The central nervous system begins to develop in the human fetus 2 to 3 weeks after conception through the folding and fusion of ectoderm to create the neural tube.³ Around the fifth week of gestation, the neural tube closes and ectodermal tissues begin to differentiate into precursors of various brain structures according to their longitudinal and circumferential locations.² The subsequent major maturational events of early brain development include the birth and death of neurons, neuronal migration, the elaboration and pruning of axons and dendritic arbors, the formation and elimination of synaptic contacts between neurons, metabolic and molecular interactions of neurons with glia, and myelination of axons. These are dynamic processes that have their own unique maturational time tables within and across brain regions (Fig. 1).

By the eighth week of gestation, stem cells in the ventricular (also called proliferative) zone, located adjacent to the precursor of the lateral ventricles, differentiate into the site for the division and origin of cortical and subcortical neurons.^5–7 Between weeks 12 and 20 of gestation, neurons migrate from the ventricular zone and a newly formed adjacent region called the subventricular zone along a scaffolding of glial cells toward their destinations in the cortex (Fig. 2).^8,9 Preliminary application of magnetic resonance imaging (MRI) to the study of human fetuses has shown visible evidence for the migration of neurons during the first half of gestation.^10,11

Between 8 and 16 weeks’ gestational age (GA), migrating neurons from the proliferative zones transiently connect to the subplate, a “waiting station” located directly underneath (i.e., on the ventricular side) of the cortical plate, the region that will become the cerebral cortex. Neurons in the subplate receive connections from afferent neurons originating in the thalamus, basal forebrain, and brainstem, temporary connections that are generated ahead of the correct cortical targets for these neural projections.^2,12–16 At 17 weeks’ GA, this layer appears on human fetal MRI as a transient layer visible beneath the cortex (Fig. 3A).¹⁰ By 20 weeks’ GA, axons within the subplate break their connections there and form new, more permanent synapses within the cortical plate, thereby triggering the death of subplate neurons. This refinement of cortical connections reaches its peak at 24 to 28 weeks’ GA¹⁶ and coincides with increased functional organization of cortical circuits, as quantified by electrophysiological measurements in human infants.¹⁷ Simultaneously, the subplate ceases to appear on fetal MRI scans.¹⁰ Neurons in the cortical plate organize into vertical layers that are distinguishable by microscopic examination of postmortem tissue but that, thus far, have not been visualized successfully by in vivo imaging techniques. Starting around 20 weeks’ GA in humans, sulci and gyri become visible on fetal MRI (Fig. 3B).¹⁰

The number of neurons in the human brain peaks around the 28th week of gestation at levels 40% greater than in the adult. Dendritic growth and arborization begin to accelerate rapidly at this time, ^1,18–26 which corresponds with the disappearance of the proliferative zone and cortical subplate on MRI and with an increase in cortical thickness visible on both postmortem analysis and MRI (Fig. 3).¹⁰ Studies of the timing of dendritic growth and synaptogenesis across different brain regions have yielded variable findings. Studies of postmortem human brains have suggested that development in the frontal cortex is later than in more posterior regions of the brain, such as the visual cortex.^27,28 In contrast, quantitative analyses of synaptogenesis in the motor, somatosensory, visual, prefrontal, and limbic cortices of macaque monkeys and human fetuses have indicated that the course of development is similar across these regions, particularly during the time of increasing synaptogenesis.^29–33 Arborization of dendrites and axons is not visible directly on MRI, but they are believed to account for a significant portion of the increase in thickness of the cortex with age. The rate of synaptogenesis also increases rapidly at this time, peaking after the 34th week of gestation in humans at a rate of 40,000 new synapses per second.² Synaptogenesis continues in postnatal life, although as the increasing rate of synapse elimination begins to overtake the declining rate of synapse formation, the net number of synapses begins to decrease at puberty.^1,18,24–26 Age- related changes in positron emission tomography (PET) measures of glucose metabolism seem to parallel the rate of synaptogenesis in humans^34,35 and in nonhuman primates, ³⁶ suggesting that PET studies may provide an indirect measure of this neuronal process.

Myelination is accompanied by the proliferation and differentiation of oligodendrocytes that are necessary for neuronal insulation and metabolism. Myelination of the optic radiations and occipital white matter begins 1 to 2 months before birth in humans and extends gradually to the frontal lobe by 9 months’ postnatal age.^2,37 Myelination in the cortex proceeds in a posterior-to-anterior direction and seems to follow maturation of functional circuits, with sensory pathways myelinating first, followed by motor pathways, and finally by as-sociation areas.²⁸

Apoptosis (programmed cell death) and synaptic elimination are distinct processes that play important roles in the development of the primate brain during the second half of gestation and through the first few years of life.² The exact timing and dynamics of these processes in humans are unknown, and in fact, no longitudinal studies of synapse formation or regression in humans have been reported. Consequently, whether synaptogenesis and the elimination of cells and synapses counterbalance one another between 2 and 7 years of age during a so-called plateau phase,² or whether the net number of synapses increases through these years and peaks between 4 and 6 years, is unclear.³⁸ What does seem clear, however, is that the regression of synapses predominates thereafter, being particularly prominent throughout adolescence.² Synaptic pruning is currently not visible directly with neuroimaging, although the prominent thinning of frontal and parietal cortices during adolescence is thought likely to be attributable to synaptic pruning.^1,39 Pathological synaptic pruning may contribute to the genesis of at least some psychiatric disorders, although this hypothesis has not yet been confirmed empirically.^2,33,40

One of the first longitudinal studies of regional brain volumes in children was conducted by investigators in the Child Psychiatry Branch of the NIMH. Growth curves of volumes in various cerebral lobes were compiled for 145 children scanned at approximately 2-year intervals.⁴⁹ These data included two scans each from 65 children, three scans each from 30 children, four scans from two children, and five scans from one child. Linear increases in white matter volume up to age 20 years were detected in all brain regions, whereas quadratic effects of age were detected for cortical gray matter volumes in frontal, parietal, and temporal lobes, indicating that gray matter volumes in those regions followed an “inverted U-shaped” developmental curve. Gray matter volumes increased before adolescence, and then they peaked at approximately 12 years of age in frontal and parietal cortices and at 16 years of age in the temporal lobe. Gray matter volumes decreased in all regions thereafter (Fig. 4). Although an oft-cited finding, the differences between these developmental peaks have not been assessed statistically and therefore require further study and confirmation.

A more recent study from the same group of investigators used a finer-grained morphological analysis to quantify cortical development.⁵⁰ Gray matter density was measured point by point on the surface of the brain in each lobe. A map was then constructed using 52 MRI scans from 13 children who were scanned at 2-year intervals during a 10-year period (Fig. 5). Changes in gray matter density, or the proportion of gray matter within a given volume of cortical tissue, were captured in a time-lapse movie for data acquired between 4 and 21 years of age (http://www.loni.ucla.edu/~thompson/DEVEL/dynamic.html). These dynamic analyses showed that cortical thickness generally decreases with advancing age in “back-to-front” progression. Thinning of gray matter occurred first in sensorimotor areas, spreading subsequently and progressively into dorsal and parietal, superior temporal, and dorsolateral prefrontal cortices throughout later childhood and adolescence. Thus, cortical areas that subserve sensorimotor processes mature earliest, whereas those that mediate higher-order functions mature last. These findings are remarkably consistent with those reported in a cross-sectional study over a wider age range.⁵¹ The cortical thinning was interpreted as a likely consequence of the pruning of neural connections that has been documented in animal models of development.⁵² Unlike the prior study that defined coarse lobar divisions, however, this study used a finer-grained method of morphological analysis, although of a largely overlapping data set. It did not report age-related “peaks” in mean global volumes of gray matter but instead reported steady decreases in cortical density across adolescent development. The NIMH investigators who conducted these studies explained that, with more spatiotemporal detail, this newer method detected maturational trajectories that varied temporally across brain subregions and likely was responsible for the differing rates of change in these subregions from those reported in their prior volumetric study. When lobar volumes were measured, rates of growth and loss were likely averaged over large expanses of tissue, and this averaging may have distorted measures of overall rates of change within brain subregions. Nevertheless, subsequent studies by the same group of investigators applied complex growth models to the same overlapping data set and reported peaks in cortical thickness, with primary sensorimotor areas attaining their peak thickness before secondary areas, followed by higher-order association areas.^53,54

Other studies have used similar, spatially fine-grained methods of morphological analysis to measure changes in cortical thickness with development.^39,55 One longitudinal study consisted of data from 45 children between the ages of 5 and 11 years who were scanned twice within a 2-year period. Changes in cortical thickness and overall brain size were assessed.³⁹ Brain growth was most prominent in the prefrontal, temporal, and occipital regions bilaterally. Cortical thinning was observed in the right dorsolateral prefrontal and parieto-occipital regions, whereas cortical thickening was detected in classic language areas (Broca and Wernicke areas) of the frontal and temporal lobes. In a follow-up study of the same sample of children, gray matter thickening in the left inferior frontal gyrus was associated with improving phonological skills, but not with improving motor skills.⁵⁶ These findings suggested that cortical thickening may represent the proliferation of synaptic connections that helps to form the neural networks that subserve the acquisition of skills required for the development of language in humans. Nevertheless, both cortical thinning and growth in regional brain volumes seemed to occur in the same regions (particularly in the right dorsolateral prefrontal and parieto-occipital cortices) and at the same time,³⁹ suggesting myelination as one possible explanation for cortical thinning. In addition to synaptic pruning, which reduces the number of synapses in the brain, ongoing myelination of axons by oligodendrocytes would likely increase overall brain size.⁴⁸ Increasing myelination would also contribute to regional cortical thinning if the axons within the cortical gray matter that were unmyelinated earlier in childhood become myelinated later with advancing age. Consistent with this interpretation of increasing myelination rather than synaptic pruning as the cause of regional thinning of the cortical mantle, thinning of the left dorsal frontal and parietal cortices correlated with improved performance on various cognitive tasks,³⁹ suggesting that progressive myelination and its associated cortical thinning could be responsible for the improvements in cognitive performance that suggested improved neural processing within prefrontal and parietal cortices.

These finer-grained morphological techniques are now used frequently to study the differences in the trajectories of brain development of healthy children from those afflicted with neuropsychiatric disorders, such as childhood-onset schizophrenia (COS)^57,58 or attention-deficit/hyperactivity disorder (ADHD).^59,60 In a longitudinal study of 12 patients with COS and 12 healthy comparison subjects who were scanned at 2-year intervals during adolescence, accelerated gray matter loss was detected, progressing in a back-to-front direction, which was similar to the pattern observed during normal brain maturation.⁵⁰ These findings suggested that gray matter loss in COS may reflect an exaggeration of the normal brain maturational process of synaptic pruning, perhaps supporting a hypothesis of “excessive pruning,”⁶¹ whereby a fault in synaptic elimination during adolescence is thought to produce schizophrenia, consistent with recent evidence that cortical thickness in COS predicts clinical outcome in adulthood.⁵⁸ In contrast, findings from a recent longitudinal study of a large sample of children and adolescents with ADHD revealed a delayed, rather than a premature, pattern of cortical development relative to healthy controls that was related to clinical outcome.⁶⁰ This maturational delay was most prominent in prefrontal regions that mediate the higher-order, cognitive functions that are known to be impaired in ADHD.⁶²

Understanding whether normative maturation of brain structure extends into later life requires studying individuals whose ages vary widely. Information gleaned from a study of brain maturation across the life span has important implications for theories of cognitive development and of cognitive decline with aging. Findings from a study of 176 healthy individuals ranging in age from 7 to 87 years revealed that the trajectories of maturational changes described above likely continue beyond adolescence and into young adulthood.⁵¹ A nonlinear decrease in cortical thickness with age was detected over the dorsal frontal and parietal cortices, with the most rapid thinning continuing until age 45 years, after which the cortical thickness remained generally constant (Fig. 6). Because the total volume of white matter continued to increase until the fifth decade, this decrease in gray matter density could have been caused by the progressive myelination of the underlying cortices, rather than by synaptic pruning. Alternatively, this decrease may have been caused by an age-related decrease in neuropil (the unmeylinated dendrites, axonal arbors, and synapses within the gray matter), consistent with findings of a rapid decrease in the total number of synapses with increasing age in nonhuman primates.⁶³ In addition, cortical thinning that began at approximately 45 years and that progressed quickly was detected in the posterior temporal cortices of the left hemisphere⁵¹ (Fig. 6), consistent with evidence that some language functions, such as language production and word retrieval, decline during normal aging.⁶⁴

In addition to reporting the regional specificity in timing of gray matter development, the first longitudinal study described above also reported evidence for a sex-specific effect in volumes of gray matter, possibly peaking 1 to 2 years earlier in girls than in boys.⁴⁹ Formal statistical analyses, however, indicated that the shapes of these developmental trajectories did not differ significantly across sexes (Fig. 4; i.e., the interaction of the quadratic age effect with sex was nonsignificant). Moreover, the study did not report overall differences in brain size between males and females, nor did it account for those differences in the comparisons of regional volumes across sexes. A follow-up study of an expanded sample from the same group of investigators compared the trajectories of regional shapes and volumes across sexes.⁷¹ This study included 829 scans from 387 subjects ranging in age from 3 to 27 years. After accounting for the effects of differing overall brain size, volumes of frontal gray matter were proportionately greater in females, and volumes of occipital white matter were proportionately greater in males, across all ages studied. The trajectories in temporal and parietal lobes, however, were similar across males and females.

Findings from a cross-sectional study of 176 individuals between 7 and 87 years of age revealed thicker cortices in females in the right inferior parietal and posterior temporal regions.⁷² These differences remained significant in a subgroup of 36 individuals who were matched for total brain volume and age (Fig. 7), confirming that thicker cortices in temporoparietal regions in females relative to males were independent of differences in brain size. Age-by-sex interactions were not significant in temporoparietal cortices, however, suggesting that thinner cortices are present in males from childhood to adulthood. Given that cortical thinning in these regions during development may arise from either greater synaptic pruning⁴³ or myelination⁷³ or both, and that cortical thinning may thereby contribute to more efficient neural processing,³⁹ thinner cortices in men in these regions may contribute to the superior visuospatial abilities that they seem to have at all ages. These findings differ from those of the NIMH longitudinal study in which sex differences in temporal and parietal gray matter volumes were not observed. Whereas that study used a fully automated method to calculate gray matter thickness, this recent study used “semiautomated” methods in which 35 sulcal and gyral landmarks were manually traced on each image from each individual. A validation study comparing the fully automated method with manual segmentation reported volumetric differences between 10% and 15%.⁷⁴ Thus, a fully automated method may be sufficiently accurate to detect large effects but is unlikely to be sufficient to detect these smaller effects of age–sex interactions within temporoparietal cortices.

The increasing directional restriction of the diffusion of water in the brain with advancing age has been a reasonably consistent finding across studies of healthy children (Fig. 8).^80–82 This restriction in the diffusion of water increases dramatically immediately before the onset of myelination is visible macroscopically, and it increases more gradually thereafter as myelin is added to axonal sheaths, suggesting that age-related changes in myelination likely produce the increasing constraints on diffusion. Myelination is thought to enhance the speed and fidelity of the transmission of information encoded in action potentials that propagate along neurons, likely contributing to age-related improvements in cognition.

Many DTI studies of infants and young children show that the brain undergoes rapid microstructural changes from birth to 5 years of age.^82–85 Findings from studies of older children suggest that the directionality of diffusion in white matter pathways continues to increase from childhood through adolescence.^86–89 For example, cross-sectional findings from a study of 34 children and adolescents ranging in age from 6 to 19 years revealed significant associations of increasing diffusion directionality with advancing age in prefrontal regions and in white matter pathways surrounding the basal ganglia.⁸⁶ Age-related changes in diffusion in the left frontal cortex have been associated with improvements in working memory capacity during childhood,⁹⁰ and increases in the restriction of diffusion in frontostriatal regions from childhood to adulthood have been associated with improvements in performance on a task requiring the engagement of cognitive control.⁹¹ These age-related changes in the organization of white matter and their behavioral correlates suggest that frontostriatal systems myelinate progressively with age, thereby increasing the effective connectivity between frontal and striatal regions and perhaps contributing to the development of the higher-level cognitive functions that these neural systems subserve.^92,93 Last, a recent DTI study of 202 individuals ranging in age from 5 to 30 years reported a regional pattern of maturation in which diffusion properties in frontotemporal pathways continue until age 30 years.⁹⁴ These findings are consistent with findings from anatomical and functional imaging studies suggesting that the development of both frontal and temporal cortices is protracted and extend well into adulthood.^50,51

Findings from MRS studies of adult patients with schizophrenia suggest decreases of NAA concentrations in the dorsolateral prefrontal cortex and hippocampus relative to control values.^103–105 Reductions in NAA ratios have been detected in these same brain regions in adolescents with schizophrenia, suggesting a biological continuum between childhood- and adult-onset schizophrenia.^106,107 Reduced NAA concentrations in frontal and temporal lobes suggesting the presence of reduced neuronal viability may serve as a biomarker for the presence of disease in children and adults with schizophrenia, consistent with anatomical findings of exaggerated thinning of these cortical regions in this population⁵⁸ and with functional findings of frontal abnormalities in high-risk individuals with prodromal symptoms.¹⁰⁸ This confluence of findings suggests that the pathogenic origins of schizophrenia occur earlier than the ages of the youngest children studied, although the precise developmental timing of these neurometabolic disturbances in schizophrenia remains to be determined.

Language development is one of the most widely studied brain functions in healthy children.^{111– 114} It is a fundamental human trait that begins developing early and rapidly, making language a sensitive index of normal brain development. Findings from studies of language comprehension,¹¹⁵ verbal fluency,^111,112 and reading^114,116 indicate that healthy individuals show age-related increases and decreases in prefrontal and temporal brain areas when engaging these various linguistic functions.¹¹⁴ Functional MRI findings suggest, for example, that age-related increases in activation of language systems in the left frontal and temporal cortices seem to support the normal acquisition of reading and phonological skills during childhood and adolescence,⁹⁸ consistent with the protracted anatomical thinning of these cortices during development.^39,50

Delineating the trajectory of brain activation association with improved reading skills during the course of typical development can serve as a reference that allows us to identify disruptions in developmental processes that may contribute to reading impairments in children with dyslexia. In addition, these normal developmental trajectories may also be used to identify adaptive, developmentally based compensatory systems that support the acquisition of reading skills in children with dyslexia whose deficits in phonological processing typically persist into adulthood.¹¹⁷ Functional MRI studies of the development of normal language functions, for example, have helped to inform our understanding that dyslexia may be a consequence of disruptions in the development of normal functioning of the left hemisphere parietal- and occipital-temporal brain systems.^118,119 These systems support reading abilities, including phonological processing, or the linking of sounds to symbols that ultimately enables the rapid perception of words in unimpaired readers. One fMRI study compared age-related changes associated with phonological processing when reading pseudowords across large samples of children and adolescents with dyslexia and normal reading abilities.¹²⁰ Individuals with dyslexia relied increasingly more on the left posterior medial occipital-temporal areas with increasing age, whereas normal readers relied on a more anterior occipital-temporal region in the left hemisphere. These findings suggest that when children with dyslexia mature, they rely on an alternative posterior neural system that is involved in recognition memory,¹²¹ likely supporting their memorization of words to compensate for their deficits in phonological processing when reading. In contrast, with advancing age, normal readers increasingly rely on an anterior temporo-occipital region that has been termed the visual word form area^122,123 and that supports phonological processing in normal adult readers. Although future longitudinal studies on individuals who are dyslexic and unimpaired readers are warranted, these findings suggest that developmental dyslexia may arise from an early functional disruption in the visual word form area that in typical readers supports development of adult-level reading skills.

Another important developmental process investigated extensively using fMRI is the ability to control behaviors that conflict with personal and societal norms.^124–126 Both cognitive and emotional maturation requires the development of this capacity for “inhibitory control,” making it one of the most centrally defining characteristics of healthy psychological development. Children must learn to engage inhibitory processes to filter and to organize their thoughts, feelings, and behaviors based on social and emotional cues, especially in the face of competing information or distracting stimuli.¹²⁷ Findings from fMRI studies of healthy individuals suggest that the maturation of these functions is associated with the development of the prefrontal cortex, along with anatomically connected, subcortical brain regions.^{124–126,128}

Many experimental paradigms have been used to study the development of inhibitory control processes. The Stroop, Simon, flanker, go/no-go, and stop-signal reaction time tasks all require participants to suppress a more automatic behavior in favor of a less automatic one in the face of cognitive conflict that arises from the presentation of competing or distracting stimuli. Inhibitory control is necessary to mobilize attentional resources toward the appropriate stimuli and thereby resolve cognitive conflict to modulate the automatic tendency to respond in one way rather than another. Findings from developmental studies reveal that performance on these tasks improves continuously with age during childhood and does not reach adult levels of performance until at least 12 years of age.^124–126 The Stroop task is one of the most commonly studied of these paradigms.¹²⁹ It requires participants to inhibit word reading in favor of a less automatic behavior, naming the color of ink in which the letters of a color-denoting word are written. When the color that a written word denotes matches the color of the ink in which the letters are printed (e.g., “R-E-D” written in red ink), children perform the task easily, as indexed by their rapid responses and infrequent errors. However, cognitive conflict occurs when the color that a word denotes does not match the color of the printed letters (e.g., “R-E-D” written in blue ink), making the task more difficult, as indicated by slower responses and more frequent errors. Imaging studies of brain activity during color naming of the mismatching compared with the matching stimuli have demonstrated activation in large expanses of anterior cingulate, prefrontal, and parietal cortices, as well as the striatum, in both adults ¹³⁰ and children.¹³¹

A recent fMRI study, for example, identified age-related differences in the brain activity generated by healthy children and adults during performance of the Stroop task.⁹² Activation of the inferolateral prefrontal cortex and lenticular nucleus increased with age, as did the speed and accuracy of response on the task, indicating that increasing activity in frontostriatal circuits with age supports the age-related improvement in inhibitory control (Fig. 9). These findings are consistent with those from prior developmental imaging studies showing age-related changes in frontostriatal recruitment during performance of other tasks (e.g., Simon, flanker, and go/no-go) that similarly require the resolution of cognitive conflict.^124,125,128 The increasing activation of prefrontal cortices during these inhibitory tasks from childhood to adulthood^92,132 likely reflects the development of cognitive control processes that typically begin to emerge during adolescence,¹³³ at a time when anatomical studies suggest that cortical gray matter thins^39,50 and when DTI studies suggest that frontostriatal fiber tracts are continuing to myelinate.⁸⁶ Nevertheless, future longitudinal fMRI studies of inhibitory control processes in healthy individuals are required to ensure that these cross-sectional findings represent true developmental changes in inhibitory control functions.⁶⁵

The protracted anatomical and functional development of the prefrontal cortices and associated subcortical structures that subserve inhibitory control processes may contribute to the development of a variety of psychiatric disorders in which children have difficulty controlling their thoughts, emotions, and behaviors. These disturbances may release from regulatory control, for example, the various underlying impulses or urges that manifest as either the tics of Tourette’s syndrome (TS), the compulsions of obsessive-compulsive disorder, or the impulsive behaviors that characterize ADHD. These neurodevelopmental disorders, particularly when they occur together, are thought to share a common underlying neural basis involving anatomical disturbances in frontostriatal circuits.^134,135 Understanding the normal development of inhibitory control functions mediated by these circuits can therefore inform our understanding of the etiology of TS, obsessive-compulsive disorder, and ADHD.

One fMRI study, for example, compared across individuals with and without TS the correlations of age with frontostriatal activations during performance of the Stroop task.¹³⁶ Behavioral performance on the task improved with increasing age in patients with TS, just as it did in non-TS controls, reflecting the maturation of neural systems that subserve inhibitory control. In contrast to the normal developmental trajectory of behavioral performance on the task, the imaging findings showed that adults with TS rely on exaggerated activation of frontostriatal regions, which was interpreted as a likely compensatory functional response that produces normal performance on the task, despite deficits in neural plasticity and inhibitory reserve in adults with TS.¹³⁷ Understanding the normal developmental pattern of Stroop-related activations was required to understand how developmental changes in activation of frontostriatal circuits diverge from normal trajectories in people with TS, and how exaggerated activity likely supports normal behavioral performance, even in the presence of an underlying anatomical hypoplasia or in the presence of impaired neural plasticity.

The prefrontal cortex is thought to modulate activity in subcortical structures,¹³⁸ including limbic areas that likely give rise to the ability to engage inhibitory control over emotions. Emotional changes during adolescence involve the increasing ability to read a wide variety of social and emotional cues, including facial expressions. Thus, fMRI studies involving the perception of (and attention to) facial expressions have been used to study emotional development and the development of emotional control.^139–141 For example, a study comparing brain activation across adolescents and adults revealed that adolescents activated the amygdala and prefrontal regions (orbitofrontal and anterior cingulate cortices ) more than adults when viewing fearful faces.¹³⁹ When instructed to focus on a nonemotionally salient feature of the face, however, only the adults engaged the orbitofrontal cortex. These findings suggested that adults but not adolescents modulate activity in prefrontal cortices in response to attentional demands, thereby engaging control over the emotionally evocative stimuli. Thus, the maturation of neural systems that subserve emotional control processes is protracted during normal development. The evolving capacity for emotional control likely derives primarily from increasing functional maturation of prefrontal cortices during adolescence.