• Elia Benhamou

Embracing complexity in neurodevelopment - Part 1


This blog is the first of a series. We are embarking on a new project and we want to take you on the journey with us. The purpose of our project is try and study neurodevelopmental diversity, by taking what we call a ‘transdiagnostic’ approach. What do we mean by that? And what practical differences might it mean for the way we study child development?


Why do we love categories?

Humans don’t like grey areas. Our brain has evolved to classify objects based on their ‘core’ features: how they sound, look, and feel. To do this the brain must create boundaries – information within a boundary belongs to the same object, and information outside the boundary is not part of that object. A fascinating review written by Mark Johnson in 2011 described the development of the brain during the first years of our life as a neural specialisation or tuning of brain activity. For instance, a brain region originally activated by a wide variety of visual objects will progressively only activate in response to upright human faces.


But categories don’t stop at objects – we seem to want to categorise people too. Categorising things can reduce our perception of the unknown, and help us generalise within a category. If we label something we feel we understand it, and can predict what it might do next. Categorising can be efficient and help us know how to interact with something. But categorising people may be an altogether different matter with its own scientific and ethical baggage. Clinically-trained specialists learn to categorise. They look for specific features that correspond to specific diagnoses. The diagnoses provided by clinical or educational specialists are a crucial step on the pathway to effective support for those with neurodevelopmental conditions. A family with a child experiencing difficulties at school might be referred to a specialist who will decide, based on results from a battery of tests, if the child fulfils the criteria for a specific disorder such as ADHD, autism or a sensory processing disorder. A diagnosis is a landmark moment for a family and is needed to start leveraging support from schools and other sources of support. The video below is a vivid demonstration of the utility of diagnosis for informing teachers and parents of what to expect in terms of behaviour and how best to manage it.



In which case categories are not so useful anymore?

However, these diagnoses do not work like others you might be more familiar with, such as for physical conditions. Neurodevelopmental conditions cannot be identified through any scans or blood tests. Instead, they are made on the basis of observations of behaviour.

The most common framework used by practitioners is the DSM-5, a huge manual which describes behavioural features that tend to cluster together, crystalising these feature-clusters into diagnostic categories.


This framework has many limitations, which have been widely acknowledged (Cuthbert and Insel, 2013; Newby et al., 2015; Astle and Fletcher-Watson, 2020; Astle et al., 2021). For instance, it fails to deal with multiple surface-level behaviours or multiple diagnoses at once.

Co-occurring features are the norm rather than the exception. For example, 22 to 83% of people diagnosed with autism (ASD) meet DSM criteria for ADHD and 30 to 65% of people diagnosed with ADHD present significant features of ASD (Leitner, 2014). Around 50% of children with dyslexia also meet criteria for DLD (Developmental Language Disorder) and vice versa (McArthur et al., 2000).
Co-occurring features are the norm rather than the exception. For example, 22 to 83% of people diagnosed with autism (ASD) meet DSM criteria for ADHD and 30 to 65% of people diagnosed with ADHD present significant features of ASD (Leitner, 2014). Around 50% of children with dyslexia also meet criteria for DLD (Developmental Language Disorder) and vice versa (McArthur et al., 2000).

Yet, co-occurring difficulties are the norm rather than the exception: they are present in 44% of children with ADHD and in 65-85% of autistic children (knowing that ADHD and autism have co-occurrence rates ranging from 30-70%). This has serious implications for children whose needs are not met and for their parents who often feel confused and helpless when their children end up with either no diagnosis despite real struggles in daily-life or with different diagnoses depending on the specialist they saw (psychiatrist, educational psychologist, speech and language therapist, neurologist, paediatrician).

What about finding the underlying mechanism?

The purpose of the framework isn’t to help researchers. In fact, it turns out that supposed categories haven’t been particularly helpful for researchers, because they often don’t line up with underlying processes and mechanisms. The truth is, there is no simple one-to-one matching between identified mechanisms and surface level features characterising a neurodevelopment condition. In fact, we now know that an emerging developmental profile can occur via multiple underlying routes (often referred to as equifinality) or, on the contrary, the same underlying route can lead to very different profiles across children (sometimes referred to as multifinality). From a researcher’s perspective, this makes neurodevelopmental conditions very hard to study.



There is no one-to-one mapping between biological routes and diagnosis. Biological routes include expression of genes and environmental factors which affect gene expression. Genes determine how the brain is structured (volume of brain regions, white fibers between brain regions to form networks) and how it functions (patterns of neurons firing activity within and between brain regions). The brain’s activity translates into actions and thoughts. Certain brain areas/networks have been linked to specific cognitive functions and to specific behaviours. These observable signs are used by health practitioners to make a diagnosis.

However, the research field is still dominated by the idea that each condition – or each individual with that diagnosis – is caused by the same underlying mechanism. This idea is often referred to as the ‘core-deficit’ framework. For instance, in the case of autism, the dominant core-deficit theory for a long time has been the ‘theory-of-mind model’ which posits that autistic people uniquely lack the ability to detect and interpret the mental states of others. However, the latest findings show that theory-of-mind difficulties are not unique to autistic people and emphasise instead problems with mutual-understanding between autistic and non-autistic people (Edey et al., 2016; Crompton et al., 2021).


Shifting from pathology to neurodiversity: the need for individual stories

Our lab is not the first to recognise the limitation of the ‘traditional’ diagnostic paradigm (Bishop & Snowling, 2004; Cuthbert & Insel, 2013; Etkin & Cuthbert, 2014; Dalgleish et al., 2020). In fact, there has been a shift in contemporary thinking in the last decade or more. One way of changing the research paradigm is instead to think in terms of ‘neurodiversity’, rather than discrete categories like ADHD. This shift has the potential to be more inclusive. Rather than needing a categorical diagnosis, this reframing includes people whose experience of the world is neurodivergent, i.e. who are ostracised or disadvantaged because of their differences. Some advocates of the neurodiversity framework thus affirm that ‘there is no such thing as a normal brain, that variation is natural and none is more right or wrong than another’ (see Jess Meadows’ latest blog). This view pushes back on a status quo that can pathologise those who are seen as different.


As Akiko Hart writes, there is not a single story, “there are in fact as many stories as there are people. The single story doesn’t honour this multiplicity and this complexity.” Neurodivergent people might experience their difficulties as a disability, others as an illness, a sickness of the brain and finally others who see it as a difference, not necessarily debilitating but actually a big part of “who they are and how they are in the world”. As Freddy’s mum rightly says in the video below: all children are different. Even when diagnosed, they will always display behaviours and have experiences in their lives that could not have been anticipated by the ‘textbook cases’ studied by specialists. Children experience their diagnosis differently and a one-size-fits-all approach would prevent them from discovering their unique strengths and having their unique needs addressed.



But how do we translate this perspective into research?

Well, first, we need to find the right question. As researchers, we must always ask ourselves, what are we really trying to do? For instance, do we want to find the cause of a behavioural feature in order to prevent it, or do we want to better understand how this feature ‘expresses’ itself in each individual depending on their brain structure, their genetic dispositions and their environmental factors (e.g. socio-economic status or parental education) with the ultimate goal of tailoring interventions that are adapted to each individual? The way transdiagnostic researchers choose a research question and try to move away from the traditional practice (i.e. looking at a single disorder, at one or two key mechanisms or focus on a single level of analysis) will be covered in our next blog.

A transdiagnostic approach should also lead to a change in recruitment. Typically, recruitment was based on ‘formal’ diagnosis so that theories about a particular diagnosis could be tested and challenged in comparison to other diagnostic groups or a control group. This type of recruitment is highly biased and automatically excludes children with no formal diagnosis but real learning and/or behavioural difficulties. Functional recruitment with community screening or large-scale population-representative cohorts are two ways to yield transdiagnostic insights. This topic will be covered in the third blog of our series.

If researchers aim to abandon diagnostic grouping, they must also devise new methods of analysis. For instance, we might want to consider behavioural or cognitive features as continuous dimensions (as opposed to discrete categories) and use clustering techniques which use the data themselves to find new groupings in this multidimensional space. These methods will be explained in more details in the fourth and fifth blogs of our series, accompanied with vivid examples of the type of insights these studies led to.

Ultimately, what we hope to achieve with this series is to help you understand the conceptual change that is occurring in multiple institutions supporting communities of people with learning difficulties, including the academia. The belief that traditional diagnostic groups are discrete and associated to pure underlying causal mechanisms slowly erodes. Reality is extremely complex and tackling the topic of development involves taking into account a tremendous number of factors. Not a single but numerous mechanisms are constantly at play and leads to a great variability in multiple dimensions. We ultimately want to understand how this variability turns into the observable outcome that we are interested in such as a barrier to learning or a daily-life difficulty.


In this blog series, we will describe how the interaction of concepts taken from psychology, brain imaging, computer science or genetics have helped to reshape the field of developmental science and sharpen up its functions, i.e. addressing individual needs by finally embracing the complex abundance of phenomena that characterise the daily experience of an individual child.


Stay tuned! Our next article will describe how a transdiagnostic researcher finds out the elements which make a research question good and pertinent, or conversely, not so useful.

168 views0 comments