public choice — Arts — Then Do Better

Introduction: I had some musings (with LLMs) on the issues/strengths of medical and social models of disability, in particular autism. And thinking about, my own “model” which I called bureaucratic model, but might be better called administrative… and then synthesising this into interactive models which are discussed briefly… and some links end for further reading.

My question: For understanding autism, what does each model explain well, what does it obscure, and what framework best guides how we treat autistic people and design support?

My answer is that both the medical and social models capture something important, and either becomes misleading when treated as the whole story. And to ignore the administrative state is to ignore a whole other layer in terms of what is happening.

For autism in particular, one could use something close to an interactional model: disability arises from the interaction between a person's underlying characteristics, the demands of their environment and the support available to them.

This is also broadly where the World Health Organization's International Classification of Functioning has ended up. The WHO explicitly treats disability as arising from interactions between a health condition and personal and environmental factors rather than locating it entirely either within the individual or within society.

I would then add two further perspectives.

Neurodiversity asks a normative question: even where someone is different, should that difference actually be regarded as something that needs correcting?

And what I will call the administrative or bureaucratic model asks a different question again: how does a continuous, complicated set of human needs become legible to a government that has to decide who receives resources?

Together these give a more complete picture of autism than any one model alone.

2. The medical model

In a simple form:

A person has an impairment or disorder that produces limitations. The appropriate response is diagnosis, treatment, rehabilitation, education, care or some other intervention directed substantially at the individual.

Applied to autism, the causal story might look like:

neurodevelopmental differences → difficulties with communication, sensory processing, executive function, adaptive functioning, etc. → disability

Consider an autistic person who cannot reliably communicate pain, experiences severe sensory distress, cannot safely navigate roads alone, has major sleep problems, cannot manage money or food preparation independently, or requires substantial help with everyday living.

Those difficulties cannot plausibly all be explained away as discrimination or poor social design.

The medical model therefore has important strengths. It takes impairment seriously. It allows us to investigate causes and mechanisms. It legitimises treatment of problems that cause suffering. And it recognises the enormous heterogeneity contained within the autism diagnosis.

This is particularly important for autistic people with intellectual disability, epilepsy, severe communication difficulties or very high support needs.

A theory of autism that leaves little conceptual room for intrinsic impairment risks marginalising precisely the people who need the greatest amount of support.

But the medical model can also (and does also) go badly wrong.

It can turn difference into pathology merely because it departs from the statistical norm. It can make normalisation the objective rather than wellbeing. And it can locate every problem inside the autistic person even when the environment is generating much of the difficulty.

A child who cannot tolerate a noisy, chaotic classroom might be described as having poor emotional regulation.

But perhaps the classroom is an extraordinarily hostile sensory environment for that child.

Teaching the child to tolerate it better is one possible intervention. Changing the classroom is another.

And the fact that a behaviour is recognisably autistic does not itself tell us whether changing it would improve the person's life.

I tend to ask the question, can we bring the world closer to the person as an intervention? And also, how much of the challenge is situations in the world / society / environment rather than the person?

The classic social-model distinction is between impairment and disability.

Very roughly:

People have bodily or cognitive differences. Society disables them when environments, institutions and expectations are designed around people without those differences.

The influential British social-model tradition emerged from disability activism in the 1970s, particularly the work of the Union of the Physically Impaired Against Segregation.

Applied to autism:

autistic characteristics + poorly matched social environment → disability

This produces a very different set of questions.

Instead of asking:

“How do we make this autistic child behave more normally?”

you might ask:

“What is it about this school, communication system or expectation that makes functioning unnecessarily difficult?”

That shift is genuinely powerful.

An autistic person may struggle enormously with an unstructured verbal meeting but function exceptionally well when information is written down beforehand.

Someone may be unable to work effectively in an open-plan office but perform extremely well remotely.

Someone who communicates unusually may have considerably more understanding than observers infer from their speech.

The social model consequently highlights reasonable adjustments, accessible environments, alternative communication, predictable routines, sensory accommodations, anti-discrimination and institutional redesign.

It also guards against an important conceptual mistake:

unusual ≠ undesirable ≠ pathological

Some (arguably many? can we argue all? ) autistic characteristics are simply differences.

Intense interests, preference for routine, unusual communication styles or reduced interest in conventional social rituals need not themselves be things that ought to be “fixed”.

4. Where the social model runs into trouble with autism

The criticism:

Some disability is intrinsic to impairment, rather than wholly socially produced.

Imagine the most autism-friendly society realistically possible.

There would still be autistic people who could not independently cross a road safely, communicate urgent medical needs, regulate overwhelming sensory experiences, manage money, prepare food or live without substantial assistance.

Changing society could reduce their disability enormously.

It would not abolish it.

This matters because an excessively strong version of the social model can inadvertently erase people with the greatest support needs.

Tom Shakespeare's influential criticism of the strong social model makes a related point: the sharp conceptual separation between bodily impairment and socially produced disability becomes difficult to maintain once we recognise that impairment itself can restrict people's lives. The social model remains an extremely useful corrective to medicalisation without necessarily being a complete causal account of disability.

Autism also creates a second problem for a purely environmental explanation: people's accommodation needs can conflict.

One autistic person requires silence; another constantly seeks sensory stimulation.

One needs rigid predictability; another finds routines oppressive.

One communicates best through writing; another cannot read.

There therefore isn't necessarily an “autism-friendly environment” analogous to installing a wheelchair ramp.

Autism involves multiple dimensions, and accommodations have to be individualised.

Where the medical model runs into trouble

The corresponding mistake is to treat observed difficulty as though it directly measures intrinsic impairment.

Conceptually:

Observed disability = individual characteristics × environmental demands

rather than simply:

Observed disability = severity of autism

The multiplication sign is metaphorical, but the interaction is the important point.

Suppose an autistic teenager functions badly in school but substantially better at home with one-to-one teaching, written instructions, predictable routines and breaks.

A crude medical interpretation might be:

severe autism causes educational dysfunction

A crude social interpretation might be:

conventional schooling creates the disability

The evidence may support something closer to:

There is a genuine neurodevelopmental difference whose disabling consequences vary dramatically depending upon the environment.

That is a much more useful causal model.

It also explains why the same person can appear profoundly impaired in one setting and strikingly capable in another.

This distinction between underlying capacity and realised performance is explicitly present in the WHO's ICF framework: what someone can do and what they actually do in their current environment are related but distinct outcomes.

6. Neurodiversity adds something different

I would distinguish the social model from neurodiversity, although they overlap.

Neurodiversity introduces an additional normative proposition:

variation in human cognition is not automatically pathology.

That seems broadly right.

But there are stronger and weaker versions.

The weaker claim seems strong to me:

Autistic traits should not be treated as deficits merely because they are atypical.

The stronger claim:

Autism is simply a difference rather than a disability or disorder

is much harder to sustain across the entire autistic population but could have a claim amongst some.

Autism is unusually heterogeneous.

Someone living independently, working successfully and mainly encountering difficulty because social conventions favour neurotypical communication has a very different relationship with autism from someone requiring lifelong continuous assistance.

Trying to force both experiences into either “disease” or “difference” throws information away.

Robert Chapman's work provides one of the more sophisticated philosophical versions of the neurodiversity argument. Rather than simply denying impairment, he argues for an ecological conception of mental functioning in which individual abilities have to be understood partly in relation to the wider social environment in which they operate.

A useful recent overview by Steven Kapp similarly argues that the neurodiversity movement and the medical model need not be simple opposites. In practice, neurodiversity proponents can support individual interventions that improve quality of life while rejecting normalisation as an end in itself.

That strikes me as a much stronger position than either “everything autistic should be treated” or “nothing autistic should be treated”.

7. A better causal model

Perhaps I would suggest an interaction model.

Intrinsic characteristics

Neurodevelopment, cognition, sensory processing, language, executive function, motor abilities, emotional regulation, intellectual ability and co-occurring conditions.

↓

Environment

Noise, complexity, predictability, communication style, social expectations, educational structure, employment practices and physical environment.

↓

Person-environment interaction

The demands placed upon this particular person relative to their capacities.

↓

Assistance and adaptation

Human support, assistive technology, education, treatment, environmental adjustments and social care.

↓

Outcomes

Independence, distress, relationships, learning, employment, safety, wellbeing and flourishing.

This gives us a much better practical question than:

“Is autism a disability or merely a difference?”

Ask instead:

“What is preventing this particular person from flourishing, and which lever is best changed?”

Sometimes the answer is the environment.

Sometimes it is skills or treatment.

Sometimes it is ongoing human support.

Sometimes there is nothing that needs changing at all.

Usually it is some combination.

8. An economic way of seeing the problem

There is also a useful resource-allocation perspective.

Suppose someone has difficulty doing X.

There are broadly three margins of adjustment:

Person → Environment → Assistance

You can invest resources in:

changing the person's capabilities;
changing the environment;
providing another person or technology to bridge the gap.

The best combination depends on the particular difficulty.

If £500 of assistive technology removes a communication barrier, insisting on years of behavioural training could be both expensive and unnecessarily burdensome.

Conversely, if learning a particular skill substantially increases someone's future autonomy and can be learnt without excessive distress, saying “society should simply accommodate this” could deprive them of something valuable.

And sometimes neither adaptation is sufficient.

Society simply needs to provide care.

That last category gets insufficient attention in some versions of the social model.

Dependence on other humans is not necessarily a policy failure. (see my podcast conversation with disability activist David Ruebain on interdependence here)

Some people genuinely need substantial lifelong assistance.

The relevant comparison is not independence versus dependence as moral categories. It is which arrangement produces the best achievable life given the person's characteristics, preferences and resources.

9. The administrative or “bureaucratic” model

There is then another layer, particularly important in welfare states such as the UK.

Disability is not only a biological condition or a person-environment interaction.

It also becomes an administrative status.

Governments have finite budgets for education, social care, disability benefits, healthcare, supported housing and specialist services.

Human need, however, is continuous and multidimensional.

Government decisions often cannot be.

At some point an institution must decide:

eligible / not eligible

benefit / no benefit

additional educational provision / ordinary provision

funding / no funding

So the administrative causal chain looks something like:

underlying characteristics → functional difficulty → evidence and classification → administrative eligibility → resources

This is not primarily a theory of what disability is.

It is a theory of how governments make disability legible enough to allocate resources.

10. Why labels become valuable

A diagnosis therefore acquires an administrative function as well as a clinical one.

Government cannot directly observe someone's sensory overload, executive dysfunction, communication difficulties or need for supervision.

It has a problem of asymmetric information.

The person and family know far more about the difficulties than the state does.

But a state allocating scarce resources cannot simply accept every claim without evidence.

It therefore uses proxies:

diagnoses;
professional assessments;
functional tests;
reports;
administrative descriptors;
eligibility thresholds.

Diagnosis can therefore act partly as a signal.

It provides a shared vocabulary and external validation that the difficulty is persistent and recognisable.

Economically, one way of interpreting this is that diagnosis can reduce the transaction costs of demonstrating need.

That does not mean diagnosis is necessarily a formal prerequisite for support (at least in theory….)

But:

“A recognised diagnosis may make it easier to demonstrate and legitimise need within bureaucratic systems”

And many people report that without a diagnosis it is very hard to get support…

11. Bureaucracy changes behaviour

Once categories determine access to resources, people and institutions inevitably respond to those categories.

This generates several predictable effects.

Threshold effects. Two people with almost identical underlying needs may receive substantially different resources because one falls just above an eligibility boundary and the other just below.

Incentives to seek diagnosis. If a diagnosis helps obtain appropriate educational provision, accommodations or benefits, the private value of obtaining a diagnosis rises. This does not imply the diagnosis is false; it changes the incentive to identify and document an existing condition.

Professional gatekeeping. Clinicians, educational psychologists and other professionals become important partly because their assessments convert private experiences into evidence that institutions recognise.

Strategic presentation. Systems organised around demonstrating impairment naturally encourage applicants to explain what they cannot do rather than what they can.

Socioeconomic inequality. Families with more time, knowledge, money and confidence may be better able to commission assessments, obtain reports, understand bureaucratic language and appeal adverse decisions.

Path dependence. Once support is conditional upon demonstrating continuing impairment, families may reasonably worry that evidence of improvement could threaten the support partly responsible for that improvement.

None of this requires fraud or bad faith.

It follows from institutional incentives. There is a large literature on public choice theory and other literature that accounts for this… (basically: Once rules attach resources to categories, rational actors respond to the rules as well as to the underlying need the rules are trying to measure.) And these impact both the state actors and the non-state actors…

12. The paradox of the deficit narrative

This creates an uncomfortable tension between the social and administrative models.

The social model says:

Do not define disabled people by their deficits. Look at capabilities and change unnecessary environmental barriers.

The bureaucracy often says:

Explain precisely what this person cannot do, demonstrate that it is serious and persistent, and provide evidence. Otherwise we cannot justify allocating resources.

So families can find themselves having to construct what might be called an administrative deficit narrative.

To obtain the resources that allow someone to flourish, you first have to document everything that prevents them from flourishing.

This can feel dehumanising.

But it is important not to caricature the state either.

There is a genuine allocation problem.

Resources are finite. Need is not directly observable. Eligibility creates incentives. Some form of assessment is unavoidable.

The administrative model points to a more general phenomenon.

Human variation is continuous. Bureaucracy needs categories.

Governments cannot allocate £7,842.37 of educational support because someone's underlying level of need happens to sit at exactly 7.84237 on some objective scale.

There is no such perfect scale.

Instead institutions construct categories, measurements and thresholds.

Once those measurements determine resources, the measurements themselves acquire value.

This has something of a Goodhart's-law character.

A diagnosis begins partly as a way of describing or identifying need.

Once diagnosis or documented impairment becomes useful in accessing scarce resources, individuals and institutions have additional incentives to obtain and preserve the category.

It means the observed system is an equilibrium produced jointly by:

human characteristics + social environments + administrative rules + behavioural responses to those rules.

That is a richer explanation than either “the disability is entirely inside the person” or “society creates the disability”.

How to synthesise?

Weight the different models according to the question being asked. So you might get…

Why does this person process the world differently?
Medical and developmental science.

Why are they struggling so much in this particular setting?
Social and interactional models.

Should this unusual behaviour actually be changed?
Neurodiversity, wellbeing and the person's own preferences.

What adjustments should a school or employer make?
Primarily the social model.

Could medication, therapy, education or skills training improve life?
Medical and developmental approaches.

How much ongoing support does this person require?
Functional and interactional assessment.

Why do diagnostic categories sometimes become so important to families?
Administrative incentives and resource allocation.

What should ultimately be optimised?
Flourishing, wellbeing, autonomy, relationships, participation and safety.

So overall synthesis is:

Autism is a real neurodevelopmental difference that can produce intrinsic impairments, but whether and how severely those impairments become disabling depends substantially on the person's environment and available support. Society then overlays an administrative system that determines which of those needs become recognised and resourced.

The normative principle I would put above all of this is:

The objective should not be normality. It should be flourishing.

That resolves quite a lot.

If an autistic person's harmless repetitive behaviour makes other people uncomfortable, there is little reason to train it away merely to make them look normal.

If inability to communicate pain is causing suffering, improving communication is enormously valuable.

If school is producing distress because it demands a particular form of social interaction that isn't necessary for learning, change the school environment.

If learning a skill would genuinely expand someone's future choices, help them acquire it.

If someone cannot safely navigate independently despite every sensible environmental adjustment, recognise the impairment and provide support rather than pretending society alone created the disability.

And if someone requires lifelong human assistance, that does not make their life less valuable or represent a failure of inclusion.

(while easy to say, m and harder to put in practice)

For autism, therefore:

The medical model is descriptively necessary but normatively insufficient.

The social model is morally and institutionally corrective but descriptively incomplete.

Neurodiversity rightly challenges the assumption that difference itself requires correction, but cannot make severe impairment disappear by changing the language used to describe it.

The administrative model explains how genuine needs become categories, entitlements and resources, while also creating its own incentives and distortions.

A defensible position is consequently pluralist:

accept difference, accommodate difference, treat suffering where treatment helps, teach useful capabilities where worthwhile, provide technology or human assistance where it bridges limitations, and provide care where substantial limitations remain.

The relevant outcome….

how good a life they can actually have.