Introduction

Adolescence can be seen as a strong contrasting period. In this period people start looking for sensation, but they are also extra sensitive to the development of depression. The question is how brain development is related to this.

Adolescence

Adolescence is the transition from childhood to adulthood. Adolescence starts around the start of puberty and ends when the individual is relatively independent of the parents. This means that the individual can live alone.

Looking at the past, adolescence is seen as a turbulent period. In this period there is a much larger amount of preventable deaths and there is a relatively large amount of psychopathology.

Understanding adolescence

There are two main approaches to understanding adolescents' behaviour and brain development:

• Translation. Behaviour that belongs to adolescence, such as increased time spent with friends, is investigated here. The age-related behaviour entails risks, but also provides important adjustment functions. The changed behavioural pattern associated with adolescence is not specific to adolescents but occurs in many mammals. Because adolescence is an important phase from an evolutionary point of view, animal studies can also contribute to our understanding.

• Transition. As mentioned, adolescence is a transition period between childhood and adulthood. Many different stages of development have been investigated by themselves, with professionals focusing on the young child, for example. However, it is difficult to understand whether a certain process belongs to adolescence when other developmental stages are not considered. The interaction between the various development stages is interesting here.

Various developmental steps are indicated in the article:

• Development non-specific for adolescence

• Development occurring in adolescence

• Development typical of adolescence

Note that the corresponding graphs can also be displayed the other way around: for example, development typically shows a peak during adolescence, but there may also be a trough. Non-specific development can increase linearly, as shown in the figure, but can also decrease linearly.

Neurobiological models

Some neurobiological models are shown in the article:

• Dual system model. The two systems model of willpower forms the basis for this model. According to this model, self-monitoring is the result of the balance between a cold and a warm system. The cool system is emotion-neutral, strategic and flexible, while the warm system is driven by fear, wishes and reflexes. Individual development and stress affect the balance between the two systems.

This model can contribute to our understanding of direct and deferred remuneration. In addition, the dual system model can be used to explain the difference between adolescents and adults in looking for sensation and making risky decisions.

• Triadic model of motivated behaviour. This model explains the differences between adolescents and adults. The limbic (emotional) system is subdivided into two parts:

  • reward (ventral striatum)

  • avoidance (amygdala)

According to this model, motivated behaviour arises from a balance of reward-driven and damage-avoiding behaviour. In adolescence, there is a stronger tendency towards reward-driven behaviour and there is, therefore, an imbalance.

• Imbalance model. Regional neurochemical, structural and functional changes in the brain cause an imbalance in brain circuits. Different brain regions mature earlier and faster than other brain regions. For example, the sensorimotor and subcortical areas are matured faster than the prefrontal cortex. In addition, the sensorimotor cortex reaches its peak in cortical volume in late childhood, while the association cortexes only reach their peak volume in adolescence. The dopamine system, involved in reward, also has a density peak of receptors in early adolescence in the striatum, while this peak in the prefrontal cortex is only reached in adulthood.

The above striking differences are not found in childhood since relatively many brain areas have not yet matured. In adulthood, however, a relatively large number of brain regions have matured, through which even such differences are no longer present.

Self-control

Definition

Self-control is being able to suppress inappropriate emotions, wishes and actions and to display appropriate behaviour instead. A classic example of this is resisting a direct reward to receive a larger reward later (delay of gratification or the Marshmallow task).

The underlying mechanism

The article identifies three areas of the brain that are important for the cognitive and motivational circuit that supports self-control:

• Amygdala. Important for associative learning and determining the value of emotional cues from the environment. This structure can additionally activate the striatum or inhibit it.

• Prefrontal cortex. Here lies the capacity for reasoning and behavioural regulation. This structure modulates the other two structures to suppress actions guided by emotion.

• Ventral striatum. This structure has to do with learning and predicting rewards.

These three structures are modulated by dopamine and the hippocampus.

The brain of the adolescent differs in many ways from that of children and adults. In adolescents, for example, there is reduced top-down regulation of increased emotional responses. In addition, motivated actions are supported by the reduced top-down modulation of the prefrontal cortex.

Incentive for self-monitoring

Self-control seems to increase linearly from childhood to adulthood. Nevertheless, something can be said about this: in the case of incitement, self-control changes. For example, a reward for performance may result in us doing better next time. There are person-specific differences in this area. Individuals can to a greater or lesser extent shift their focus from direct information to a focus on later pay.

Attractive cues

One of the first investigations into reward processing looked at differences in the level of financial remuneration. Both the ventral striatum and the orbitofrontal cortex were found to be sensitive to indications that predicted the greatest reward. This sensitivity was much higher in adolescents than in children and adults.

Adolescents appear to take more risk when they receive direct feedback than adults. Adolescents have a peak in reward sensitivity around the age of fifteen.

Adolescents are less good at suppressing a response to an unexpected positive cue compared to children and adults. This can be seen neurologically in increased activity in the ventral striatum in adolescents.

Incentives based on performance

In research, participants were told that for some achievements, if they were performed properly, they received a financial reward. The expectation of a reward led to a greater improvement in performance in adolescents than in adults. This is represented in the brain as increased activation in the ventral striatum in adolescents. Unfortunately, baseline measures were not taken into account in these studies. Perhaps adults were already performing so well that it was virtually impossible for them to improve their performance in anticipation of a reward. In addition, the reward for adolescents may be subjectively higher for adolescents than for adults.

Research with a point system instead of a financial reward had to circumvent this last problem. This research showed that adolescents with relatively high rewards are good at not responding impulsively.

Social environment

The social environment influences behaviour throughout life. It seems that the influence of the social environment during adolescence is by far the greatest. For example, a large increase in risky behaviour can be observed in adolescents, and not in adults, when peers are present. This can be indirectly associated with the dopamine-rich part of the ventral striatum. It seems that the presence of peers has a behavioural effect. A study into the extent to which peers ensure behavioural reinforcement has been investigated. Every form of positive reinforcement led to faster response time. In adolescents, there was an increased activation pattern in the premotor circuit with positive social feedback, regardless of the outcome. This was not observed in adults and children.

Self-Control in the event of a threat

Adolescents are not fearless: they overestimate the chance of a negative outcome after risky behaviour. Yet they cannot take this into account sufficiently at the moment, probably due to a combination of peers, the environment and their own emotions.

Indications of danger

The best indication of danger comes from the frightened facial expression of another person. Research shows that we learn fears at an early age. How we can express and suppress this depends on development. The first neuroimaging studies on anxiety focused on the amygdala, the structure that external cues must interpret on emotional value. When a fearful cue is found, the individual is prepared to fight or flee (or even freeze). The lateral nucleus projects onto the central nucleus. This then projects onto the brainstem, the hypothalamus, and the autonomic nervous system. This causes the expression of fear. When the danger cue is no longer present, the fear response is suppressed.

Research into the amygdala among adolescents and adults indicates that adolescents have a greater activation pattern with possible danger expressions. Follow-up research indicates that adolescents also have increased activation compared to children. Adolescents also respond more slowly to anxious cues than to neutral or cheerful cues.

Recent studies have shown that adolescents find it difficult to suppress responses to emotional stimuli. Adolescents, especially men, respond more impulsively to threatening cues than neutral cues compared to children and adults. This can also be seen from increased activity in the limbic cortical areas and the ventral striatum.

The increased level of impulsivity in adolescents could be the result of a not yet sufficiently mature prefrontal top-down regulation of the amygdala.

Conditioned fear

The traditional conditioning process involves the combination of a neutral cue with a negative stimulus. After repeated combining, the neutral cue receives a negative charge. Research into conditioned anxiety among children, adolescents and adults showed that adolescents appeared to be less able to extinguish conditioned fear than children and adults. Similar results have been found in research with mice.

Adolescence can be seen as a period in which anxiety associations are less well regulated, which can lead to reduced self-control when the anxiety is present.

Contextual fear

Contextual anxiety depends on being able to learn from threats in the environment. Research into this has been done among mice. Mice in the adolescence period seemed to suffer little from context-dependent anxiety, compared to younger and adult mice. What is special is that when the adolescent mice grow up, they do show the context-dependent fear (which they had already been taught in adolescence) as adult mice. An explanation may be that the amygdala activity has been dulled by changes in the hippocampus during context-dependent conditioning.

Conclusion

The ability to suppress inappropriate emotions, wishes and actions and replace them with appropriate behaviour diminished when striking environmental cues are present. Both in behaviour and in the brain this leads to an increased reactivity in adolescents compared to children and adults.

Why are the brains programmed that way?

It seems logical that brain mechanisms have evolved around socially relevant cues since social status in evolution is very important for survival.

The adolescence period is by no means unique to humans but is observed in many different mammals. During adolescence, it is important to acquire skills to survive independently of others in adulthood. The search for new things means that new sources can be explored and new (types of) relationships can be entered into. Everything leads to the ultimate goal of adolescence: becoming an adult.

The changes in adolescents' development are helped because adolescents are extra sensitive to socially important cues and are less troubled by potential dangers. Together this causes exploration in the world, beyond the safe environment of a home.

Mental health

One in five adolescents has a mental illness, for example, substance abuse. Early substance abuse is a good predictor of later addiction. In addition, narcotics affect the dopamine system. This reinforces the rewarding effect of the dopamine system, which is very active in adolescence.

In addition to substance abuse, there are also many adolescents (around 10 per cent) with an anxiety disorder. This is because being unable to suppress an emotional response when no danger threatens becomes pathological. Adolescents with an anxiety disorder have increased activity in the amygdala and there is a decreased connectivity in the front-amygdala circuit. Desensitization is the most common treatment for an anxiety disorder. This technique works with fear cancellation. Unfortunately, this form of treatment only works in half of the patients. The studies discussed earlier in this article seem to indicate that anxiety killing for adolescents is not the right way to reduce anxiety disorder.

Laws and policies

In recent years, an enormous number of laws and policy documents have been amended in juvenile criminal law. The studies aimed at the development of adolescents and the increased tendency towards risky behaviour and impulsiveness were used. This research indicates that young people should be held responsible for their behaviour, but that there should be a shared responsibility.

Further neuroimaging research into the brain-behavioural relationship in adolescents can make a major contribution to the legal system:

• Immature brain structures during adolescence affect judgment, decision-making, risk-taking, and criminal behaviour.

• Research can contribute to the treatment and rehabilitation of criminal youth.

What is this article about?

Humans are highly social compared to other species. Even 1-year-old toddlers help others. This social behaviour keeps developing and becomes increasingly complex with age. For example, preschoolers mainly use lies for their own benefits, while school-aged children start to lie to protect other’s feelings (white lies).

Developmental changes in social behaviour are related to developmental changes in cognitive functions, such as perspective taking and impulse control. In the current chapter, evidence for the hypothesis that the gradual development of impulse control and perspective-taking skills are related with more strategic thinking and increased incorporation of other’s intentions in social decision-making.

Why use economic games?

Economic games were introduced by behavioural economics. These are used to study psychological and neural mechanisms that underlie social decision-making. Examples are the Ultimatum Game and the Dictator Game. In these games, there are two players. In the Ultimatum Game, one player is given a reward (money, food or anything else). This person is the ‘proposer’, and proposes to split his reward with another player, the ‘responder’. If the responder accepts the proposal, both players receive what was proposed. If the responder rejects, then none of the players receives any reward. In the Dictator Game, the responder is not able to reject the proposal, and thus only receives the amount that the proposer transfers. An assumption of Game theoretical models is that all players should accept all proposals that are above 0, because any free reward would be beneficial. However, this is not true. These games have shown that people do not only care about maximizing their outcomes, they also seem to care about others. For example, in the Dictator Game, the proposer often transfers on average 20-30% of their stakes to the responder, even though they do not have to. In the Ultimatum Game, proposers often choose for an equal split of their stake and responders usually reject offers that are smaller than 20% of the stake. It has also been shown that people do not like to receive less than the proposer, which is called disadvantageous inequity aversion. When people then reject the offer, this might serve as a way to ‘correct’ the inequity.

These paradigms are useful, because they can be used to determine differences between age groups. They also make it easier to quantify complex social behavior which is useful for neuroimaging research. Furthermore, these games can be manipulated so that researchers can gain more understanding of the subcomponents of social decision-making.

How does a preference for fairness develop?

Toddlers are reluctant to engage in prosocial behavior when it is costly to themselves. Studies employing the Dictator Game have shown that although children tend to keep most of the resources to themselves, the size of their donations increase between the ages of 3 and 8. By the age of 9, their donations do not differ from donations made by adults. Thus, there is an increase in costly sharing. This is not due to differences in knowledge about what is fair; infants as young as 15 months already expect resources to be distributed equally. Even though 3-year-olds do not differ from 8-year-olds in beliefs about equal division of rewards, they still keep more than 50% of the rewards to themselves in a Dictator Game. During this age, the willingness to send rewards increases. This means that a developing sense of fairness leads to that children engage in more equal splits.

In addition to fairness considerations, there are also strategic concerns. For instance, adults offer higher shares of the stake when they know that the second player can punish (reject) unfair offers. One can thus look at differences in the Ultimatum Game and Dictator Game as a measure of strategic social behavior. During late childhood (7-10 years), children make higher Ultimatum Game proposals compared to Dictator Game proposals, but their proposals are still smaller than those proposed by adults. During adolescence, the difference between the two games becomes greater, which means that there is a developmental increase in strategic behavior across adolescence. In sum, the prosocial tendency to share with others emerges early, but social behavior becomes increasingly strategic across childhood and adolescence.

What are the cognitive mechanisms underlying developmental change in strategic social behavior?

It has been shown that strategic social behavior is dependent on the capacity to implement behavioral control; a selfish impulse (keeping all resources to oneself) has to be controlled. Proposers also need to be able to take the responder’s perspective into account, to predict what proposals they might accept or reject. This perspective-taking is often called ‘theory of mind’, which young children often are not capable of. In sum, behavioral studies that have employed these game paradigms show that cognitive development which are both related to impulse control and perspective taking play a crucial role in understanding age-related changes in social behavior.

What neural networks are involved in social decision-making?

There are three brain networks which are important in social decision-making: a basic affective network, a cognitive regulatory network, and a ‘mentalizing’ network. The basic affective network consists of the anterior insula, ventral striatum, and the amygdala. These determine whether social stimuli should be approached (positive affect) or avoided (negative affect). Increasing activity in the insula has been related to unreciprocated trust and receiving unfair offers. These brain regions interact with a cognitive regulatory network which consists of the dorsal anterior cingulate cortex (dACC), and regions in the prefrontal cortex (PFC), such as the ventrolateral prefrontal cortex (vIPFC) and the dorsolateral prefrontal cortex (dlPFC). These areas are related to cognitive control over impulses and can help individuals to act in a goal-directed manner when they experience a conflict between self-interest and social norms. These control-related areas are thus crucial for the regulation of strategic social behavior. Both the affective and control regions interact with a third system, the ‘mentalizing’ network. This includes the left and right temporoparietal junction (TPJ), superior temporal sulci, ventral, and dorsal regions of the medial PFC and the temporal poles. These regions are often active in tasks that involve thinking about other people’s mental states, when people have to infer other people’s thoughts, beliefs or desires. Lastly, taking other people’s perspectives in economic games has been associated with activation in regions of the mentalizing network, such as the TPJ and the dorsomedial PFC.

How can we understand social behavior from a developmental neuroscience perspective?

It has been shown that different brain regions mature at different ages. Sensorimotor regions in the occipital and parietal lobes mature first. The dlPFC and the TPJ are the latest brain regions to fully mature, which might explain the protracted developmental pattern in associated functions such as cognitive control and perspective taking. Affective networks mature early, while regions of the cognitive regulatory network mature late. This can explain why affective reactions to unfairness are visible early in development and why strategic considerations during social decision-making develops later. This is in line with studies that have shown that detection of violations of fairness and neural responses in the insula and the dACC mature before adolescence, and that intentionality understanding in fairness decisions develops across adolescence and is related with neural activity that are important for perspective taking (TPJ) and impulse control (dlPFC). It has been suggested that early adolescents tend to make more self-oriented choices, while late adolescents tend to make more other-oriented choices. This is due to differences in the development of the dmPFC and the TPJ.

Introduction

Adolescents behave more at risk than children or adults. This risky behaviour can be characterized as behaviour that can be prevented (for example, binge drinking or traffic accidents due to wild driving). Many professionals state that this behaviour poses the greatest threat to the well-being of adolescents. The risky behaviour that adolescents display differs from the risky behaviour of adults in terms of social background (adolescents often behave more at risk when peers are present) and the quantity. Recent research shows that the risky behaviour of adolescents is directly influenced by the presence of peers: as soon as peers are present, adolescents are more at risk. This has not been found in adults. It appears that two neuronal systems are involved in the tendency toward risky behaviour in adolescents:

• The spur processing system (ventral striatum including the nucleus accumbens and the orbitofrontal cortex).

• The cognitive control system (lateral prefrontal cortex).

Previous research into impulsive and risky behaviour has often observed increased activity in the brain areas involved in incentive processing. In addition, studies found reduced activity in brain structures involved in cognitive control. Both brain systems are subject to many changes during adolescence. The first system undergoes many changes, particularly in the early adolescence period, due to the distribution and density of dopamine receptors. This also explains the greater sensitivity to reward in this period.

The cognitive control system matures more slowly and also takes longer to do so. This is partly due to the reduction of grey matter and the increase of myelin layers. This process starts before the onset of adolescence and lasts until at least the 25th year of life. Due to maturation, adolescents are becoming better at positions related to executive functioning and flexibility.

It seems that the risky behaviour of adolescents is caused by an imbalance between the maturation of the aforementioned systems. In the current study, adolescents, young adults and adults measured brain activity when making decisions in a simulated traffic light task. The ultimate goal of this game is to achieve the highest possible financial reward in the shortest possible time. Every participant played the game with and without peers looking at them.

Methods

Attendees

40 people participated in this study: 14 adolescents between 14-18 years (including 6 women), 14 young adults between 19-22 years (including 7 women) and 12 adults between 24-29 years (including 6 women).

Procedure

The traffic light task is a driving task where the participant must keep control of the vehicle from the driver's position. Each participant did the game four times (twice with, and twice without spectators). Each round had twenty intersections (so there were twenty trials per round) and lasted less than six minutes. By pressing a button, the participant could indicate whether or not to stop at the intersection, while the traffic light went from green to orange to red. By varying the timing of the traffic lights and the chance of a collision when driving through red, the game remained unpredictable. Risky behaviour was rewarded financially (by a higher amount if the participant was faster). Waiting for green light resulted in less time due to the time delay. However, when the risky behaviour caused a collision, the loss of time was greater (and the financial reward smaller) than if the participant waited for the green light.

Each participant had to bring two friends of the same gender and in the same age group. In the spectator condition, the participant was told that his or her friends were watching in an adjacent room. This entailed a surprise effect since the participants were not informed of this in advance. During the break for the spectator condition and between the two spectator rounds, the participant and the spectators were able to speak to each other via the intercom. The peers were instructed about what they were allowed to say and what they were not allowed to say to prevent behavioural influences.

After the traffic light task (where the participant was in an MRI scanner), the participants were asked to complete questionnaires. These questionnaires investigated the degree of impulsiveness, sensation seeking and sensitivity to the influence of peers.

Results

Behavioural

Adolescents and (young) adults showed similar behaviour in the condition without spectators. However, this was not the case in the condition with spectators. The participants in the adolescent group were the only ones who took more risk when their friends watched. They drove more often through red and also had an accident more often.

Regression analysis showed that behaviour was significantly related to the degree of sensation seeking. There was no significant relationship between behaviour and impulsivity. Task performance is therefore mainly influenced by the extent to which someone seeks to reward and sensation.

fMRI

In the current study, particular attention was paid to those areas of the brain, which left the main effect of age or an interaction effect of age and condition.

In areas in the left prefrontal cortex, there was more brain activity in adults than in adolescents.

The influence of peers on adolescents was evident in increased activation in the ventral striatum and the orbitofrontal cortex. This indicates that the presence of peers makes the incentive processing more sensitive. These results were not found in (young) adults.

In addition, different regions of interests (ROIs) were looked at to make the decision-making process dependent on the riskiness of the given situation. In adolescents, there was more activation in the ventral striatum and the orbitofrontal cortex, related to risky decisions. This was not found with the (young) adults.

The self-report of sensitivity to peers turned out to be significantly related to the activation patterns in the ventral striatum. This applied to all age groups. 

Discussion

The results from the current study replicate earlier research findings. The unique way of measuring the influence of peers in the current study contributes to a better understanding of the effect of peers' influence on the decision-making process during adolescence. The riskier behaviour when peers looked at it cannot be explained by the encouragement of these peers, as there was no contact during the task and the contact between the tasks was subject to strict rules. An explanation that may be possible is a neural sensitivity due to the incomplete maturation of the brain.

The difference in risky behaviour between adolescents and (young) adults in the current study can only be attributed to the presence of peers since all three groups showed similar behaviour in the condition where no spectators were present. In adults, the presence of peers does not even seem to be rewarding at all. In addition, they appear to be better able to use the left prefrontal cortex to inhibit the reward system to make strategic decisions. Regardless of the presence of peers, there was more activation in the aforementioned brain area in adults than in adolescents.

The social influence on decisions made by adolescents can be seen as an interaction between social information processing and the processing of reward.

Adolescents have increased activation in the brain areas involved in incentive processing in response to social stimuli such as face recognition, compared to children and adults. Because of the rewarding effects of the presence of peers, this can lead to a reinforcing stimulus towards the incentive processing system, whereby risky behaviour is experienced as extra rewarding.

It cannot be claimed with certainty that the increased activation in the ventral striatum in adolescents in the current study is purely due to the presence of peers. An alternative explanation can be given, for example, by talking about social uncertainty or greater sensitivity to distraction. These explanations are less likely since other brain regions should also show different activation patterns during the current study.

Bullet point summary:

• Regulating aggression after social feedback is an important pre-requisite for developing and maintaining social relations, especially in the current times with larger emphasis on online social evaluation.
• Studies in adults highlighted the role of the dorsolateral prefrontal cortex (DLPFC) in regulating aggression. Little is known about the development of aggression regulation following social feedback during childhood, while this is an important period for both brain maturation and social relations.
• The current study used a longitudinal design, with 456 twins undergoing two functional MRI sessions across the transition from middle (7 to 9 y) to late (9 to 11 y) childhood. Aggression regulation was studied using the Social Network Aggression Task.
• Behavioral aggression after social evaluation decreased over time, whereas activation in the insula, dorsomedial PFC and DLPFC increased over time.
• Brain–behavior analyses showed that increased DLPFC activation after negative feedback was associated with decreased aggression. Change analyses further revealed that children with larger increases in DLPFC activity from middle to late childhood showed stronger decreases in aggression over time.
• These findings provide insights into the development of social evaluation sensitivity and aggression control in childhood.

When do children develop emotion regulation skills?

Regulating emotions during social interactions is one of the most important requirements for developing social relationships in childhood. With increasing age, children become better at regulating their emotions. Even though several studies have examined regulation processes in the context of social evaluation in adolescence, few studies have investigated the development of social emotion regulation during childhood. The current study builds upon new insights in the neural processing of social emotion regulation by examining change in neural and behavioral social control in a longitudinal functional MRI (fMRI) study in middle-to-late childhood.

What does social evaluation entail?

Neuroimaging research has shown that the significance of social evaluation is deeply rooted in our brain. Social evaluation, including social acceptance and rejection, has previously been studied using ecologically valid social judgment paradigms. It is well documented that social rejection can lead to ag- gression and retaliation. Controlling emotions elicited by social evaluation feedback relies on cognitive control, that is, individuals with better cognitive control functions show less aggression following rejection. Increased activation in the dorsal ACC and AI was related to less aggression after social rejection in adults with high executive functioning, whereas adults with low executive functioning showed increased aggression with increasing neural activation. Prior studies in adults further showed that the dorsolateral prefrontal cortex (DLPFC) might serve as a regulating mechanism for aggression after social evaluation, such that increased DLPFC activity after social rejection was related to less behavioral aggression. Interestingly, prior theoretical perspectives have suggested that DLPFC maturation is an important underlying mechanism for developing a variety of control functions in childhood. However, no study to date has examined longitudinal developmental changes in these brain regions in childhood in the context of social evaluation.

What did the current study measure?

The current study makes use of a developmental twin sample. This ongoing longitudinal twin study examines the development of social evaluation and behavioral control in 7- to 13-year-old children. The current study includes the first two fMRI assessments, separated by 2 years. Using linear mixed-effects modeling, it was first investigated how behavioral aggression after positive, negative, and neutral social feedback changed over time. Next, it was investigated changes in brain responses related to positive, negative, and neutral social feedback longitudinally and examined brain–behavior associations.

What were the findings of the current study?

The current study revealed three main findings:

• Behavioral aggression after social evaluation decreased over time, and this decrease was most pronounced for behavioral responses after positive and neutral social feedback.
• Confirmatory ROI analyses showed that increased activity in AI was related to more aggression following social feedback (regardless of its valence), whereas increased activity in DLPFC was correlated with less aggression.
• Bilateral DLPFC activity was correlated to less subsequent aggression following negative social feedback. Longitudinal comparisons confirmed that larger increases in DLPFC activity across childhood were related to larger decreases in behavioral aggression, in particular after negative social feedback.

The behavioral results confirmed the initial hypothesis that behavioral aggression decreases over time, consistent with prior reports on age related increases in behavioral control. Taken together, this study set out to test longitudinal changes in neural systems underlying social evaluation and aggression regulation and their relation to behavioral outcomes. We found an increase in behavioral control across childhood, as behavioral aggression decreased over time. Moreover, DLPFC activation was related to a decrease in behavioral aggression. Notably, children with larger increases in DLPFC activity across 2 years displayed the largest decrease in behavioral aggression over time. These results contribute to our understanding of how the developing brain processes social feedback and suggest that the DLPFC might serve as emotion regulation mechanisms in the context of negative social feedback.

What were the main conclusions of the current study?

The current study investigated 456 twins undergoing two functional MRI sessions across the transition from middle (7 to 9 y) to late (9 to 11 y) childhood. During this time, their behavioral aggression after social evaluation decreased, whereas activation in the insula, dorsomedial PFC and DLPFC increased over time. Brain–behavior analyses showed that increased DLPFC activation after negative feedback was associated with decreased aggression. Change analyses further revealed that children with larger increases in DLPFC activity from middle to late childhood showed stronger decreases in aggression over time.

Bullet point summary:

• Adolescence is the peak period for the incidence of anxiety disorders. Recent findings have revealed the immaturity of neural networks underlying emotional regulation in this population. Brain vulnerability to anxiety in adolescence is related to the unsynchronised development of anxiety-relevant brain functional systems.
• However, our current knowledge on brain deficits in adolescent anxiety is mainly borrowed from studies on adults. Understanding adolescent-specific brain deficitsis essential for developing biomarkers and brain-based therapies targeting adolescent anxiety.
• This article reviews and compares recent neuroimaging literature on anxiety-related brain structural and functional deficits between adolescent and adult populations, and proposes a model highlighting the differences between adolescence and adulthood in anxiety-related brain networks.
• This model emphasises that in adolescence the emotional control system tends to be hypoactivated, the fear conditioning system is immature, and the reward and stress response systems are hypersensitive.
• Furthermore, the striatum’s functional links to the amygdala and the prefrontal cortex are strengthened, while the link between the prefrontal cortex and the amygdala is weakened in adolescence.
• This model helps to explain why adolescents are vulnerable to anxiety disorders and provides insights into potential brain-based approaches to intervene in adolescent anxiety disorders.

What are anxiety disorders?

Anxiety disorders mainly manifest as excessive fear, worry and avoidance that induce severe emotional distress, somatic diseases, and cognitive and behavioural impairments, and in turn damage normal social functioning and negatively affect quality of life. Often, anxiety disorders have an early onset, which may be related to the developmental trajectory of the adolescent brain. The brain structure changes significantly from childhood to adolescence in terms of myelination and synapse pruning. Hormones of puberty, together with pressures from the external environment, reshape the central neural system. These developmental processes and abnormalities may trigger and/or mediate the onset and progression of anxiety disorders in adolescence.

What is the goal of the current article?

In this article, we would review anxiety-relevant abnormalities in the developing brain and attempt to propose a psychopathological model of neural systems underlying anxiety disorders for adolescence, emphasising the differences between adoles- cents and adults.

What brain structures are associated with anxiety disorders?

Previous studies have shown that the amygdala, prefrontal cortex, bed nucleus of the stria terminalis (BNST), hippocampus, striatum, anterior insula, anterior cingulate cortex and hypothalamus were closely related to anxiety disorders. Notably, the functional connectivity between the amygdala and the anterior insula is associated with the degree of avoidance. These structures, which are closely related to anxiety, may play unique roles in the development of cognitive and emotional capabilities among adolescents.

• Previous literature supports that the amygdala is related to fear learning and that its pathological increase in volume is a sign of anxiety disorders.
• The activation of the medial prefrontal cortex (mPFC) in adolescents is weakened during extinction recall. The activation of the mPFC of anxious adolescents, however, is stronger than that of the healthy group during negative emotion processing. High levels of anxiety symptoms are related to delayed development of the neural circuits including the prefrontal cortex in children and adolescents.
• In adolescents, decreased hippocampus volume is one of the risk factors for anxiety disorders.
• The BNST mediates the regulation of anxiety by neural circuits underlying substance abuse.
• The striatum contributes to anxiety symptoms and anxiety-related bias in emotional, motivational and attentional processes, since it has vital contributions to these processes, and striatum-based functional connectivity differs between anxious adolescents and healthy adolescents.
• The hypothalamus is a critical structure involved in the anxiety circuit. It is also a central part of the HPA axis responsible for regulating emotions, defensive behaviour, aggression and stress responses. The dysregulation of HPA axis activities during adolescence increases the risk of developing anxiety disorders.

What neural networks underlie adolescent anxiety?

Several neural networks underlie anxiety disorders in adolescents:

• The neural circuit between the mPFC and the amygdala is closely related to cognitive control. In adolescence, the maturation of the mPFC is later than that of the amygdala, and the ability of top-down regulation is immature and weak. Hence, insufficient inhibition of the amygdala neurons may provide a potential neural basis for adolescent anxiety disorders.
• The ventral hippocampus, basolateral amygdala and the mPFC form an interconnected circuit that plays a vital role in fear learning and extinction. The ventral hippocampus and the basolateral amygdala rely on each other coding fear-related memories. Glutamatergic inputs from the basolateral amygdala to the ventral hippocampus pyramidal neurons increase individual anxiety, while inhibition of this projection reduces anxiety-related behaviours.
• The neural networks that include the BNST, amygdala and prefrontal cortex participate in the anticipation of uncertain threats. The excessive and persistent anxiety response of patients with anxiety disorders to uncertain information may be underlain with neural projections from the basolateral amygdala to the BNST.
• Reward representations are available to the ventral striatum that participates in forming motivational and goal-oriented behaviours. The amyg- dala actively regulates the striatum. The direct projection from the amygdala to the striatum supports ‘fight or flight’ motor responses, as well as avoidance learning.
• Under stressful conditions, amygdala- to-hypothalamic outputs maintain anxiety behaviours.

How do differences in brain networks vary across subtypes of anxiety disorders?

Morphologically, patients with social anxiety disorder (SAD) have a larger grey matter volume in the dorsal striatum. They have reduced frontal lobe volume and increased amygdala volume relative to healthy controls. Patients with generalised anxiety disorder (GAD) have reduced ventromedial prefrontal cortex volume and hypothalamus volume. Patients with panic disorder (PD) have smaller grey matter volumes in the amygdala, the hippocampus, the prefrontal cortex and the bilateral striatum.

How does adolescence influence the increased risk on anxiety disorders?

Several developments in adolescence influence the increased risk on anxiety disorders in this period.

• From adolescence to adulthood, the neural connections in anxiety neural networks change dramatically. Functional and structural connections between the prefrontal cortex and the amygdala are reorganised during adolescent development. Reduced functional connectivity between the amygdala and the prefrontal cortex indicates failure of top-down control in the prefrontal cortex, which is a potential pathological factor for anxiety disorders in both adolescents and adults.
• Hippocampal volume increases and then decreases from adolescence to adulthood. During adolescence, the structural connections between the amygdala and the mPFC develop, which is related to fear conditioning. For adolescents, excessive functional connectivity between the prefrontal cortex, the amygdala and the hippocampus is associated with a higher risk of anxiety disorders. For adults, the lack of hippocampal inhibition of the prefrontal cortex, which is common in typically developing adolescents, may increase the risk of mood disorders in adults.
• The relationship between the development of the BNST and anxiety disorders in adolescents, as well as the neural connections of the BNST and their functions during adolescence, is underdocumented and needs further investigation.
• Then, the functional connectivity between the amygdala and the striatum in response to emotional cues is declining during development, and is associated with enhanced cognitive control. Adolescents with GAD show increased functional connectivity between the striatum and the amygdala, accompanying higher sensitivity to reward-related stimuli.
• Lastly, for adolescents, experiencing stress leads to overexpres- sion of corticotropin-releasing factor receptor in the hypothalamus, the amygdala and the prefrontal cortex, causing abnormalities in the HPA axis that will persist into adulthood. Significantly increased responses to stress in adolescents compared with adults, as evidenced by prolonged hormonal exposure, are possibly responsible for the increased vulnerability to psychiatric disorders seen during adolescence.

What are the most important conclusions about the developing adolescent brain in relation to anxiety disorders?

This article reviewed and compared recent neuroimaging literature on anxiety- related brain structural and functional deficits between adolescent and adult populations, and proposes a model highlighting the differences between adolescence and adulthood in anxiety-related brain networks. This model emphasises that in adolescence the emotional control system tends to be hypoactivated, the fear conditioning system is immature, and the reward and stress response systems are hypersensitive. Furthermore, the striatum’s functional links to the amygdala and the prefrontal cortex are strengthened, while the link between the prefrontal cortex and the amygdala is weakened in adolescence. This model helps to explain why adolescents are vulnerable to anxiety disorders and provides insights into potential brain-based approaches to intervene in adolescent anxiety disorders.