There Is No Tesco Molecule
Updated: Sep 20
What ADHD, brain injury and Tesco can teach us about what counts as real
There is no Tesco molecule.
Nobody has isolated one. There is no Tesco gene, no Tesco region of the brain, no blood test that comes back positive for Tesco. You could take a superstore apart brick by brick, interview every member of staff, read every supplier contract and count every tin of chopped tomatoes, and at no point would you encounter the thing itself. It isn't in the bricks. It isn't in the staff. It is definitely not in the beans.
And yet Tesco exists. It employs people. It changes what a high street looks like. It takes money out of my account most weeks, which feels like a fairly robust demonstration of existence.
I should be careful here, because this is where I first went wrong with my own analogy. The building is real in a completely unmysterious way. So are the lorries, the freezers and the tins. What depends on shared agreement is Tesco as an institution: the name, the legal arrangements, the fact that we all treat a particular scatter of buildings and contracts and people as one thing with a continuous identity that persists when the staff change and the buildings are rebuilt. That part is held together by collective recognition. Withdraw the recognition and the bricks stay exactly where they are, but Tesco doesn't.
I started thinking about this because of a Channel 4 documentary, The Great ADHD Myth?, and an article written in response to it by Alastair Mordey, arguing that ADHD looks more like a social construct than a real condition, and that de-diagnosing people would be a sensible place to start.
I am not primarily interested in defending ADHD. I am interested in a move the argument keeps making without pausing over it. When someone says that a thing is socially constructed, what exactly do they take themselves to have disproved?
What "socially constructed" actually tells us
The honest answer is: less than the phrase is usually asked to carry, and also more than its critics allow.
Tesco is a constructed institution. Dissolve the company law, the registration, the shared habit of treating those buildings as one enterprise, and there is nothing left over that was Tesco. The construction goes all the way to the bottom of the thing itself.
ADHD is not that. ADHD is a constructed classification. It is a way of grouping and naming a set of difficulties. And a great deal about it genuinely is constructed, more than defenders of the diagnosis sometimes like to admit. The terminology has changed during my lifetime, as have the boundaries and criteria. So has the threshold at which difficulties are treated as clinically significant, shaped partly by expectations about how long a nine-year-old should sit still, how much administrative complexity an adult should absorb without help, and how much of life we are expected to organise unaided. All of it is a human arrangement, and a fairly recent one.
But the classification did not manufacture what it classifies. Abolish the manual tomorrow and the recurring difficulty with starting, organising and completing the homework does not disappear. What remains open is how best to explain it — whether the grouping is the right grouping, whether the boundary is in the right place, whether we have one thing here or several. Those are real questions. They are not the same question as whether anything is happening.
The sceptical phrasing blurs three questions that need to stay apart: whether the category is constructed, whether it can be worn as an identity, and whether it is a medical illness.
A category can be the first two and still be organising something real. Money is constructed. Universities are constructed. Professions, borders, credit ratings, the offside rule. None of them dissolves when we recognise that its boundaries, rules and meanings were made by people.
This is the first appearance of a move that keeps recurring in different costume. A limitation gets converted into a conclusion it cannot carry: “we cannot establish it in this particular way” becomes “there is nothing there to establish.”
Constructed becomes unreal.
Continuously distributed becomes trivial.
Requiring expert interpretation becomes invented by the experts.
And uncertainty about how we classify something becomes certainty about the character of the people classified.
Those are four separate questions, and they get run together into one question with one answer. What follows is an attempt to take them apart.
Where the sceptics are right
Before going further I want to concede properly, because a piece that concedes nothing is not worth reading.
On assessment and prescribing: there is no biomarker. No scan, no blood test, nothing that identifies ADHD in an individual person. This is true, it is uncontested, and nobody serious claims otherwise. If adults are being diagnosed on the strength of a questionnaire and a brief uncritical interview, that is a failure and it matters — the standard requires specialist assessment, developmental history, evidence across settings, and states plainly that a diagnosis should not rest on rating scales alone. Prescribing has risen substantially and deserves scrutiny. Stimulant misuse and diversion are real: a 2025 systematic review of people prescribed stimulants for ADHD found past-year misuse of their own medication running at around 23 per cent, and past-year diversion at around 18 per cent.
On what diagnosis and support can do: a diagnosis can become totalising. It can turn from one useful account of a person into the whole account, which forecloses questions rather than opening them. And some accommodations reduce opportunity rather than creating it. Support that removes participation rather than making participation possible can narrow a person's opportunities. That risk is real too.
On reviewability: diagnoses should be reviewable, and I can think of no good argument against careful, voluntary reassessment for someone who thinks theirs no longer fits. The long-term evidence on medication is thinner than the short-term evidence, and the researchers who produce it say so themselves.
And the concession that matters most, which the sceptics are simply right about: improvement on a stimulant does not establish that a person has ADHD. Nor does the way the drug works tell us what caused the difficulty in the first place. Paracetamol relieving a headache does not reveal a paracetamol deficiency. Response to treatment is not a diagnostic test and should never be used as one. Stimulants do have effects in people without ADHD, though those effects are more modest and more uneven across different kinds of thinking than the confident version of this claim suggests.
That is a substantial amount to grant. But notice what kind of thing it is. These are reasons for caution about how ADHD is diagnosed and treated. They are not evidence that the difficulties are invented, that continuously distributed traits cannot become disabling, or that people who need expert assessment were created by the assessment.
If everyone is a bit like this
The strongest version of the sceptical case is not about biomarkers at all. It is this: inattention, restlessness, disorganisation, boredom and difficulty finishing dull tasks are everywhere. Everyone has them. So what are we actually diagnosing?
The premise is correct, and I want to concede it fully rather than fight it, because fighting it is a losing position and an unnecessary one. Taxometric studies — the statistical work that asks whether a population contains two distinct groups or one continuous distribution — have looked at ADHD in children, in adolescents and in adults, and keep finding the same thing. There is no natural category. There is no break point separating people with ADHD from everyone else. Haslam and colleagues, who did much of this work, drew the obvious conclusion: the threshold has to be set on pragmatic grounds.
This is not the first question wearing a different hat. If ADHD is a constructed classification, we should expect its threshold to require judgement; that is what constructed classifications are like. The absence of a natural boundary tells us how carefully the category has to be used. It does not tell us that nothing lies beneath it.
Here is what I find interesting. The same methods, applied to psychotic experiences in the general population, find the same thing. No natural break between milder and more severe experiences; a continuous distribution; thresholds that have to be set on grounds other than a discontinuity in the data. And nobody concludes from this that severe psychotic experiences are not worth identifying, or that their continuous distribution makes them unreal.
To be precise about what that does and doesn't show: it concerns psychotic experiences as continuously distributed phenomena, not schizophrenia as a clinical construct, and it does not make ADHD and schizophrenia equivalent in validity or anything else. What it removes is the contrast the argument depends on: the idea that proper conditions come with a natural edge and ADHD doesn't.
So what does distinguish a clinical pattern from ordinary variation? Not any single trait. It is the configuration: several of them together rather than one in isolation, persisting rather than passing, present from childhood rather than arriving with a bad year, showing up across settings rather than in one, producing consequences, and not better explained by something else. Every one of those is a matter of degree, and every one of them requires judgement.
Which is why the comparison usually offered is not a comparison at all. Mordey's version sets someone smoking a crackpipe for twenty-three hours a day against someone who does not want to make their bed. That is not a contrast between two constructs. It is the most extreme presentation of one condition against the most trivial manifestation associated with another. A symmetrical version would put mild against mild, or severe against severe. Nobody has suggested that failing to make the bed warrants a diagnosis.
At the severe end, the meta-analytic evidence associates these difficulties with impairment in education and employment, with risky driving and accidental injury, and with difficulties in social functioning. Associated with — these are population-level patterns, not a claim that ADHD alone produces any of them in any particular person. But they are not nothing, and they are not about bedmaking.
What anyone with eyes can and cannot see
There is a fallback position, and it is the most interesting thing in the sceptical case. Mordey's version runs like this. Fine, addiction has no biomarker either — but addiction is obvious, you can see it, and so is schizophrenia, whereas ADHD needs a psychologist to come along and find it. That, he says, is the difference.
The comparison with schizophrenia has already failed once, as an argument about natural boundaries. It fares no better as an argument about visibility. What a clinician observes is speech, behaviour, and what a person reports about their own experience. The diagnosis is then arrived at through history, assessment against criteria, judgements about duration and functional effect, and the exclusion of substances, medical conditions and mood disorders. Nobody sees schizophrenia. They see evidence, and they interpret it.
The move being made here is that needing interpretation counts as evidence against reality, and it is worth taking seriously, because it does more work than it appears to.
Start with the standard itself. Observable to anyone with eyes is not a scientific criterion. It is a criterion about observers. It favours difficulties that are dramatic, physically apparent, culturally familiar and already accepted as legitimate — and disfavours everything else, regardless of what is actually happening to the person.
Then notice that addiction does not meet it either. Substance use is observable. Addiction is an interpretation of a pattern. DSM-5 sets out eleven criteria across four clusters — impaired control, social impairment, risky use, and the pharmacological ones — and requires at least two of them within a twelve-month period, with the count determining severity. So distinguishing use from disorder means judgements about control, craving, persistence in the face of harm, and how the whole thing has moved over time. Tolerance and withdrawal are only two of those eleven, and even they — the most physiological-looking of the set — need context rather than observation: when they arise in someone taking a medication solely under appropriate medical supervision, they do not count towards a diagnosis. None of that arrives through the eyes. It arrives through history, report, observation and inference — the same process the argument treats as suspect when applied to ADHD.
But the strongest reason to distrust the visibility standard comes from somewhere else entirely, and none of it is about ADHD. It comes from neurology, including cases where an injury has been independently established. Three cases, each doing one job. An accepted cause does not make its consequences visible. Even a visible lesion still requires behavioural interpretation. And a real material change and a socially constructed label can occupy the same case at the same time.
Consider someone left with executive difficulties after a stroke or traumatic brain injury. The frontal regions and their circuitry are particularly vulnerable to head injury, so this is common rather than exotic, and cognitive difficulty is the leading cause of disability after traumatic brain injury. Such a person may speak fluently. They may walk normally. They may do perfectly well on anything simple. What has gone is the capacity to initiate, to sequence, to hold a plan while executing it, to notice their own errors, to switch when circumstances change.
We can observe that a task was not completed. Observation alone cannot tell us whether the reason was difficulty initiating, sequencing, remembering, understanding, or something else entirely.
There is a study I find clarifying here. After concussion, people were tested on a task-switching paradigm. Their performance on the trials that did not require switching was unaffected — indistinguishable from controls. It was the switching that cost them, and it went on costing them for at least a month.
The narrow point is this. Even where an injury is established, and sometimes plainly visible on imaging, the cognitive consequences do not become visible along with it. They emerge only through history, structured assessment, informant accounts, performance on complex tasks, and comparison with how the person used to be.
Not that brain injury is always believed. Qualitative research with survivors records the opposite: being told they look fine, that there is nothing wrong with them, that they are imagining it. Mild injury may leave nothing on a structural scan at all. This is not a contrast between difficulties that get taken seriously and difficulties that don't. It is narrower, and it holds anyway. If expert assessment is necessary after a brain injury that everyone agrees happened, then needing expert assessment cannot by itself make a developmental difficulty suspect.
Hemispatial neglect makes the same point more sharply, because here the lesion is not merely present but often plain. After a right-hemisphere stroke, a person may attend to only one side of space: eating from one half of the plate, drawing only half a clock, shaving one side of the face, not responding to someone approaching from the left.
The temptation is to call this a vision problem. It usually isn't. Neglect is primarily a disorder of attention, usually associated with cortical injury. Hemianopia is loss of part of the visual field, caused by damage somewhere along the visual pathway. They are different phenomena, although they can resemble one another from across a room and can occur together in the same person.
Here is the part that matters. Standard visual field testing may not distinguish them, because it cannot tell you whether a stimulus was missed through inattention or through blindness. The scan shows the lesion. The scan does not show what the person perceives. To find that out you need behavioural work: line bisection, cancellation tasks, copying, and watching what someone actually does over breakfast.
The lesion is visible. The deficit requires interpretation. There is no way round this.
Then there is foreign accent syndrome, which is the reason I find this whole area so difficult to dismiss.
Foreign accent syndrome is rare. In acquired neurological cases, speech can change after a stroke or head injury — in articulation, timing, stress and intonation — in ways that listeners hear as a foreign accent. Those cases are usually caused by small focal lesions in the network that produces speech.
The person has acquired real changes in speech production, but not the foreign accent or linguistic history that listeners attribute to them. The speech typically does not match any actual foreign accent, which is why researchers prefer the term pseudo-foreign accent: a constellation of changes that listeners resolve into a category they already have. When Belgian researchers embedded recordings of five people with the syndrome among five genuinely foreign-accented speakers and five native controls, the listening panels sorted the last two groups without difficulty: the native speakers were recognised as native, the foreign-accented speakers reliably heard as non-native. The people with foreign accent syndrome landed nowhere stable. They were heard as partly foreign and partly native, never as foreign as someone with a real foreign accent, and assigned a scatter of different origins.
One finding does more than any other. In a case study from 1992, reported by later researchers rather than read here at first hand, listeners rated speech recorded before and after onset; the post-onset speech was judged no less intelligible and no more disordered than before. It was judged significantly more foreign. The acoustic change was real. But the category of foreignness, and which foreign accent listeners believed they heard, was supplied by the people listening.
Real neurological event. Constructed label. Both at once, in the same case.
What these cases have in common is not a conclusion about ADHD. It is a conclusion about observers. Visibility is partly a property of the person doing the looking, and absence of visibility is not absence of difficulty. This is the basis of hidden disability, and it applies with particular force to people who mask, who compensate, who appear to be managing because somebody else is quietly structuring their life, or because they are spending unsustainable effort and doing their recovering in private. When the scaffolding goes or the demands rise, what looked like coping stops looking like coping.
I want to state the limit of all this plainly, because it would be easy to overreach and I would rather not.
These examples do not show that ADHD is a valid construct, or that any particular person has it. They are not evidence about ADHD at all. What they show is narrower and, I think, sufficient: that requiring expertise to recognise and classify a phenomenon does not mean the expertise created it. You can see that the form was not returned. You cannot see why by looking harder.
So we have three things. A constructed classification. A threshold set by judgement rather than read off a natural boundary. And difficulties that can only be identified through interpretation. The sceptical argument treats each as a proof of unreality and stacks them until the conclusion feels overwhelming.
They are not proofs of unreality. They are three reasons to be careful about what we claim to know, and they apply across a great deal of medicine. What keeps happening is that a limit on our certainty is converted into certainty that there is nothing there.
Which would be an argument about evidence, if it stopped at evidence.
From uncertainty about diagnosis to certainty about character
It does not stop at evidence. This is where an argument that began with uncertainty about classification becomes remarkably certain about character.
The destination runs roughly like this. Diagnosis produces an identity built around being unwell. That identity produces accommodations. Accommodations produce reduced expectations. Reduced expectations produce withdrawal from work and education, which becomes permanent. Therefore de-diagnosis.
Every arrow in that chain requires evidence, and the statistics offered cannot supply it. Counts of how many young people are not in education, employment or training, and what health conditions they report, are snapshots. They cannot tell you whether a diagnosis came before or after someone disengaged, whether anyone's expectations changed, or what would happen if the diagnosis were removed.
It would be easy to run the chain backwards — unmet need and exclusion cause disengagement, therefore support is vindicated — and it would be just as unsupported. The available statistics cannot determine which way the causation runs, or whether it runs both ways at once. That is an uncomfortable place to stand, and the only one the data allow.
One detail shows how easily this material misleads. Just over half of young people classified as NEET reported having a health condition. Long-term or temporary sickness was the stated reason for being NEET for around one in five of them. Those are different measures, and treating the first as the second produces a considerably more alarming picture than the figures support.
What I think is actually going on is less dramatic and harder to legislate for.
The same underlying difficulty produces wildly different consequences depending on what is being demanded, what resources are available, and how much room there is for things to go wrong. This is not a claim that ADHD is caused by circumstances. The underlying difficulty may remain, but the extent to which it becomes disabling depends partly on the demands, barriers, support and resources around the person.
Consider two people with much the same trouble organising themselves. One has an accountant, some flexibility about when work gets done, a partner who handles the post, savings that absorb a missed deadline, and the option of taking a fortnight off when things get bad. The other has shift work with a rota that changes weekly, a benefits system that punishes a missed appointment, no financial margin at all, and nobody to hand anything to. The difficulty has not changed. Its exposure has, its available compensation has, and its consequences have.
It is worth being precise about what the first person's apparent independence is made of. An accountant. A partner handling the post. Savings. The executive load has not disappeared; a good deal of it has been transferred to somebody else, some of it paid for and some of it not. That is not a criticism of the arrangement. It is a description of what competence often consists of, and of what its absence costs.
One of them will look eccentric and the other will look like a problem.
I am not saying that money removes the difficulty, or that wealth protects anyone from being severely disabled by it — it plainly does not. I am saying that resources change what a difficulty costs, and that a great deal of what gets read as character is actually the cost being visible.
This is also where the argument about accommodations goes wrong, and I say that as someone who agrees that some accommodations are badly designed. The premise smuggled into the sceptical case is that support means removing expectations. But good support may change demands, provide structure or remove barriers. Removing unnecessary administrative complexity, allowing breaks, or reducing an unsustainable workload can be exactly what makes participation possible. The test is not whether demand went up or down. It is whether the support expands a person's access, agency and sustainable participation, or merely makes their world smaller. Poor support is an argument for designing support better, not for withdrawing it.
Underneath all of it is something I notice constantly, and it is the reason I wanted to write this.
Executive difficulty translates into moral language with almost no resistance. Difficulty starting a task becomes lazy. Attention that varies with interest becomes only trying when they want to. Losing track becomes careless. Needing structure becomes dependent. And asking for an adjustment becomes wanting special treatment.
Every one of those translations is available instantly, requires no evidence, and feels to the person making it like plain common sense rather than an interpretation at all.
Which is the thing the argument misses about itself. Removing a diagnosis does not remove interpretation. It restores an older one. The space that opens when you announce there is no biomarker does not stay empty; it fills with character judgement. And character judgement is not the neutral, evidence-free default it presents itself as. It is a theory of the person, held with far more confidence than the evidence for it would ever justify.
Diagnostic humility without disbelief
I have not argued that every ADHD diagnosis is correct. I do not believe that, and it would be a strange thing to believe about any diagnostic category.
What I think can be held together is this. We can insist that assessments meet the standard that already exists on paper, and watch prescribing carefully. We can resist diagnoses that become the whole account of a person. We can design accommodations that increase capability and sustainable participation rather than removing people from the situations where support is needed, and notice when we have got that wrong. We can review a diagnosis that no longer fits. And we can keep our expectations high.
None of that requires treating people's difficulties as invented, or the people themselves as morally suspect. Diagnostic humility should produce better formulation and better support. It should not produce disbelief, blame, or the quiet withdrawal of reasonable adjustments from people who need them.
It is worth putting these things together, because separately each one sounds like a concession and together they are the argument. The category can be constructed. The traits can be continuously distributed. The threshold can require judgement. The difficulties can be observable only indirectly. And assessment can still, in practice, be done badly.
All of that is true. None of it establishes that anything has been fabricated, and none of it establishes anything whatever about whether a person is trying.
So, back to Tesco.
There is no Tesco molecule, and there is no scan that can decide whether a particular person has ADHD. The absence of either does not settle the question of reality. What it tells us is that we have been asking one kind of evidence to answer a question it was never going to answer on its own.
Which leaves two questions rather than one, and I think we need both. What does the evidence justify us in believing? And what follows for the person we are talking about?
The first without the second can produce an account that is technically careful and humanly disastrous. The second without the first can turn compassion into another form of certainty unsupported by evidence. Held together, they change the argument. The question is no longer simply whether a difficulty is real, but how our account of it is justified, and what that account allows us to do next.
Sources and further reading
Mordey, A. "Is the ADHD myth unravelling?" The Spectator Australia, 24 August 2026 — the version cited and linked throughout is the page at spectator.com.au dated 24 August 2026. The Great ADHD Myth?, Channel 4, broadcast 18 August 2026.
Diagnostic standards: NICE guideline NG87, Attention deficit hyperactivity disorder: diagnosis and management (2018, subsequently updated).
Terminology: the sequence runs from DSM-II (1968) to DSM-III (1980) to DSM-III-R (1987), with later editions revising subtypes, presentations and criteria.
Distribution: Haslam, N. et al. (2006), Australian & New Zealand Journal of Psychiatry 40:639–647; Marcus, D.K. & Barry, T.D. (2011), Journal of Abnormal Psychology; Marcus, D.K. et al. (2012), Journal of Psychiatric Research. Psychotic experiences: Taylor, M.J., Freeman, D. & Ronald, A. (2016), Dimensional psychotic experiences in adolescence, Psychiatry Research 241:35–42.
Outcomes: Willcutt, E.G. et al. (2012), reported within Faraone, S.V. et al. (2021), The World Federation of ADHD International Consensus Statement, Neuroscience & Biobehavioral Reviews 128:789–818.
Medication: Cortese, S. et al. (2018), Lancet Psychiatry 5:727–738; Ostinelli, E.G. et al. (2025), Lancet Psychiatry. Effects in people without ADHD: Ilieva, I.P., Boland, J. & Farah, M.J. (2015), Neuropsychopharmacology. Misuse and diversion: Forrest et al. (2025), systematic review and meta-analysis.
Addiction criteria: American Psychiatric Association, DSM-5-TR, substance use disorder.
Brain injury: Rabinowitz, A.R. & Levin, H.S. (2014), Cognitive Sequelae of Traumatic Brain Injury, Psychiatric Clinics of North America. Task-switching after concussion: Mayr, U. et al. (2014), PLoS ONE 9(3):e91379. Survivor accounts of disbelief: qualitative study of community-based rehabilitation following traumatic brain injury, International Journal of Speech-Language Pathology (2024).
Hemispatial neglect: Sarwar, A. & Emmady, P.D., Spatial Neglect, StatPearls.
Foreign accent syndrome: Verhoeven, J. et al. (2013), Accent attribution in speakers with Foreign Accent Syndrome, Journal of Communication Disorders 46:156–168. The before-and-after listener ratings are Ingram, J.C.L., McCormack, P.F. & Kennedy, M. (1992), Journal of Phonetics 20:457–474, as reported in Kanjee, R. et al. (2010), Journal of Neurolinguistics 23:580–598.
NEET figures: Department for Education, NEET age 16 to 24, 2025.
David Corcoran is a Chartered Psychologist and the founder of Beyond Thrive. He writes about how psychological ideas are interpreted and translated into practice.
Comments