Professor Uta Frith has recently argued that the modern autism spectrum has become so broad that the diagnosis risks becoming scientifically meaningless. She questions whether some people diagnosed in adulthood are actually autistic and argues that there is now an enormous difference between people with severe developmental disabilities diagnosed in childhood and articulate, independent adults receiving diagnoses much later in life. (The Times)
There is a legitimate question at the heart of Frith’s argument.
Autism is extraordinarily heterogeneous. A non-speaking autistic adult with severe intellectual disability who requires continuous care has very different needs from an autistic adult of above-average intelligence who lives independently.
It is reasonable to ask whether a single diagnosis adequately describes both people. It is also important that those with the greatest support needs are not forgotten as public understanding of autism broadens.
Where I disagree with Frith is in what appears to follow from that observation.
What are we actually calling autism?
The first problem is the tendency to confuse autism with conditions that frequently accompany it.
Autism is not the same thing as intellectual disability.
Autism is not the same thing as language disorder.
And neither intellectual disability nor language delay is necessary for someone to be autistic.
They occur together disproportionately because neurodevelopmental conditions frequently overlap, but correlation does not make them the same condition.
A person can have:
autism without intellectual or language impairment;
autism with intellectual disability;
autism with language impairment;
or autism with both.
Likewise, someone can have a severe intellectual disability or language disorder without being autistic.
This matters because otherwise we risk using a person’s overall level of disability as a measurement of how autistic they are.
Those aren’t necessarily the same thing.
The “profound autism” debate illustrates the problem
The Lancet Commission proposed “profound autism” as an administrative term for autistic people likely to require very high levels of lifelong support. Its definition relies heavily on intellectual disability, very limited language and inability to manage basic daily living independently. (Wiley Online Library)
The motivation is understandable. Someone requiring continuous lifelong care has needs that the word “autism” alone doesn’t adequately communicate.
But critics have identified an important conceptual problem: the proposed category is largely defined by intellectual and language disability rather than by autistic characteristics themselves. (PubMed Central (PMC))
Consider two autistic people with similarly pronounced autistic sensory, social and repetitive characteristics.
One has an IQ of 120 and fluent language.
The other has an IQ of 45, severe language impairment and epilepsy.
The second person is undoubtedly more disabled overall.
But are they necessarily more autistic?
Those are different questions.
It may therefore be more informative to describe the second person as having autism plus severe intellectual disability plus severe communication impairment plus epilepsy and very high support needs.
That tells us what that person actually needs.
Calling all of those additional difficulties “profound autism” risks attributing every disability the person experiences to autism itself.
The historical evidence also creates problems for a narrow definition
If articulate and intellectually able autistic people are principally a product of modern diagnostic expansion, we should expect them to be absent from the earliest descriptions of autism.
They aren’t.
Donald Triplett was Leo Kanner’s Case 1 — the first person in Kanner’s foundational 1943 autism series.
Yet Donald did not remain profoundly intellectually or linguistically disabled.
He went to college.
He drove.
He travelled.
He worked successfully in a bank.
He participated in his local community.
Kanner subsequently followed his progress into adulthood. Donald’s development and eventual competence did not cause Kanner to decide that his original diagnosis had been mistaken. (The Washington Post)
This matters enormously.
The first recognised autistic person himself demonstrates that autism does not inevitably mean profound lifelong intellectual, linguistic or adaptive disability.
Asperger’s children present the same historical problem
Hans Asperger’s descriptions included intellectually able children with normal or even precocious language — the children subsequently associated with the “little professor” description.
Their difficulties concerned social reciprocity, understanding other people, unusual communication, restricted interests and other characteristics rather than global intellectual or language impairment.
Historical reviews describe this population as having normal or precocious language acquisition, sometimes above-average linguistic ability, alongside marked abnormalities of social and non-verbal communication. (Frontiers)
Ironically, Frith herself played an important role in making Asperger’s work accessible to English-speaking autism researchers.
So:
autistic + intelligent + articulate
is not a twenty-first-century invention.
It has existed within the clinical literature almost from the beginning.
Language delay is particularly problematic as a boundary
Language delay can accompany autism, but childhood language development is not necessarily predictive of adult communication ability.
Children continue developing.
An autistic child with substantial language delay at three may eventually become a fluent adult.
Research comparing intellectually able autistic children with and without early speech delay has even found that early speech delay did not necessarily predict meaningful differences in later autistic characteristics or expressive language. (PubMed)
Conversely, some autistic children acquire vocabulary and grammar normally or precociously.
Therefore:
language delay ≠ autism
and:
fluent language ≠ absence of autism.
Speech, language and autistic social communication are different things
This becomes particularly important when Frith reportedly describes Chris Packham’s excellent communication as evidence against autism. (The Times)
Someone can have exceptional vocabulary, grammar and factual communication while still experiencing significant autistic social-communication difficulties.
There is a difference between being able to deliver an eloquent television presentation and intuitively processing:
facial expression;
body language;
conversational turn-taking;
implicit meaning;
social context;
other people’s intentions;
and the unwritten rules governing ordinary interaction.
A highly articulate person can still struggle profoundly with some of these things.
Good expressive language cannot therefore be treated as evidence of typical social processing.
Autism is neurodevelopmental — and development doesn’t stop
This is perhaps the most important problem with comparing autistic children with autistic adults.
Calling autism a neurodevelopmental condition doesn’t mean that autistic development freezes when autism first becomes observable.
An autistic four-year-old continues developing into an autistic 10-year-old, an autistic 20-year-old and eventually an autistic 60-year-old.
During those decades the person accumulates:
language;
education;
experience;
knowledge;
coping strategies;
social rules;
self-awareness;
and thousands upon thousands of social interactions.
We should therefore expect the outward expression of autism to change considerably with age.
Developmental continuity does not require behavioural immutability.
Intelligence changes the presentation even further
Intelligence can give an autistic person powerful compensatory tools.
An intelligent autistic child who doesn’t intuitively understand a particular social convention can potentially learn it explicitly.
Something that another child acquires automatically may become a rule:
When somebody tells me about something that happened to them, I should ask a follow-up question rather than immediately tell them about my similar experience.
After decades of practice, that behaviour may become extremely polished.
An observer sees successful reciprocal conversation.
What the observer cannot see is the cognitive process producing it.
One person may be doing it intuitively.
The autistic person may be consciously monitoring and calculating the interaction.
Successful behaviour therefore does not necessarily demonstrate typical underlying processing.
But intelligence cannot compensate for everything
This is particularly important to my own experience as a late-diagnosed autistic adult.
I am intelligent.
Over a lifetime, I have used both intelligence and experience to understand people better. I have learned why people behave in certain ways. I have learned social expectations and consciously changed aspects of my behaviour.
That undoubtedly changes how autistic I appear to other people.
But it hasn’t made me non-autistic.
I still struggle with conversational turn-taking.
Socialising remains stressful.
I cannot simply use intelligence to make myself automatically notice and interpret body language.
I struggle to recognise people’s faces.
There are things I can learn around, things I can partially compensate for, and things that intellectual reasoning simply cannot replace.
Knowing that information exists doesn’t necessarily give my brain automatic access to it.
That distinction is crucial.
Learned competence is not necessarily typical processing
Imagine two people conducting apparently identical successful conversations.
Person A automatically processes facial expression, tone, body language, conversational timing and social context.
Person B has spent decades learning explicit rules and is consciously monitoring some of those variables while talking.
The observable outcome may be virtually identical.
But the cognitive process is not.
Person B may also finish the interaction considerably more exhausted.
If we judge autism purely by observable adult performance, something paradoxical happens:
successful compensation for autism becomes evidence against having autism.
That is circular reasoning.
Intelligence might actually contribute to late diagnosis
This also offers an alternative explanation for some adult diagnoses.
Frith appears concerned that someone who reaches middle age before receiving an autism diagnosis may belong to a fundamentally different population.
But consider another possibility:
higher intelligence
↓
greater ability to analyse difficulties and develop compensatory strategies
↓
less obvious external presentation
↓
reduced likelihood of childhood recognition
↓
late diagnosis
Under this model, late diagnosis isn’t evidence against autism.
For some people it may partly be a consequence of the very cognitive abilities Frith interprets as making autism less plausible.
Age of diagnosis is not age of onset
An autism diagnosis at 50 does not mean:
“This person became autistic at 50.”
A proper adult assessment asks whether the developmental pattern extends backwards into childhood.
Therefore the important question isn’t:
“Does this 50-year-old behave like the autistic children Frith studied?”
Of course they don’t.
They have had another 40 years of neurological development, learning and experience.
The appropriate question is:
“Does this person’s developmental history and current cognitive profile make sense as the adult outcome of an autistic neurodevelopmental trajectory?”
That is fundamentally different.
Historical autism research also had an ascertainment problem
There is another reason earlier autistic populations appeared more profoundly disabled.
Which children were most likely to reach specialist psychiatric services in the mid-twentieth century?
A non-speaking child with intellectual disability, conspicuous repetitive behaviour and an inability to cope in ordinary school was very likely to attract clinical attention.
An intelligent, articulate child who struggled socially, intensely pursued unusual interests, found change difficult and experienced sensory problems but nevertheless achieved academically was much easier to overlook.
That creates a potentially self-reinforcing cycle:
The most visibly disabled children are referred.
↓
Researchers study those children.
↓
Those children define autism.
↓
Less visibly disabled children don’t resemble the established definition.
↓
They aren’t referred or diagnosed.
↓
Researchers continue seeing predominantly highly disabled autistic children.
↓
High global disability appears intrinsic to autism.
That is an ascertainment problem.
Expanding recognition can therefore increase prevalence without necessarily creating a new condition.
Frith nevertheless has an important argument
The strongest response to everything above would be:
“If these people’s presentations are so different, how do we know that they actually have the same underlying condition?”
That is a legitimate scientific question.
Perhaps we don’t.
Autism may eventually prove not to be one biologically homogeneous condition.
Future genetics and neuroscience may divide what we currently call autism into several different developmental pathways or conditions.
Frith could ultimately be right that the spectrum needs breaking apart.
But there is a missing step in her argument.
Nothing currently demonstrates that the biologically meaningful boundary will be:
speaking vs non-speaking;
high IQ vs intellectual disability;
diagnosed at 4 vs diagnosed at 54;
or:
independent vs requiring lifelong care.
Those characteristics may turn out to cut across entirely different biological autism subtypes.
Heterogeneity does not establish misdiagnosis
This is therefore the central logical problem.
Frith begins with an observation I accept:
Modern autism encompasses an extraordinarily heterogeneous population.
But the stronger conclusion appears to be:
Therefore some articulate, intellectually able or late-diagnosed people probably aren’t autistic.
The first proposition does not establish the second.
Something is missing:
evidence showing where the true boundary of autism lies.
Until we have that evidence, intellectual ability, language ability and overall disability cannot simply substitute for it.
There is a better way of describing heterogeneity
Rather than imagining autism as a single line:
mild autism ————— profound autism
we could describe several dimensions separately:
AUTISTIC PHENOTYPE
social communication, repetitive behaviour, sensory characteristics, need for sameness, focused interests, etc.
INTELLECTUAL ABILITY
LANGUAGE ABILITY
ADAPTIVE FUNCTIONING
CO-OCCURRING CONDITIONS
ADHD, epilepsy, genetic conditions, motor disorders, anxiety and others.
ENVIRONMENT
INDIVIDUAL SUPPORT NEEDS
This acknowledges the enormous difference between two autistic people’s lives without assuming that the person with the greatest overall disability necessarily has the greatest quantity of “autism.”
Where I agree with Frith
The autism spectrum is extraordinarily broad.
People requiring lifelong 24-hour support deserve dedicated research, resources and representation.
Current autism terminology often tells us very little about an individual’s actual abilities and support needs.
Autism may eventually need scientifically meaningful subdivision.
And researchers should absolutely be permitted to question whether current diagnostic boundaries are correct.
Where I disagree
Intellectual disability is not autism.
Language impairment is not autism.
Neither is necessary for autism.
Adult competence doesn’t tell us what someone’s childhood development looked like.
Childhood language delay doesn’t determine adult communication ability.
Intelligence can allow an autistic person to compensate for some difficulties without eliminating the underlying neurodevelopmental differences.
Some autistic difficulties cannot simply be overcome through intelligence or experience.
Early autism history itself includes intellectually able and verbally competent autistic people.
And late recognition doesn’t establish late onset.
Most importantly:
difference in overall disability does not prove difference in diagnostic authenticity.
The question I would put to Professor Frith
If we remove from the discussion:
intellectual disability,
language impairment,
adaptive disability
and
overall support needs,
what does Uta Frith believe autism itself actually is?
Because until we answer that question, we cannot logically decide that the person who speaks less, learns more slowly and requires more assistance is somehow more authentically autistic than the intelligent adult who has spent a lifetime learning how to navigate an autistic brain.
The profoundly disabled autistic person deserves recognition of all of their disabilities and the support those disabilities require.
The intellectually able autistic person deserves recognition that intelligence and accumulated experience can alter the appearance of autism without eliminating it.
Neither requires us to deny the autism of the other.
And perhaps the most important principle is the simplest:
Autism is neurodevelopmental. Development still happens.
An autistic child should grow into an adult who looks different from the child they once were.
If they learn, compensate, mature and adapt, that isn’t evidence against a developmental condition.
It is evidence of development.
Leave a comment