Rapid Infant Prefrontal Cortex Development and Sensitivity to Early Environmental Experience

Abstract

Over the last fifteen years, the emerging field of developmental cognitive neuroscience has described the relatively late development of prefrontal cortex in children and the relation between gradual structural changes and children’s protracted development of prefrontal-dependent skills. Widespread recognition by the broader scientific community of the extended development of prefrontal cortex has led to the overwhelming perception of prefrontal cortex as a “late developing” region of the brain. However, despite its supposedly protracted development, multiple lines of research have converged to suggest that prefrontal cortex development may be particularly susceptible to individual differences in children’s early environments. Recent studies demonstrate that the impacts of early adverse environments on prefrontal cortex are present very early in development: within the first year of life. This review provides a comprehensive overview of new neuroimaging evidence demonstrating that prefrontal cortex should be characterized as a “rapidly developing” region of the brain, discusses the converging impacts of early adversity on prefrontal circuits, and presents potential mechanisms via which adverse environments shape both concurrent and long-term measures of prefrontal cortex development. Given that environmentally-induced disparities are present in prefrontal cortex development within the first year of life, translational work in intervention and/or prevention science should focus on intervening early in development to take advantages of this early period of rapid prefrontal development and heightened plasticity.

Human brain development is not a linear process. Widespread recognition of the extended development of prefrontal cortex in human children has led to the overwhelming perception of prefrontal cortex as a “late developing” region of the brain. Mapping the protracted trajectory of prefrontal cortex development has been central to explanations of age-related changes in children’s cognitive development. Rapid improvements in children’s ability to regulate their behavior and emotions in a goal-directed fashion, commonly referred to as executive function (EF) skills (Fuster, 2002), rely on the development of prefrontal cortex (e.g. see Best & Miller, 2010). Slow refinement of prefrontal circuits necessary for decision-making and cognitive control is presumed to underlie the vulnerability of adolescents to making risky choices (Casey et al., 2011). Early individual differences in prefrontal-dependent behaviors also show predictive power over the lifespan; EF skills at preschool-age are predictive of long-term measures of well-being, including academic achievement, social competence, stress resilience, externalizing disorders, divorce rates, and adult body mass index (Ayduk et al., 2000; Casey et al., 2011; Eigsti et al., 2006; Mischel, Shoda, & Peake, 1988; Mischel, Shoda, & Rodriguez, 1989; Schlam, Wilson, Shoda, Mischel, & Ayduk, 2013; Shoda, Mischel, & Peake, 1990).

The characterization of frontal lobe as late developing has been useful in conveying to the general public one reason why children and adolescents think and behave differently from adults, and has even generated important dialogue regarding the application of developmental science to the field of law (Steinberg, 2009). It has also resulted in an unintended and unfortunate cost: a perception of prefrontal cortex as structurally and functionally undeveloped in young infants and toddlers. Neuroimaging techniques have advanced remarkably over the last 10 years and prefrontal cortex development can now be investigated in extremely young infants. These new studies purport that prefrontal cortex development advances most rapidly within the first two years of life, and that frontal lobe regions organize and direct cortical development in intriguing ways.

If prefrontal cortex development is uniquely precocious during early infancy, this has important implications for how children’s early environments shape the development of frontal circuits important for complex cognitive skills. Animal models and human studies suggest that development of frontal lobe structure, function, and behaviors are permanently shaped by, and may be uniquely susceptible to, early adverse experiences. Fortunately, there is growing awareness across the scientific community, government organizations, private corporations, and the general public that children are not ‘just resilient’: adverse early experiences can lead to a myriad of harmful outcomes at both the individual and societal level. The extensive media coverage of the ACE (Adverse Childhood Experiences) Study has been particularly instrumental in demonstrating the importance of early experiences for health-related outcomes in adulthood (e.g. Anda et al., 2006). Vulnerable young children are exposed to adverse environments in high numbers; even in highly industrialized countries like the United States, one in five children live in poverty (Jiang, Ekono, & Skinner, 2016). Understanding the mechanisms via which experiences shape the development of prefrontal cortex is critical to the design of intervention programs that effectively ameliorate the impact of adversity early in life when the brain remains most malleable. Unfortunately, because the rapid development of infant prefrontal cortex has been underappreciated, our field lacks a comprehensive understanding of how experiences impact prefrontal development during the earliest years of life.

Aims and Structure of Review

The overarching goal of this review is to provide an integrated evidence base for considering how and why the rapidly developing infant prefrontal cortex is highly susceptible to variations in early life experience. “Early” prefrontal cortex development is conceptualized as approximately the first two years of life, as postmortem studies suggest that basic brain structure and connectivity is present by this age (e.g. Huttenlocher & Dabholkar, 1997) and deviations in brain development associated with developmental disorders can already be detected (e.g. Hazlett et al., 2011). In doing so, this review has three unified aims to advance our understanding of the impact of early adversity on prefrontal cortex development from a biopsychosocial perspective that are briefly summarized below.

Normative development of prefrontal cortex dependent behaviors has been well-described in children and adolescents (Best & Miller, 2010; Zelazo & Müller, 2002). However, the role of infant prefrontal cortex in guiding complex behaviors has only been recently summarized (Grossmann, 2013a, 2015; Hendry, Jones, & Charman, 2016) and has not been discussed in tandem with new neuroimaging research. As such, the first aim of this review is to provide a comprehensive summary of unique temporal and organizational features of prenatal and infant prefrontal cortex development, across neurobiological processes and metrics of structural brain development. In sum, the evidence presented suggests that prefrontal cortex shows particularly precocious development in the first years of life.

A rich legacy exists in our field of conceptualizing the child’s environment (Bronfenbrenner, 1999) and considering its influence on brain development (Greenough, Black, & Wallace, 1987). In fact, multiple lines of research have already convincingly demonstrated the negative impacts of adversity on childhood or adult prefrontal cortex development, yet this work is not unified across risk factors. The second aim of this review is to provide a novel integration of research in three previously separate domains (maltreatment, poverty, and preterm birth) to illustrate that disparate types of early adversity result in a similar neural and behavioral phenotype of impaired prefrontal cortex development – and that these negative impacts can be observed even in the first years of life.

An understanding of the mechanisms via which early experience shapes prefrontal cortex development bolsters effective translation of this work. However, it is unlikely that one mechanism specifically explains how early experiences are biologically instantiated in later frontal lobe development. The third aim of this review is to identify candidate processes by which prefrontal cortex development may be impacted by the early environment. In doing so this review highlights gaps in our abilities to measure and quantify experience and its impacts on prefrontal cortex development and closes with a discussion of future directions and challenges.

Frontal Lobe Contributions to Infant Cognition

Commonalities in behavior of adult prefrontal lesion patients and young infants and toddlers are often used as evidence of the relative immaturity or “functional silence” of prefrontal cortex early in life (see Zelazo & Müller, 2002 for review of this argument). However, one of the first developmental neuroimaging studies demonstrated that metabolic activity in frontal lobe changes dramatically over the first year of life (Chugani & Phelps, 1986). To date, there has been a dramatic convergence of evidence from behavioral studies and the application of neuroimaging techniques to the study of cognition in young infants and toddlers to suggest that frontal lobe functioning is actively changing, even early in infancy. This section briefly highlights examples of diverse infant cognitive and socioemotional processes supported by early prefrontal cortex development to make the argument that even in the first year of life, infant prefrontal cortex is already implicated in a diverse set of cognitive behaviors. For more comprehensive recent reviews of prefrontal-dependent behaviors across the infancy and toddler years, please see Grossmann (2015) or Hendry, Jones, and Charman (2016).

Early Executive Function (EF) Skills

EF refers to a group of interrelated cognitive processes responsible for the regulation of thoughts, actions, and goal-directed behavior (Miyake & Friedman, 2012; Miyake et al., 2000). Although the underlying structure of EF emerges across development (e.g. see Best & Miller, 2010 for a review of this literature), in adults this skill set has been clasically decomposed into three components (Miyake et al., 2000): working memory (holding information in mind and manipulating it), inhibitory control (refraining from executing a prepotent attentional or behavioral response), and cognitive flexibility (avoiding perseveration by shifting attention flexibily between cognitive tasks and responses). Both infant behavioral, infant neuroimaging, and non-human primate lesion studies convincingly demonstrate that infant prefrontal cortex supports early EF development. For example, young infants’ failure to pass Piaget’s classic A not B task, a widely-accepted measure of early inhibitory control and working memory abilities, is presumed to reflect the developmental immaturity of infant prefrontal cortex (Diamond, 1990b). In this task infants are asked to retrieve a hidden object from one of two locations. Dramatic age-related improvements occur on this task between 7–12 months, as infants become able to tolerate longer delays before searching (Diamond, 1990b). The dependence of flexible searching behavior on prefrontal circuits has been demonstrated in non-human primates, where dorsolateral prefrontal cortex lesions selectively disrupt task performance (Diamond, 1990a). Frontal lobe function is also implicated in human infants; age-related improvements in performance are correlated with changes in frontal electroencephalographic (EEG) activity (Bell & Fox, 1992, 1997; Cuevas, Bell, Marcovitch, & Calkins, 2012) and measures of frontal lobe activation using functional near-infrared spectroscopy (fNIRS; Baird et al., 2002). Even younger infants can exhibit behaviors reflecting some knowledge of the true hiding location of the object if subtle task features are manipulated (e.g. measuring looking direction rather than reaching; see review in Luciana, 2003). Taken together, these brain-behavior relationships indicate that infant prefrontal cortex is functional earlier in development than previously recognized, and likely underlies early forms of EF skills.

Longitudinal studies have also convincingly demonstrated that infant measures of complex attention, memory, and inhibition skills are predictive of EF measures in later childhood. For example, individual differences in infant measures reflecting the ability to both sustain and flexibly shift attention are predictive of EF during childhood and adolescence (Cuevas & Bell, 2014; Johansson, Marciszko, Gredebäck, Nyström, & Bohlin, 2015; Papageorgiou et al., 2014; Sigman, Cohen, & Beckwith, 1997; Sigman, Cohen, Beckwith, Asarnow, & Parmelee, 1991), both in typically developing infants and in at-risk populations (Hitzert, Van Braeckel, Bos, Hunnius, & Geuze, 2014; Rose, Feldman, Jankowski, & Van Rossem, 2012). These predictive relationships typically remain even after controlling for general intellectual ability (e.g. IQ) and/or processing speed. Complementary findings in the behavioral literature also suggest that normative variations in dopaminergic genes with receptors in prefrontal and fronto-striatal brain regions are related to individual differences in a variety of infant and toddler cognitive behaviors, including diverse forms of attention (Holmboe et al., 2010; Markant, Cicchetti, Hetzel, & Thomas, 2014a, 2014b; Voelker, Sheese, Rothbart, & Posner, 2009). Furthermore, new research which has bridged a gap in this field by validating developmentally appropriate measures of EF for older infants and toddlers (e.g. Mulder, Hoofs, Verhagen, van der Veen, & Leseman, 2014) has demonstrated that these early EF skills, especially when considered as a latent construct, are predictive of behavior in real-world contexts (see Hendry et al., 2016 for review), much like measures of EF in older children. Although frontal lobe function cannot be directly inferred from these behavioral studies, they provide striking evidence of developmental continuity in complex cognitive skills, emerging prior to 12 months of age, that likely involve prefrontal circuitry (Hendry et al., 2016).

Diverse Cognitive and Socioemotional Functions

In addition to its role in early EF, recent functional neuroimaging studies indicate that infant prefrontal cortex is functionally active across a variety of cognitive and socioemotional tasks, even when activation would not be predicted from the adult neuroimaging literature (Grossmann, 2013a; Johnson, 2011). For example, 2-month-old infants selectively recruit the left inferior frontal gyrus when processing face, but not object, stimuli, although this brain region does not show face-selective activity in adults (Tzourio-Mazoyer et al., 2002). Frontal activation has also been reported in infants during sensory processing, speech and language processing, early forms of attention and novelty preference, and later emerging inhibitory control and working memory abilities (e.g. Dehaene-Lambertz & Dehaene, 1994; Grossmann, 2013; Nakano, Watanabe, Homae, & Taga, 2009; see Dehaene-Lambertz & Spelke, 2015 for recent review of speech and language processing literature). Why frontal lobe is implicated in so many behaviors in young infants is unclear. Classic frontal-lobe functions such as working memory may be implicated in cross-task activation (e.g. holding the previous sensory stimulus in mind to judge whether it is novel or familiar) or the frontal lobe may support more general, top-down learning processes in young infants (Dehaene-Lambertz & Spelke, 2015).

For example, although the A not B task is clearly dependent on early EF skills, the context of the task administration (typically a highly social interaction with an adult experimenter), also activates social learning processes in young infants (Topal, Gergely, Miklosi, Erdohegyi, & Csibra, 2008). Related research has demonstrated that prefrontal cortex is highly sensitive to social stimuli during early development (see Grossmann, 2015 for recent review), perhaps even showing early functional specialization for detecting social signals (Jones, Venema, Lowy, Earl, & Webb, 2015). fNIRS studies indicate that infant prefrontal cortex is implicated in detecting and processing affect in human voices (Blasi et al., 2011), and in comprehending the communicative intent of adults through detection of both eye contact and sensitivity to the infant’s own name in preverbal infants (Grossmann, Parise, & Friederici, 2010). Although some of these purportedly social functions of infant prefrontal cortex may reflect early EF skills such as sustained attention to relevant stimuli in the face of distraction, joint attention has been argued to be fundamentally social in nature (Grossmann, 2015), is known to relate to left dorsolateral prefrontal activity in infants (Caplan et al., 1993; Grossmann & Johnson, 2010) and may represent a uniquely human prefrontal-dependent skill (Saxe, 2006).

Functional Divisions of Infant PFC

The growing fNIRS literature examining prefrontal involvement in infant cognitive and socioemotional processing supports a broad division of frontal lobe into two gross anatomical regions with differing functional properties: a medial region primarily responsible for affective processing (Saito, Aoyama, et al., 2007; Saito, Kondo, et al., 2007), and a more lateral region primarily responsible for cognitive processes such as attention and memory (Baird et al., 2002; Nakano et al., 2009; Watanabe et al., 2013; Watanabe, Homae, Nakano, & Taga, 2008). This division generally coincides with organizational principles of prefrontal function from adult lesion studies and adult functional magnetic resonance imaging (fMRI) studies, suggesting a relatively continuous developmental organization of prefrontal cortical function (Grossmann, 2013a). Implications for the diverse roles of prefrontal cortex in infant cognition will be more extensively discussed later in this paper as a potential mechanism by which variations in early experience may impact long-term prefrontal development. However, it is evident that the popular conception of infant prefrontal cortex as remaining “functionally silent” should be definitively overturned.

Neurobiological Development of PFC

A role for infant prefrontal cortex across multiple cognitive and socioemotional processes suggests that important developmental changes occur in this brain region during the first years of life. A full review of embryonic and postnatal neurobiological brain development processes that support early frontal lobe functional maturation is not possible here (see Markant & Thomas, 2013 for more details on early neurobiological development). Instead, this review gives special attention to unique temporal and organizational features of prefrontal cortex development.

Neuroanatomical Definition

The frontal lobe is one of the cortical regions to undergo the greatest expansion within the trajectory of an individual’s brain development, representing approximately one-third of an adult’s neocortex (Fuster, 2002). The frontal lobe is defined as cortex that lies anterior to the central fissure (Goldman-Rakic & Porrino, 1985), or as the projection field of the dorsal medial thalamus (Goldman-Rakic & Porrino, 1985; Kolb et al., 2012). Anatomical subdivisions in primates based on histology include dorsolateral, ventromedial (cingulate), and orbitofrontal regions (Stuss & Benson, 1986). Divisions are also characterized based on differing patterns of connectivity, with dorsolateral regions demonstrating increased connectivity with the basal ganglia (Fuster, 2002) and posterior parietal cortex (Petrides et al., 1984), while medial and orbital regions show increased connectivity with the hypothalamus and limbic structures (Fuster, 2002). Ultimately, prefrontal cortex is purported to be the best connected of all cortical structures, receiving input from all other cortical regions (Fuster, 2002; Kolb et al., 2012) and may be central to coordinating the functioning of multiple, broad cortical networks (Vincent, Kahn, Snyder, Raichle, & Buckner, 2008).

The homology of prefrontal cortex anatomy and connectivity across mammalian species is well-established but remains somewhat controversial and is beyond the scope of the current paper (see Fuster, 2002 for brief review). Much of our understanding of early human brain development is derived from studies of cortical development in primates, with the inference that processes proceed in a similar order in human fetuses and neonates, although on a different timescale due to differences in gestational length.

Birth, Differentiation, and Migration of PFC Neurons

Brain development begins early in pregnancy with the formation of the neural plate shortly after conception. Over several days, the neural plate folds and elongates to form the neural tube. Along the anterior end of the neural tube neuronal and glial precursor cells rapidly proliferate in the subplate zone. The subplate zone subserving prefrontal cortex is proportionally larger than that of other regions of the brain (e.g. four times wider than the subplate of occipital cortex; Rakic, 1995). It also remains present in frontal cortex longer than in any other region of the brain, persisting up to 13 months in the gyral crowns of frontal cortex (vs. 6 months in other associative cortex areas; Pandit et al., 2013). The expansion in size and extended maturation of the frontal lobe subplate zone likely reflect the complex connectivity of prefrontal cortex, even in the infant brain (Rakic, 1995).

Neurons destined for prefrontal cortex are already developing by embryonic day 50–100 in non-human primates (Schwartz, Rakic, & Goldman-Rakic, 1991). Prefrontal neurons arrive at their final locations via migration from the subplate zone to cortex in a series of waves that results in the adult-like, six-layered neocortex. Migration occurs primarily along radial glia fibers, specialized glial cells that have long processes extending to the edge of cortex (Kriegstein & Gotz, 2003). Radial organization is prominent throughout the brain during the second trimester of gestation, but persists longer in ventral frontal areas than posterior brain regions (Takahashi, Folkerth, Galaburda, & Grant, 2012). Migratory pathways for prefrontal cortical neurons are also physically longer in distance than those of other cortical neurons, increasing the risk for migration errors (Rakic, 1995). Despite these longer pathways, prefrontal neurons are among the first to arrive in cortex due to a general anterior to posterior progression of neurogenesis and migration, and are thus more likely to support intrinsic firing patterns in the developing brain (Cahalane et al., 2011; see review in Johnson, 2012).

As migration progresses, cells acquire specific fates and differentiate to take on their mature characteristics, including the development of rudimentary dendrites and axons (Pandit et al., 2014). Elongation of axons leads to the formation of major anatomical tracts during the second trimester of gestation (e.g. limbic and thalamocortical fibers; Huang et al., 2006). Importantly, the uncinate fasciculus, a large white matter tract connecting limbic regions to prefrontal cortex, is one of the few long-range association pathways that can be reliably detected in young fetal brains (Takahashi et al., 2012). Although this major frontal lobe tract develops early, within lobe pathways are present first in occipital and parietal cortex, and emerge later in frontal regions (Takahashi et al., 2012). Across cerebral cortex, this initial series of neurobiological processes occurs primarily during the prenatal period, with cell migration ending by the first week of postnatal life; however, axon and dendritic growth proceeds well into the postnatal period as neurons begin to establish and refine connectivity patterns (see Markant & Thomas, 2013 for review).

Synaptogensis and Synaptic Pruning in PFC

Synaptogenesis (formation of synapses) and synaptic pruning (elimination of synapses) begin prenatally but extend well into postnatal life, with differential rates of change across regions of the brain. Prenatal and postnatal connectivity is generally established through exuberant development, which is then followed by a phase of selection and pruning of connections (Collin & van den Heuvel, 2013). Although this over production and subsequent pruning may seem inefficient, it allows for the individual organism’s neural circuitry to be shaped by unique environmental experiences.

Unlike primary sensory regions which reach peak synaptic density in the first months of life, post-mortem studies indicate that human prefrontal cortex attains its maximum number of synapses after 15 months of age, with some prefrontal regions not reaching peak density until early childhood (Huttenlocher & Dabholkar, 1997; Petanjek et al., 2011; although see Bourgeois, Goldman-Rakic, & Rakic, 1994 for contradictory results). Although prefrontal regions take longer to reach peak synaptic density, histological studies of superior and middle frontal gyri tissue have revealed rapid sculpting of dendritic trees in prefrontal pyramidal neurons during the first months of life. Basal dendritic trees of prefrontal pyramidal neurons in cortical layers IIIc (predominantly long-range association projections) and V (basal ganglia projections) reach 60–80% of total adult size by only three months of age. Interestingly, layer IIIc neurons display a second period of exuberant dendritic growth beginning at the end of the second year and continuing into the third year of life (Petanjek, Judaš, Kostović, & Uylings, 2008).

Prefrontal cortex is widely recognized to show extended synaptic pruning into adolescence (e.g. Woo, Pucak, Kye, Matus, & Lewis, 1997). However, most developmental psychologists are unaware that pruning is also extensive during early infancy, particularly in layer V subcortical projection neurons (Petanjek et al., 2008); advances in prefrontal-dependent behaviors during early infancy may therefore be linked to early refinement of fronto-striatal circuitry.

Myelination in PFC

Cortical myelin (also commonly referred to as white matter) is produced by oligodendrocytes, specialized glial cells that first emerge prenatally (as “premyelinating oligodendrocytes”) when axons are initially sprouting and elongating (Volpe, 2009). Actual myelination (wrapping of connections with a fatty membrane) occurs almost exclusively after birth (Pandit et al., 2014); as such, minimal myelinated white matter is present in the fetal brain until late in gestation, when development suddenly accelerates with a five-fold increase over the last weeks of human pregnancy (Hüppi et al., 1998).

Pre-myelination during prenatal development and actual postnatal accrual of myelin follows a posterior to anterior gradient (Dubois et al., 2014), with a growing literature indicating that myelination changes in frontal lobe extend well into late adolescence (e.g. Asato, Terwilliger, Woo, & Luna, 2010; Nagy, Westerberg, & Klingberg, 2004). Frontal lobe myelination during the early postnatal period is unique from that of other brain regions due to the longer persistence of pre-myelinating oligodendrocytes within this region. Pre-myelinating oligodendrocytes, which are more vulnerable than mature oligodendrotyes to perinatal insults, remain predominant in frontal lobe regions at birth (Back et al., 2001), making early frontal lobe white matter uniquely susceptible to injury.

Neuroimaging Measures of Early PFC Development

The neurobiological processes reviewed in the previous section suggest an important role for early-migrating prefrontal neurons in organizing early brain activity and a unique vulnerability of frontal lobe white matter to early injuries. Unfortunately, these neurobiological processes are not currently measurable in the living human brain. However, neurobiological processes contribute to structural brain changes that can be observed at a larger scale in human fetuses, infants, and toddlers using non-invasive imaging techniques.

Although quite different in level of analysis from the histological studies of neurobiological processes, pediatric neuroimaging research provides important information about trajectories of prefrontal cortex gray and white matter development that may be easier for developmental scientists to link to measurable changes in infant behavior. Because EEG measures do not provide conclusive information about the location of source activity (Michel et al., 2004) this review is restricted to structural neuroimaging measures where spatial resolution is improved. Critically, examination of these noninvasive indices convincingly demonstrates that extensive prefrontal cortex structural development occurs in the first years of life and that this region of the brain plays a unique role in shaping whole-brain network connectivity. See Figure 1 for a summary of infant prefrontal cortex development.

Figure 1.

Spatial distribution of cortical thickness, inferred myelination (via myelin water fraction – MWF), and default mode network in neonates, infants, and toddlers (left hemisphere). Medial prefrontal regions are among the thickest cortex at birth and across the first years of life. Similarly, medial prefrontal involvement in the default mode network is present even in neonates. In contrast, myelin accrual in frontal lobe regions is minimal earlier in development but increases rapidly over the first year of life.

Adapted from Li, Lin, Gilmore, & Shen (2015); Deoni, Dean, Remer, Dirks, & O’Muircheartaigh (2015); Gao, Zhu, Giovanello, Smith, Shen, & Gilmore (2009)

Gray Matter Volume

Growth of prefrontal gray matter volume during the early childhood period is astoundingly rapid. However, controversy exists regarding the age window in which growth rates are greatest and the extent to which prefrontal growth outstrips similar development in other brain regions. Although all studies converge on rapid growth occurring sometime within the first two years of life, differences in timing effects reported across studies likely reflect variation in scan parameters, scan segmentation methods, study samples (e.g. inclusion of preterm infants or infants undergoing magnetic resonance imaging (MRI) to rule out a neurodevelopmental problem; size of samples), study design (cross-sectional vs. longitudinal), and differences in covariates (e.g. adjustment for total brain size).

Multiple structural MRI studies with newborns and young infants have failed to document the expected adult pattern of prefrontal structural asymmetry (where the right hemisphere is greater than the left in adults and older children) (Gilmore et al., 2007; Li, Nie, et al., 2014; Matsuzawa et al., 2001; Tanaka, Matsui, Uematsu, Noguchi, & Miyawaki, 2013), which may reflect the relative structural immaturity of the newborn prefrontal cortex. Cross-sectional studies (Nishida et al., 2006) of infants born between 31–42 weeks gestation and imaged in the first months of life have reported that the fastest rate of cortical gray matter growth occurs in the frontal lobe of the brain (4.44 ml/week), followed by the parietal (2.93 ml/week), temporal (2.29 ml/week), occipital (2.40 ml/week) and limbic regions (.58 ml/week). Rapid growth in frontal lobe gray matter continues over the next two years of life, surpassing changes observed in other lobes of the brain (Matsuzawa et al., 2001). In contrast, longitudinal studies of full-term infants report a slightly different trajectory of frontal lobe volume gray matter increases. Here, increases are fastest in sensory and motor regions in the early postnatal period (Gilmore et al., 2007), with prefrontal gray matter growth rates accelerating between 1–2 years of age and continuing at a rapid pace into the second year of life (Gilmore et al., 2012).

Surface Complexity

Growth of gray matter volume reflects separable contributions of change in cortical surface area and cortical thickness, both influenced by changes in surface complexity as the brain grows from a smooth surface to a complexly folded cortex. Extensive cortical folding occurs in the third trimester of pregnancy, but folding in the frontal lobe begins later (Dubois, Benders, et al., 2008) and progresses at a slower rate (Abe, Takagi, Yamamoto, Okuhata, & Kato, 2003). However, in the first two years of life, measures of cortical surface complexity such as gyrification index (Li, Wang, et al., 2014) and the depth of sulcal pits (Meng, Li, Lin, Gilmore, & Shen, 2014) change most rapidly in frontal regions.

Although the major gyri and sulci are established by viable preterm birth, cortical surface area remains only one-third the size of the adult brain (Li et al., 2013). Cortical surface area expansion during infancy is highly non-uniform, with regions of prefrontal cortex expanding twice as much as primary visual areas (Hill et al., 2010). Longitudinal imaging studies of typically developing infants have more clearly delineated the timing of rapid prefrontal surface area increases. The highest expanding regions of the brain in the first year include parts of orbitofrontal and lateral anterior prefrontal cortex, with rapid expansion of superior and middle frontal regions during the second year of life (Li et al., 2013; Lyall et al., 2015). Although this growth is incredibly precocious, surface area at two years is only 70% of its eventual size in the adult brain (Lyall et al., 2015).

In contrast, cortical thickness reaches more adult-like values across the brain by two years of life (Lyall et al., 2015). Areas in the infant brain where surface expansion occurs at the most rapid pace are also generally the thickest (Li, Lin, Gilmore, & Shen, 2015). At birth, regions of medial prefrontal cortex are among the thickest of all cortex in the brain (Geng et al., 2016; Li et al., 2015; Lyall et al., 2015). Thick prefrontal regions also show the fastest rate of growth in the first year of life, along with disproportionately rapid thickening in other frontal subregions (inferior frontal operculum, superior frontal gyrus) during the second year (Li et al., 2015; Lyall et al., 2015). Contrary to reports that cortical thickness peaks later in childhood (Raznahan, Greenstein, Lee, Clasen, & Giedd, 2012; Shaw et al., 2008), several subregions of the frontal lobe begin to show thinning over the second year of life (Li et al., 2015; Lyall et al., 2015; although see Croteau-Chonka et al., 2016). Changes in cortical thickness can only be used to make inferences about underlying neurobiological processes like synaptic pruning. However, these provocative results suggest an early wave of pruning may occur in frontal lobe regions during childhood, long before the classically described frontal lobe pruning in the adolescent brain.

White Matter Organization

White matter grows more slowly than gray matter in absolute volume (Knickmeyer et al., 2008), but global changes in white matter organization measured via diffusion tensor imaging (DTI) are the most rapid in the first postnatal year (Geng et al., 2012; Sadeghi et al., 2013; see Qiu, Mori, & Miller, 2015 for review). Frontal lobe peripheral white matter shows the largest rate of increase in fractional anisotropy, a global measure of white matter organization, across the infant brain (Gao, Lin, et al., 2009; McGraw, Liang, & Provenzale, 2002); this tissue also shows large reductions in axial diffusivity, likely due to continued axon growth, over the second year of life (Gao, Lin, et al., 2009). Smaller tracts involving projections to and from the frontal lobe such as the fronto-occipital pathways are apparent by at least three months after birth, although they may take until one year of age to be well delineated on structural MRI scans (Collin & van den Heuvel, 2013). Studies examining white matter microstructure metrics and/or graph theory derived measures of connection strength along anatomical tracts have reported that frontal lobe pathways are more immature at birth than sensory and motor tracts (Dubois, Hertz-Pannier, Dehaene-Lambertz, Cointepas, & Le Bihan, 2006; Dubois, Dehaene-Lambertz, et al., 2008; Geng et al., 2012; Pandit et al., 2014) but then show the fastest rate of maturational change over the first year of life.

As previously discussed, the accrual of myelin occurs predominantly after birth (Dubois et al., 2014). Similar to the gradient of myelination observed in post-mortem fetal brains (i.e. caudal to rostral; central to peripheral; e.g. Takahashi et al., 2012), in-vivo myelin-specific MRI techniques report that the frontal lobe is slower to show measurable myelination onset than other lobes of the brain (Deoni et al., 2011; Deoni, Dean, O’Muircheartaigh, Dirks, & Jerskey, 2012; Deoni, Dean, Remer, Dirks, & O’Muircheartaigh, 2015; O’Muircheartaigh et al., 2014; Oishi et al., 2011). However, frontal regions then counter intuitively show a higher rate of myelin accrual over the first years of life in comparison to regions that began myelination earlier in development (Deoni et al., 2011).

Emergence of Functional Networks

Functional development of prefrontal cortical networks has recently been investigated using resting state functional connectivity MRI, which provides information about correlated activity between networks of brain regions at rest, and generally coincides with structurally connected regions of the brain (Vincent et al., 2007). Amazingly, frontal lobe networks are present extremely early in infant development and are similar in regional involvement and connectivity to what is observed in the adult brain.

Gao and colleagues first provided evidence that the default mode network, a set of regions in the adult brain that show more correlated activity at rest than during task completion, is present in a primitive state in 2-week-old infants (Gao, Zhu, et al., 2009). Remarkably, like network structure in older children and adults, the newborn version of the default mode network also includes substantial medial prefrontal cortex involvement. By one year of age, medial prefrontal cortex emerges as a hub region of the network, and by two years of age, the default mode network is similar in regional involvement and organization to the adult state (Fransson et al., 2013; Gao, Alcauter, Smith, Gilmore, & Lin, 2015; Gao et al., 2011, 2013; Gao, Zhu, et al., 2009). Fronto-parietal attention networks, which tend to be negatively correlated with the default mode network and involve substantial anterior prefrontal cortical regions, are also present in neonates and appear highly similar in typology to the adult network state by the end of the first year of life (Gao et al., 2013).

Graph theory analyses (see Cao, Huang, Peng, Dong, & He, 2016 for more general review) of frontal lobe organization over the first two years of life indicate there are decreases in local efficiency and degree among frontal regions, potentially representing regional specialization and removal of redundant and/or spurious connections (Gao et al., 2011; Gao, Zhu, et al., 2009). In particular, the existence of medial prefrontal cortical regions serving as hubs this early in brain development is striking. Hub regions typically show a large degree of anatomical and functional connectivity, and are hypothesized to play a strong role in both coordinating and integrating information within and across networks of the brain (Gao, Zhu, et al., 2009). Frontal hubs in the neonate brain suggest major frontal lobe functional development has already occurred by term birth (Doria et al., 2010; Gao et al., 2011; Smyser et al., 2010) and that frontal regions may organize infant brain activity more so than posterior cortical regions (Grossmann, 2013a).

Interim Summary: Early PFC Development

Prefrontal cortex is often referred to in the developmental psychology literature as a “late developing” region of the brain. However, rapid expansion of frontal lobe gray matter volume occurs within the first two years of life, along with dramatic changes in surface complexity. Studies mapping developmental trajectories of whole-brain surface area and cortical thickness metrics have characterized prefrontal subregions as “fast growing” areas of the brain, exemplified by thick cortex at birth and high growth rates of both surface area and cortical thickness over the first years. Both histological and neuroimaging studies of infant cortical thickness suggest frontal lobe may undergo an early wave of pruning, likely in fronto-striatal circuits, much earlier in development than previously recognized. Long-range connections involving frontal lobe are present from the second trimester of gestation; continued sculpting of frontal-cortical connectivity occurs after birth, with myelination of frontal pathways beginning later in development but then changing more quickly in comparison to other regions of the brain. Histological studies also reveal that frontal lobe white matter may be uniquely susceptible to injury. Last, the early arrival of frontal neurons via migration and the presence of medial prefrontal cortex as a “hub region” in developing brain networks suggests that prefrontal activity can influence network functionality to a greater extent than previously conceptualized. Disruptions to the development of this important, organizing region could therefore have broad, cascading effects across multiple brain networks.

Given this evidence supporting rapid development of prefrontal cortex during infancy, along with its broad role in multiple aspects of infant cognitive and socioemotional development, it is appropriate to revise the conceptualization of prefrontal cortex as a “late” developing region of the brain. Instead, prefrontal development is better described by a trajectory that shows rapid change during early infancy, along with later periods of extended refinement (Blakemore & Choudhury, 2006).

Since prefrontal cortex plays a role in some of the more unique and complex aspects of human cognition that must be learned during infancy and early childhood (Saxe, 2006), it is intuitive that this region is highly sensitive to modifications of the early environment, which is the topic of the remainder of this paper. Early and extended sensitivity of prefrontal cortex to environmental experience conveys maximum potential for the developing organism to benefit from environmentally-induced plasticity. However, if the early environment is not advantageous and/or if adaptations to the early environment are maladaptive in later contexts, this early sensitivity could serve as a risk factor for long-term neurobehavioral development.

Defining Early Experience

Converging evidence demonstrates that diverse types of early experience impact prefrontal development in young mammals, including human infants. However, experience is ubiquitous and thus challenging to empirically define. The difficulty in accurately measuring and characterizing individual differences in environmental context across the lifespan has long been recognized in developmental psychology (e.g. Bronfenbrenner, 1999) and across other related fields. Embryologists examine distinctions between errors of commission vs. omission and their impact on brain development (e.g. see Cheatham, Sesma, & Georgieff, 2010 for summary). Epidemiologists use case-control designs to distinguish the impact of genetic risk vs. environmental exposure in the development of complex diseases (e.g. Clayton & McKeigue, 2001). Even cellular and molecular neuroscientists consider the impact of variation in local features of the neuronal environment within which brain cells develop (e.g. see cortex transplantation studies in Luo & O’Leary, 2005).

Characterization of the timing and nature of early experiences is necessary to more precisely link individual differences in environmental variation to concurrent and subsequent measures of brain development. Furthermore, a better understanding of the aspects of the environment that are related to the most variance in brain development could have important policy-related implications for early intervention services. Many frameworks have been developed across disciplines interested in early childhood and developmental outcomes to characterize the quality and nature of early environments. Although several of these models lack explanations for how experience is instantiated at the neural level and most are not specific to prefrontal cortex development, they provide frameworks for conceptualizing the range of experiences that might shape concurrent prefrontal cortex development, how this may occur biologically, and why effects may carry forward over development.

Relation to Neurobiological Processes

Greenough and colleagues (1987) have provided a classic model of how different aspects of environmental experiences (expectant vs. dependent processes) impact specific neurobiological mechanisms of brain development. Experience-expectant processes rely on the developing organism being exposed to basic environmental information (e.g. patterned light, sequential sounds), so rudimentary that it should be universally experienced by all members of a species across variable environments. Experience-expectant processes are linked to sensitive periods: windows of time in which a developing system is highly plastic and most open to influence by the environment (e.g. Hensch, 2005). Tuning of the brain to the expectable environment during sensitive periods is manifested as retention of a subset of necessary synapses among those that were initially overproduced (Greenough et al., 1987), reducing the amount of brain development that must be precisely specified by genetics.

In contrast, experience-dependent processes are not limited to early development and instead reflect exposure to environmental information that is unique to the organism over the lifespan (e.g. learning rules within a new social group, learning new fine motor skills associated with a musical instrument). Experience-dependent plasticity thus allows for learning across the lifespan, providing a powerful mechanism for individual differences in brain development. Typically studied through environmental enrichment paradigms in animals, experience-dependent processes are associated with generation and retention of new synapses (Greenough et al., 1987) and myelination of connections (Lovden et al., 2010). However, there is no abrupt transition from reliance on experience-expectant processes to utilization of experience-dependent; instead, both processes shape neural development early in life, followed by life-long maintenance of established neural systems via experience-dependent processes.

What constitutes expectable input and the timing and existence of sensitive periods for higher-level cognitive and/or socioemotional processes in human infants remains controversial (e.g. see Fox, Levitt, & Nelson, 2010). Most developmental psychologists recognize that an expectable feature of the environment for a human infant includes a close relationship with a caregiver (Humphreys & Zeanah, 2014; Nelson, 2007); beyond fulfilling basic safety and nutritional needs, caregivers ideally regulate the physiology of the infant and provide an environment with developmentally appropriate cognitive and social stimulation. Although deprivation of expectable input is clearly detrimental, what constitutes the “optimum” level of experience is unclear; excess enrichment may also result in negative developmental effects. Furthermore, expectant and dependent processes are not completely separable, as a deficit in expectable input will also likely impact subsequent experience-dependent plasticity.

Dimensions of Experience

Although it was initially assumed that drastic variations in environmental experience would be required to alter cortical development (e.g. extreme sensory deprivation, such as dark-rearing of infant animals), animal models suggests that relatively subtle variations in the environment can induce developmental changes in broad regions of the brain (Mychasiuk, Gibb, & Kolb, 2012). Frameworks that describe the characteristics of adverse environments hypothesize that different types of experience (e.g. exposure to inadequate vs. harmful input; Humphreys & Zeanah, 2014) will impact brain development in unique ways. McLaughlin and colleagues recently proposed a model differentiating depriving environments (those with low complexity) from those characterized by high levels of threat (to the organism’s physical well-being) at the behavioral and neural level (McLaughlin, Sheridan, & Lambert, 2014; Sheridan & McLaughlin, 2014). At the neurobiological level, depriving environments are predicted to cause early pruning or over-pruning of synapses, including in prefrontal regions. Environments with high levels of threat are hypothesized to primarily alter fronto-limbic circuitry, driven by stress-induced changes in dendritic structure and arborization in limbic and ventromedial prefrontal regions. However, this approach of identifying unique characteristics of various experiences is challenged by the fact that multiple aspects of environmental risk often co-occur (Anda et al., 2006) and by a difficulty in conceptualizing how features of adverse environments are perceived and/or experienced by young infants and toddlers (e.g. see Graham et al., 2016 for discussion of normative development of fear during infancy).

Cumulative Nature of Experience

Rather than separating experiences by dimensional features, other models focus on measures of cumulative environmental adversity, combining separate environmental risk factors into an aggregate or composite risk score (Evans, Li, & Whipple, 2013). Exposure to multiple vs. single risk factors is more detrimental for a variety of developmental outcomes and these models outperform those that consider only single exposures (Evans et al., 2013). Bronfenbrenner’s bioecological systems model (Bronfenbrenner, 1999) is also a useful heuristic for conceptualizing how risk may aggregate across multiple levels of a child’s environment. Here the individual child is viewed as nested within a multi-leveled environment, with the most proximal environments embodying every day exposures (e.g. familial home) and more distal environments representing broader systems that impact the sociocultural context in which development occurs (e.g. government policies that impact the availability of a caregiver). Although environmental risks most proximal to the infant would be predicted to have a greater negative effect on development, distal influences that shape functioning of the caregiver are also recognized.

These models are largely atheoretical from a neurobiological perspective, but are broadly congruent with conceptions of toxic stress and/or allostatic load (Hertzman & Boyce, 2010; Lupien, McEwen, Gunnar, & Heim, 2009; McEwen & Morrison, 2013; Shonkoff et al., 2012), which describe the impact of increasing risk exposure on the development of the stress system, including fronto-limbic circuitry. Incorporating cumulative risk exposure across levels of the environment better reflects the co-occurring risk factors experienced in many adverse environments. However, this approach often fails to consider intensity of individual risk factors and the potential for risk factors to produce interactive, cascading effects on neurobehavioral development.

Adaptation Following Early Experience

Canalization theories (Gottlieb, 1991; Waddington, 1942) predict that even in the case of exposure to adverse environments, most developmental processes can only be temporarily derailed, instead quickly returning to and/or remaining on their previous developmental pathway. However, not all individuals respond similarly to the same early experiences, a challenge that has been addressed in recent theories describing individual differences in sensitivity to the environment. Theories rooted in evolutionary biology have posited that high sensitivity to the environment may be selected for over evolutionary history. Within this framework, individuals vary not only in susceptibility to risk, but more broadly in sensitivity to both positive and negative environmental contexts (Belsky & Pluess, 2009; Boyce & Ellis, 2005). Experience-induced changes in neurobehavioral development can be viewed as the organism’s attempt to adapt to the current environment (e.g. see adaptive calibration model; Del Giudice, Ellis, & Shirtcliff, 2011). Rather than adverse environments producing detrimental outcomes, deficits may arise when the environment changes and previous adaptations become problematic. To date these models have primarily focused on genetic differences in biological reactivity and/or stress neurobiology. However, individual differences in environmental sensitivity may also be related to neurobiological mechanisms that promote homeostasis (e.g. maintaining certain synaptic connections; Johnson, Jones, & Gliga, 2015), including within prefrontal cortex.

Importance of Timing

The title of this paper implicitly suggests that early experiences show some privilege in shaping developmental trajectories. However, the broad effects of timing (including developmental timing of exposure, duration of exposure, and frequency of exposure) of experiences on subsequent development are not well understood. Sensitive periods are widely recognized to represent the developmental window within which the impact of environmental variations will be strongest, implying that early experiences are the most transformative. Sensitive periods for higher-level cognitive functions are presumed to occur later than those for sensory or perceptual functions (Fox et al., 2010), but whether they exist for prefrontal-dependent behaviors is unclear. The impact of timing is also difficult to discern because early experience may impact prefrontal development in a way that is not easily measurable or quantifiable during early life. Deficits in prefrontal-dependent behaviors induced by early adversity may not emerge until the demands on children’s cognitive processing increase during the later childhood years, a pattern that would be consistent with ‘growing into deficit’ (Sesma & Georgieff, 2003) or a ‘sleeper effect’.

Early Adverse Experiences and PFC Development

Although defining both the timing and nature of experiences remains challenging, many studies have examined long-term impacts of adverse early environments on later prefrontal cortex development (see Mackey, Raizada, & Bunge, 2012 for review of additional environmental variations on prefrontal development). However, this work has generally examined different types of risk factors in isolation, and as such lacks an integrated perspective on why diverse forms of early risk converge to impact prefrontal cortex development. Furthermore, few studies have examined whether the impacts of early adversity on prefrontal development can be detected early in development: in young infants and toddlers.

This next section provides an overview of research on the impacts of early adversity on prefrontal cortex development across three risk factors: maltreatment, poverty, and preterm birth. These three types of adverse experiences were selected because they represent common risk factors for children across the world that are disappointingly prevalent, and because they differ in temporal and dimensional characteristics that may uniquely challenge the development of prefrontal cortex. As in the section on normative prefrontal cortex development, this review focuses on the impact of early experience on structural neuroimaging metrics because of their high spatial resolution and stronger theoretical relation to neurobiological processes. Select examples of functional effects from behavioral and/or functional imaging studies are also provided to illustrate how impacts of experiences are measurable at different neurobehavioral levels. Last, because few studies have concurrently measured the impact of adverse environments on neurobehavioral development in young infants or toddlers, an overview of follow-up studies is also provided to demonstrate that negative impacts persist over development.

Maltreatment and PFC Development

Diverse forms of childhood maltreatment are associated with pervasive, long-term impacts on prefrontal cortex structure and function (see Bick & Nelson, 2016; Hart & Rubia, 2012; Humphreys & Zeanah, 2014 for recent reviews that span multiple brain systems). The impact of early maltreatment on human brain development is not surprising, given that abuse or neglect by a close caregiver represents an intense deviation from the biologically expected infant-parent caregiving relationship and likely alters both experience-expectant and experience-dependent brain development processes. Although duration of maltreatment experiences and type of maltreatment may differentially impact brain development, most children who are exposed to maltreatment experience multiple subtypes (e.g. physical abuse, neglect) of recurring maltreatment across childhood. Studies examining long-term impacts of maltreatment on prefrontal cortex development have generally compared individuals with a history of maltreatment at any point in childhood to non-maltreated comparison groups. In contrast, post-institutionalized (PI) or orphanage-reared children experience a time-limited period of early neglect (i.e. the institutional care environment) followed by adoption into well-resourced families. Because the neglectful environment is limited to early life, studies with PI children demonstrate the lasting impacts of adversity limited to early childhood and the ability of the brain to adapt to the new, post-adoptive environment.

Few studies have examined the impact of maltreatment and/or institutional care on prefrontal cortex development early in life. However, because PI children are adopted at different ages, the impact of duration of exposure to early adverse environment can be assessed. This section provides a broad overview of the impact of childhood maltreatment on later prefrontal cortex development, with a primary focus on studies of PI children as this allows the timing of neglect to be constrained to the early childhood period. Preliminary evidence indicates that altered prefrontal-dependent behaviors can be observed in the first years of life in children exposed to diverse forms of maltreatment; however, because this literature is quite small, findings from both maltreated and PI infants and toddlers are summarized, with special attention devoted to two longitudinal studies that have carefully delineated the impact of timing of early adversity in PI children.

Maltreatment across childhood

Following childhood exposure to maltreatment (see Pechtel & Pizzagalli, 2011 for review) or in the years following adoption from orphanage care (see Merz, Harlé, Noble, & McCall, 2016 for review) children show poorer EF skills. In PI children there is some evidence that longer duration of orphanage care is predictive of worse EF and regulatory abilities (Bos, Fox, Zeanah, & Nelson, 2009; Colvert et al., 2008; Loman et al., 2013; McDermott, Westerlund, Zeanah, Nelson, & Fox, 2012; Merz, McCall, Wright, & Luna, 2013; Pollak et al., 2010; Tottenham et al., 2010). Psychopathology is common across both groups of children, but higher rates of attention and hyperactivity problems in PI youth, likely reflecting poorer frontal-lobe functioning, have been argued to represent deprivation-specific patterns of functioning (e.g. see Kumsta et al., 2008). Behavioral differences in EF related to childhood maltreatment experiences are instantiated in altered prefrontal cortex functioning as indexed via fMRI. Maltreated children and adolescents show atypical functional activation during cognitive control tasks (Bruce et al., 2013; Carrion, Garrett, Menon, Weems, & Reiss, 2008; de Bellis & Hooper, 2012; Mueller et al., 2010), as well as increased frontal activation during emotion regulation (de Bellis & Hooper, 2012; McLaughlin, Peverill, Gold, Alves, & Sheridan, 2015), and error processing (Lim et al., 2015).

Neuroimaging studies of both maltreated and PI children indicate that these forms of adversity converge to impact fronto-limbic and fronto-striatal circuitry. Prefrontal cortex volumes measured at adolescence are disproportionately smaller in PI youth (Hodel, Hunt, et al., 2015) and a recent meta-analysis suggests ventro-lateral prefrontal regions are consistently smaller in maltreated children (Lim, Radua, & Rubia, 2014; see Bick & Nelson for summary of contradictory findings). Studies of both maltreated and PI youth have also documented reductions in prefrontal surface area (Hodel, Hunt, et al., 2015), folding complexity (Kelly et al., 2013), and cortical thickness measures (Hodel, Hunt, et al., 2015; McLaughlin, Sheridan, Winter, et al., 2014). Both maltreatment and early institutional care are associated with broad changes in fronto-limbic connectivity during childhood and adolescence, including reduced white matter integrity (Eluvathingal et al., 2006; Govindan, Behen, Helder, Makki, & Chugani, 2010; Hanson, Adluru, et al., 2013; Huang, Gundapuneedi, & Rao, 2012) and decreased resting functional connectivity (Burghy et al., 2012; Herringa et al., 2013). Poor white matter organization also extends into fronto-striatal circuits in PI children (Behen et al., 2009; Kumar et al., 2014). Impacts of childhood maltreatment on EF (Gould et al., 2012; Nikulina & Widom, 2013), prefrontal cortex gray matter volume (Ansell, Rando, Tuit, Guarnaccia, & Sinha, 2012; Kitayama, Quinn, & Bremner, 2006; Tomoda et al., 2009), frontal white matter organization (Choi, Jeong, Rohan, Polcari, & Teicher, 2009), and connectivity across frontal-lobe networks (Cisler et al., 2012; Jedd et al., 2015; Philip et al., 2013; van der Werff et al., 2013; Wang et al., 2014) have been documented in adults, suggesting prefrontal disruptions associated with adversity persist over development: a topic that has not been investigated in adults formerly adopted from orphanage care.

Maltreatment during infancy

A smaller literature has documented the impact of maltreatment during infancy on prefrontal cortex function. Measures of the impact of institutional care during a similar time range are lacking (as most infants are still in the orphanage), but a set of recent studies has investigated prefrontal-dependent behaviors soon after children are adopted. Structural brain development early in life has not been investigated in these populations (except see Graham, Pfeifer, Fisher, Carpenter, & Fair, 2015 for a recent report of altered default mode network connectivity in infants exposed to early life stress), and is not included in this overview. Instead the impact of timing of exposure to neglect on prefrontal cortex development is reviewed in a subsequent section.

PFC dependent behaviors

Alterations in fronto-limbic functioning are present early in life in maltreated children. Infants exposed to higher levels of non-physical interparental conflict show increased activity in anterior cingulate cortex when processing angry vs. neutral voices (Graham, Fisher, & Pfeifer, 2013). Differences in frontal lobe activity are also present in EEG studies with toddlers exposed to maltreatment as infants; 15 month-olds with a history of maltreatment show increased amplitude at frontal sites when attending to negative (angry) emotions in comparison to non-maltreated controls (Curtis & Cicchetti, 2013), a pattern that persists across early childhood (Curtis & Cicchetti, 2005, 2011). Deficits in EF have been documented in PI children as young as two years of age (Doom et al., 2014; Hostinar, Stellern, Schaefer, Carlson, & Gunnar, 2012). Poorer early EF skills in young PI children are related to longer duration of institutional care (Doom et al., 2014), less time spent in early family care (Hostinar et al., 2012), and poorer quality orphanage care (Hostinar et al., 2012), and are predicted by measures of frontal lobe electrical activity at 18 months (Tarullo, Garvin, & Gunnar, 2011).

Duration of early neglect

Given the lack of human studies describing early impacts of maltreatment on prefrontal cortex development, studies that carefully examine the impact of timing of exposure to early adversity provide crucial evidence about the importance of early experiences. Although other studies have also documented timing effects (e.g. see Cowell et al., 2015 for evidence that maltreatment during infancy has the strongest impact on later EF), this review focuses on two exemplar, longitudinal projects, both conducted with cohorts of institutionally reared children: the ERA study and the BEIP.

The English and Romanian Adoptee (ERA) study has followed the development of a random sample of Romanian children, adopted from institutional care into families in the United Kingdom in the 1990s (Rutter, Sonuga-Barke, & Castle, 2010). After the fall of Ceausescu’s government, children were quickly adopted out of institutions, resulting in a natural experiment in distribution of age at exit from orphanage care (0–42 months of age at adoption). The Bucharest Early Intervention Project (BEIP) is a randomized control trial of the benefits of foster care for Romanian children initially placed in institutional care as young infants (Zeanah et al., 2003). Because children were randomly assigned into foster care at different ages (6–31 months of age), the impact of duration of early adversity could be examined in a controlled design. Both studies have demonstrated that earlier timing of exit from orphanage care is associated with improved catch-up. However, the threshold at which timing matters varies across developmental domains, and results suggest that in many cases, normative prefrontal cortex development trajectories are altered for children who spend as little as six months of care in the institution.

PFC dependent behaviors

Both studies examined the relationship between length of deprivation and later Attention Deficit Hyperactivity Disorder (ADHD) symptom severity, a marker of altered frontal lobe development. In the ERA study, ADHD symptoms showed a threshold relationship with duration of early institutional care, with a steep increase in symptomatology for children adopted after six months of age (Kennedy et al., 2016; Kreppner, O’Connor, & Rutter, 2001; Stevens et al., 2008). The BEIP study demonstrated that EEG markers of delayed cortical development at study onset, including in frontal regions, (Marshall & Fox, 2004; Marshall, Reeb, Fox, Nelson, & Zeanah, 2008), underlie the later development of higher rates of ADHD symptoms (McLaughlin et al., 2010). Children placed in foster care before 24 months showed normalization of frontal EEG activity by 8 years of age (Vanderwert, Marshall, Nelson, Zeanah, & Fox, 2010). Intervention effects for event related potential (ERP) markers of higher-level attentional processes have also been observed in the BEIP cohort (see Bick & Nelson, 2016 for review). Interestingly, despite an impact of early placement into foster care on frontal EEG activity, ADHD symptomatology was not similarly remediated (Humphreys et al., 2015; Zeanah et al., 2009). EF deficits in the BEIP cohort were also not remediated by foster care (Bos et al., 2009) and are linked to higher ADHD symptoms (Tibu et al., 2016), perhaps because all children in the BEIP study were older than six months of age when randomization to foster care occurred.

PFC gray matter volume

The BEIP study did not report strong relationships between duration of early neglect and later measures of prefrontal cortex structure. At age 8, children who had spent any time in institutional care had smaller cortical gray matter volumes (Sheridan, Fox, Zeanah, McLaughlin, & Nelson, 2012). Reduced cortical thickness was also observed in dorsolateral and orbitofrontal regions (McLaughlin, Sheridan, Winter, et al., 2014).

PFC connectivity

In contrast to patterns of prefrontal cortex gray matter development, the BEIP study did show positive impacts of randomization into foster care on overall white matter development (Sheridan, Fox, et al., 2012). Improvements in white matter organization at age 8 in fronto-limbic and fronto-striatal tracts were present in children assigned to foster care (Bick et al., 2015).

Poverty and PFC Development

Beyond the effects of extreme deprivation experienced by PI children, recent studies suggest that variations within a more typical range of environments also relate to long-term measures of prefrontal development (see Hackman, Gallop, Evans, & Farah, 2015; Johnson, Riis, & Noble, 2016; Pavlakis, Noble, Pavlakis, Ali, & Frank, 2015 for reviews across multiple brain systems). Family socioeconomic status (SES) is a multifaceted construct, typically characterized by parental educational attainment, family income, and parental occupation. Specific aspects of SES shape different proximal features of the child’s environment, and as such, may relate to different developmental outcomes (Duncan & Magnuson, 2012). For example, micronutrient deficiencies may be more related to insufficient familial income, while parent-child interactions may vary as a function of parental education. Variance within both of these facets of SES are likely reduced in magnitude in comparison to the extreme deprivation experienced by PI children, and as such, may not alter experience-expectant brain development processes to the same degree. However, unlike the deprivation experienced by PI children, poverty is typically not restricted to the early childhood years. Growing up in poverty also increases children’s risk for experiencing physical maltreatment and neglect (Maguire-Jack, Lanier, Johnson-Motoyama, Welch, & Dineen, 2015). As such, children living in poverty commonly experience aggregating risk, both across individual risk factors and across developmental periods.

An extensive literature supports that childhood poverty (broadly defined) impacts the development of prefrontal cortex structure, function, and dependent behaviors. The majority of these studies have investigated concurrent impacts of SES in school-aged children or aggregate measures of SES across longer developmental time periods. This section highlights the concurrent and long-term impacts of poverty on children’s development, and subsequently devotes more extensive coverage to the smaller literature describing the impact of “early” SES on prefrontal cortex development.

Poverty across childhood

EF skills correlate with concurrent measures of poverty across a wide variety of EF tasks, during the preschool period through the adolescent years (e.g. see Hackman & Farah, 2009; Hackman et al., 2015). Of note, in studies where additional cognitive and/or socioemotional processes have been measured, the negative impact of poverty on EF skills is disproportionately strong (e.g. Farah et al., 2006; Noble, McCandliss, & Farah, 2007). Behavioral differences in EF are reflected in fMRI measures of prefrontal functional activity. Lower SES during middle childhood is associated with reduced prefrontal engagement during complex learning tasks (Sheridan, Sarsour, Jutte, D’Esposito, & Boyce, 2012) and studies using resting EEG or ERP measures of attention or EF have also widely documented frontal-lobe activity differences in lower-SES children and adolescents (see review in Raizada & Kishiyama, 2010).

A growing number of cross-sectional and longitudinal studies using large, population-based pediatric samples have linked measures of childhood poverty to changes in prefrontal cortex structure. In the NIH MRI Study of Normal Brain Development, lower SES during childhood predicted reduced frontal lobe gray matter volume (Hair, Hanson, Wolfe, & Pollak, 2015; although see The Brain Development Cooperative Group, 2012) and decreased cortical thickness in prefrontal sub-regions including the right anterior cingulate cortex and left superior frontal gyrus (Lawson, Duda, Avants, Wu, & Farah, 2013). Associations between SES and measures of surface area, rather than cortical thickness, were detected in the Pediatric Imaging, Neurocognition, and Genetics (PING) data set across a broadly distributed set of frontal lobe regions (Noble, Houston, et al., 2015). Smaller studies with samples of children and adolescents have provided converging results, documenting SES impacts on inferior frontal (Noble, Houston, Kan, & Sowell, 2012; Raizada, Richards, Meltzoff, & Kuhl, 2008) and superior and middle frontal gyri (Jednoróg et al., 2012) volumes. Different aspects of prefrontal structural development may relate to specific constructs underlying SES (e.g. parental education vs. familial income; Lawson et al., 2013; Noble, Houston, et al., 2015). Depth of poverty is also related to prefrontal measures, where frontal lobe volume decreases (Hair et al., 2015) and reductions in surface area (Noble, Houston, et al., 2015) are non-linearly related to childhood SES. Critically, longitudinal studies have documented that SES measures collected during childhood are predictive of prefrontal cortex development measured in adulthood, suggesting early differences can persist. Lower SES measured during childhood predicts poorer EF (Evans & Schamberg, 2009), lower orbitofrontal cortex volume (Holz et al., 2015), altered frontal connectivity in the default mode network (Sripada, Swain, Evans, Welsh, & Liberzon, 2014), and lower prefrontal activation and/or functional connectivity during emotion regulation tasks (Javanbakht et al., 2015; Kim et al., 2013; Liberzon et al., 2014) in adulthood.

Poverty during infancy

Until recently it was unclear at which age SES disparities in brain development first emerged, as very little research has been conducted in young children or using measures of SES restricted to when children were infants. Recent research suggests that the impacts of living in poverty on prefrontal cortex emerge shocking early in development.

PFC dependent behaviors

Infants living in lower-SES families perform more poorly than their higher-SES peers on a variety of behavioral tasks that reflect early EF, including the classically prefrontal-dependent A not B task (Clearfield & Niman, 2012; Lipina, Martelli, Vuelta, & Colombo, 2005) and problem solving tasks (Clearfield, Stanger, & Jenne, 2015). Deficits in higher-level memory skills are also present in the first two years of life (Markant, Ackerman, Nussenbaum, & Amso, 2016; Noble, Engelhardt, et al., 2015). Studies collecting longitudinal measures of early EF skills in infants have demonstrated that low-SES infants show a developmental lag of up to 3 months behind their higher SES peers across the first year of life (Clearfield & Niman, 2012; Clearfield et al., 2015). An EEG study of preschool aged-children also documented that lower-SES during the toddler years predicts a developmental lag in frontal power during early childhood (Otero, 1997). Developmental trajectories of higher-level attention skills required to sustained attention (Clearfield & Jedd, 2013) and explore objects (Clearfield, Bailey, Jenne, Stanger, & Tacke, 2014; Tacke, Bailey, & Clearfield, 2015) are qualitatively different in lower-SES infants than their higher-SES peers, reflecting poorer focused attention and less advanced exploration of objects across the first year of life. Behavioral differences in attention processes may be related to differences in frontal resting EEG power that are present in lower-SES infants by 6–9 months of age (Tomalski et al., 2013).

PFC gray matter volume

Hanson and colleagues (2013) were the first to document SES-related differences in the rate of frontal lobe grey matter developmental across the first years of life (Hanson, Hair, et al., 2013). Divergence of the low-SES group from normative frontal lobe volumetric development was present within the first year of life and increased in magnitude across the toddler years (Hanson, Hair, et al., 2013). A recent study indicates that SES-differences in total cortical gray matter volume are present by 5 weeks of life (Betancourt et al., 2015), although regional lobular effects were not examined. Evidence suggests that early differences in frontal lobe gray matter volume likely persist across development; familial SES measures collected at 3 months of age are predictive of orbitofrontal cortex volumes in adulthood (Holz et al., 2015), even after controlling for current SES.

PFC connectivity

A recent study reported SES-based differences in resting state functional connectivity over the first year of life (Gao, Alcauter, Elton, et al., 2015). Lower-SES infants showed alterations in default mode network connectivity measures at 6 months of age, including higher within-network connectivity and lower outside-network connectivity. However, these findings should be interpreted with some caution as they were marginally significant and were not detected when infants were re-scanned at 9 and 12 months.

Prematurity and PFC Development

Since SES is associated with prefrontal cortex development in young infants, it is possible SES differences begin even earlier in development, during the prenatal period. Studies with preterm (PT; birth before 37 weeks gestation) infants have clearly demonstrated that the early prenatal and postnatal environment dramatically shapes prefrontal cortex development. Unlike children exposed to maltreatment or growing up in poverty, the environment of early brain development for PT infants may be one of “excessive” stimulation for an under-developed brain. Furthermore, unlike most early risk factors, the exact timing/duration of exposure is known, based on the number of weeks PT infants are born early.

Extensive alterations in frontal lobe development have been reported in children born very PT (<32 weeks gestation) and/or very low birth weight (VLBW; <1500 grams; see de Kieviet, Zoetebier, van Elburg, Vermeulen, & Oosterlaan, 2012 for a general overview of brain development in very PT children). Although medical care has improved dramatically for these young and vulnerable infants, rates of both short- and long-term major medical complications, including brain injury, remain problematic (Volpe, 2009). The prevalence of significant brain injuries is drastically reduced in healthy PT infants born closer to term (Kinney, 2006; Sannia et al., 2013). However, a growing literature suggests that even healthy infants born PT are at risk for subtle, long-term alterations in prefrontal cortex development. To date, most long-term follow-up studies of PT birth have focused on higher-risk PT children. As such, this review includes literature on higher-risk PT birth, but emphasizes that lower-risk PT birth (e.g. moderate-to-late PT infants born between 32–36 weeks without major medical complications and/or children born more PT with major neurological conditions) is associated with similar, albeit smaller in magnitude, impacts on frontal lobe development. The following sections provide an overview of the long-term correlates of PT birth, followed by a more extended discussion of emerging impacts that are detectable during the infant and toddler years.

Prematurity and later development

Specific deficits in EF in very PT and/or VLBW children, not explained by general cognitive impairments (see review in Mulder, Pitchford, Hagger, & Marlow, 2009), have been consistently documented across the preschool, childhood, and adolescent years (Aarnoudse-Moens, Duivenvoorden, Weisglas-Kuperus, Van Goudoever, & Oosterlaan, 2012; Aarnoudse-Moens & Smidts, 2009; Anderson, Doyle, & Victorian Infant Collaborative Study Group, 2004; Bayless & Stevenson, 2007; Bohm, Smedler, & Forssberg, 2004; Harvey, O’Callaghan, & Mohay, 1999; Loe, Lee, Luna, & Feldman, 2012; Luu, Ment, Allan, Schneider, & Vohr, 2011; Marlow, Hennessy, Bracewell, & Wolke, 2007; Narberhaus, Segarra, Cald, & Gim, 2008; Ritter, Nelle, Perrig, Steinlin, & Everts, 2013; Saavalainen et al., 2007). Children and adolescents born very PT also show reduced prefrontal engagement during EF tasks (Griffiths et al., 2013; Mürner-Lavanchy, Ritter, et al., 2014; although see de Kieviet et al., 2014) and atypical patterns of functional activation (Nosarti et al., 2006), with group differences extending into fronto-striatal regions (Curtis, Zhuang, Townsend, Hu, & Nelson, 2006) and fronto-parietal attention networks (Carmody et al., 2006).

Higher-risk PT birth is also associated with decreased prefrontal gray matter volume (Ball et al., 2012; Nagy et al., 2009; Nosarti et al., 2008; Peterson et al., 2000; Zhang et al., 2015) and surface area (Grunewaldt et al., 2014; Lax et al., 2013; Sølsnes et al., 2015) throughout childhood and adolescence. Both prefrontal thinning (Lax et al., 2013; Nagy, Lagercrantz, & Hutton, 2011) and thickening have been reported in very PT children and adolescents vs. full-term controls (Bjuland, Løhaugen, Martinussen, & Skranes, 2013; Grunewaldt et al., 2014; Martinussen et al., 2005; Mürner-Lavanchy, Steinlin, et al., 2014; Sølsnes et al., 2015), with some suggestion that cortical development may be delayed in higher-risk PT children (Mürner-Lavanchy, Steinlin, et al., 2014; although see Rimol et al., 2016). Decreased frontal white matter volume (Giménez et al., 2006; Kesler et al., 2008; Nosarti et al., 2008) and poorerer WM integrity in frontal-lobe tracts (Constable et al., 2008; Duerden, Card, Lax, Donner, & Taylor, 2013; Mullen et al., 2011; Skranes et al., 2007; Sølsnes et al., 2016) are also present in children and adolescents born very PT. Studies examing integrity of whole brain networks report aberrant connectivity patterns across the default mode and dorsal attention networks in higher-risk PT individuals (see Kwon et al., 2016 for recent review). Alterations in both behavioral and fMRI measures of frontal lobe volume persist into adulthood (Daamen et al., 2015; Furre et al., 2016; Kalpakidou et al., 2012; Lawrence et al., 2009; Nosarti et al., 2009), as do changes in frontal lobe volume (Nosarti et al., 2014), cortical thickness (Nam et al., 2015), surface area (Furre et al., 2016), and connectivity (Eikenes, Lohaugen, Brubakk, Skranes, & Haberg, 2011), suggesting effects persist across development.

Lower-risk PT children also have poorer EF skills in comparison to full-term children during the preschool years (Baron et al., 2009; Baron, Kerns, Muller, Ahronovich, & Litman, 2012; Baron, Weiss, Litman, Ahronovich, & Baker, 2014; Brumbaugh, Hodel, & Thomas, 2014; Caravale, Tozzi, Albino, & Vicari, 2005; Hodel, Brumbaugh, Morris, & Thomas, 2015; Vicari, Caravale, Carlesimo, Casadei, & Allemand, 2004; although see Baron, Erickson, Ahronovich, Litman, & Brandt, 2010). Critically, although PT birth does occur at higher rates in low-SES families, negative impacts on frontal lobe development are detectable even in middle to upper class samples of lower-risk PT children (Brumbaugh et al., 2014; Hodel, Brumbaugh, et al., 2015), suggesting unique impacts of prematurity. Impacts of lower-risk PT birth on EF are generally smaller in magnitude than those observed in higher-risk PT cohorts, less consistent across EF tasks, and have not been detected after early school age (Baron et al., 2014; Brumbaugh et al., 2016; Gurka, LoCasale-Crouch, & Blackman, 2010; Tideman, 2000).

Few studies have examined long-term structural correlates of lower-risk PT birth, and those published are restricted to follow-up during the late school age-early adolescent period. A recent study suggests lower-risk PT children have smaller cortical surface area than full-term controls and may show delayed cortical development (Brumbaugh et al., 2016), as neither cortical thickness nor surface area measures showed expected relationships with PT children’s age. However, a second study did not report differences in frontal lobe volume or overall cortical surface area and thickness in lower-risk PT children (Rogers et al., 2014). Prefrontal functional connectivity was also altered in default mode and attentional networks in two small studies examining the same sample of lower-risk, PT children (Degnan et al., 2015a, 2015b).

Prematurity and infant development

Most differences in prefrontal cortex development described during the childhood years can be observed in PT infants within the first two years of life. When examining infant outcomes, correction for degree prematurity is important in order to match “developmental” time for PT children to their full-term peers. As such, PT infants are typically compared to full-term infants based on their “corrected” age (age adjusted for degree of prematurity), and corrected-ages are routinely used at least through age 2, including in the studies described below.

PFC dependent behaviors

The development of early attention skills that underlie later EF abilities have been well-characterized in PT infants and toddlers (see review in van de Weijer-Bergsma, Wijnroks, & Jongmans, 2008). In a series of elegant, longitudinal studies beginning in infancy, Rose, Feldman, and Jankowski demonstrated that higher-level attentional processes are altered across the first years of life in higher-risk PT infants, contributing to later differences in adolescent EF skills (see Rose, Feldman, & Jankowski, 2016 for summary). Interestingly, lower-risk PT infants may show a slight “benefit” in attentional processing earlier in development; lower-risk PT infants are faster at disengaging and shifting attention than their full-term peers, although this “benefit” in attention typically disappears by 4–6 months of age (Hitzert et al., 2015, 2014; Hunnius, Geuze, Zweens, & Bos, 2008). Other studies have also found that lower-risk PT infants perform better on early attention measures than their higher-risk PT peers (Landry & Chapieski, 1988; Reuner, Weinschenk, Pauen, & Pietz, 2015). However, by 18 months, lower-risk PT children show poorer orienting and alerting than full-term toddlers (Jong, Verhoeven, & Baar, 2015), suggesting deficits may emerge over time or may induced by atypical early attention patterns.

Early EF differences have been detected in very PT infants (Sun, Mohay, & Callaghan, 2009) and toddlers (Clark, Woodward, Horwood, & Moor, 2008; Pozzetti et al., 2013) across the first two years of life in comparison to full-term infants (see review in Gonzalez-Valenzuela et al., 2015). A small study using the A not B task found that lower-risk PT infants outperformed full-term infants when tested at their corrected-age and performed equivalently on other early measures of EF (Matthews, Ellis, & Nelson, 1996); however, this advantage for lower-risk PT infants disappeared when comparing groups based on chronological age (i.e., age from birthdate) and other studies have found variation in gestational age in lower-risk PT infants is correlated with early EF skills (Hodel et al., 2017). By toddlerhood, poorer performance on measures of inhibitory control and working memory are present in both higher-risk (Lejeune, Tolsa, Graz, Hüppi, & Barisnikov, 2015; Phillips, Ruhl, et al., 2011; Woodward, Edgin, Thompson, & Inder, 2005) and lower-risk PT children (de Haan et al., 2000; Voigt, Pietz, Pauen, Kliegel, & Reuner, 2012; although see Evrard et al., 2010). Neuroimaging measures suggest higher-risk PT infants show atypical brain activity in comparison to their full-term peers (see Mento & Bisiacchi, 2012 for review of recent literature) and this effect may also be present in lower-risk PT infants in orbitofrontal cortex regions (Wu et al., 2016). However, most studies have focused on whole-brain rather than frontal-specific differences and have not imaged frontal-dependent tasks.

PFC gray matter volume

Studies with higher-risk PT infants overwhelmingly indicate that PT birth disrupts normative trajectories of prenatal brain development (Kapellou et al., 2006), including that of prefrontal cortex gray matter. At term-equivalent age (see Anderson, Cheong, & Thompson, 2015 for review), higher-risk PT infants show regionally specific volumetric reductions in orbitofrontal cortex volume (30% decrease in higher-risk PT vs. full-term infants), and additional decreases in dorsal prefrontal regions in infants with documented brain injuries (Thompson et al., 2007). Individual differences in term frontal lobe volume are predictive of long-term neurodevelopmental outcomes (see Anderson et al., 2015 for review), including toddler measures of EF (Woodward et al., 2005). Volumetric alterations in higher-risk PT toddlers are driven by overall thicker cortex that is smaller in surface area (Phillips, Montague, et al., 2011), a pattern consistent with later childhood and adolescent brain measures in this population.

At term-equivalent age, lower-risk PT infants also show widespread changes in gray matter development, including smaller overall brain size (Munakata et al., 2013; Walsh, Doyle, Anderson, Lee, & Cheong, 2014; although see Mewes et al., 2006) and immature gyrification (Walsh et al., 2014). Similar to higher-risk PT infants, volumetric differences in lower-risk PT infants at term-equivalent age are related to later neurodevelopmental outcomes (Cheong et al., 2016; Walsh et al., 2014). However, alterations in gray matter volume in lower-risk PT infants appear to be global, rather than specific to frontal lobe.

PFC connectivity

At term-equivalent age, higher-risk PT infants show overall reductions in white matter volume and poorer white matter integrity in comparison to their full-term peers (see Pannek, Scheck, Colditz, Boyd, & Rose, 2014 for review), including poorer organization of frontal lobe white matter (Anjari et al., 2007; Rose et al., 2008). Individual differences in white matter integrity at term-equivalent age are predictive of long-term neurodevelopmental outcomes (see Pannek et al., 2014 for review), including EF during the toddler (Edgin et al., 2008) and preschool years (Woodward, Clark, Pritchard, Anderson, & Inder, 2011). Several studies have investigated functional connectivity in the higher-risk PT infant brain (see review in Kwon et al., 2016). Resting-state networks are present in PT infants by term-equivalent age (Fransson et al., 2013), but network organization is less complex in higher-risk infants born PT (Pandit et al., 2014; Smyser et al., 2010, 2016), including within frontal-lobe networks (Pandit et al., 2014; Smyser et al., 2016).

At term-equivalent age lower-risk PT infants also have decreased white matter volume (Mewes et al., 2006) and poorer white matter organization across the majority of the brain’s major white matter pathways (Kelly et al., 2016), including in fronto-limbic and fronto-striatal tracts. White matter microstructure in the lower-risk PT infant’s brain is characterized by changes in several diffusion metrics, suggesting broad alterations in myelination, axonal packing, and/or axon diameter following even lower-risk PT birth.

Interim Summary: Early Experience Shapes PFC Development

Across diverse variations in early life experience (maltreatment, poverty, PT birth) a common theme emerges: early differences in the environment impact prefrontal cortex development. Alterations in prefrontal cortex development are present within the first year of life, show continuity over developmental time, and are detectable across multiple levels of neurobehavioral development including measures of behavior, gray matter volume, white matter microstructure, and network organization. Interestingly, most long-term impacts of early life experience on prefrontal cortex development appear to be quite similar, despite vast differences in the characteristics of early risk exposure. Poorer EF skills, beginning during late infancy, are consistently reported in children exposed to adverse early environments. Similarly, reductions in prefrontal gray matter volume were common across all the experiences reviewed, along with changes in cortical thickness and surface area, as well as poorer white matter microstructural organization within fronto-limbic and/or fronto-striatal circuits.

Some diverging impacts across the three types of experience reviewed also warrant attention. Both PI children and children growing up in poverty generally had thinner frontal cortex, while higher-risk PT children had thicker frontal cortex, suggesting different disruptions in underlying neurobiological processes. Although differences in fronto-striatal functions appeared to be consistent across all three risk factors, different patterns of fronto-limbic findings emerged. Children and adults exposed to childhood maltreatment generally showed increased activation of frontal circuits during emotion regulation tasks, while adults with a history of childhood poverty showed opposing effects. Differential connectivity and functioning of fronto-limbic circuits across experiences may relate to differing dimensions of early adverse environments (McLaughlin, Sheridan, & Lambert, 2014; Sheridan & McLaughlin, 2014) and/or changes in brain development that convey an advantage in certain adverse environments (Del Giudice et al., 2011). However, even in these cases where effects were not comparable in direction among the three experiences reviewed, an overall consistent pattern remained: across development, individuals exposed to early life adversity show broad alterations in frontal lobe structure, function, and dependent-behaviors in comparison to their lower-risk peers.

Potential Mechanisms for PFC Impacts

Given that relatively diverse variations in early life experience all impact long-term prefrontal cortex development, it is unlikely that there is one specific mechanism by which early experience shapes prefrontal circuits. Studies of human infants and children clearly provide little control in investigating mechanisms. Across the three example experiences previously reviewed (maltreatment, poverty, PT), early environmental risks typically overlap; for example, children who are maltreated are also more likely to live in poverty, and poverty itself is a risk factor for PT birth. Animal models of early experience are promising in their ability to delineate separable mechanisms. However, even in these cases of greater experimental control, one change in the early rearing environment (e.g. maternal deprivation) may produce additional confounding experiences (e.g. changes in nutrient impact and thermoregulation for deprived infant animals). This review draws together three potential mechanisms via which early experience may produce long-term impacts on prefrontal neurobehavioral development, with the explicit recognition that these likely act in combination to impact later development.

1. Rapid Development Conveys Vulnerability

It is widely hypothesized that the fastest structurally growing regions of the brain are the most metabolically active, and may thus be the most susceptible to the negative effects of early environmental and/or biological insults (Hüppi et al., 1998; Volpe, 2009). Evidence was presented in the first section of this review that prefrontal cortex exhibits rapid development during early infancy, including changes in structure, connectivity, and function, often at a greater relative rate than other regions of the brain. Basic neurobiological processes including synaptogenesis, synaptic pruning, and myelination within prefrontal cortex also extend well into the early childhood period. Since prefrontal cortex grows at a rapid or perhaps even an accelerated rate in comparison to other brain regions during early life, it may be particularly sensitive to adverse early environments.

Small, early changes in rapidly developing prefrontal white matter are known to result in larger structural and functional deficits that emerge over time. Preterm infants with relatively minor white matter lesions at birth have abnormal cortical folding patterns in the central sulcus and frontal lobe sulci by term-equivalent age, but not in other regions of the brain, in comparison to PT infants without such lesions (Dubois, Benders, et al., 2008). White matter injury in neonates also causes direct effects on pre-myelinating oligodendrocytes, which remain predominant in frontal lobe regions at birth (Back et al., 2001), as well as general axonal damage and gliosis (see Smyser et al., 2013). Early, relatively minor disruptions to vulnerable prefrontal cortex white matter may result in delayed or atypical myelination effects (Nishida et al., 2006), providing a neurobiological explanation for the emergence of “sleeper effects” associated with early adverse environments.

Prefrontal cortex may also be vulnerable to variations in early experience due to its broad interconnectivity with other rapidly developing, subcortical brain regions. The putamen shows the highest relative growth rate over the first year of infancy in comparison to all other regions of the brain (Choe et al., 2013), and has extensive connections with prefrontal cortex through fronto-striatal tracts. Vulnerability of subcortical gray matter to early insults has been widely demonstrated in PT infants (Baldoli et al., 2014; Ball et al., 2013; Toulmin et al., 2015). Experience-induced changes in the development of subcortical regions could result in aberrant input to prefrontal cortex, as structural and functional connectivity between subcortical and prefrontal regions increases with developmental time.

Because much of the basic structure and connectivity of prefrontal cortex is determined prenatally, variations in experience during this time period could impact the gross architecture of prefrontal cortex structure and connectivity. Studies examining birth weight variation within the normative range, a proxy for “optimal” prenatal experience, have shown that higher-normal birth weights are associated with increased surface area in the superior and inferior frontal gyri from childhood through adolescence (Raznahan et al., 2012), suggesting that prenatal experiences can have long-term impacts on frontal-lobe development. Although joint effects of variations in prenatal and postnatal environments have not been extensively explored, cumulative exposure to negative environments would likely have the largest impact on prefrontal development. Studies with PI children provide preliminary insight into the extreme vulnerability of the developing brain when exposed to both pre- and postnatal insults. The BEIP study documented that children who were both low birth weight and experienced longer institutional care showed the most extreme cognitive vulnerability as measured by lower developmental quotient/IQ score (Johnson et al., 2010). Although IQ is not strictly prefrontal-dependent (Roca et al., 2010), in typically developing children IQ and prefrontal-dependent EF skills overlap. Global deficits in IQ in these higher-risk PI children highlight the cumulative and interactive effects of disruptions to both the prenatal and postnatal environment on rapidly developing frontal circuits.

2. Sensitivity to Stress

An extensive animal literature supports the argument that stress is a form of environmental variation that can differentially alter prefrontal cortex structure and function. All of the variations in early experience reviewed in this paper share the potential to be biologically and/or psychologically challenging to the developing organism. Infants who have experienced maltreatment (Cicchetti, Rogosch, Toth, & Sturge-Apple, 2011) and children currently residing in orphanages (Carlson & Earls, 1997) have altered cortisol rhythms, as do infants exposed to prenatal risk factors (Osterholm, Hostinar, & Gunnar, 2012) and/or born very PT (Grunau et al., 2007). Altered cortisol production is present in lower-SES infants within the first year of life (Blair, Raver, Granger, Mills-Koonce, & Hibel, 2011; Clearfield, Carter-Rodriguez, Merali, & Shober, 2014) and is predictive of later individual differences in early childhood measures of EF (Blair, Granger, et al., 2011; see Berry et al., 2016 for review). It is probable that, at least for some early variations in environmental experience, the physiological and/or subjective experience of stress provides a final common pathway to altered prefrontal development. For example, models of toxic stress and/or allostatic have been widely used in characterizing the converging impact of multiple stressors present in the lowest-SES environments on children’s development (Evans et al., 2013), and are consistent with previously discussed studies documenting non-linear relationships between depth of poverty and prefrontal cortex development.

Animal models of early life stress allow for a more thorough investigation of how the timing and nature of specific stressors impact prefrontal cortex development. Since parent-offspring interactions are comprehensive and necessary for the survival of offspring in many mammalian species, these studies typically use paradigms involving temporary separations of young animals from parental care. Maternal deprivation in rodents is associated with an increased rate of apoptotic cell death in frontal cortex of neonatal animals, suggesting specific vulnerability of this brain region (Zhang et al., 2002). Deprivation-induced changes in prefrontal gray matter volume have also been observed in primates (Lyons, Afarian, Schatzberg, Sawyer-Glover, & Moseley, 2002; Spinelli et al., 2009). Parental deprivation in highly social Octodon degus results in a cascade of prefrontal alterations including increased synaptic density in medial prefrontal cortex (Ovtscharoff & Braun, 2001), lowered spine density and reduced dendritic length in orbitofrontal cortex (Helmeke et al., 2009), and a shift in the balance between serotonergic and dopaminergic innervations in both medial (Braun, Lange, Metzger, & Poeggel, 2000) and orbital (Poeggel, Nowicki, & Braun, 2003) frontal regions. This pattern of results suggests that early deprivation interferes with both typical prefrontal synapse formation (Bock et al., 2005) and pruning (Ovtscharoff & Braun, 2001), with effects persisting into adulthood.

The impact of early life stress on prefrontal cortex development varies based on the age at which effects are assessed and the timing and nature of the stressor. For example, when rodents were exposed to mild prenatal stress, their offspring showed increases in spine density in both medial prefrontal and orbitofrontal cortex as juveniles; however by adulthood there was no effect of prenatal stress on orbitofrontal cortex spine density and the medial prefrontal cortex effect was now reversed in direction (Muhammad, Carroll, & Kolb, 2012; Mychasiuk et al., 2012). These results suggest a potential neurobiological correlate of “sleeper effects” sometimes observed in studies of human children, as long-term impacts of early life experience may take time to fully emerge. Timing of stressor exposure also predicts neurodevelopmental outcomes, with some suggestion that prenatal stressors are more disruptive to long-term development than stress during the early postnatal period (Muhammad et al., 2012). Even prenatal stressors that are distal to the actual pregnant animal (e.g. stressing a neighboring cage mate) have been associated with changes in microstructural properties of prefrontal cortical neurons in offspring (Mychasiuk, Gibb, & Kolb, 2011), an interesting parallel to models of human experience that consider the impact of risk at proximal vs. distal levels of the environment. Since prenatal stress occurs before dendritic growth and synaptic formation have plateaued, it has a stronger impact on the formation of neural networks (Kolb et al., 2012), and may set the range of physiological systems for the later course of development (McEwen, 2012). For human children exposed to early life adversity, prenatal and postnatal stress likely co-occur, a topic discussed at the end of this review.

Animal studies also indicate that prefrontal cortex is sensitive not only to parental deprivation, but also to other forms of atypical social experience that activate stress biology. Mice who experience social isolation at any age show similar patterns of alterations in prefrontal microstructure that persist following social reintegration. However, younger animals show more pronounced changes in prefrontal-dependent behaviors and oligodendrocyte microstructure than those observed in adults (Liu et al., 2012; Makinodan, Rosen, Ito, & Corfas, 2012), suggesting a sensitive period for early social interaction exists. In comparison to rodents, humans experience a much longer period of extensive social interaction with their caregivers and a higher reliance on learning from others in social contexts (Csibra & Gergely, 2006), indicating that impacts of atypical social environments on prefrontal cortex development may be even greater for human infants and toddlers.

Ultimately, the importance of stress as a major mechanism by which early experience impacts prefrontal development cannot be ignored. However, there is a substantial literature indicating that prenatal and early postnatal stress in both animals and human infants does not uniquely impact prefrontal development. Stress has well-known effects across multiple regions of the brain, with an extensive literature documenting its unique and differential impacts on hippocampal and amygdala development (see review in McEwen, 2012). All of the subcortical regions that have been shown to be “stress-sensitive” project directly or indirectly to prefrontal cortex. It is unclear if the vulnerability of prefrontal cortex to early life stress is due to: 1) intrinsic properties of prefrontal cortex itself, 2) receipt of aberrant input from other stress-sensitive brain regions, or 3) some combined, interactive effect. In the context of understanding early life stress and prefrontal development, it is likely best to view prefrontal cortex as part of several integrated systems that are highly reactive to stressful early experiences.

3. Importance in Early Learning

If prefrontal cortex constitutes a critical hub region for early learning and brain development, small disruptions to normative processing could have cascading impacts across multiple cognitive and brain systems. The traditional view of infant brain development as strictly hierarchical, with organization and specialization moving from lower-level sensory circuits to higher-level cortical regions, does not appropriately capture the early functional characteristics of infant frontal lobe (Dehaene-Lambertz & Spelke, 2015). Instead, evidence has already been presented earlier in this review that infant prefrontal cortex is broadly active across a variety of infant cognitive and social information processing tasks, suggesting this brain region plays an important role in early learning and in shaping patterns of cortical development.

Studies of learning in adults routinely report increased prefrontal activity during skill acquisition and learning of new motor behaviors, with prefrontal activity decreasing once skills have become well-learned (Fletcher et al., 2001; Sigman et al., 2005). Similarly, fMRI studies with young children report that diffuse patterns of activation, including in frontal regions, decrease with development to be replaced by more task-specific activation (Casey et al., 1997; Durston et al., 2006). Much of this apparently age-related change might be due to underlying learning processes. In infants, broad prefrontal cortical activity observed across diverse tasks could reflect active learning from predictable, yet complexly organized, environmental stimuli. For example, a recent study demonstrated that 8-month old infants are able to spontaneously learn hierarchical rule structures, and that this learning is related to eye-blink rate, an indirect measure of early fronto-striatal dopaminergic functioning (Werchan, Collins, Frank, & Amso, 2015). Many cognitive and socioemotional processes likely place high attentional and memory demands on the young infant brain. There is also increasing recognition that even young infants exhibit sophisticated information processing capabilities that serve as precursors to later EF (Hendry, Jones, & Charman, 2016; Rose, Feldman, & Jankowski, 2012), which may be broadly engaged across a variety of learning contexts. As attentional and memory demands of learning become less effortful with development, prefrontal-involvement likely decreases.

Considering learning more broadly at the circuit level, Grossman (2013) has argued that a major task of prefrontal cortex in infancy is to learn to select the appropriate pattern of posterior regional activity for a given cognitive or motor task. Accordingly, this “efficient selection” cannot happen until better prefrontal long-range structural and functional connectivity develop with time and experience (Grossmann, 2013a). A role for prefrontal cortex in coordinating network activity is consistent with conceptualizations of medial prefrontal cortex as an early “hub-region” for brain development, playing a strong role in coordinating information within and across brain networks (Gao, Zhu, et al., 2009). Although this account has not been tested empirically, it would predict high levels of prefrontal activity across tasks during early infancy, followed by decreases in prefrontal activation across tasks as children age. Decreasing prefrontal activity would be replaced by activation in more posterior regions, along with concomitant changes in integrity of long-range prefrontal structural and functional connections, a pattern that could be tested in a longitudinal fNIRS/DTI data set.

The ability of humans to efficiently learn from others is a key adaptation of the human species (Csibra & Gergely, 2006), and developing an understanding of other people is a fundamental task for the human infant in learning about the environment (Grossmann & Johnson, 2007). For example, infant errors on the classic A not B task are known to be influenced by the social context in which the task takes place; searching errors traditionally attributed to failure of working memory or inhibition may instead reflect a bias by the infant to interpret the experimenter’s social intent (Topal et al., 2008). In the adult brain, a network of brain regions commonly referred to as the “social brain” exists which specializes in learning from and processing information related to social agents (Adolphs, 2009). Portions of infant medial prefrontal cortex also demonstrate functional specialization for detection of social signals early in development (Grossmann, 2013a, 2013b, 2015) and early attentional mechanisms exist which enhance infant attention to social stimuli (Johnson, 2005). Differences in frontal lobe processing of social information for infants who experience atypical social environments (i.e. maltreatment) are present with the first years of life (Curtis & Cicchetti, 2013). Furthermore, because infants develop in socially complex environments, developmental pathways for what appear to be basic cognitive skills (e.g. sustained attention) may also be dependent on normative social experiences (Yu & Smith, 2015) and functioning of early social information processing networks that include medial prefrontal cortex.

It is also possible that early learning experiences during infancy “prime” or shape later learning capabilities within prefrontal circuits. Multiple studies have reported that training in young animals improves learning measured on equivalent tasks at adulthood, even though young animals may not show evidence of learning during training (see review in Bock, Poeggel, Gruss, Wingenfeld, & Braun, 2014). For example, on an avoidance learning paradigm young rats who were trained between postnatal days 17–21 did not show behavioral evidence of learning, but did show changes in spine formation rates in ventromedial and lateral orbitofrontal cortex (Bock et al., 2014). Changes in spine formation appeared to reflect a form of “tagging” functionally relevant synapses, protecting them from synaptic elimination during later development. As adults, these rodents are able to activate circuits that were “pre-shaped” by early training experience, accelerating learning in comparison to animals who were not trained during early life (Bock et al., 2014). Environmental entraining of frontal circuits has also been demonstrated in the adult human neuroimaging literature using measures of functional connectivity (Kelly & Castellanos, 2015). Viewed from this perspective, infant learning experiences broadly optimize the prefrontal cortex for adult learning, due to microstructural changes in prefrontal circuitry as a result of early experience.

In combination, these results broadly demonstrate the underappreciated importance of prefrontal cortex for infant learning across multiple contexts, including in behavioral measures of learning and in organizing and constraining brain development at both the synaptic and network level. Many of the hypotheses put forward here require further testing. For example, whether the absence of normative learning experiences (i.e. neglect) or presence of highly aberrant learning experiences (i.e. maltreatment) impact the ability of prefrontal hub regions to organize and/or constrain trajectories of brain development at the synaptic or circuit level has not been empirically tested. However, from this broader perspective, relatively small changes in prefrontal cortex development could have immediate impacts on vast aspects of infant neurobehavioral development and could initiate cascading changes in future cognitive abilities that build on earlier skills.

General Discussion

To summarize, infant prefrontal cortex shows precocious structural, connectivity-based, and functional development during the late prenatal period and into the first years of life. Sensitivity of this rapidly developing brain region to environmental influences allows for efficient tuning of prefrontal circuits to the early environment. Unfortunately, evidence from both human and animal studies highlights the vulnerability that goes along with this normally adaptive process. Exposure to biological and environmental risk factors during this time period of rapid frontal-lobe development induces deficits within the first year of life that may persist over time.

The urgent societal need to provide normative environments that support prefrontal cortex development in young infants is quite clear. Despite massive recovery in children adopted from institutional care, six months of institutional rearing in infancy can produce permanent changes in prefrontal cortex development. Near the end of the first year of life, lower-SES infants show a developmental lag of approximately three months on measures of early EF skills in comparison to their higher-SES peers. Differences of only 4–8 weeks of gestation in lower-risk PT infants are associated with alterations in brain volume and white matter connectivity at term-equivalent age, and lasting differences in prefrontal-dependent behavior when children reach preschool. The similar neural and behavioral phenotype of diverse forms of early risk suggests a final common pathway of altered prefrontal cortex development. However, important differences in prefrontal development across early risk exposure (e.g. frontal lobe cortical thinning following neglect but thickening in PT children) emphasize that the timing and nature of early adverse experiences matters. Characterizing longitudinal trajectories of normative prefrontal cortex development and development following adversity is difficult for a myriad of reasons. However, this work is critical to understand both why prefrontal circuits may be vulnerable to early experiences, and how or if development may recover following early adversity.

Challenges and Recommendations

Measuring infant outcomes

Innovations in infant multimodal imaging have been critical in determining the neural correlates of early experience during the infant and toddler years. Advances in the development of sensitive behavioral measures of early prefrontal cortex skills have also been dramatic, but measurement of these higher-level constructs in young infants remains challenging (see Hendry, Jones, & Charman, 2016 for recent discussion). An important future direction will be to determine the degree to which various infant and toddler measures of early EF abilities show developmental stability (e.g. Rose, Feldman, & Jankowski, 2012) and to more carefully test their reliance on prefrontal circuitry using functional neuroimaging techniques. Most studies examining prefrontal functioning in young infants have used fNIRS, which has limited spatial resolution in comparison to fMRI and is restricted to measurement of superficial cortical activity. Cognitive neuroscience approaches that more carefully segregate measures of infant sensory, cognitive, and social information processing into constituent components and/or parametrically vary task processing demands will be useful in determining if infant prefrontal cortex subserves unique functions, or if broader cross-task activations are due to the increased attention and memory demands for young infants inherent in processing complex stimuli.

The development and utilization of sensitive measures of both behavioral and brain differences in prefrontal circuits are likely especially important to detect altered functioning in children who have experienced more subtle forms of early adversity. For example, in lower-risk PT children, recent studies suggest that behavioral EF impairments do not persist into the school years (e.g. Baron, Weiss, Litman, Ahronovich, & Baker, 2014). However, this is not consistent with large, population-based studies demonstrating higher rates of attention problems and poorer school performance in lower-risk PT individuals across development (e.g. Chan, Leong, Malouf, & Quigley, 2016; Chan & Quigley, 2014; Talge et al., 2010). Sensitive measures across various levels of neurobehavioral development may be particularly helpful to determine if, even in the context of equivalent behavior, prefrontal circuitry has been re-sculpted by early life experience, and to determine how early adversity impacts different levels of organization within prefrontal systems in tandem (e.g. reductions in regional prefrontal volume but increased fronto-limbic connectivity, etc).

Characterizing early and later environments

Defining and measuring experiences remains an inherent difficulty in this field. Fortunately, substantial progress has been made in developing frameworks that characterize differences in the timing and nature of early experiences. Critically, some of these perspectives have also highlighted the importance of viewing changes induced by the early environment in prefrontal circuitry not as “deficits” in brain development, but as attempts to support adaptation within the context of adversity (Callaghan & Tottenham, 2016; McEwen, 2012). For example, a history of early institutional care is associated with atypical patterns of functional connectivity within fronto-limbic systems. Rather than “deficits” in connectivity, fronto-limbic circuits may instead mature early, perhaps reflecting an adaptive trade-off for life in the orphanage environment that may have negative consequences later in the post-adoptive environment (Callaghan & Tottenham, 2016; Gee et al., 2013). Considering the nature of environments that follow exposure to early adversity may lead to important advances in our understanding of how delayed or “sleeper” effects of early adversity emerge over developmental time.

The corollary of understanding the sensitivity of prefrontal cortex to negative environmental experiences is considering the role of both normative and positive experiences in its development. Longitudinal studies have demonstrated that normative variations in maternal sensitivity during infancy are predictive of individual differences in children’s EF skills later during toddlerhood and into the preschool years (Bernier, Carlson, Deschenes, & Matte-Gagne, 2012; Bernier, Carlson, & Whipple, 2010; Blair, Granger, et al., 2011; Cuevas et al., 2014). However, there is little information about how normative or positive environmental variations influence prefrontal cortex structure and/or connectivity over development. Rodent studies of enriched neonatal tactile stimulation produce positive changes in prefrontal cortex structure and prefrontal-dependent behaviors at adulthood (Kolb et al., 2012). One recent study demonstrated that early individual differences in human parenting behaviors, within the normative range, were prospectively related to EEG measures of infant frontal lobe development (Bernier, Calkins, & Bell, 2016). Similarly, interventions for PT infants that minimize mother-infant separation in the Neonatal Intensive Care Unit are associated with more typical patterns of frontal lobe EEG activity and development (Myers et al., 2015; Welch et al., 2014). Investigating how normative and positive environments influence developmental trajectories of prefrontal cortex has important implications for both describing the biological mechanisms via which experiences are instantiated in the brain, and for considering how normative or positive experiences may partially ameliorate the impacts of exposure to adverse environments.

Considering timing and plasticity

An implicit assumption across many studies examining neural correlates of early adversity is that early experiences show some privilege over later experiences in shaping developmental trajectories. In addition to denoting impacts of timing, longitudinal study designs are needed to determine the trajectory of prefrontal cortex development following adversity. In studies which examine outcomes years following early adversity it is impossible to distinguish if subsequent exposure to positive environments has reduced the negative impact of early adversity over time, if impacts have remained relatively constant over development, or if impacts are emerging or increasing as children age and the environment becomes more cognitively demanding. There is some evidence that early experiences do matter more in shaping trajectories of prefrontal cortex development. Measures of childhood poverty are predictive of adulthood prefrontal cortex structure (Holz et al., 2015), connectivity (Sripada et al., 2014) and function (Javanbakht et al., 2015; Muscatell et al., 2012) even after controlling for concurrent, adulthood measures of social status and/or income. Rare studies that have used random assignment out of early adverse environments (e.g. BEIP study) allow for casual inferences to be made about the impact of subsequent exposure to positive environments, although ethical considerations make this type of work inappropriate in almost all contexts (Millum & Emanuel, 2007).

A very limited number of studies have specifically compared the impact of early vs. later experiences on trajectories of prefrontal cortex development, few studies have longitudinal measures, and even fewer examine the impact of “early” experiences in the first years of life vs. childhood more broadly. Many children exposed to early adverse environments also unfortunately go on to experience risk factors across subsequent environments. As such, the strength of casual inferences regarding impacts of early experience on long-term prefrontal development is quite tenable. Integration of results from correlational studies in humans with animal models of early adversity, in which temporal and dimensional characteristics of experience can be tightly controlled, is an extremely valuable approach, although cross-species comparisons of animal and human experiences are not without difficulties (Brett, Humphreys, Fleming, Kraemer, & Drury, 2015).

Ultimately, whether sensitive periods exist for higher-level prefrontal-dependent behaviors in human infants and/or toddlers is unclear (Fox et al., 2010), although emerging evidence supports the existence of early sensitive periods for the development of emotion regulatory systems that include fronto-limbic networks (Callaghan & Tottenham, 2015). Adaptive calibration models (Del Giudice et al., 2011) also posit that there are multiple developmental periods in which biological systems are highly sensitive to environmental input, potentially including prefrontal systems that undergo continued refinement during adolescence (Fuhrmann, Knoll, & Blakemore, 2015). As such, important future directions to better characterize the impact of timing of early adversity on prefrontal development include: 1) utilizing longitudinal study designs, beginning early in development, to determine the trajectory of neurobehavioral changes, 2) integrating results with animal studies in which timing and dimensional characteristics of adverse environments can be tightly controlled, and 3) determining when prefrontal systems are more or less sensitive to the environment across development.

Themes for Future Inquiry

Systems perspectives

The human brain develops as an integrated system, across neural circuits, across biological systems, across levels of environment, within the growing individual. An important theme for future research is to better integrate findings at the “systems” level. For example, describing how rapidly changing subcortical and prefrontal cortical systems interact across development in the infant and possibly even fetal brain (Thomason et al., 2015) will likely improve our understanding of how neural systems organize and reciprocally shape one another’s early functional activity. In tandem, delineating the impacts of early experience on separable frontal circuits (e.g. fronto-limbic vs. fronto-striatal) is an important step in documenting specificity and/or convergence of effects across frontal systems and fits well within existing frameworks that characterize the nature of early adverse environments (Humphreys & Zeanah, 2014; Sheridan & McLaughlin, 2014). An additional “system” of development that has been too often ignored in the literature on early adversity is the prenatal period. Impacts of variations in the prenatal environment in shaping long-term physiology of the organism are well-known in epidemiology (Godfrey & Barker, 2001), and accordingly, shape children’s later brain development (e.g. Raznahan, Greenstein, Lee, Clasen, & Giedd, 2012; Sarkar et al., 2014; Sowell et al., 2008). As prenatal risk and postnatal risk are often confounded in human children, the prenatal period must be incorporated as another system of influence on children’s development that may shape trajectories of how young infants and toddlers respond to early adversity (Fisher et al., 2016). Last, the brain does not develop in isolation from the body. Understanding interactions between multiple biological systems sensitive to adversity, including physical growth, immunological functioning, and epigenetic modifications will more clearly illuminate how broader prefrontal cortex development is sensitive to environmental variations.

Individual differences

Resilience processes explain the variability of outcomes in individuals following exposure to past or concurrent adversity (Masten, 2001). Most studies investigating either normative infant prefrontal cortex development or development of this region following adversity have focused on age-related or group-level changes, rather than characterization of individual differences (although see Humphreys & Zeanah, 2014). As previously discussed, evolutionary biology theories posit that individuals vary in their degree of sensitivity to both positive and negative features of the environment (Belsky & Pluess, 2009; Boyce & Ellis, 2005). Although individual differences may exist, prefrontal cortex development is sensitive to the environment in all individuals and this sensitivity continues across the entire lifespan at the neurobiological level (Greenough et al., 1987; Lovden et al., 2010). Understanding normative individual differences in prefrontal development, as well as individual differences in prefrontal circuitry following exposure to adverse environments, is an important theme for future research; however, in tandem, we must also recognize that prefrontal cortex is always plastic to some degree, for every individual, at every point in development.

Translational implications

The public health and policy implications of this growing body of literature are quite clear. Disparities in prefrontal cortex development induced by diverse early risk factors (e.g. maltreatment, poverty, PT birth) are observable shockingly early in development, and infants at the highest levels of biological and environmental risk are disproportionately impacted. Waiting to intervene until children are preschool-aged misses an early window of opportunity, when prefrontal circuits organize and are especially sensitive to the environment. Although much remains to be learned about the mechanisms underlying sensitivity of prefrontal cortex to the early environment, intervention programs designed to target environmental disparities may produce a better return on investment earlier in development: in the toddler, infant, or even prenatal periods (see Doyle, Harmon, Heckman, & Tremblay, 2009 for a discussion on timing).

There is also little evidence to suggest that differences in prefrontal cortex development induced by early experience are completely immutable. Designing interventions to promote typical prefrontal cortex development in infants is an emerging science (Wass, Porayska-Pomsta, & Johnson, 2011; Wass, Scerif, & Johnson, 2012; Wass, 2015). Although few training programs have been designed for infants and toddlers, interventions using innovative methods such as eye-gaze control (in which stimulus presentation is contingent on where infants are looking) provide promising evidence that training early EF skills alters the way infants experience and interact with their world; even small doses of this training are effective in the first year of life (Wass et al., 2011, 2012) and show transfer across cognitive domains (Wass et al., 2012). Additionally, because early prefrontal-dependent skills develop in the context of the parent-child relationship (Swingler, Perry, & Calkins, 2015), interventions to improve infant prefrontal cortex development can also be targeted at the dyadic level. Sensitive parenting has been linked to better EF development in younger children (Bernier, Calkins, & Bell, 2016; Bernier, Carlson, Deschenes, & Matte-Gagne, 2012; Bernier, Carlson, & Whipple, 2010), and longitudinally to adolescent structural development of prefrontal cortex (Whittle et al., 2014). Thus, interventions that target the development of sensitive, responsive caregiving in parents may positively shape infant prefrontal cortex development. Ultimately, the untapped potential for intervention during this early window of prefrontal cortex development, a time in which both behavioral and neural plasticity of prefrontal cortex is heightened, and during which transfer of training across cognitive domains may be greater than later in development (Wass, 2015), cannot be ignored. Fortunately, experiences or interventions that occur beyond early childhood can also have profound impacts on prefrontal cortex development (Diamond & Lee, 2011). However, the possibility of intervention should not diminish the severity of potential long-term disruptions in prefrontal cortex development (DiPietro, 2000). Even following years of exposure to positive environments, the literature reviewed in this paper suggests that some differences in prefrontal cortex development are not fully ameliorated. To fully reduce the impact of adverse early experiences on subsequent prefrontal cortex development, changes to broader social policies that reduce disparities are likely necessary.

Conclusion

Once characterized as “late-developing”, human prefrontal cortex is a “rapidly maturing” region of the brain, which functions earlier and shapes cortical development more broadly than previously appreciated. Substantial evidence indicates that diverse forms of early risk converge to similarly impact the development of prefrontal cortex structure, function, and dependent-behaviors within the first years of life. Multiple mechanisms exist via which the timing and nature of early experiences may shape early prefrontal cortex development, although this work has not sufficiently investigated correlates of normative and positive variations in the environment. Pervasive, early emerging impacts of adverse environments on prefrontal cortex development in young infants highlight the importance of understanding: 1) interactions among rapidly developing biological systems, 2) differences in sensitivity of prefrontal cortex to the environment both across individuals and over developmental time, 3) and when and how to target interventions to reduce the negative impacts of early adverse experiences by capitalizing on the heightened behavioral and neural plasticity of prefrontal cortex during infancy.

Highlights.

• Prefrontal cortex develops rapidly between 0–2 years of age.

• Diverse forms of early adversity converge to disrupt prefrontal circuits.

• Understanding neural mechanisms of early adversity remains difficult.

• Infancy is a time of both vulnerability and opportunity in prefrontal development.