snyder et al., 2019). However, the peak of antisocial behavior is approximately 19 years (Snyder et al., 2019), which is after puberty typically ends (Dorn et al., 2006), but before hormones have finished increasing (Havelock et al., 2004; Kelsey et al., 2014). Furthermore, in the most consistently, severely, and chronically antisocial youths, antisocial behavior begins before puberty (see Frick, 2012). In short, there is a shortage of data that clearly links hormonal changes associated with puberty to an increased risk of antisocial behavior. A better understanding of the specific biological mechanisms linking hormonal changes and antisocial behavior is needed.

The legal system has long considered development as a key factor in determining culpability and what appropriate legal dispositions following adjudication should be. For over 100 years, the United States has implemented various juvenile justice systems with the intention of developmentally appropriate legal recourse for juvenile offenders (Bartol, 2002). Notably, while the developmental science of the past several decades has indicated that brain development is not complete until the early- to mid-20s (Gogtay et al., 2004; Cohen et al., 2016), the justice system (in some jurisdictions) has lowered the age at which juveniles can be tried as adults to as low as 13 years (Griffin et al., 2011). The legal system’s approach cannot be justified on the basis of developmental science.

The conflict between the developmental science and the legal system has been noted elsewhere (e.g., Steinberg, 2009; Cohen and Casey, 2014). Developmental scientists have recommended that juveniles 15 years old and younger should not be treated as adults by the justice system and that the maturity of older youth be considered on a case-by-case basis when considering transfer to adult court (e.g., Steinberg, 2009). Notably, the studies informing this position have considered age, but not the role of puberty and pubertal hormones. Furthermore, the developmental science would suggest that, as development continues into the early- to mid-20’s, more consideration of the developmental status of individuals in their 20’s should take place in the justice system. For immature individuals in their early 20’s, the more treatment-oriented accountability options of the juvenile justice system may be more appropriate than the adult justice system.

The onset of puberty has a normative range of approximately 5 years (Dorn et al., 2006). As such, it is normal for there to be substantial differences in levels of pubertal development between individuals, particularly in the early teenage years. The range of height and physical development seen in any junior high school classroom is a testament to this. Despite this normal reality, legal statutes and guidance typically invoke only age (e.g., age of majority, to drive, buy alcohol, etc.). As noted above, there are some youth that are capable of making adult-like decisions in their mid-teens, but this is not ubiquitous (Steinberg, 2009). Considering development more holistically, using behavioral markers, and potentially endocrine markers, would allow the legal system to operate in a fashion more consistent with developmental science.

Importantly, however, a substantial amount of additional scientific work is needed before developmental science can provide adequately validated behavioral or endocrinological markers. A well-validated measure of development/maturity using behavioral markers seems conceivably within reach with sufficient investment, given the research scales current in use (e.g., the Psychosocial Maturity Index; Cauffman and Steinberg, 2000). The gap between research tools and validated instruments ready for use in clinical or legal contexts is immense, however, and substantial work is needed. Endocrinological markers are further away than behavioral markers. Extensive work is still needed to establish reference norms for children and adolescents and to deal with lingering methodological issues with respect hormone assays (e.g., Granger et al., 2004). Moreover, the convergent validity between the behavioral and endocrinological markers of development/maturity will need to be established.

Adolescence is a time of great change, which conveys the opportunity for extensive change, both positive and negative (Dahl, 2004). The goals of the juvenile justice system are more weighted toward treatment and rehabilitation relative to the adult justice system, with the idea that youth are more malleable than adults (Bartol, 2002). The brain is relatively plastic and changing until the early- to mid-20s (Gogtay et al., 2004; Cohen et al., 2016), which suggests that malleability is present until a later age than currently assumed. Increasing the effort placed on treatment in individuals who were previously not considered treatable may have a positive influence on outcomes. The converse may also be true. By treating 18-year-olds as adults, despite the fact that they are not likely finished maturing, the known detrimental effects of incarceration (e.g., higher levels of depression, substance abuse, reduced life satisfaction; Yi et al., 2017) may be exacerbated.

Issues of development in the justice system cannot be considered in a vacuum; individual differences in development interact with other factors that influence the justice system. For example, systemic racism continues to plague the American justice system (Tonry, 2011). Early onset of puberty is associated with increased likelihood of antisocial behaviors throughout adolescence, particularly low-level offenses such as truancy, substance abuse, vandalism, etc. (for a review see Mendle and Ferrero, 2012). African-American youth are rated as less “innocent” and are viewed as older and less “child-like” than same-age White youth (Goff et al., 2014). Critically, police officers over-estimated the age of both African–American and Hispanic youth, suggesting that this bias exists within law-enforcement (Goff et al., 2014). African-American and Hispanic youth, particularly males, are at increased risk for arrest relative to European-American youth (Kirk, 2008; Brame et al., 2012). The interaction between early puberty and being African-American and/or Hispanic on risk for justice-involvement has not been examined explicitly to our knowledge. However, it seems likely that the risk of justice-involvement in early developing African-American and Hispanic youth will be even higher than the risk of justice-involvement for early developing White youth. This interaction is just one example where the justice system must better integrate developmental science into legal practice and understand the interactions between development and societal issues that also influence the justice system.