Show the full module text
Module 3 — The Criteria, in Plain Language
What the diagnostic criteria for autism and ADHD actually say, translated, with the parts a questionnaire cannot reach made concrete.
Self-Identification · Part Two — The evidence
The big idea
Short on capacity today? The big idea: Both screener courses say the same thing over and over, in slightly different words: a questionnaire cannot establish the diagnostic requirements. It is an honest sentence and a useless one on its own, because almost nobody has been told what the requirements are. This is the module where they get said. It is the most factual thing in this course, and the one you should read most slowly.
Step 1 — The lesson
A. What a criterion set actually is
The criteria are not a description of autistic or ADHD people. They are a decision procedure — an agreed list of conditions that must all hold before a clinician writes something down. That distinction explains most of the confusion that follows.
A description can fit you loosely and still be true. A decision procedure either resolves or it does not. So you can recognise yourself entirely in the picture the criteria are pointing at and still not meet them as written, and you can meet them as written while feeling the description misses the thing that actually defines your week. Both happen. Neither is a scandal.
The criteria for both conditions live in the same book — the fifth edition of the psychiatric diagnostic manual, text-revised in 2022, usually written DSM-5-TR. It is copyrighted, so what follows is a translation, not a reproduction. Where to read the original is in the sources.
B. Autism, translated
The autism criteria have five parts, lettered A to E. All five have to hold.
A — social communication and social interaction. Persistent differences in this area, present across multiple contexts rather than in one bad relationship or one bad job. There are three sub-areas, and all three must be met, either now or historically:
- back-and-forth in social exchange — the reciprocity of a conversation, sharing interests or feelings, initiating and responding;
- the non-verbal layer — eye contact, gesture, facial expression, body language, the way those integrate with speech;
- relationships — forming them, keeping them, adjusting behaviour to different social settings, and understanding what a relationship is asking of you.
The phrase or historically is doing real work. A trait you have solved for does not stop counting.
B — restricted, repetitive patterns of behaviour, interests or activities. Here there are four items and at least two must be met, again currently or by history:
- repetitive movement, speech, or use of objects;
- insistence on sameness — routines, rituals, distress at small changes;
- intense, narrow interests, unusual in their focus or their grip;
- sensory difference — over-reacting, under-reacting, or an unusual pull towards sensory input.
Notice that sensory difference sits in this domain rather than in its own. A great many people who arrive at this material through sensory overwhelm are looking for a category that has been filed one shelf along.
C, D and E — the three requirements that are not about symptoms at all.
- C, onset. The features must be present in the early developmental period — with the caveat that they may not fully show themselves until the social demands of life exceed what a person can meet, and that they may be hidden by strategies learned later. The publisher’s own explainer for these criteria says the wording lets in people whose symptoms “may not be fully recognized until social demands exceed their capacity”. That caveat is inside the criteria. It is not a loophole somebody found.
- D, impairment. The features must cause clinically significant impairment in social, occupational or other important areas of current functioning.
- E, differential. The picture must not be better explained by intellectual disability or global developmental delay.
Severity levels. There are three — requiring support, requiring substantial support, requiring very substantial support, usually written as levels 1, 2 and 3. Two things about them are routinely misreported. They describe support needed, not how autistic somebody is. And they are rated separately for the two domains, so a single number was never the right shape for the answer.
Diagram — A · Five pins, one shear line. The autism criteria have five parts, lettered A to E, and all five have to hold — which is why the figure draws a lock rather than a list: nothing turns until every pin is on the line. Part A, social communication and social interaction: persistent differences present across multiple contexts, with three sub-areas of which all three must be met, either now or historically — back-and-forth in social exchange, the non-verbal layer, and relationships. Part B, restricted and repetitive patterns of behaviour, interests or activities: four items of which at least two must be met, again currently or by history — repetitive movement, speech or use of objects; insistence on sameness; intense, narrow interests; and sensory difference. Parts C, D and E are the three requirements that are not about symptoms at all: C, onset in the early developmental period; D, clinically significant impairment in current functioning; and E, a picture not better explained by intellectual disability or global developmental delay. There are three severity levels, describing support needed rather than how autistic somebody is, and they are rated separately for the two domains, so a single number was never the right shape for the answer.
C. ADHD, translated
The ADHD criteria have the same architecture: symptom lists, then four requirements about the shape of a life.
A — the symptoms. Two lists, nine items each. One for inattention: sustained attention, careless errors, listening, following through, organisation, avoiding effortful tasks, losing things, distractibility, forgetfulness. One for hyperactivity and impulsivity: fidgeting, leaving your seat, restlessness, difficulty being quiet, being driven as if by a motor, talking excessively, blurting, difficulty waiting, interrupting.
You need six or more from a list, sustained for at least six months, out of step with your developmental level and directly affecting how you function. You can qualify on either list, which is what produces the three presentations: predominantly inattentive, predominantly hyperactive-impulsive, and combined.
And then the number changes. From age 17, the threshold drops from six to five.
B — onset. Several symptoms present before age 12. Not diagnosed before 12, not noticed before 12. Present.
C — settings. Several symptoms present in two or more settings — home, work, school, with friends, with family, in other activities.
D — interference. Clear evidence that the symptoms interfere with, or reduce the quality of, social, academic or occupational functioning.
E — differential. The symptoms must not occur exclusively during a psychotic illness, and must not be better explained by another mental disorder.
Diagram — B · The bar drops at seventeen. Part A is two symptom lists of nine items each — which is what the nine notches on the post are. One for inattention: sustained attention, careless errors, listening, following through, organisation, avoiding effortful tasks, losing things, distractibility, forgetfulness. One for hyperactivity and impulsivity: fidgeting, leaving your seat, restlessness, difficulty being quiet, being driven as if by a motor, talking excessively, blurting, difficulty waiting, interrupting. You need six or more from a list, sustained for at least six months, out of step with your developmental level and directly affecting how you function; and you can qualify on either list, which is what produces the three presentations - predominantly inattentive, predominantly hyperactive-impulsive, and combined. From age 17 the threshold drops from six to five, which is the second bar. Part B, onset: several symptoms present before age 12 - not diagnosed before 12, not noticed before 12, present. Part C, settings: several symptoms present in two or more settings. Part D, interference: clear evidence that the symptoms interfere with or reduce the quality of functioning. Part E, differential: not occurring exclusively during a psychotic illness and not better explained by another mental disorder.
D. The four gates a questionnaire cannot reach
Every screening instrument in the sister courses measures traits now, from one source — you. Four of the requirements above are not about now, and not answerable by you alone. This is what that means in practice.
Diagram — C · What the beam does not reach. Every screening instrument in the sister courses measures traits now, from one source - you - which is the lit circle in the drawing. Four of the requirements are about neither. One, onset: a screener cannot ask your Year 4 teacher anything, and the evidence that satisfies this gate is mundane and it decays. Two, more than one setting: if the only place your attention collapses is the job you started in March, the criteria are pointing at the job; if it is the job and the kitchen table and the group chat, that is two or more settings. Three, impairment: the gate is not is this hard, it is what has this cost - letters unopened until they become a fine, a degree abandoned, three jobs in four years, friendships that faded because you never initiated; and its mirror image, the person whose visible life is intact because they spend every evening recovering from having held it together. Four, differential: not better explained by something else, which is a comparison a clinician runs across the whole history, and not something you can run on yourself in an evening.
1 · Onset. A screener cannot ask your Year 4 teacher anything. The evidence that satisfies this gate is mundane and it decays: school reports, the comment that keeps recurring in them, a parent who remembers you lining things up, an old diary, a sibling, a photograph of you at a birthday party standing three feet outside the group. Adults routinely fail this gate not because the symptoms were absent but because everyone who witnessed them is dead, estranged, or was never paying attention. That is an evidence problem, not a verdict.
2 · More than one setting. If the only place your attention collapses is the job you started in March, the criteria are pointing at the job. If it is the job and the kitchen table and the group chat you have not replied to since April, that is two or more settings. This gate is the one most often failed honestly by people who have arranged their entire life around a single sheltered role — and most often passed by people who assumed one bad job disqualified them.
3 · Impairment. The gate is not is this hard. It is what has this cost. Concretely: letters unopened until they become a fine. A degree abandoned. Three jobs in four years, each ended by the same conversation. Friendships that faded because you never initiated. And its mirror image, which the criteria handle less well: the person whose visible life is intact because they spend every evening recovering from having held it together. Cost paid in exhaustion is still cost, but you will have to name it explicitly, because nothing on a form asks.
4 · Differential. Not better explained by something else is the requirement that does the most quiet damage, because people read it as nothing else present. It is not. Anxiety, depression, trauma, disrupted sleep, thyroid problems, hearing loss and grief all generate scores on these instruments, and they also co-occur with both conditions at high rates. The question a clinician is asking is which account better organises the whole history — not whether a second thing exists. You cannot run this comparison on yourself in an evening, and the strength of your recognition is not evidence about how it would come out.
Diagram — D · Two lenses on the same page. The differential requirement does the most quiet damage, because people read not better explained by something else as nothing else present. It is not. Anxiety, depression, trauma, disrupted sleep, thyroid problems, hearing loss and grief all generate scores on these instruments, and they also co-occur with both conditions at high rates — which is why both lenses are lying on the page in the drawing rather than one of them being taken away. The question a clinician is asking is which account better organises the whole history, not whether a second thing exists. You cannot run this comparison on yourself in an evening, and the strength of your recognition is not evidence about how it would come out. Co-occurrence is the norm, not the exception.
E. Where the criteria fit badly, and are known to
This is not a settled instrument applied to a solved problem.
The clearest admission is the one printed in the ADHD criteria themselves: the threshold falls from six symptoms to five at age 17. A list that needs a different pass mark for adults is a list that was written about children. The wording carries the same fingerprint — several hyperactivity items describe a classroom, and the adult version of climbing on the furniture is something you have to translate for yourself before you can answer.
The autism criteria concede their own limit in criterion C, which anticipates a person whose features were invisible until demand outgrew capacity, and which explicitly allows for learned strategies obscuring them later. That is masking, named inside the diagnostic requirements. And masking is exactly what makes the criteria hard to apply to adults: the domain A features are observable behaviours, and a person who has spent thirty years constructing a passable version of those behaviours will present as someone who does not have the trait. Camouflaging has been studied in relation to sex, gender identity and how late a person is diagnosed, and it is one of the working explanations for why so many women reach the criteria only in adulthood.
Two consequences follow, and it is worth holding both. The first is that failing to meet the criteria as an adult is weaker evidence than it looks. The second, which cuts the other way, is that the finding from Module 1 still stands: within a clinically assessed sample, self-reported traits had no significant relationship with what a clinician rated by observation. The gap between how you feel and how you present runs in both directions.
Diagram — E · A photograph, coming apart. The onset requirement asks for several symptoms present before age 12 - not diagnosed before 12, not noticed before 12, present. The evidence that satisfies it is mundane and it decays: school reports, the comment that keeps recurring in them, a parent who remembers you lining things up, an old diary, a sibling, a photograph of you at a birthday party standing three feet outside the group — which is the photograph in the drawing, and the edge of it is already gone. A screener cannot ask your Year 4 teacher anything. Go and collect the onset evidence now, while it still exists: school reports go missing, parents die, and this is the only gate where delay destroys the evidence. Ask for the reports. Ask the awkward question of the relative who remembers. Adults routinely fail this gate not because the symptoms were absent but because everyone who witnessed them is dead, estranged, or was never paying attention. That is an evidence problem, not a verdict.
F. What helps
1. Turn the four gates into four questions and answer them on paper.
What is the evidence this was present before I was twelve? Where else does it happen? What has it actually cost me? What else could account for this? Written answers are worth more than an hour of thinking about it, and they are the raw material for Module 4.
2. Go and collect the onset evidence now, while it still exists.
School reports go missing. Parents die. This is the only gate where delay destroys the evidence, and it is the gate adults fail most often. Ask for the reports. Ask the awkward question of the relative who remembers.
3. Track a fortnight before you judge the impairment gate.
Not a diary of feelings — a log of consequences. What was late, what was missed, what was re-done, what you cancelled because you had nothing left. People consistently underestimate this in retrospect, because the cost of an adapted life is invisible from inside it.
4. Do not disqualify yourself on the differential requirement.
The most common way to talk yourself out of this is to notice you are also anxious, or also depressed, or slept badly for a decade, and conclude the question is closed. Co-occurrence is the norm, not the exception. Better explained by is a comparison, and you are not the one who has to run it.
5. Hold the criteria as a description of a procedure, not a verdict on you.
You can meet every requirement and still be waiting three years for anyone to say so, and you can fail one on missing paperwork. Module 5 is about what a screener adds to this, and Module 11 is about whether you want an in-depth assessment at all. Neither answer makes the other one wrong.
Up next
Module 4 — Reading Your Own History