Neurodevelopmental disorders


  • Hana D'Souza,

    Corresponding author
    1. Sensorimotor Development Research Unit, Department of Psychology, Goldsmiths, University of London, London, UK
    2. Centre for Brain and Cognitive Development, Birkbeck, University of London, London, UK
    3. The London Down Syndrome Consortium (LonDownS), University College London, London, UK
    Search for more papers by this author
  • Annette Karmiloff-Smith

    1. Centre for Brain and Cognitive Development, Birkbeck, University of London, London, UK
    2. The London Down Syndrome Consortium (LonDownS), University College London, London, UK
    Search for more papers by this author

  • This article is part of a special collection on developmental systems designed to explore the powerful array of forces shaping the individual development of brains, bodies, and behavior. The collection was created and edited by Mark S. Blumberg (University of Iowa), John P. Spencer (University of East Anglia), and David Shenk (Author, The Genius in All of Us), in conjunction with the DeLTA Center at The University of Iowa. View the full collection: How We Develop – Developmental Systems and the Emergence of Complex Behaviors
  • Conflict of interest: The authors have declared no conflicts of interest for this article.


Recent technological advances allow us to measure how the infant brain functions in ways that were not possible just a decade ago. Although methodological advances are exciting, we must also consider how theories guide research: what we look for and how we explain what we find. Indeed, the ways in which research findings are interpreted affects the design of policies, educational practices, and interventions. Thus, the theoretical approaches adopted by scientists have a real impact on the lives of children with neurodevelopmental disorders (NDDs) and their families, as well as on the wider community. Here, we introduce and compare two theoretical approaches that are used to understand NDDs: the neuropsychological account and neuroconstructivism. We show how the former, adult account, is inadequate for explaining NDDs and illustrate this using the examples of Williams syndrome and specific language impairment. Neuroconstructivism, by contrast, focuses on the developing organism and is helping to change the way in which NDDs are investigated. Whereas neuropsychological static approaches assume that one or more ‘modules’ (e.g., visuospatial ability in Williams syndrome) are impaired while the rest of the system is spared (e.g., language in Williams syndrome), neuroconstructivism proposes that basic-level deficits have subtle cascading effects on numerous domains over development. Neuroconstructivism leads researchers to embrace complexity by establishing large research consortia to integrate findings at multiple levels (e.g., genetic, neural, cognitive, environmental) across developmental time. WIREs Cogn Sci 2017, 8:e1398. doi: 10.1002/wcs.1398

For further resources related to this article, please visit the WIREs website.


Dramatic increases in the availability of data on neurodevelopmental disorders (NDDs; see Table 1) are providing new and exciting opportunities for understanding the infant brain and the factors that shape its development. Much of the excitement in this field can be traced to technological advances that, in ways that were not possible just a decade ago, allow us to measure how the infant brain functions (see Bick and Nelson, Early experience and brain development, WIREs Cogn Sci, also in the collection How We Develop). Although these methodological advances are exciting, we must also consider how theories guide research: what we look for and how we explain what we find. Here we compare two very different theoretical approaches for explaining NDDs—the neuropsychological account and neuroconstructivism—and explain why we believe that the latter approach is more likely to lead to better treatments and outcomes.

Table 1. A List of Neurodevelopmental Disorders, Ordered by Prevalence
Condition Prevalence per 100
  1. Prevalence has been divided by two if reported for one sex only. (Reprinted with permission from Ref [4]. Copyright 2010 Public Library of Science; modified)
Lesch-Nyhan syndrome0.0005
Lowe syndrome0.0005
Rubinstein-Taybi syndrome0.0008
Cornelia de Lange syndrome0.0014
Cri du chat syndrome0.0020
Angelman syndrome0.0040
Williams syndrome0.0044
Marfan syndrome0.0067
Prader-Willi syndrome0.0067
Rett syndrome0.0080
Duchenne muscular dystrophy0.0143
Tuberous sclerosis0.0167
Trisomy 180.0250
Velocardiofacial syndrome0.0250
Neurofibromatosis type 10.0308
Turner syndrome0.0400
Noonan syndrome0.0571
Fragile X syndrome0.0615
Klinefelter syndrome0.0860
Fetal alcohol syndrome0.1000
Cerebral palsy0.1500
Down syndrome0.1667
Tourette syndrome0.5000
Autistic spectrum disorder0.6500
Developmental dyscalculia3.0000
Attention deficit hyperactivity disorder5.0000
Intellectual disability5.5000
Developmental dyslexia6.0000
Developmental coordination disorder6.5000
Specific language impairment7.4000
Speech sound disorder10.0000

To illustrate the differences between the two approaches, consider Williams syndrome, a rare genetic disorder. Researchers have documented the fact that children with Williams syndrome have problems with number processing. If we adopt a neuropsychological approach[5, 6] and posit that the brain is composed of independently functioning modules, one of which is a number-processing module, then interventions will be focused specifically on the number domain once the deficit is identified in middle childhood. In contrast, if we adopt a neuroconstructivist (or other developmental) approach and posit that number processing is an emergent property of a self-organizing system that interacts with the environment, then we will focus our intervention efforts much earlier in development to identify those influences that set number processing on an atypical developmental trajectory.[7] Indeed, differences in basic-level processes (e.g., poor visual attention) might affect the way in which infants with Williams syndrome scan numerical displays, faces, and other visual stimuli. So, adopting a neuroconstructivist or other developmental approach, we would target those basic-level processes with the aim of improving developmental processes and outcomes.


Understanding NDDs requires not only describing differences between typically developing individuals and individuals with NDDs, but also understanding how these differences emerge over developmental time. This entails tracking differences as early in development as possible and on multiple levels of description (e.g., genetic, neural, cognitive, environmental). Up until recently, progress has been limited by the absence of suitable methodologies for infants and young children. Recent technological advances (e.g., new infant-friendly electroencephalography [EEG, Figure 1]) allow us to measure how the infant brain functions (for a comparison of different methods used to measure infant brain activity, see Figure 2). This enables researchers to elucidate how changes on the neural level are associated with changes on other levels of description—such as cognition or behavior. Furthermore, EEG and neuroimaging approaches have the potential to help identify important early markers of NDDs of unknown origin (e.g., attention-deficit/hyperactivity disorder, autism) before the behavioral symptoms emerge later in development. This could potentially have high clinical significance since it would provide an opportunity for timely interventions. EEG and neuroimaging approaches can further be used as tools for assessing the impact of these interventions.

Figure 1.

An infant with Down syndrome wearing a cap that records electrical activity in the brain. Electroencephalography (EEG) is one of the most commonly used methods to measure brain responses in infants.

Figure 2.

This figure compares the spatial and temporal resolution of different methods used to measure brain activity in infants. It also illustrates the relative degree of tolerance needed from the infant for each method, ranging from yellow (low) to red (high). EEG, electroencephalography; ERP, event-related potential; MEG, magnetoencephalography; NIRS, near infrared spectroscopy; fMRI, functional magnetic resonance imaging; DTI, diffusion tensor imaging; PET, positron emission tomography. (Reprinted with permission from Ref [8]. Copyright 2010 Elsevier)

Although methodological advances are exciting, we must also consider how theories guide research: what we look for and how we explain what we find. The interpretation of the research findings then affects the design of policies, educational practices, and interventions. Thus, the theoretical approaches adopted by scientists have a real impact on the lives of many children with NDDs and their families, as well as on the wider community.

The neuropsychological account views the brain as if it were a Swiss Army knife,[9] containing built-in, separate, special-purpose tools. Proponents of this approach argue that the brain has discrete parts, often referred to as modules, each of which has a specific, evolved cognitive function. They use the existence of adult neuropsychological patients to support the modularity view. One example given by advocates of this approach invokes patients with damage to the fusiform face area (FFA), which is a part of the cerebral cortex located in the temporal lobe (see Haist and Anzures, Functional development of the brain's face-processing system, WIREs Cogn Sci, also in the collection How We Develop). The FFA is selectively activated when adults perceive faces.[10, 11] When the FFA has been damaged in adult patients, the ability to identify faces is particularly impacted to the point that otherwise normally functioning patients are unable to recognize their partner or even themselves in the mirror[12, 13] (Prosopagnosia Research Centers: This is the basis for the claim that there exists an independently functioning face-processing module.

Indeed, theorists using data from neuropsychological patients and embracing the modularity hypothesis take the existence of double dissociations in adults to make strong claims about the functional architecture of the human mind/brain (e.g., Refs [14-16]; see discussion in Shallice[17]). A double dissociation is established when a specific brain lesion X in Patient 1 relates to poor performance on Task A but not Task B, whereas a different brain lesion Y in Patient 2 relates to poor performance on Task B but not Task A.[18]

The usual assumption is that the brain lesion is the direct cause of poor performance, leading to the conclusion that the cognitive function necessary to perform well on Task A is linked to the brain region damaged in Patient 1, and vice versa for Patient 2. Hence, double dissociations for different cognitive abilities in brain-damaged adults are used to relate brain and cognitive data directly, thereby elucidating the architecture of the adult cognitive system and how brain lesions affect it.

Returning to the Swiss Army knife analogy, according to the neuropsychological account, it is possible to selectively break one of the parts (e.g., the can opener) without affecting the other parts (e.g., the corkscrew). But does this analogy provide insight into the functioning of the brain? We think not. Indeed, a growing body of evidence suggests that adult brain networks display large-scale patterns of interconnectivity that are inconsistent with the notion that there exist independently functioning modules per se (for more details, see Ref [19]). Because it is questionable that the neuropsychological framework is useful for explaining even adult neural and cognitive processes, it is unlikely to have any explanatory or predictive power when applied to NDDs in children.

When the adult neuropsychological approach is applied to children with NDDs, it is assumed that individual differences in cognitive ability are the result of a deficit in one or more innately specified modules, perhaps due to faulty genes. For example, a ‘specifically impaired’ phonological processor has been purported to be the root cause of dyslexia[20] and an impaired ‘theory of mind’ module has been suggested to be a major causal factor in autism[21] (see Wozniak, et al., The development of autism spectrum disorders: variability and causal complexity, WIREs Cogn Sci, also in the collection How We Develop).

The static neuropsychological approach contrasts with developmental approaches such as the developmental systems approach,[22] the dynamic systems approach,[23] and neuroconstructivism.[24] These developmental approaches share the view that development is a process of self-organization that results from interactions between multiple subsystems within a context. Intrinsic factors (e.g., physiological, psychological, neural) as well as extrinsic factors (e.g., informational cues, social context) constrain each other and shape the developmental process. Neuroconstructivism is the developmental approach that has most often been applied to explain NDDs, and thus will be the one we focus on hereinafter.

In contrast to the neuropsychological approach described above, proponents of the neuroconstructivist approach argue that adults with acquired brain lesions cannot be compared to children with NDDs. Although specific patterns of network activation in the brain are relatively stable (highly specialized) in adults, they do not necessarily start out that way. Patterns of activation become increasingly specialized over developmental time (see Figure 3) through interactions between various brain regions and through processing different types of input.[26-29] Thus, although any brain injury is likely to impair multiple cognitive functions, it is more likely to cause a relatively specific deficit in the highly specialized adult brain than in the less specialized (and more plastic) infant brain.[24, 29] Therefore, the brain of a child with an NDD cannot be described as composed of a set of damaged versus intact parts (see discussions in Refs [30-32]). Instead, it can be better characterized as an atypical system developing under different constraints.

Figure 3.

Functional magnetic resonance imaging (fMRI) data showing the neural response to faces in 10–12-year-olds and adults. Children show more distributed and bilateral activation than adults. The colored bar to the left represents the percent increase in intensity of activation in the experimental task, as compared to the control task. Note that according to the radiological convention the left side of the brain represents the right hemisphere (RH), and the right side of the brain represents the left hemisphere (LH). (Reprinted with permission from Ref [25]. Copyright 2003 John Wiley and Sons; modified)

One surprising consequence of the neuroconstructivist view is that even behavioral performance that falls within the normal range may be supported by atypical brain processes in children with NDDs.[33] For example, despite their low IQ, individuals with Williams syndrome fall in the normal range on tests measuring face processing.[34, 35] Yet the cognitive and neural processes that support this performance in the Williams syndrome children are very different from those used by typically developing children.[33] Furthermore, task performance by children with NDDs can sometimes surpass performance by typically developing children on the same task (e.g., Ref [36]). Children with autism spectrum disorder are actually faster and more accurate than typically developing children in finding hidden shapes embedded in larger meaningful pictures (e.g., finding a triangle within the drawing of a pram). Why? Because those with autism spectrum disorder tend to focus on local details and are less distracted than neurotypicals by the meaning of the whole picture.[36]

If the developing infant brain starts out highly interactive, then an initial impairment in one cognitive component is likely to have cascading effects on other parts of the developing system. That is, a basic-level deficit in the cognitive system will constrain the emergence of several higher-level cognitive functions, because these functions emerge from complex interactions in the brain.[30, 32] This early basic-level cognitive deficit may be underpinned initially by small variations in one or more factors including gene expression, neuronal growth and migration, synaptogenesis, and synaptic pruning[26, 31, 37] (see Darnell and Gilbert, Neuroembryology, WIREs Dev Biol, and Jernigan and Stiles, Construction of the human forebrain, WIREs Cogn Sci, also in the collection How We Develop). These initial small alterations interact with other genetic and environmental events that, over time, give rise to the resulting phenotype. Such complexities make it challenging to elucidate developmental causes in complex, dynamic, multilevel systems. To understand NDDs, it is thus critical to account for these dynamic, complex interactions between and within all levels of organization across time, from genes to environment.[38] Simple explanations may seem more tractable but they do not suffice; we must embrace complexity (see Appendix for an example of complex interactions).

As a result of cascading effects and multilevel interactions, children with NDDs are likely to develop atypical neural and cognitive trajectories with numerous widespread impairments,[30, 32] rather than a set of impaired and intact modules. A single deficit may have differential cascading effects on the system, with some functions being more affected than others. Thus, some atypicalities may be subtle and difficult to detect, using standardized measures designed for children with typically developing brains.[39] Because these subtle differences are hard to identify, they lead some theorists to claim that impaired modules co-exist with ‘intact’ modules in NDDs, even though fine-grained analyses have revealed this not to be the case.


One example that illustrates the advantages of the neuroconstructivist approach is that of Williams syndrome (Williams Syndrome Association:; Williams Syndrome Foundation:

Within the neuropsychological framework, Williams syndrome is often invoked as a model NDD. This is because of its known genetic etiology and its uneven cognitive profile, with seemingly intact and impaired components. Proponents of the neuropsychological approach characterize this syndrome as having specific impairments in spatial and numerical cognition modules alongside ‘intact’ language and face recognition modules.[40-47] Consequently, Williams syndrome is often compared with other NDDs in an attempt to delineate double dissociations not only in adults but also in children.

For example, a double dissociation has been repeatedly proposed between Williams syndrome and ‘specific language impairment’ (SLI), a developmental disorder that—as its name suggests—appears to target language development in a highly specific manner. Researchers point to this double dissociation as evidence that language is the product of a stand-alone cognitive module (e.g., Refs [44, 48, 49]). Accordingly, proponents of this double dissociation describe Williams syndrome as a disorder in which the ‘system for language is selectively spared’ (Ref [42], p. 193) and SLI as a disorder characterized by a single impairment in the language domain in an otherwise intact brain.[43, 44, 50-52] Furthermore, the proposed double dissociation is used to support the notion that ‘a grammar module’ develops normally in Williams syndrome alongside low IQ while being selectively impaired in SLI alongside normal IQ, suggesting that syntax is a cognitive module operating independently of nonverbal cognitive abilities in children (e.g., Refs [44, 48, 49]).

In fact, however, the purported double dissociation between Williams syndrome and SLI does not exist. Rather, as proposed by the neuroconstructivist approach, several aspects of the brain and cognitive system in both Williams syndrome and SLI develop atypically.[31, 53-55] For example, when the purported double dissociation was tested by directly comparing individuals with Williams syndrome and SLI on a battery of verbal and nonverbal tests, no double dissociation emerged. Although children with Williams syndrome did perform significantly worse on nonverbal tasks, their performance in the verbal domain did not differ from that of individuals with SLI.[56] Furthermore, despite the fact that language is a relative strength in individuals with Williams syndrome, it still falls well below the level of proficiency expected of a typically developing child of a similar chronological age. In other words, within Williams syndrome, language ability may be strong relative to other domains (e.g., visuospatial ability), but it cannot be considered to represent an intact language module because it develops atypically. Most aspects of language, including the lexicon,[57] morphosyntax,[58] and pragmatics,[59] are atypical in Williams syndrome when compared to typically developing children.

Even the claim that SLI entails a selective impairment of language alongside preserved nonverbal intelligence (e.g., Refs [60-62]) is controversial. Although individuals with SLI perform within the normal range on nonverbal intelligence tests, they have been shown to perform significantly worse than their siblings,[53] which points to subtle, more widespread impairments.[63] Moreover, SLI has been associated with numerous other (subtle) cognitive, sensory, and motor difficulties.[53, 64-70] Finally, the double dissociation is proposed purely on behavioral evidence whereas, as predicted by the neuroconstructivist approach, widespread brain atypicalities are found in both Williams syndrome and SLI (see review by Ref [55]; see also Ref [71]).


As mentioned earlier, the theoretical approaches that scientists adopt influence the type of research questions they ask, which in turn determine the studies they carry out and the way that they interpret their findings. These decisions have consequences for health and social policies as well as for education and intervention (see Appendix for an overview of how different theories affect assessment and intervention).

The neuropsychological approach generates research questions such as ‘By what age does each module come online?,’ ‘Which modules are impaired and which are intact?,’ and ‘Where are these modules located in the brain?’ Such questions necessarily lead to the search in children with NDDs for specifically impaired modules in an otherwise intact brain. This type of research tends to generate interventions that target one specific cognitive system in isolation of others (e.g., language in SLI; number in dyscalculia). However, as argued above, it cannot be assumed that the infant brain is modular. Brain circuits interact across different regions at every age, but especially in early development when they are more plastic and less specialized. Thus, the static research questions that are generated by the neuropsychological approach are now giving way to new, more dynamic questions such as ‘How do neural circuits and cognitive functions emerge and change over developmental time?,’ ‘Will an early, low-level deficit be followed by compensation or compounding of effects?,’ ‘Which domains interact across developmental time?,’ and ‘What aspects of the ever-changing environment affect gene expression and cognitive development?’

Neuroconstructivists are helping to change the way in which NDDs are investigated. Whereas neuropsychological approaches assume that basic-level deficits directly impair one or more ‘modules’ while sparing the rest of the system, neuroconstructivism suggests that basic-level deficits have cascading effects that alter interactions within and between networks. These cascades act to constrain the emergence of many higher-level (interactive) functions. In other words, a basic-level deficit that affects one functional domain (e.g., visual attention) may constrain the emergence of functions in other domains (e.g., language) because the developing system is highly interactive; it is not composed of isolated, minimally interactive brain parts. Thus, neuroconstructivism requires that researchers adopt a truly developmental approach, focused on change over time, to gain insight into interactions among the genetic, cellular, neural, cognitive, behavioral, and environmental levels of description. This is critical in order to discover how an initial perturbation in one domain, and at one level of analysis, may over time influence other domains at other levels of analysis.

The neuroconstructivist approach has also helped to engender a shift in research from small domain-focused research teams (focused, for example, on numerical cognition) to large collaborative networks of multidisciplinary teams that include geneticists, cellular and molecular biologists, neuroscientists, social scientists, and psychologists as well as computational modelers and others. To facilitate this collaboration, large cross-scientific consortia are being established (e.g., British Autism Study of Infant Siblings [BASIS,]; European Autism Interventions - A Multicentre Study for Developing New Medications [EU-AIMS,]; London Down Syndrome Consortium [LonDownS,]; and Pediatric Imaging, Neurocognition and Genetics [PING,]).

Furthermore, the neuroconstructivist view, like all developmental approaches, highlights the importance of tracking developmental trajectories,[72] beginning as early in development as possible, in order to unpack how interactions at multiple levels give rise to the resulting NDDs. For example, in the case of Williams syndrome, the genetic deletion likely affects basic-level processes that have cascading effects on various domains over developmental time.[31, 54] Being able to ascertain how tiny variations interact with factors early in development provides greater insight into the emerging phenotype. Examining developmental trajectories is also important for identifying possible protective factors (factors that improve outcome) and risk factors (factors that worsen outcome). Uncovering the cascading effects of a basic-level perturbation, as well as any protective and risk factors, are critical for the design of early diagnostic tests and intervention strategies.

To be able to tailor early diagnostic tests and interventions, it is necessary to compare children with NDDs not only with typically developing children, but also with other NDDs, conducting cross-syndrome comparisons.[7, 73-77] This is especially important because initial perturbations may be very similar across syndromes, yet tiny subsequent differences may ultimately yield very different developmental outcomes.


The authors of this paper are funded by Wellcome Trust Strategic Grant No. 098330/Z/12/Z conferred upon The LonDownS Consortium UK.



Williams syndrome (WS) is caused by the deletion of some 28 genes on one copy of chromosome 7 (7q11.23), 22 of which are expressed in the brain.[78] This initial deletion at the genetic level is linked to a number of atypicalities at the level of the brain (brain chemistry, brain anatomy, brain volume, hemispheric asymmetry, temporal patterns of brain activity, etc.[31]), and these are associated with various cognitive and behavioral abnormalities.[54] However, although the genetic deletion is the root cause of WS, this simple fact belies the complexity of how WS emerges over developmental time.

For instance, alterations in many brain areas have been reported in WS. Changes in one of them—the hippocampus—have been linked to deficits in spatial navigation and long-term memory.[79] While hippocampal changes may be in part due to genetic mutations, it is important to highlight the fact that environmental factors such as chronic stress and anxiety also affect neural development.[78, 80-82] It is known that stress and anxiety are especially high in WS.[83] Therefore, it is not necessarily the case that it is the genetic mutation that directly affects brain development in WS; it could also be a combination of emotional perturbations that also impact the brain. Thus, it is impossible to understand how the WS neuro-cognitive profile emerges without studying it very early in development, across time, and at multiple interacting levels.


Neuropsychological Approach

  • Assessments focus on the identification of impaired and intact parts of cognition and brain.
  • Once the ‘broken’ module has been identified, the intervention targets this module specifically (e.g., practicing language in the case of specific language impairment or number in the case of Williams syndrome).
  • If the module is identified as being underpinned by specific genes, then gene therapy may be considered the only solution.

Neuroconstructivist Approach

  • Assessments are broad, mapping the profile of the child's strengths and weaknesses over time.
  • No specific isolated impairment is assumed. Rather, many domains will be more or less affected.
  • The ideal intervention is based on prior in-depth syndrome-specific research and usually takes place early in the developmental trajectory, enhancing protective factors and/or reducing risk factors in the child's development.
  • The targeted intervention need not be in the most visibly problematic domain, like language or number. Rather, it can be targeted at basic-level processes that underlie the domain, such as memory, attention, etc. It is these processes that would be likely to have the most cascading impact on other domains.
  • Timely intervention is key. Effective intervention ultimately depends on the timing of various interactions among domains, which is why it is crucial to base the intervention on syndrome-specific research that identifies how systems change over developmental time in a given neurodevelopmental disorder.
  • Although the earliest intervention will not necessarily be the most successful one, younger brains are typically more plastic than older ones. Also, early in development, even relatively small perturbations can have cascading effects over developmental time.