Introduction

Some of you may have noticed another small rhythm running alongside the longer work here: most school mornings, over on Substack Notes, I post a photograph under the thoroughly uncomplicated heading Today’s Outfit.

It began as exactly what it sounds like. I was dressed. I liked what I was wearing. I took a photograph. Then, because apparently I am incapable of leaving an ordinary thing entirely ordinary, I wrote something about it. A necklace became a thought about colour. A skirt opened onto memories of Highland Games and years spent kilted. Flowers led towards Alba. A photograph from the side made me wonder why plus-size clothing so often seems designed to hide the body wearing it. Pearls were sometimes simply pearls because… why not?

But repetition has a way of revealing function. After doing this day after day, I began to understand that Today’s Outfit had quietly become part of my daily writing practice—a wee field notebook built around something concrete, embodied, and already present. The photograph gives me somewhere to begin. The writing lets me discover what else was alive around it. For someone for whom writing and pragmatics require substantial work, that regular exercise matters. These are muscles I need to use.

And then I realised something rather obvious: not everyone who reads The AutSide uses Notes. Some subscribers may never see these small daily pieces at all. Yet collectively they have become connected to the larger work I am doing here—gestalt processing, embodiment, language, autistic support, queerness, visibility, and the strange recursive business of discovering what I know by giving the knowing somewhere to rest.

So I am bringing some of them over.

Not every outfit. Not every morning. Think of these instead as occasional dispatches from the field—a handful of photographs gathered together, with some of the writing that grew around them and, occasionally, a look underneath the practice itself.

The longer essays are where I build the framework.

These are where you can sometimes watch me finding it.

The Complete Gestalt

At first
I thought I was writing
about clothes.

A skirt.
A tunic.
Pearls, because why not.
Flowers, because they make me happy.
Purple nails like wee sweets.
Beads, the colour of salmon.
Big white trainers
under all of it.

Today’s outfit.

Something small enough
to begin.

Because writing
is not small for me.

Language is not small.

There is knowing
and there is saying,

and between them
there can be an enormous distance.

There is the whole
already alive,

body, memory, colour,
history, feeling,
Alba, childhood,
gender, safety,
someone’s face,
something somebody said
thirty years ago,

all of it present

before I know
which sentence
comes first.

Speech wants the sentence.

Now.

Writing waits.

So every morning
I give myself something
to begin with.

Here is what I wore.

There.

A handhold.

A photograph
holding the whole still
long enough
for me to return.

I like these beads.

Why?

Almost.

Not quite that.

Quarter-turn.

There.

And suddenly
I am not writing
about beads anymore.

I am writing about colour.

About allowing myself
to be delighted.

About a body
I spent years
trying not to draw attention to.

About the bullied child
who learnt that perhaps
if she could make herself smaller,

quieter,

less noticeable,

there might be
less of her
for cruelty to find.

It found her anyway.

So now,

Today’s outfit.

Again.

And again.

And again.

Not because yesterday’s writing
failed.

Because muscles
are maintained
through use.

Writing.

Pragmatics.

Locating the feeling.

Finding the signal
without murdering everything else
as noise.

Working out
what another person needs
in order to meet me here

without reorganising
the whole of myself
for their comfort.

This is occupational work.

This is field work.

This is soul work.

The photograph
becomes a field note:

I was here.

My body was here.

Not the body
I might have had.

Not the body
someone thinks
a woman ought to have.

This one.

Tall.

Broad.

Athletic history
still living in it.

Shoulders and hips
and bicepts and legs
and feet that were never
going to disappear politely
beneath the furniture.

I’m not hiding my body.

The clothes make me happy.

That is enough.

I’m not hiding my voice.

It is not some untouched
museum-piece Argyll,

as though migration
and time
and a life lived elsewhere
could pass through a voice
without leaving anything behind.

But Alba remains there.

The West Highlands remain there.

Because something
in those sounds
still tells my body:

safe.

This is how safety sounded
when I was a wee bairn.

And I will not
make that voice tiny
so someone else
can recognise it
as feminine.

Who decided
femininity should sound
like a six-year-old lassie?

Small.

Light.

Careful.

Acoustically apologetic.

As though womanhood
reaches its purest form
at the edge of disappearance.

My Stone Borne ancestors
were not tiny people.

People sailed those coasts.

Crossed water.

Worked land.

Raised children.

Carried what needed carrying.

Laid stones
against earth and sky

and watched the heavens
move through them.

These are the exploits
of wee bairns?

No.

I will not become wee
so that you can recognise me
as a woman.

I’m not hiding my feelings.

This delights me.

That hurt.

I love this.

I miss that.

I am angry.

I feel beautiful.

I like my butt.

Sometimes the knowing
arrives long before
I can name it.

So the page waits.

It holds the feeling
without demanding
immediate identification.

Almost.

Not that.

Again.

Quarter-turn.

There.

I’m not hiding
that I’m trans.

Adjective.

Meaning: not cis.

I have never been cis.

Not before I knew the word.

Not when other people
used another one.

Not when hiding
was safer.

Not ever.

I’m not hiding
that I’m queer.

And I mean queer.

Not merely
a convenient umbrella.

The old disturbance
of the expected line.

The oblique angle.

The thing that turns
just enough
for the structure underneath
to become visible.

Every quarter-turn
of it.

And yes,

you can see the skirt.

I have worn them
for more than half my life.

Why were you comforted
when you could call it
a kilt?

Same legs.

Same body.

Same fabric
falling from a waist.

But kilt returned me
safely to a story
you already understood.

Alba.

Athlete.

Tradition.

Masculinity restored.

Skirt

leaves the question open.

I rather like it open.

Because I am not trying
to arrive at the opposite
side of somebody else’s line.

I am not trying
to disappear successfully
into another category.

That would still make
disappearance
the achievement.

My quarter-turn
is elsewhere.

What happens
if I become

more precisely visible
as myself?

Body.

Voice.

Feeling.

History.

Strength.

Flowers.

Pearls.

Queerness.

Transness.

Alba.

Skirt.

Trainers.

All informationally alive.

No requirement
that one cancel another.

No requirement
that femininity erase strength,

that transness erase history,

that queerness erase belonging,

that adornment erase the athlete,

that the woman apologise
for the amount of space
her body occupies.

The whole
can hold them.

Perhaps that is why
I keep taking the photographs.

Perhaps that is why
I keep writing.

The page gives coherence
somewhere to rest
whilst language catches up.

But after all these mornings,

I think it gives
me
somewhere to rest too.

A field notebook.

A practice.

A small daily occupation
in making myself available
to myself.

Here I am.

Here is what I wore.

Here is what delighted me.

Here is what I noticed.

Here is where it led.

Here is what became knowable
because I stayed with it
long enough.

Tomorrow
I will exercise
the muscles again.

Not because today
was insufficient.

Because living things
are maintained
through use.

And because I spent
quite enough of one life
practising disappearance.

So—

you can see my body.

You can hear my voice.

You can know my feelings.

You can know I’m trans.

You can know I’m queer.

You can see the skirt.

You can see
all of me
that happens to be visible today.

I am not failing
to conceal it.

I am not failing
to pass.

I am not failing
to become smaller.

I have stopped accepting
diminution
as the price
of being recognised.

Today’s outfit.

Something small enough
to begin.

Something large enough
to hold a life.

I’m not hiding.

I’m here.

Field Notes

These are the Field Notes behind the Today’s Outfit pieces I’ve been dropping daily in Notes.

The photograph is the easy part. Stand in front of the mirror before school. Phone up. Click. There I am.

The writing around it is something else.

Writing is difficult for me. Pragmatics is difficult for me. Both are part of my ASD Level 2 support needs, and I have come to think of these little daily pieces as a kind of occupational practice—not therapy assigned by someone else, but meaningful work I have built for myself. Muscles need use. So I begin somewhere deliberately ordinary.

What am I wearing today?

This week, the first answer has been: whatever permits me to survive Los Angeles.

The heat has been relentless—day after day above 100 degrees, concrete throwing yesterday’s heat back into today, sun pressing down, and humidity making even shade feel temporary. Under those conditions a skirt is as defensive as it is decorative. Air moves. Fabric moves. I can move. The flowers, polka dots, beads, and pearls may be adornment, but the architecture underneath them is practical.

And yes, there are rather a lot of skirts.

I know professional dress standards have become less important to many younger people, and I am certainly not proposing their return. I am hardly a strict observer myself. But somewhere in my earliest gestalt of teacher are the women who taught when I was young. They wore skirts. They seemed impossibly elegant and put together to me—not because a skirt made them better teachers, but because that was one of the ways I first learnt what a teacher could feel like in a room.

Apparently some part of me kept that.

So I borrow from it now.

Not as rule.

As pleasure.

Long skirts that move when I walk. Floral tops. Pearls against black. Salmon-coloured beads against green. Polka dots answering polka dots because sometimes the correct amount of pattern is more. And when I wear trousers, they tend to be linen, loose, and chosen with the same attention.

A girl must try to be her best self, I guess.

But then comes the part these photographs cannot show.

I sit down with the image and begin asking what remained alive around it.

Why did I choose this?

What pleased me?

What did my body know about the heat before I had language for it?

Why pearls?

Why flowers?

Why did seeing myself from this angle produce a thought about clothing manufacturers, or teachers I knew decades ago, or Alba, or femininity, or the long history of trying not to be noticed?

Almost.

Not quite.

Quarter-turn.

Try again.

The photograph becomes a handhold.

This is the craft of the daily writing: beginning with something bounded enough that I can enter it, then allowing the relational field around it to become available without demanding that I know beforehand where it leads. The whole is already there. Writing gives me time to differentiate what is inside it.

And across days, something larger accumulates.

Black and flowers.

Green and salmon.

Pink blossoms and cream.

Pearls and polka dots.

The same bathroom mirror.

The same woman.

Not hiding her body.

Not dressing to make herself smaller.

Not trying to disappear successfully into somebody else’s idea of what a woman, a trans woman, a teacher, or a fifty-something woman ought to look like.

Just practising.

Writing.

Pragmatics.

Feeling.

Embodiment.

Being seen.

Being able to look back at myself and say, yes, that felt like me today.

That is why these are Field Notes.

The clothes give me somewhere to begin.

The photograph holds the moment still.

The writing lets me discover what else was there.

Movement VIII—Level 2

I am Level 2—requiring substantial support.

I began with those words because I could not quite reconcile them with the life I could see around me.

I return to them now.

Same words.

Different object.

Or perhaps the object has not changed at all. I have simply walked around it.

At the beginning, I could see the apparent contradiction easily enough. I teach. I research. I write. I have spent my professional life inside complicated systems, first in forensic science and now in education. I work with language constantly. I work with evidence. I manage enormous amounts of information. I recognise patterns, construct arguments, solve problems, teach other people how to solve problems.

And somewhere beside all of that sat the diagnostic phrase:

requiring substantial support.

Where?

That was the question.

I was not interested in proving the diagnosis wrong. Nor was I willing to turn myself inside out to produce enough suffering to make it believable. Both responses accepted the same premise—that competence and substantial support were somehow contradictory and that one therefore had to explain the other away.

The first quarter-turn was simply to stop asking how disabled I appeared.

Instead:

What has to be present around me for the competence people observe to become available?

Now I can begin to answer.

The support is everywhere.

Not everywhere in the sense that I am helpless without assistance. Everywhere in the more interesting sense that the conditions through which I function have become densely scaffolded, much of that scaffolding assembled by me over decades without ever calling it disability support.

It is technological.

It is environmental.

It is temporal.

It is informational.

It is relational.

It is the original document rather than somebody’s summary of it. It is the email that remains available after the conversation would have vanished. It is search when a missing piece of context prevents the rest from settling. It is a screenshot preserving something before I know why it matters. It is external memory. Multiple windows. Notes. Histories. Archives. The possibility of comparison. Provenance.

It is time.

Time to reread.

Time to discover what the question actually is.

Time for language to become available after knowing has already arrived.

Time for the whole to settle enough that I can begin taking it apart.

And increasingly, it is technology capable of holding unresolved information outside my body whilst I determine its relationships.

That may be one of the most important things I have discovered in writing this series.

Too many things remain informationally alive.

Before the gestalt resolves, I cannot always safely determine what carries high evidential weight and what is contextual; what is redundant and what is corroborating; what contradicts something else; what only appears to contradict it; what belongs to the source; what somebody else has inferred; what will remain peripheral; what will become the detail that reorganises everything.

What can safely die.

I had initially thought about this as signal-to-noise ratio.

I still do.

But the phrase changed as I turned it.

The problem is not simply that signal exists alongside noise and I struggle to filter one from the other. That assumes the categories have already stabilised.

Often they have not.

Sometimes what looks from outside like an inability to ignore irrelevant information is, from inside, an inability to establish that the information is irrelevant yet.

The distinction matters because it changes what support does.

If the problem is unwanted noise, support should remove it.

If the problem is unresolved informational weighting, removing information too early may remove the relationships through which signal and noise would have become distinguishable.

Sometimes less helps.

Sometimes less makes everything harder.

Sometimes the whole is the accommodation.

That recognition sent me backwards through my own technological life.

I was never the autistic child whose support disappeared when she became an adult.

There was no support to disappear.

No IEP. No accommodations. No scaffold built around my processing because anyone understood what my processing required.

I built.

And technology kept giving me better materials.

Word processing allowed language to become revisable rather than immediate.

Email allowed communication to become asynchronous.

Search made context recoverable.

Smartphones made the scaffold portable.

Screenshots allowed unresolved information to remain unresolved without being lost.

Searchable histories externalised memory.

Multiple windows allowed relationships to remain simultaneously available.

AI gave me something stranger still—a place where an unresolved informational field could be held, manipulated, questioned, corrected, turned, and returned to without requiring premature closure.

None of those technologies belongs uniquely to disability.

That is partly why I had not recognised them as support.

Everyone uses them.

But everyone wearing shoes does not mean shoes perform the same function for every body. The ubiquity of an object tells us almost nothing about the function that object performs within a particular system.

A phone can be an ordinary consumer device and an accessibility technology at exactly the same time.

A tiny object can perform an enormous amount of support work.

And support does not cease to be substantial because it is inconspicuous.

That distinction may be the one I needed all along.

Substantial support is not the same thing as conspicuous support.

I had been looking for the person beside me.

The prompt.

The aide.

The visible accommodation.

Instead, much of my support had been distributed into the architecture of my life.

And because I administer so much of it myself, successful functioning can easily be mistaken for evidence that the support need is absent.

But that inference runs backwards.

If I perform well under conditions of support, the performance tells us that I can perform well under those conditions.

It cannot tell us that the conditions were unnecessary.

The competence is real.

So is the scaffold.

This is where I return to language.

My relationship with language has followed me through diagnostic systems. Under DSM-IV, my scripts helped establish the presence of language and therefore mattered to the distinction that placed me on the Asperger’s side of the boundary. Under DSM-5, the boundary moved, and my substantial difficulties with language became part of understanding the support level I required.

The processing did not suddenly change.

The diagnostic architecture around it did.

And now, looking from inside, I encounter another uncertainty.

I am autistic.

I am a gestalt processor.

I know both things about myself.

What I cannot cleanly establish is where one stops and the other begins.

Some of the behaviours through which autism becomes diagnostically legible are also things I recognise as downstream expressions of my gestalt processing. Scripts. Context dependence. The relationship between parts and wholes. Difficulties producing language before enough of the whole has resolved. The instability of informational hierarchy. The social consequences of being required to decompose before decomposition is available.

I cannot use the behavioural diagnosis to separate those mechanisms cleanly because the possible confound is already inside the diagnostic signal.

I do not need to solve that here.

That may be another form of premature compression.

Perhaps the useful finding is simply that the question exists.

My diagnosis names autism.

My Level 2 designation names substantial support needs.

My lived investigation identifies the management of unresolved informational weighting—signal and noise, context and hierarchy, preservation and decomposition—as one place where I can finally see some of those support needs operating.

Whether that mechanism belongs to autism, gestalt processing, their intersection, or a relationship for which our existing categories do not yet have adequate language remains open.

n=1.

Me.

That is enough.

Because I can now return to the question with which I began:

Where, exactly, is the support?

It is in the conditions.

It is in the architecture.

It is in the tools through which I preserve the field long enough to understand it.

It is in the time between receiving language and having to return language.

It is in access to the source before somebody compresses it for me.

It is in being able to keep something whose relevance I cannot yet explain.

It is in externalising what would otherwise have to remain alive inside me.

It is in the systems I have built so successfully that their support function can disappear beneath the competence they enable.

And that leaves me with a different question from the one with which I started.

Not:

How can I possibly be Level 2 when I can do all of this?

But:

How much of what gets called my independence is support that has become so continuous, ordinary, and successfully incorporated into my environment that nobody recognises it as support?

Even me.

Perhaps the apparent contradiction was never between disability and competence.

Perhaps it was between two ways of measuring the same life.

One looks at the finished performance and asks what I can do.

The other looks at the entire system and asks what makes doing it possible.

I know which one interests me now.

Because the competence from which someone might infer that I do not require substantial support is occurring under conditions of substantial support.

The support works.

That is not evidence against the support need.

It may be the clearest evidence I have of what the support need actually is.

And underneath all of it remains the thing I recognised when I first turned towards signal and noise.

Before the hierarchy.

Before the decomposition.

Before I know what matters and what can be released.

Too many things remain alive.

Until they don’t.

Movement VII — The Diagnostic Signal

Eventually the signal-to-noise problem turns back on the diagnosis itself.

I have been trying throughout this series to understand something apparently straightforward: what does Level 2—requiring substantial support mean for me?

Language keeps appearing.

It was there when I was diagnosed under DSM-IV, when the presence of scripts became evidence that language was present and helped place me on the Asperger’s side of the diagnostic boundary. It was there again when DSM-5 reorganised that boundary and my profound difficulties with language—social language, pragmatic language, everyday language—became part of determining the level of support I required.

And now it appears for a third time.

Because the closer I look at what I understand as gestalt processing, the harder it becomes to know where gestalt processing ends and autism begins.

Not because I think they are the same thing.

I do not know that.

The problem is more fundamental.

Autism is substantially identified through observable behaviour. A clinician encounters a person, gathers developmental history, observes communication and interaction, administers instruments, listens to accounts from the person and sometimes from others, and asks whether the resulting constellation meets diagnostic criteria.

The movement is something like:

processing architecture → lived experience → observable behaviour → clinical description → diagnostic criterion → autism

By the time we arrive at autism, something enormous has been compressed.

Necessarily so.

A diagnosis has to reduce. It gathers a sprawling, uneven, context-dependent human life into a category sufficiently stable to communicate something useful about it.

That does not make the category false.

But it does make me curious about what disappeared during the compression.

Because some of the behaviours through which my autism becomes legible are also behaviours I recognise when I describe my gestalt processing.

Scripts.

The relationship between parts and wholes.

Difficulty producing language before enough of the whole has resolved.

The strange instability of main idea and supporting detail before informational weighting settles.

The need for context.

The delay between knowing and being able to say what I know.

The social consequences when another person requires decomposition before decomposition is available.

These things enter the diagnostic field as autistic characteristics.

Then I turn the question around and ask why I experience them.

And the answer can become circular remarkably quickly.

Why do I do this?

Because I am autistic.

How do we know I am autistic?

Among other things, because I do this.

There is nothing inherently wrong with that circle if what I am doing is classification. A constellation of observable characteristics can identify a useful category without the category itself being the mechanism producing every characteristic within it.

But classification and explanation are different operations.

That distinction matters enormously here.

I can say that a particular relationship with language is part of how my autism was diagnostically recognised. I cannot therefore assume that autism explains the cognitive architecture producing that relationship with language.

The diagnostic label is downstream of the observation.

And now gestalt processing enters upstream.

Perhaps.

That perhaps has to remain.

I am not proposing that autism is actually gestalt processing underneath. I am not proposing that diagnostic criteria have mistakenly classified gestalt processors as autistic. I do not have evidence for either claim, and my own experience could not establish them anyway.

But I cannot solve the problem in the opposite direction either.

I cannot cleanly separate my autism from my gestalt processing by returning to behavioural criteria when some of the behaviours contained within those criteria may themselves be expressions of the processing architecture I am trying to separate from autism.

The possible confound is already inside the signal.

For me, the narrow claim is enough:

When I examine the behaviours that constitute diagnostic evidence for my own autism, I cannot always determine whether I am observing autism independently of gestalt processing, because some of those diagnostic behaviours may also be observable consequences of the processing architecture I am trying to describe.

And suddenly I am back with cassette tapes, VHS, slides, summaries, and copies of copies.

The diagnosis is another transformation.

Not a degraded copy in the simple analogue sense. Not an inaccurate one. Certainly not a useless one. Autistic names something real and consequential about my relationship with the world, and Level 2 names support needs I am trying, finally, to understand from inside.

But the diagnosis is not the original field.

It is a representation produced from it.

Certain things have been selected as signal. Others have been given less weight. Distinct phenomena have been gathered beneath common headings. Behaviour has been made legible enough to classify.

And perhaps I have been making the same mistake with my diagnosis that I resist when somebody sends me the slides.

I have been asking the compression to contain information that existed only in the field from which it was made.

Autism can tell me where a recognisable constellation exists.

Level 2 can tell me that substantial support is required.

Neither necessarily tells me what processing architecture makes that support necessary.

So the quarter-turn changes the question again.

Not:

What does my diagnosis explain?

But:

What information did my diagnosis compress?

And what becomes visible if I go back to the original?

Movement VI — Can’t Ignore Noise

There is a phrase in the research that catches me because I recognise the behaviour immediately.

Can’t ignore noise.

Yes.

And then almost as quickly:

No.

Or, more precisely, perhaps—but I am no longer certain that ignore describes what is happening.

This is the difficulty of reading research from inside an experience that has usually been measured from outside. The observed phenomenon can be perfectly accurate whilst the mechanism implied by its description feels slightly displaced.

Someone gives me information that does not matter to the task.

I continue processing it.

From outside, the interpretation is straightforward: I have failed to filter irrelevant information. The signal is available, the noise is present, and my attention has failed to suppress the noise efficiently enough.

That may sometimes be exactly what happens.

But there is another possibility I recognise much more intimately.

I do not yet know that it is noise.

Those are not the same problem.

In the filtering account, informational status is already established:

This is irrelevant → I cannot inhibit it.

In the account I am trying to describe:

Its relevance remains unresolved → I cannot yet safely classify it as irrelevant.

The behaviour at the surface can look almost identical.

I continue attending to the detail.

I include something another person would omit.

I hesitate before answering.

I appear unable to find the point.

I hold possibilities open after someone else has already discarded them.

But from inside, I may not be struggling to suppress something whose irrelevance I understand. I may still be determining what kind of information it is.

The kiwi problem makes this beautifully small.

Ten kiwis weigh 190 grams each. What is their total weight?

Then comes the extra information: five of the kiwis are smaller than average.

Noise.

The answer remains 1,900 grams.

But the additional sentence becomes noise only because the task has already established the relationship within which it carries no useful weight. I know what I am being asked to calculate. I know that the mass of each kiwi has already been stipulated. Individual variation cannot change the result.

The question makes the sentence disposable.

Remove the question and tell me only that ten kiwis weigh 190 grams each and five are smaller than average.

Now what?

The second fact remains alive.

Perhaps you are about to ask about variation. Perhaps the individual weights matter. Perhaps “190 grams each” means an average rather than a stipulated individual mass. Perhaps the apparent contradiction is itself the point.

I cannot yet know.

The information has not changed.

Its status has.

That distinction matters to me because it describes something I experience repeatedly before a gestalt resolves. Information remains alive not because I have decided that everything is equally important, but because the hierarchy through which importance becomes knowable has not yet stabilised.

Then the gestalt arrives.

And the change can be astonishingly fast.

Something that had been occupying enormous cognitive space becomes irrelevant. Five possibilities collapse into one. Details that seemed impossible to release suddenly require no effort to release at all. The field reorganises, informational weights settle, and what had been an undifferentiated demand becomes structure.

I can tell you the main idea.

I can summarise.

I can discard.

I can move.

From outside, perhaps I have finally succeeded in filtering the noise.

From inside, the noise has finally become noise.

This is where n=1 becomes important to me—not as an apology for the limits of what I can claim, but as protection against claiming the wrong thing.

I do not know whether this is how autistic / gestalt attention works.

I do not know whether another autistic / gestalt processor who appears unable to ignore irrelevant information is experiencing anything remotely similar.

I know something smaller.

When I encounter descriptions of autistic people as unable to ignore noise, I recognise the observable behaviour. I do not always recognise the mechanism presumed beneath it.

That gap matters.

Because if we mistake unresolved relevance for failed inhibition, the support we offer may work directly against the processing need. We remove information. Simplify the field. Redirect attention. Tell the person what matters.

Sometimes that may help.

Sometimes we may be taking away the very relationships through which what matters would have become knowable.

The behaviour cannot tell us which.

And this is where the quarter-turn becomes methodological rather than merely personal.

Behaviour tells us what happened at the surface. It does not uniquely identify the processing architecture that produced it.

I can look as though I cannot ignore the noise.

Sometimes I am still waiting to know what noise is.

Movement V—The Copy of a Copy

I knew what a copy of a copy meant before information became weightless.

Cassette tape made the lesson audible.

Take an original recording and duplicate it onto another cassette. The copy is recognisably the same recording, but it is not the same signal. There is a little more hiss. A little less frequency information. The noise floor rises. Copy that cassette again and the losses accumulate.

VHS made the same thing visible.

Original to copy. Copy to another copy. Each generation softened the image, introduced noise, lost detail, changed colour, destabilised edges. Eventually there came a point where something that had unquestionably existed in the original could no longer be recovered from the descendant.

The information had not moved intact.

It had passed through a transformation.

In my first years in multimedia forensics, this distinction mattered enormously. The gold standard was the true and exact copy of an original. Not something that represented what the original contained. Not someone’s description of it. Not a reconstruction assembled from what another person considered important.

The relationship to the original mattered.

Provenance mattered.

Because every generation away from the source introduced another place where information could be lost, altered, obscured, or mistaken for something that had been there all along.

I have been thinking about that again.

I was explaining to a colleague why the apparently helpful phrase I’ll send you the slides is often much less helpful to me than people expect.

The slides are already a copy.

They are not the meeting. They are not even necessarily the presentation. Someone decided what belonged on them, what could remain in speech, what needed emphasis, what everyone presumably knew already. Then the presentation happened around them—questions, explanations, disagreements, examples, departures, repetitions, moments when something apparently minor became important because of what somebody else said next.

Then I receive the slides.

The information looks clean.

That may be precisely the problem.

With an old cassette, the hiss told me that I was listening to a degraded signal. VHS generation loss announced itself in the picture. But informational copies can lose context without displaying the loss.

Meeting → presentation → slides → colleague’s notes → summary → me.

At every generation somebody performs a transformation.

Something survives.

Something disappears.

And unlike analogue noise, what disappears is not random. It is conditioned by another person’s judgement about what mattered.

This is where the problem becomes Bayesian for me.

I am not merely receiving fewer data. I am receiving data that have already passed through somebody else’s priors—their expectations about relevance, context, importance, and what could safely be discarded.

They have weighted the field.

Then they give me the compression.

But I do not know their weighting function.

A copy of a copy can therefore preserve noise whilst deleting signal. Worse, it can remove the contextual information I would have needed to discover which was which.

Less information can create more uncertainty.

This is why the usual accessibility instinct towards simplification can sometimes work backwards for me.

Summarise it.

Reduce it.

Give me the important bits.

But important is exactly the problem under investigation.

If the gestalt has resolved, compression can be wonderful. Once I understand the structure, enormous quantities of information can fall away. I can summarise ruthlessly because I know what the summary must preserve.

Before that point, someone else’s summary can become another object I have to reverse-engineer.

What did this come from?

Why did they retain this?

What was beside it?

Was this statement in the original, or is it an inference?

What disappeared between generations?

I find myself reaching instinctively for the old forensic principle.

Show me the original.

The email thread rather than someone’s account of it. The report rather than the slide about the report. The quotation in context rather than the paraphrase. The source beside the interpretation. Give me provenance. Give me enough of the chain that I can tell what happened to the signal on its way here.

Not because I need everything forever.

Because I need to know what can safely disappear.

Perhaps that is another part of what substantial support means for me: access to information sufficiently close to its source that I can establish its structure before someone else’s compression becomes my starting point.

Sometimes the whole is the accommodation.

I do not need less information.

I need enough information to know what less information would be.

Cathy and I began this week with a question that has been following much of my recent writing: what do we actually mean when we say evidence? Not simply whether evidence exists, but who gets to decide what counts as evidence, who is permitted to produce it, and whose observations are allowed into the record. Around gestalt language processing, those questions have become particularly sharp. Again and again, people ask, Where is the evidence? Yet gestalt processors are describing our experiences, parents are describing their children, clinicians have accumulated decades of observation, and researchers across different disciplines and countries have encountered related patterns. At some point, the question begins to change. If all of these people are telling you something is happening, what would you need before you were prepared to believe them?

For me, that question is inseparable from my own life. I am autistic, with substantial language and pragmatic support needs, yet I can sit with Cathy and have this conversation, teach high school, write, research, and speak publicly. Those things can look contradictory only if the support underneath them disappears from view. I remain a gestalt processor. I can take the whole that arrives and disassemble it into increasingly precise language when the conditions and supports allow me to do so. Under enough emotional or informational load, however, I can lose access to that granularity—sentence structures become unavailable, then smaller structures, sometimes until I am relying upon a handful of familiar scripts. That is not regression into some earlier kind of person. It is the same architecture operating with different levels of available support.

This led us towards another way of asking about evidence: what became possible? I came to occupational therapy relatively late, at a time when there were few obvious language services for an autistic adult who already had a career and children. A creative occupational therapist eventually led me towards theatre and improvisation. Where is the randomised controlled trial proving that improv theatre would improve my pragmatic communication? I do not have one. What I have is the life afterwards. Conversation became easier. Teaching became possible. Standing in federal court and being cross-examined about scientific evidence became possible. Refusing something I did not want to do became easier. Eventually, even performing in a musical became possible. Nobody needed me to disappear in order for those things to happen. The evidence is partly in what the support made available.

Cathy brought the same question into her work with children and families. Parents know things because they are there: they see how their children communicate across days and years, what changes when they feel safe, what supports communication, what closes it down, and what becomes possible in relationships where someone is genuinely curious about the person in front of them. Clinicians and educators accumulate another kind of knowledge through repeated encounters. And autistic people bring knowledge from inside the phenomenon itself. None of these perspectives is infallible, but dismissing them because they do not resemble a controlled clinical trial does not make the resulting position more scientific. It simply narrows in advance the kinds of reality science is prepared to notice.

By the end of our conversation, the scale had widened considerably. Gestalt processors are finding one another across countries, cultures, disciplines, languages, and writing systems, and the similarities are difficult to dismiss as an internet fashion. Psycholinguists, philologists, speech and language practitioners, teachers, parents, and gestalt processors may enter through different doors, but we keep encountering pieces of the same problem. That does not mean every claim made about gestalt language processing has been proven, nor that frameworks such as NLA should be exempt from empirical scrutiny. It means something more foundational: evidence does not begin only when somebody designs an intervention study. It begins with noticing, describing, listening, comparing, and asking what changes because this knowledge exists. Cathy and I kept arriving back there. The evidence is distributed amongst us. The harder question may be who has been taught to recognise it.

Movement IV — The Support Became Infrastructure

There was no support to disappear.

I had no IEP. No accommodations. No classroom aide recognising where language or information had become too much, no clinician explaining why I used scripts, no one arranging the field differently because my relationship to it was different.

I was simply expected to manage it.

When autism finally acquired a diagnostic name for me, in my mid-twenties, the architecture was DSM-IV. Asperger’s disorder still existed as something separate from autistic disorder, and language mattered enormously to the distinction. My scripts—language that was present, available, apparently evidence of language having developed—were part of what placed me on the Asperger’s side of that diagnostic boundary.

I now understand those scripts through my gestalt language processing.

That makes what happened later particularly interesting.

DSM-5 removed the Asperger’s/autism division and brought both beneath Autism Spectrum Disorder (ASD). Now the diagnostic architecture asked a different question. And there, language mattered again.

My substantial struggles with language—including social and pragmatic language—became central to the Level 2 support-needs designation I carry now.

The processing had not suddenly changed.

The diagnostic meaning assigned to it had.

Under one architecture, my scripts helped establish that language was present. Under another, my functional relationship with language helped establish how much support I required.

Language was inside the diagnostic signal both times.

And social language, pragmatic language—these phrases can make the thing sound narrower and more clinical than it is. Pragmatic language is asking a question at work. Understanding what somebody actually wants from an email. Knowing what matters in a meeting. Working out how much context another person needs. Recognising what has been implied but not said. Repairing the misunderstanding. Knowing when the answer I have given is not the answer the other person believes they asked for.

It is everyday language.

So perhaps it should not surprise me that much of the infrastructure I eventually built around myself is infrastructure for language too.

I am old enough to remember the world before most of it.

That gives me something like an accessibility archaeology—not a nostalgic history of gadgets, but the possibility of looking backwards and asking, with each technological arrival:

What became externally supportable here?

Word processing meant language no longer had to leave me in the order it arrived. I could put something down, move it, return to it, discover that the whole was not what I had first thought, and reconstruct the parts around what had finally become clear.

Email changed time.

Someone’s language could remain still.

I could reread it. Determine what they were asking. Notice what I had missed. Formulate a response without another human being waiting inside the silence for language to become available.

Search changed context. Missing information no longer had to remain missing merely because I did not possess it at the moment I discovered I needed it. I could move outward until the relationship appeared.

Then the smartphone gathered the scaffold into one object and put it permanently within reach.

Notes. Photographs. Messages. Calendars. Maps. Search. Screenshots.

A photograph could preserve something before I knew why I needed it. A screenshot could hold a piece of information whose weight had not yet resolved. Searchable histories meant context did not have to survive perfectly inside memory. Multiple windows allowed things that might belong together to remain simultaneously present whilst I worked out whether they did.

None of this necessarily gave me less information.

Often it allowed me to retain more information safely outside myself.

That distinction matters.

If too many things remain informationally alive before the gestalt resolves, then one form of support is somewhere for those things to remain alive without all of them having to remain equally alive inside me.

Technology became that somewhere.

And now there is AI.

Its usefulness to me is not primarily that it gives me answers. It gives me another place in which an unresolved informational field can remain available whilst I work upon its relationships. I can bring the source, the second source, the screenshot, the thing somebody told me, the thing that does not fit. I can correct. Compare. Return. Turn the object. Ask what changes if this detail is signal rather than noise.

Language can remain provisional whilst meaning resolves.

For someone whose relationship to language has been diagnostically significant across two different diagnostic architectures, that does not feel incidental.

It looks increasingly like support.

But because these technologies are ordinary, their support function disappears into their ordinariness. Everyone uses email. Everyone searches. Everyone carries a phone. Everyone takes screenshots. Increasingly, everyone has access to AI.

The ubiquity of a tool tells us nothing about the function it performs for a particular person.

I was not a supported autistic child who became an unsupported autistic adult.

I was an unsupported child who kept encountering technologies through which cognitive operations I had always been required to perform internally, immediately, and alone could gradually be moved into the environment.

So I am increasingly suspicious of the story that says I became more independent.

I became extraordinarily good at supporting myself.

I built the scaffold as I went.

Eventually, it became infrastructure.

Movement III — Main Idea and Supporting Details

I write this IEP Reading Goal.

Not these exact words every time, but some version of them appears constantly in special education:

Given a grade-level passage, identify the main idea and supporting details.

It is so ordinary that I had stopped seeing what it asks a student to do.

We call it reading comprehension. We teach strategies for it. We measure accuracy, collect data, establish baselines, write annual goals. If a student cannot reliably identify the main idea, we recognise that something is happening between encountering the language and extracting what matters from it.

Then I turn the object a quarter-turn.

What, exactly, are we asking the student to do?

A passage presents an informational field. Sentences arrive one after another, each comprehensible perhaps, each carrying something that might matter. The student must establish hierarchy among them—weight propositions against one another, recognise relationships, distinguish central claims from evidence, determine which details support which ideas, assign lower weight to information that does not constrain the emerging whole.

Then something even more difficult.

The student must determine what can safely be discarded.

Only after those operations can the passage become compressible:

The main idea is…

That little sentence at the bottom of the worksheet is not retrieval. It is the output of an enormous reduction.

The field has become a hierarchy.

Seen through signal and noise, the familiar school task becomes stranger still.

The main idea is something like the highest-weight signal—not necessarily a sentence sitting intact somewhere in the passage, but the representation that best preserves what the passage is doing. Supporting details matter because of their relationship to that signal. An incidental detail becomes low-weight information for this particular interpretive task.

The categories are relational.

Which means the student has to construct them.

And when a disabled student struggles to do that, we support the operation.

We highlight important sentences. We chunk passages into smaller sections. We provide graphic organisers. We reduce the amount of text. We offer headings and guiding questions. Sometimes we explicitly tell the student which information matters.

These can be excellent supports.

But I have begun wondering what, exactly, each one supports.

If I highlight the important information before the student reads, I have reduced the signal-to-noise problem—but I have done so by performing part of the weighting operation myself. I have already decided what matters.

The student is now processing my compressed representation of the field.

That may provide access to the content. It may be exactly what is needed at that moment. But it is not necessarily the same thing as supporting the student to establish informational weight.

The distinction matters to me because I recognise the operation.

Not because I believe every student with a main-idea goal is a gestalt processor. Not because difficulty identifying main ideas somehow demonstrates autism. The causes of that difficulty can be many.

I recognise something narrower.

We already understand that constructing informational hierarchy can be a disability-related support need.

We understand it well enough to write it into an IEP.

And then something peculiar happens.

The student becomes an adult.

There is no moment at which the underlying cognitive operation necessarily changes, but the language around it does. Difficulty determining what matters becomes poor organisation. Needing additional context becomes inefficiency. Holding too many possibilities open becomes overthinking. Difficulty extracting the important points from a meeting becomes inattention. Needing time to work out what has informational weight becomes failure to prioritise.

Eventually it becomes professionalism.

Personal responsibility.

Something I should simply know how to do.

The support need has crossed an institutional boundary and changed names.

That recognition turns me back towards my own Level 2 designation differently.

Perhaps the question is not whether I somehow outgrew the need to construct hierarchy from an informational field. Perhaps I became extraordinarily skilled at building conditions around myself in which that construction could happen.

The child gets a graphic organiser.

The adult is expected to organise herself.

And somewhere between those two sentences, support becomes invisible.

Movement II – Signal to Noise Ratio

The problem begins before I know there is a problem.

Information arrives.

A sentence. A gesture. A number. Something somebody said ten minutes ago. The order in which two things happened. A detail in a document that does not quite agree with another detail three pages later. The thing someone emphasised. The thing they conspicuously did not. Something I already know that changes what this new thing might mean.

They arrive without necessarily arriving with instructions about their relative importance.

Too many elements remain informationally alive.

This is difficult to describe because, from outside, it can look like a familiar autistic problem: difficulty filtering irrelevant information. Too much detail. Too little hierarchy. Attention captured by things that should have been discarded whilst everybody else gets on with finding the point.

But irrelevant is doing an extraordinary amount of work there.

Irrelevant to what?

Before the gestalt resolves, I cannot safely know what carries high evidential weight and what is merely contextual; what is redundant and what is corroboration; what contradicts something else and what only appears to; what will remain a peripheral detail and what, three pieces of information later, will reorganise the whole.

I cannot always know what belongs to the source and what somebody has inferred from it.

I cannot always know what can safely be discarded.

That last word matters.

Safely.

Because information does not arrive for me already divided into signal and noise. Its weight emerges through relationship. A detail that is noise for one question can become signal for another. Something apparently incidental can remain incidental—or it can be the pebble that turns the entire field.

So when I talk about signal-to-noise ratio, I have to correct myself almost immediately.

Noise is not necessarily a property of the information.

It is a relationship between information and an organising question, purpose, or whole.

If someone asks me the total mass of ten kiwis and tells me that each weighs 190 grams, the fact that five are smaller than average is noise for that calculation. I can discard it because the problem has already supplied the structure that establishes its irrelevance.

But remove the question.

Now tell me about the kiwis.

Five being smaller than average remains alive because I do not yet know what I am going to need it for.

That difference feels very close to the centre of my processing.

The difficulty is not necessarily that I know something is irrelevant but cannot inhibit my attention to it. Sometimes I have not yet established that it is irrelevant. The hierarchy has not settled. The informational field remains open, and whilst it remains open, premature disposal carries risk.

This is where gestalt processing enters for me—not as everything arriving simultaneously in some magical flash, but as the dependence of the parts upon a whole that gives them their weight.

Once that whole resolves, something changes.

Information that had remained insistently alive can become quiet. Relationships stabilise. Some pieces move towards the centre; others become supporting structure; still others become genuinely disposable. I can decompose. I can sequence. I can summarise. I can tell you what matters.

From outside, it may look as though I have finally filtered the noise.

From inside, something different has happened.

The information has become filterable.

That distinction changes what support might mean.

If the problem were simply failure to suppress irrelevant information, support might reasonably focus on reducing distraction, removing detail, simplifying the field, telling me which bits matter.

But if the difficulty lies partly in unresolved informational weighting, removing information can remove precisely the relationships through which relevance would have become knowable.

Less is not automatically clearer.

Sometimes I need enough of the field for the field to settle.

And this is the first place where requiring substantial support begins to acquire a shape I recognise—not help because I cannot comprehend the information, but support around the conditions under which information can become organised enough to use.

The support need isn’t simply filtering the noise.

It is establishing a sufficiently reliable signal-to-noise ratio that decomposition can proceed.

Movement I—Requiring Substantial Support

There is something almost accusatory in requiring substantial support when the person carrying the designation appears not to require very much at all.

I know how my life reads from outside.

I teach. I research. I write books and essays. I work with complicated systems and complicated ideas. Before education, I spent thirty three years worth of eight hour days in forensic science, in a profession where uncertainty had to be examined rather than wished away, where the distinction between what the evidence contained and what someone believed it contained mattered enormously. I can stand in front of a classroom and teach mathematics. I can construct an argument across thousands of words. I can take a problem apart until I find the assumption hiding inside it.

And I am autistic, Level 2—requiring substantial support.

There is a temptation, when those two things are placed beside each other, to resolve the apparent contradiction in one of two directions.

Perhaps the diagnosis overstates my disability.

Or perhaps I need to produce the evidence—to inventory everything difficult, expose enough private struggle, make the cost sufficiently visible that Level 2 becomes believable again.

I am interested in neither.

This is n=1. I am not trying to establish what Level 2 means for autistic people, nor what substantial support ought to look like. I am asking something much narrower and much more difficult to answer from a diagnostic manual:

What does requiring substantial support mean in my life?

Part of the difficulty is that we tend to imagine support as a person.

Someone prompts. Someone explains. Someone reminds. Someone simplifies the instructions, organises the task, helps initiate it, interprets the social situation, redirects attention, supervises completion. Support becomes visible through the presence of a supporter, and substantial support becomes imaginable as more of that presence.

I do not move through most of my life with such a person beside me.

So where is the support?

I think the question becomes distorted if I begin by counting the things I can do independently, because independently has already smuggled the conclusion into the measurement.

There is a difference between capacity, supported performance, and cost.

Capacity asks whether I can do something.

Performance asks what I actually accomplish under particular conditions.

Cost asks what the accomplishment requires—before, during, and after—and what happens when the conditions that make it possible are removed.

Those three can produce very different portraits of the same person.

If I write something complex, the writing is mine. If I successfully navigate a complicated professional problem, that competence is mine. Support does not make the accomplishment counterfeit. But neither does accomplishment establish the absence of support.

A wheelchair does not make the distance travelled unreal.

Glasses do not make the reading someone else’s.

And support does not cease to be support because it has become so thoroughly incorporated into the conditions of a life that nobody notices it being used.

That is the direction in which I want to turn the object.

Away from the question how disabled do I appear?—a question that makes my outward performance the measure of an internal support need—and towards something environmental, relational, almost ecological:

Under what conditions does my functioning become possible?

That question does not begin with deficit.

It begins with architecture.

What must remain stable? What must be available? What must be externalised? What forms of time, information, technology, predictability, language, memory, and relational access have accumulated around me? Which of them have become so ordinary that I no longer register them as support at all?

And what happens when they disappear?

I am not ready to answer those questions yet.

For now, I want to leave the contradiction intact: the apparently competent woman and the diagnostic phrase requiring substantial support, occupying the same body without either cancelling the other.

The first quarter-turn is simply this:

What has to be present around me for the competence people observe to become available?

Prologue—Where Is the Support?

I have been carrying the phrase Level 2—requiring substantial support for years without really knowing what to do with it.

Not doubting it. Not trying to escape it. Something stranger than that.

I could not see where the support was.

I can see the life built around it easily enough. I teach. I write. I research. I move between bodies of information that are sometimes enormous, hold relationships across them, find patterns other people have missed, and make language from those patterns. I have spent much of my working life being trusted with complexity. None of this cancels autism, of course, but it makes requiring substantial support difficult to locate if support means what we usually imagine it means—another person beside me, prompting, translating, simplifying, helping me do what I cannot do alone.

And then I began looking somewhere else.

Not at what I can do, but at what has to be present for me to do it.

That quarter-turn changes the object.

My phone is there. My computer is there. Search is there. Email is there. Screenshots, notes, searchable histories, original documents, multiple windows, preserved conversations, the ability to return to something hours or days later. Increasingly, AI is there—not replacing thought, but holding information still whilst I work out its relationships.

And underneath all of these I began recognising the same problem.

Too many things remain informationally alive.

Before the gestalt resolves, I cannot safely know what carries the greatest weight and what merely provides context; what is redundant and what will become the detail that reorganises everything; what belongs to the source and what somebody else has inferred from it; what contradicts something I already know; what can be ignored; what can be discarded.

The usual language for this is filtering.

Signal and noise.

But that description assumes something I am increasingly uncertain I possess at the beginning of the process: a stable distinction between the two.

I cannot cut through the noise if I do not yet know what the noise is.

And that recognition sends me back towards Level 2 with a different question. Perhaps substantial support does not mean someone helping me understand because I cannot understand. Perhaps some part of my support need lies further upstream—in preserving, arranging, comparing, contextualising, and holding an informational field long enough for its internal weighting to resolve, until signal becomes signal and noise can finally become noise.

This series is an attempt to stay with that possibility.

But it is no longer happening alone.

As these pieces began resolving, another series resolved with them: What Remains Alive Between Us, over in Sensual Residue. Not afterwards. Not as application, companion commentary, or the personal version of an AutSide theory.

At the same time.

That distinction has become important enough that I have started thinking of the form itself as superposition.

The same object occupies more than one interpretive state. On AutSide I turn towards information, support, diagnosis, language, technology, signal, and noise. On Sensual Residue I turn towards another person inside precisely that same field—care, desire, reciprocity, misrecognition, relational meaning, what happens when the thing whose significance has not yet resolved has already touched me.

Neither series explains the other.

Neither is the original.

They remain simultaneously available, and something becomes visible in the interference between them that neither could quite produce alone.

There is an intellectual ancestor hovering around this experiment for me: Robert Anton Wilson.

I keep wondering what Wilson might have done with Substack.

Not what he would have written about on Substack, nor whether he would have liked the platform, but what someone so interested in reality tunnels, model agnosticism, recursive uncertainty, and the instability of the observer might have done with a publishing form no longer constrained by the singular procession of a printed book.

A book eventually makes you choose an order.

This chapter, then that chapter.

This argument, then its complication.

Even a book designed to destabilise certainty remains bound between two covers, its multiplicities carried sequentially through the same physical object.

But here?

I can publish two movements on the same morning.

The same object can turn simultaneously in two places.

A reader can encounter Requiring Substantial Support on AutSide and Requiring Someone Else on Sensual Residue—not as argument and response, but as two states of the same investigation. Signal/noise can exist beside signal/desire. Infrastructure beside reciprocity. The forensic demand for an original beside the relational recognition that between-us may have no master tape.

So much possibility.

And the form begins doing something the argument itself requires.

I do not have to decide too early which interpretation is the main idea and which is supporting detail.

I can let both remain alive.

That matters particularly because this investigation is deliberately n=1. I am not proposing a new account of autism, nor claiming that my processing architecture explains somebody else’s. I am taking the diagnostic object attached to me—autistic, Level 2, requiring substantial support—and turning it, a quarter-turn at a time, to ask what those words describe when observation begins inside the person rather than outside her.

That immediately creates another superposition.

I am autistic.

I am also a gestalt processor.

And the closer I look, the harder it becomes to separate the diagnostic signal of autism from the processing phenomena I understand as gestalt. Some of the behaviours through which autism becomes clinically legible may also be downstream expressions of the very processing architecture I am trying to examine.

The diagnosis can tell me that support is needed without necessarily telling me what cognitive mechanism makes that support necessary.

I do not want to collapse that uncertainty.

Not into this is autism.

Not into this is gestalt processing.

Not even into some satisfying synthesis in which both turn out to have been different names for the same thing.

Perhaps they are not.

Perhaps the uncertainty is information.

Perhaps keeping both models alive long enough to see where they converge, interfere, contradict, and reorganise one another is not a failure to reach the argument.

Perhaps that is the argument.

And so the form repeats the processing.

AutSide and Sensual Residue.

Autism and gestalt processing.

Signal and noise.

Self and field.

Support and competence.

Not binaries waiting for me to select the correct side, but simultaneous models whose relationships may tell me more than premature resolution could.

From here, I want to follow the support itself—through signal and noise, main ideas and supporting details, copies of copies, the absence of childhood accommodations and the expectations placed upon autistic adults, analogue life and digital infrastructure, the apparently simple act of deciding which piece of information matters and which can safely be allowed to die.

Sensual Residue will be turning the same objects at the same time.

There, another person is already in the field.

Because perhaps I have been asking the wrong question.

Not:

How can someone who does all of this require substantial support?

But:

How much of what I can do is possible because I have spent a lifetime building the conditions under which I can do it?

And, simultaneously, somewhere beside it:

What changes when some of those conditions cannot exist around me because they can only exist between us?

I do not need to make one question wait for the other.

That is the experiment.

The competence is real.

So is the support.

The autonomy is real.

So is the interdependence.

Let them remain alive at the same time.

Introduction

I came to these questions at a quarter turn.

I had been following the work of Dr. Amy Maslen-Miller—the Samoan Scientist—because one part of her work intersects strangely with one part of my own life: her attention to nutrition, culture, and Type 2 diabetes amongst Samoan communities in New Zealand met my own long nutritional journey, and my attempts to understand what allows my body to thrive within an industrial food environment that I increasingly experience not as neutral abundance, but as a particular historical arrangement of food.

The subjects are not the same. I do not need them to be.

I recognised the shape.

More recently, Dr. Maslen-Miller posted about presenting science through connection first—locating the scientist, their relationship to the community, and why their research matters there. That was the turn. I want to acknowledge her explicitly because her framing provided the point of contact from which this piece moved, even though what followed moved somewhere else.

It moved into autism and gestalt processing.

Because autism and gestalt processing research are not something I encounter from one stable position. I am autistic. I am a gestalt processor. I write and conduct research about autistic and gestalt experience. I read other people’s research and review manuscripts. I listen to researchers present what they believe they have learned about people like me. I work in education, where research becomes practice and practice lands on autistic and/or GLP children. I chair an IRB, where proposed knowledge about human beings arrives before it has been made and somebody has to ask what its making will require of them.

Researcher. Research subject. Reader. Reviewer. Practitioner. Community member.

The positions keep crossing.

And somewhere amongst those crossings, three questions have begun to stay with me.

Who are you?

What is your relationship to us?

What changes for us because this knowledge exists?

Not a checklist.

A three-body problem.

Three Bodies – The Complete Gestalt

Who are you?

Not your affiliation.
Not the letters after your name.
Not the laboratory
that makes the question respectable.

Who are you
when you enter this field?

What do you already believe
when you call something autism?

What have you learned
to notice?

What have you learned
not to see?

—

What is your relationship to us?

Not how many autistic people
were in your sample.

Not whether one of us
read the questionnaire.

Relationship.

Who was there
when the question became a question?

Who could tell you
you had misunderstood?

Who could change the direction
rather than comment upon it?

Who are we to you
before we become data?

—

What changes for us
because this knowledge exists?

Not what changes
for the literature.

Not the citation count,
the grant,
the next study,
the instrument refined enough
to measure us more precisely.

For us.

What becomes easier?

What becomes possible?

What are we permitted
to refuse?

Who listens differently?

Who teaches differently?

Who stops trying
to make us disappear?

What becomes ours
that was not ours before?

—

Who are you?

What is your relationship to us?

What changes for us
because this knowledge exists?

Three bodies.

None holding still.

Each changing the gravity
of the other two.

And there I am too—

researcher,
autistic person,
gestalt processor,
teacher,
reviewer,
the one being described
and the one doing the describing—

unable to stand outside the field

and no longer convinced

that anyone should.

Field Notes

The researcher does not disappear

There is a particular trick that scientific writing can perform: by the time I reach the methods section, nobody appears to be doing the looking.

Participants were recruited. Measures were administered. Responses were coded. Associations were identified.

The researcher recedes, and the findings begin to sound as though they were simply there.

But somebody decided what counted as a question.

Somebody decided what counted as autism, communication, impairment, success, reciprocity, behaviour, language. Somebody selected what could be measured and therefore what could become visible. Somebody encountered what remained and decided what it meant.

Who are you?

The question puts the researcher back into the field.

Not to invalidate the research.

To complete the information.

Relationship is not participation.

Autistic people, gestalt processors, the neurodivergent, we can be everywhere in the research and still have almost no relationship to its production.

We can supply the blood, complete the inventories, enter the scanner, answer the interview questions, perform the tasks, provide the language samples, wear the sensors, and become the n from which knowledge about autism is extracted.

Presence is not relationship.

So the second question goes further.

What is your relationship to us?

Did autistic knowledge shape what you thought was worth asking?

Could a gestalt processor tell you that the phenomenon you thought you were measuring was not the phenomenon occurring?

Could we change the interpretation?

Could we change the question?

Could we stop the study going somewhere it should not go?

Relationship changes what can happen between bodies.

That is precisely why it matters.

Knowledge does things

The third question is the one I cannot put down.

What changes for us because this knowledge exists?

Research on autism and gestalt processing produces consequences even when it does not produce benefit.

It produces categories. Interventions. Assessments. Professional practices. Educational expectations. Clinical language. Ways for parents to understand their children. Ways for teachers to interpret behaviour. Ways for institutions to decide what an autistic person needs, lacks, can do, cannot do, should become.

Knowledge moves.

So I want to follow it.

If this finding becomes accepted, where does it travel?

Whose power increases?

What becomes possible for autistic people?

What becomes possible to do to autistic people?

Those are not the same direction.

Gestalt processing makes the problem difficult to ignore

Perhaps this is why the three questions settled so quickly for me.

I do not experience knowledge as a collection of independent pieces that can be cleanly separated from the relations amongst them.

The whole matters.

Context matters.

Relationship changes meaning.

Who is speaking changes what is being said. Where they stand changes what they can see. What they intend to do with the knowledge changes the meaning of producing it.

The three questions therefore refuse to stay separate.

Who you are changes your relationship to us.

Your relationship to us changes what you can know about us.

What you intend that knowledge to do changes the questions you choose to ask.

And what the resulting knowledge does to our lives changes the relationship again.

Three bodies.

There is no neutral place outside their gravity from which we can simply be observed.

At the quarter turn

This began somewhere else.

That matters to me too.

Maslen-Miller’s work did not need to be about autism for me to recognise something in it. Her questions crossed mine at an angle, and the recognition arrived before I knew what I was recognising.

That is familiar.

It is gestalt recognition.

The categories remain different. The histories remain different. The communities remain different. I do not collapse one into another.

I turn the thing.

And suddenly a relation that was invisible from one position becomes obvious from another.

This time, what appeared was not an answer.

It was three questions.

Who are you?

What is your relationship to us?

What changes for us because this knowledge exists?

I think I will be carrying them for a long time.

Opening — This One Came Backwards

Usually, these pieces move the other way.

Something begins on AutSide as architecture. Language. Memory. Assessment. A pattern I can see before I quite know what it means.

Then, eventually, it follows me across to Sensual Residue.

There, the architecture acquires a body.

This one came backwards.

I was writing about attraction—about the strange persistence of desire across ageing, and its equally strange capacity to disappear when almost everything that once seemed to constitute it remains.

The face remains.

The voice remains.

The history remains.

I can identify every part.

And still, the whole is gone.

That stopped me.

Because I know this problem.

I know what it is to take something whole and begin naming what is inside it. To find the words, the pieces, the relationships between them. To become increasingly precise about everything that appears to constitute the thing.

And then discover that precision has not preserved it.

The parts can all be there.

The whole does not have to return.

Somewhere between Sensual Residue and AutSide, the question turned.

What if this is not only something desire does?

What if I have been watching it happen in language all along?

The Whole Arrives First

A gestalt is not necessarily a collection of smaller things arriving together.

That sounds obvious to me now.

It did not always.

When we talk about gestalt language processing, particularly through Natural Language Acquisition, there is an understandable tendency to begin with what can be heard. The longer utterance. The stored phrase. The prosody that holds it together. Later, the mitigations. The pieces becoming increasingly mobile until language can be generated with greater flexibility.

There are stages.

There are parts.

We can watch some of the movement between them.

But the whole came first.

And first may mean something more consequential than chronology.

The whole may be doing something.

A phrase arrives with a voice attached to it. With rhythm. Affect. A place. A person. What happened immediately before it. What usually happens afterwards. The feeling in my body when it was heard. The relationship in which it became meaningful. The purpose it served before purpose could necessarily be named.

These are not decorations around the language.

They may be part of what makes the language available as language.

Later, we can take the whole apart.

We can identify words. Syntax. Pronouns. Intonation. Communicative functions. We can locate repetitions and substitutions and mitigations. We can become extraordinarily precise about what the gestalt contains.

That precision matters.

But something happens very easily once the pieces are visible.

We begin to imagine that we have discovered how the whole was made.

Word plus word.

Meaning plus prosody.

Language plus context.

Person plus memory.

As though the gestalt had been assembled somewhere out of independently meaningful components and delivered to us already glued together.

I am no longer sure that is the right direction of travel.

Sometimes the whole is what tells the pieces what they mean.

A word inside one gestalt is not necessarily the same thing, experientially or functionally, as the same word inside another. A phrase carried in one person’s voice may not be interchangeable with the identical phrase spoken by someone else. Remove the relationship, the affect, the timing, the bodily state, and all the lexical material may remain whilst something essential has disappeared.

The pieces survive the analysis.

That does not mean the architecture survives with them.

We can know exactly what is there.

We can name every part.

And still not know what made them cohere.

The Seduction of Taking Things Apart

Taking things apart works.

That is what makes this difficult.

Teachers do it. Speech and language therapists do it. Psychologists do it. Researchers do it. I do it.

We separate words from utterances, phonemes from words, grammatical structures from language, communicative functions from conversations. In classrooms, we break complex tasks into smaller steps. In research, we isolate variables so that something impossibly entangled can become measurable.

There is nothing inherently wrong with this.

Often it is how we learn.

The danger begins slightly later.

Once the parts are lying separately on the table, we forget that we put them there.

Someone left a comment on one of my recent videos that inadvertently gave me a rather lovely example of this. They had noticed that sometimes I say cannae and sometimes I say can’t. There was a faint quality of having caught something—an inconsistency, perhaps. If I say cannae, why don’t I always say cannae?

The trouble was that I hadn’t really noticed I was doing it.

I knew, obviously, that both forms existed in my language. I could tell you what they meant. Asked directly, I could probably produce either one.

But that wasn’t what was happening when I spoke.

Cannae belonged somewhere.

So did can’t.

And once the commenter separated the two forms and placed them beside one another, I could suddenly feel something that had been almost invisible whilst I was speaking. Cannae was connected to a relational whole—to particular voices, histories, rhythms, people, ways of being with people—and sometimes that whole connected with the new now strongly enough that cannae came forward.

Other times, the present context connected elsewhere.

And can’t arrived.

I wasn’t standing somewhere behind the sentence selecting between two lexical alternatives.

The relationship to the present seemed to be doing some of the selecting.

There is an older piece of evidence that keeps returning to me here.

In 1994, Simon Baron-Cohen and Ruth Staunton studied autistic children growing up in England whose mothers had non-English accents. They found that 83.3% of the autistic children acquired their mother’s non-English accent, compared with 12.5% of their non-autistic siblings. They interpreted this through the social framework available to them—autistic children might identify less strongly with peers and therefore converge less towards peer-group phonology.

I wrote about this recently in The Dharma of Gestalt Processing: Hidden in the Evidence, because the observation becomes rather different if we rotate it slightly. What if the family voice is not merely phonology that failed to be replaced by the socially dominant accent? What if the voice itself belongs to the relational whole through which language became meaningful—the people who comforted, laughed, reassured, regulated, repeated, and welcomed the child into language?

Perhaps the accent is carrying more than accent.

Perhaps cannae is too.

Once I extract cannae and can’t from the moments in which they occurred, I can compare them beautifully.

Same semantic territory.

Different phonology.

Different dialect.

Different histories.

I can analyse the pieces until there is almost nothing left to analyse.

But that does not tell me why one arrived then.

For that, I have to put back something analysis removed.

The person.

The relationship.

The old now touching the new one.

The whole.

This is the seduction of decomposition. It gives us extraordinary precision, and precision feels so much like explanation that we can easily mistake one for the other.

The parts really are there.

The analysis really can be correct.

But the boundaries around those parts may belong to the analyst rather than to the phenomenon.

And once we forget who made the cuts, it becomes very easy to believe the whole was assembled along the same lines.

Teaching the Pieces Back

Then we teach the pieces back.

This is where the question becomes practical.

A child produces a long, recognisable gestalt. The adults around her hear words inside it. Perhaps the utterance is mitigated. Perhaps there are pieces beginning to move. An SLT can identify pronouns, verbs, noun phrases, communicative functions, places where language might become more flexible.

This is useful information.

And then there is a temptation.

Teach the pieces.

Model the smaller unit. Practise the pronoun. Substitute the noun. Elicit the verb. Build enough flexible components and eventually, surely, the child will be able to assemble them herself.

Sometimes she can.

But notice the assumption hiding inside the intervention.

We have taken something apart and presumed that development will run backwards through our analysis.

The classroom does this constantly too.

Break it down.

It is one of the most persistent pieces of advice we give teachers working with students who are struggling. Make the task smaller. Reduce the number of steps. Teach each skill separately. Practise until secure. Then put everything back together.

And often this is excellent teaching.

I do it.

But I have also watched a student understand the large thing and become increasingly lost as I make it smaller.

That can look almost impossible from the outside.

Surely one step is easier than six.

Surely one paragraph is easier than an essay.

Surely identifying the operation is easier than solving the whole problem.

Surely the individual words are easier than the utterance.

Unless the whole was carrying information the pieces needed.

A maths problem, for example, may become intelligible because I can see where it is going. The structure tells me what the individual operations are doing there. Remove that structure and give me one isolated step at a time, and I may now have to remember a sequence of apparently arbitrary instructions whilst also inferring the thing those instructions once made visible.

The task has become smaller.

The cognitive demand has not necessarily become lighter.

Language can do something similar.

A gestalt may carry prosody, relationship, affect, anticipated response, and communicative purpose together. Extract a pronoun from it and the pronoun is certainly smaller than the original utterance.

But smaller is not the same as simpler.

Now the learner may have to understand the isolated form without the relational architecture that previously made it meaningful.

Then we hand her the next piece.

And the next.

Eventually we ask her to put them together.

From our perspective, we have scaffolded the task.

From hers, we may have removed the scaffold that was already there.

This does not mean we should never decompose language or classroom tasks. It does not mean NLA should somehow preserve every stored gestalt intact, or that explicit instruction is inherently hostile to gestalt processing.

It means something narrower.

We need to know what the whole was doing before we decide the pieces are easier.

Because sometimes the learner is not struggling to assemble the parts.

She is struggling because we took away the thing that told the parts how to belong.

Analysis and Synthesis Are Not Inverse Operations

There is a simple assumption underneath all of this.

If I can take something apart, I should be able to put it back together.

Analysis in one direction.

Synthesis in the other.

Except they are not necessarily inverse operations.

A teacher can establish that a student can perform A, B, and C separately. Each skill can be demonstrated. Each box can be checked.

Then ABC is presented.

Nothing.

This is often interpreted as a problem of transfer. Or working memory. Or task persistence. Or insufficient mastery of one of the component skills. We go back through the pieces looking for the weak one.

Sometimes that is exactly what we should do.

Sometimes all the pieces really are there.

What is missing is not another piece.

It is the organisation.

An SLT can encounter the same thing from the other direction. A child can be prompted to produce a word, a pronoun, a grammatical form, perhaps even several increasingly flexible combinations. The elicitation succeeds.

Then the spontaneous context arrives.

The language does not.

Again, the pieces have not disappeared. We just demonstrated them.

What changed was the whole.

And the reverse can be equally misleading.

A learner may fail to produce an isolated component when asked for it directly, yet use something carrying that component successfully inside a meaningful activity. If our assessment privileges the isolated demonstration, we may conclude that the capacity is absent.

But perhaps isolation changed the task.

This matters because decomposition feels evidential.

If I have watched a student perform every step individually, it seems reasonable to say that she can do the whole thing.

If she cannot do the whole thing, something must therefore have gone wrong with effort, attention, compliance, memory, motivation.

And if she can do the whole whilst struggling to demonstrate one of its supposed prerequisites separately, we become suspicious in the other direction.

How can you do the harder thing if you cannot do the easier one?

The question assumes the hierarchy.

Part.

Part.

Part.

Whole.

But for a gestalt processor, difficulty may not increase neatly in that direction.

Sometimes ABC makes A possible.

The whole provides orientation. It constrains possibility. It tells me what kind of thing I am doing, where I am inside it, what this piece is doing here, and what should come next.

Remove the whole and A may become harder, not easier.

This is why successful analysis does not establish successful reconstruction.

We can identify every component of a performance.

We can teach every component.

We can measure every component.

And still not have explained how the performance becomes available.

Decomposition gives us knowledge.

It does not guarantee reversibility.

Once a whole has been taken apart, putting the pieces beside one another is not necessarily the same thing as putting the whole back.

Accuracy at the Level of the Fragment

There is another way we take language apart.

We listen.

A student says something. A client says something. Perhaps the exchange is difficult. Perhaps something unexpected happens.

Later, we remember the pieces.

The words.

The tone.

The pause before the answer.

The repetition.

The thing she said immediately afterwards.

The fact that she looked away.

We may remember all of it accurately.

That matters.

But accuracy at the level of the fragment does not guarantee fidelity to the whole.

This is where I think teachers and SLTs need to be particularly careful, because listening is already interpretation. We are not recording devices. We hear language through what we know about communication, development, behaviour, disability, pragmatics, intention, and the particular person in front of us.

Usually, that expertise helps.

Sometimes it makes a reading feel more certain than it is.

A student says, I don’t care.

The words are real.

Perhaps the flat prosody is real too. Perhaps she turns away. Perhaps five minutes earlier she refused the task.

Every part of the account can be accurate.

Then the parts begin to cohere for the listener.

She doesn’t care.

A whole has been made.

Perhaps it is the right one.

But perhaps I don’t care was stored language for a moment in which caring had become unbearable. Perhaps it meant I cannot do this in front of you. Perhaps the phrase came forward because another relational whole had been activated. Perhaps it was the language available when more precise language was not.

I don’t know.

That is precisely the point.

The professional listener may know exactly what was said without yet knowing what whole produced it.

This is how a transcript can be accurate and the reading wrong.

And I do not want teachers or SLTs to become Wrong Readers.

Not because they should distrust their observations. Not because every utterance conceals some inaccessible meaning that only the gestalt processor can ever know. Communication would become impossible if that were the standard.

Something simpler.

Leave a little space between observation and interpretation.

She said, “I don’t care.”

That is an observation.

She did not care.

That is a reading.

The distance between those sentences can become very small.

Sometimes it disappears entirely.

For a gestalt processor, I think we have particular reason to keep it visible.

Because the language we hear may carry more context than remains audible once we have extracted the words. Prosody may belong to another moment. A phrase may carry relational history. What sounds like a complete proposition may be one available route through a much larger internal whole.

The words matter.

The tone matters.

The behaviour matters.

Write them down.

Notice them carefully.

Just do not assume that because the fragments are accurate, the whole we construct from them must belong to the speaker.

We can hear every word correctly.

And still read the person wrong.

The Listener Makes a Gestalt Too

There are always at least two gestalts in the room.

Mine.

And yours.

You do not receive the whole that exists inside me. You cannot. You receive what becomes available across the distance between us—words, prosody, movement, expression, timing, silence, whatever history we already share.

Then you make meaning.

Of course you do.

That is communication.

You hear what I say through everything you already know. Through your history. Your expectations. Your emotional state. Your professional training. Your culture. Your understanding of autism. Your experience of me.

Even through what happened ten minutes ago.

None of this makes the listener unreliable.

It makes the listener human.

The teacher listening to a student is making meaning.

The SLT listening to a client is making meaning.

The psychologist observing behaviour is making meaning.

So is the student.

So is the client.

So am I.

Meaning does not travel intact from one nervous system into another.

Something crosses.

Something is received.

Something is made.

Usually, this happens so seamlessly that we never need to think about it. Your gestalt and mine are sufficiently coherent with one another. I say something, you understand something close enough to what I meant, and the conversation carries on.

The seam disappears.

Until it doesn’t.

The difficulty begins when the meaning I made from your communication becomes indistinguishable, in my own mind, from the communication itself.

You said this.

Perhaps I did.

And what you meant was this.

Perhaps.

But something has happened between those two sentences.

Interpretation.

A whole has formed in the listener.

And because that whole feels coherent—because the words fit, the behaviour fits, the context fits—it can become difficult to remember that coherence was produced partly on this side of the exchange.

The listener’s gestalt feels discovered.

It may actually have been made.

This does not mean every interpretation is equally plausible, or that intention is unknowable, or that communication dissolves into competing private realities.

It means we need to preserve one small distinction.

What you said.

What I understood.

Most of the time they travel together.

When they separate, that little distance matters enormously.

Especially when only one of us knows it is there.

Your Whole Becomes My Parts

This is where it becomes strange.

I give you a whole.

Not deliberately. I am not standing inside myself thinking here comes the gestalt. I speak. I respond. I do something. Perhaps several things happen across an exchange.

You receive it.

Then, later, you take it apart.

You said this.

Yes.

Then you did this.

Yes.

You used that word.

I did.

And before that, you said this other thing.

Also true.

The pieces are mine.

I recognise them.

Then something shifts.

So you meant this.

And suddenly I don’t know what we are talking about anymore.

From your side, the movement may feel almost seamless. A led to B. B makes sense beside C. D clarifies something ambiguous in A. The pieces form a pattern.

A whole emerges.

But it is not necessarily the whole I gave you.

It is a new one.

Whole₁ → A + B + C + D → Whole₂

The difficulty is that Whole₂ can look remarkably like an explanation of Whole₁.

It contains my words.

My actions.

My tone.

My history.

Nothing has been invented.

And yet something has been made.

Then Whole₂ is handed back to me.

Explain.

This is the moment I think we routinely misunderstand.

Because I may recognise every piece and still not recognise the object you have assembled from them.

Yes, I said that.

Yes, I did that.

Yes, those things happened in that order.

But no, I do not know how to explain the meaning you are asking me to defend.

That meaning was not inside my whole.

It emerged inside yours.

And now I am being asked to travel backwards through your analysis—to take the fragments you extracted from my experience, discover how you connected them, reconstruct the gestalt that formed for you, and then explain why I produced something I did not produce.

From the outside, my difficulty can look suspicious.

The evidence is right there.

A.

B.

C.

D.

How can I possibly deny the whole?

But I am not necessarily denying the pieces.

I may be telling you something much more precise.

Those are mine.

That isn’t.

“Yes, I Said That”

Yes.

I said that.

There is a peculiar helplessness in being shown your own words and recognising them completely.

I remember saying them.

I may even remember the room, the conversation, what happened immediately before, the feeling of the exchange as it moved through me.

I am not disputing the record.

Then comes the question.

So why did you mean this?

And I cannot find it.

Not the answer.

The premise.

I search backwards through what I remember, trying to locate the thing you are describing. I can find my words. I can find what happened. I can find pieces of what I was feeling and what I was trying to do.

I cannot find your whole.

That absence can look extraordinarily suspicious from the other side.

Perhaps I am contradicting myself.

Perhaps I am avoiding the obvious implication.

Perhaps I lack insight into my own behaviour.

Perhaps I am being deliberately literal to escape responsibility.

Perhaps I am manipulating language.

Perhaps I simply do not want to admit what everyone else can plainly see.

There are many available explanations for my failure to recognise the meaning being attributed to me.

Very few begin with the possibility that the meaning was never mine.

This matters because recognising the pieces is not the same as recognising the whole someone else has assembled from them.

I can accept responsibility for what I said.

I can accept responsibility for what I did.

I can hear that something hurt you.

I can discover that my words produced a meaning I did not anticipate and still take seriously the fact that the meaning occurred.

None of that requires me to say that your reconstruction was what existed inside me all along.

Sometimes I genuinely cannot make that move.

And the harder I try, the stranger my own words can begin to feel.

They have been removed from the whole in which I knew them and returned carrying another architecture.

They are familiar.

The arrangement is not.

So I keep saying the only thing I know to be true.

Yes, I said that.

And perhaps what the listener hears is evasion.

But sometimes it is precision.

I recognise the evidence.

I do not recognise the reading.

Then Power Enters the Room

So far, this could simply be misunderstanding.

Two people.

Two gestalts.

A seam between them that did not hold.

Given enough time, curiosity, and mutual trust, perhaps we can go back. You can tell me what you heard. I can tell you what I remember. We can put both wholes beside one another without requiring either to disappear.

Sometimes that is enough.

Then power enters the room.

Teacher and student.

SLT and client.

Psychologist and patient.

Administrator and student.

Parent and child.

Employer and employee.

Clinician and person being assessed.

Now the two meanings do not necessarily have equal weight.

One of us may have the authority to decide what happened.

To grade.

To diagnose.

To document.

To discipline.

To determine whether behaviour was intentional, oppositional, manipulative, inappropriate, unsafe.

To write the sentence that someone else will read later.

This changes the exchange.

Because Whole₂ is no longer merely your interpretation of my Whole₁.

It can do things.

Your meaning can become a consequence for me.

The teacher’s reconstruction can become a referral.

The clinician’s reconstruction can become a diagnostic observation.

The psychologist’s reconstruction can become evidence about insight, affect, or credibility.

The administrator’s reconstruction can become discipline.

The parent’s reconstruction can determine whether the child is believed.

At that point, my inability to recognise your whole is no longer an interesting difference in meaning-making.

It is dangerous.

I need you to understand me.

More precisely, I need you not to misunderstand me in a way that can hurt me.

And that need changes what happens next.

The conversation is no longer only about what I meant.

Now I am also trying to work out what you think I meant, how firmly you believe it, what evidence you think supports it, what consequences might follow, and what combination of words might make your gestalt loosen enough for mine to become visible again.

That is a different language task.

A different relational field.

A different body.

Power has entered the meaning-making process.

And once it does, misunderstanding can become threat.

Power, Threat, Meaning

There is a framework for this.

Power Threat Meaning.

PTMF.

I am cautious about bringing frameworks into experiences too early. Once a framework has named the pieces, it becomes very easy to see only the pieces it knows how to name.

But here the sequence has already appeared.

Power.

Threat.

Meaning.

Power is not simply who has the louder voice.

It is who can make an interpretation consequential.

A teacher can decide that a student’s words constitute defiance. An administrator can decide that an explanation is an excuse. A clinician can describe a person’s account as lacking insight. An SLT can decide what communicative function an utterance served.

The interpretation travels further because of who made it.

That creates threat.

Not necessarily physical threat. Not necessarily even conscious fear.

Something more immediate.

This meaning can do something to me.

And now I have another problem to solve.

I began with:

What did I mean?

Now I need to know:

What do you think I meant?

How did you get there?

Which pieces are you using?

How certain are you?

What can happen to me if you remain certain?

What do I need to say now?

Meaning-making has become protective.

For a gestalt processor, that can be an enormous demand.

I am trying to remain connected to my original whole whilst simultaneously entering yours. I need to understand the architecture you have built from my fragments, locate the point where our meanings separated, and generate language precise enough to reopen that separation.

Quickly.

Because power rarely experiences itself as waiting.

The teacher needs an answer.

The clinician has another question.

The meeting has forty minutes.

The administrator wants to know what happened.

Why did you say it?

What were you trying to do?

Do you understand why that was inappropriate?

The clock enters the field too.

And somewhere in all of this, my original task—telling you what happened for me—has almost disappeared.

I am no longer only communicating.

I am managing your interpretation of my communication.

Trying to make meaning quickly enough to reduce the threat created by the meaning you already made.

PTMF asks us to look differently at what happens next.

Not:

What is wrong with her?

But:

What power is operating here?

What threat does that create?

What meaning is she making of what is happening?

And then, crucially:

What is she doing to survive it?

Because once misunderstanding becomes consequential, the response to being misunderstood becomes part of the story too.

And that response is very easy to read wrong.

Defending a Gestalt That Was Never Yours

And then comes the impossible instruction.

Explain yourself.

It sounds simple.

It isn’t.

Because I am no longer being asked only to explain my own meaning. I am being asked to defend myself against yours.

To do that, I have to keep several things available at once.

My original whole.

The pieces you extracted from it.

The way you appear to have connected those pieces.

The meaning that connection produced for you.

The consequence attached to that meaning.

And somewhere inside all of this, the language that might undo it.

That is an extraordinary generative demand.

I have to enter a gestalt I never possessed.

I have to reverse-engineer it.

Perhaps A and B became important to you whilst C, which organised the whole for me, barely registered. Perhaps my prosody meant something different to you. Perhaps a phrase I experienced as relationally stored sounded deliberate and newly generated. Perhaps you connected something I said today with something I said three weeks ago and produced a continuity I never experienced.

I have to find that architecture before I can challenge it.

Then I have to build another one.

Quickly.

No, that’s not what I meant.

But now you want to know what I did mean.

Fair enough.

Except the original whole did not necessarily arrive with an analytic explanation attached.

It arrived whole.

Now I must take it apart under pressure, translate it into language that makes sense inside your framework, identify where your reconstruction diverged, and offer an alternative sufficiently coherent that you will relinquish the one you already have.

And I have to do this whilst knowing that failure may itself mean something.

Too slow.

Evasive.

Contradictory.

Overexplaining.

Literal.

Defensive.

Not taking responsibility.

The demand for clarification can therefore become the condition that makes clarification less available.

This is where the difference between capacity and availability becomes difficult to ignore.

Perhaps I could explain it tomorrow.

Perhaps I could write it.

Perhaps I could explain it to someone with whom the field is different.

Perhaps the language exists.

But you need it now.

And now is the problem.

Threat does not make generative language cheaper.

It makes every new sentence carry more weight.

So I reach for what is available.

A familiar phrase.

A repetition.

A fragment.

That’s not what I meant.

I don’t know.

I already told you.

Perhaps silence.

And now those responses enter your gestalt too.

More pieces.

More evidence.

The harder I try to defend myself against a whole that was never mine, the more material I may give you from which to build it.

When Language Becomes Evidence Against You

Then the language changes.

Of course it does.

Generative language is expensive. Under threat, it can become more expensive still.

For a gestalt processor, that may mean older language becomes easier to reach than language that has to be assembled now. Scripts return. Repetition increases. A phrase begins doing more work than its literal meaning can carry.

Or there is less language.

I don’t know.

That’s not what I meant.

I already told you.

Again.

Again.

Perhaps I agree because agreement appears to end the questioning.

Perhaps I go silent.

Perhaps I leave.

Perhaps I become angry because every attempt to correct the reconstruction seems to strengthen it.

Perhaps my account changes—not because the event changed, but because I am searching for another route into a whole I cannot make available under these conditions.

And now something dangerous happens.

The response to the threat becomes evidence for the interpretation that created the threat.

She’s defensive.

She keeps changing her story.

She can’t explain herself.

She’s refusing accountability.

Each observation may even contain something true.

I may be defensive.

My language may be changing.

I may be unable to explain myself.

I may be refusing to agree with the account being placed in front of me.

But none of those things establishes that the original interpretation was correct.

They tell us something about what is happening now.

That distinction matters.

A student questioned by an administrator is not communicating under the same conditions as the student who spoke in the classroom.

A client challenged by a clinician is not producing language inside the same relational field as the one in which the original language occurred.

The act of investigation has changed the conditions.

And then we use the communication produced under those changed conditions to infer what happened before them.

This is particularly dangerous when we assume that more pressure produces more reliable language.

Think carefully.

I need you to tell me the truth.

That doesn’t match what you said before.

Try again.

For some people, perhaps pressure sharpens recall.

For a gestalt processor whose generative language is becoming less available as threat increases, pressure may do almost the opposite.

The demand for greater precision can produce less flexible language.

The demand for consistency can produce increasingly desperate attempts to find words.

The demand for explanation can make explanation unavailable.

Then the resulting difficulty enters the record as though it were independent evidence.

The field produces a change in language.

The listener observes the change.

And the change is attributed entirely to the person.

This is how language becomes evidence against its own conditions of production.

The Wrong Reader

This is where the Wrong Reader appears.

Not a bad reader.

Not necessarily an unkind one.

That distinction matters.

The Wrong Reader may be extraordinarily careful.

She may have written down exactly what I said. She may remember the sequence correctly. She may quote me verbatim. She may be able to point to the pause, the repetition, the changed wording, the moment I became angry.

She may have receipts.

Every fragment can be accurate.

The error happens somewhere else.

She mistakes the gestalt she produced from the evidence for the gestalt that produced the evidence.

That is a very easy mistake to make because her reconstruction is coherent.

A fits beside B.

B explains C.

C makes the later D suddenly legible.

Once those pieces have cohered, the whole begins to feel inevitable.

Of course that’s what happened.

And once a whole feels inevitable, alternative readings begin to look like evasions from something obvious.

This is why the Wrong Reader does not need bad intentions.

A teacher can care deeply about a student and still become the Wrong Reader.

An SLT can understand gestalt language processing and still become the Wrong Reader.

A psychologist can follow professional standards meticulously and still become the Wrong Reader.

I can become the Wrong Reader.

The problem is not located in a particular profession or neurotype.

It is a problem of interpretation forgetting itself.

I heard.

I separated.

I connected.

I made meaning.

Then somewhere along the way, I made meaning from what you said became this is what you meant.

The first statement leaves a seam.

The second closes it.

For gestalt processors, that closure can be particularly consequential because the reconstruction may be made from language whose relational architecture is no longer visible. The words survive. The context is summarised. The behaviour is recorded.

The whole that made those pieces meaningful may already be gone.

Then another whole takes its place.

And because the new whole contains accurate pieces, challenging it becomes extraordinarily difficult.

How do I say your reading is wrong when every quotation is correct?

How do I explain what has disappeared between the transcript and the meaning?

Perhaps this is the sentence I need to keep returning to:

Accuracy at the level of the fragment does not guarantee fidelity to the gestalt.

Not in therapy.

Not in classrooms.

Not in assessment.

Not in relationships.

The Wrong Reader is not the person who failed to hear the words.

Sometimes she heard every word.

She just assembled them into the wrong whole.

When the Wrong Reading Becomes the Record

Then someone writes it down.

That changes things.

Clinical notes.

IEPs.

Behaviour reports.

Disciplinary referrals.

Psychological assessments.

HR records.

Case histories.

She stated…

She admitted…

She was unable to explain…

She became agitated when challenged…

Each sentence may begin with an observation.

But once interpretation enters the record, it acquires durability.

I know something about this from another life.

I spent decades in forensic science, where language could become evidence in a particularly literal way. I testified. I was cross-examined. And however many times I did it, there was a particular terror in the testimonial experience—the knowledge that another person could take something technically precise I had said and move it into an architecture I did not recognise.

The other side had a theory of the case.

My side might have one too.

Sometimes the evidence supported one.

Sometimes it supported another.

And sometimes the evidence supported no conclusion at all.

That last position could be surprisingly difficult to hold.

But, Dr Hoerricks, you just said that…

Yes.

I did.

And then the fragment would return.

A statement about what an image could establish. A limitation. An observation. Something technically accurate within the boundaries in which I had offered it.

Now placed beside another fragment.

Then another.

A different whole beginning to form.

Cross-examination taught me something I did not yet have language for: an accurate statement can be moved into an inaccurate architecture without changing a single word.

The transcript remains correct.

The meaning shifts around it.

Institutional records can do something similar, except the reconstruction survives the room.

A teacher writes that a student “refused to explain her behaviour”.

Perhaps she did.

But that sentence does not preserve the relational field in which explanation became unavailable.

A clinician writes that a client “became agitated when challenged.”

Perhaps she did.

But challenged has already organised the scene. So has agitated. Neither tells the next reader what meaning the client was trying unsuccessfully to protect.

A psychologist records that someone “demonstrated limited insigh.”

An administrator writes that a student “changed her account.”

An employer notes that an employee “became defensive when concerns were raised.”

The fragments harden.

The interpretation becomes grammar.

And then someone else reads it.

This is where the Wrong Reader becomes more than an interpersonal problem.

The next teacher may encounter the report before she encounters the student.

The next clinician may know that the client has “limited insight” before asking her what happened.

The next administrator may already know that this student “changes her story.”

The reconstruction arrives first.

Then the person walks into it.

The old reading has become part of the new relational field.

And now the GLP is no longer only trying to make herself understood by the person in front of her.

She may be speaking into a gestalt that was assembled months or years ago by someone who is not even in the room.

The wrong reading has become the record.

And records have power.

The Recursive Field

And then it happens again. The next person has read the record. Not maliciously. She is preparing. A teacher reads the previous teacher’s notes because she wants to understand the student before school begins. An SLT reads the previous assessment. A therapist reads the intake. An administrator reviews the disciplinary history. This is what responsible professionals are supposed to do.

But the record arrives before the person.

Limited insight. Task refusal. Becomes defensive when challenged. Inconsistent reporting. Difficulty accepting responsibility.

These phrases begin organising attention. Then the GLP arrives and something ambiguous happens. Perhaps she pauses before answering. Perhaps she says I don’t know. Perhaps a familiar script appears, or she corrects the teacher with more intensity than the moment seems to require. Perhaps she gives one account, then finds different language for it ten minutes later. The new reader does not encounter any of this without context. She already has a gestalt.

And the new fragments fit.

There it is.

This is how an interpretation becomes recursive. The previous field produced a reading. The reading became a record. The record enters the next field, which means the next field is not entirely new. Expectation has arrived first.

Threat may arrive earlier too, because the student has a history. She recognises the question, the tone, the particular shift in an adult’s face when an answer has somehow become the wrong answer. She may already know what happens next. Language becomes expensive sooner. Scripts become available sooner. The body protects itself sooner.

The student described as defensive enters a field already organised around the expectation of defensiveness, encounters the threat embedded in that expectation, and becomes defensive.

The record appears to have predicted her.

Perhaps it helped produce her.

This is where an individual misunderstanding becomes systemic. No single professional has to behave terribly. No one has to decide to misread an autistic person. Each person can act reasonably from the information available to her—read the file, notice the pattern, document what happened, pass the information forward—and still, across those individually reasonable acts, a wrong reading can acquire institutional momentum.

The GLP becomes increasingly difficult to encounter outside it. Every new interaction supplies more fragments, every fragment is interpreted through the whole already present, and every resulting interpretation makes that whole feel more certain. What began as somebody’s reconstruction gradually acquires the appearance of longitudinal evidence.

The circle closes.

And somewhere inside it is a person whose original gestalt may never have been recovered at all.

The Wrong Reader no longer needs to remain in the room.

She has reproduced herself.

What This Means for SLT and Classrooms

None of this means we stop taking things apart. That would be an easy conclusion, and the wrong one. Decomposition is necessary for teaching, speech and language therapy, assessment, and research. I need to be able to isolate a mathematical operation, an SLT needs to be able to notice what is happening inside an utterance, and a researcher needs variables sufficiently bounded to study. The question is not whether we analyse. It is what we believe analysis has given us.

We need a little more epistemic humility around the pieces.

When a student produces language we do not understand, we can ask what whole might have given those words meaning before deciding what they mean in isolation. When a phrase appears somewhere unexpected, we can become curious about where else it belongs. Whose voice does it carry? What relational history travels with it? What happened before it became available? What does this new now share with an older one?

The same applies when we break down a classroom task. Smaller may be easier. Often it is. But when a learner becomes more confused as the task becomes more granular, we should resist assuming that she has exposed some previously hidden deficit. Perhaps we removed the organising structure she was using. Put the whole back. Show where the pieces live. Let the destination remain visible whilst we work on the road towards it.

And when we reconstruct meaning from a student’s language, the distinction becomes more consequential. We can bring our interpretation back as a question rather than a verdict. This is what I understood—does that fit? We can return the pieces to context rather than asking the learner to defend them after we have rearranged them. We can allow her to reconstruct her own whole, even when it takes longer, arrives in writing, requires another person, or does not resemble the explanatory sequence we expected.

This matters particularly when the interaction has already become difficult. An inability to explain under pressure is information, but it is information about the entire condition in which explanation is being requested. It does not automatically establish poor insight, dishonesty, avoidance, or lack of understanding. If generative language is becoming less available as threat rises, asking the same question more forcefully may simply move the language further away.

Sometimes the most useful intervention is not another prompt.

It is to reopen the field.

Change the time. Change the person. Change the mode. Give the student space to write. Return tomorrow. Restore enough safety that the original whole has some possibility of becoming available again.

And then listen carefully to what returns.

The practical shift is small. We still analyse. We still teach components. We still document observations. We still make professional judgements.

We simply remember that the pieces are not innocent.

Someone made the cuts.

Someone made the reconstruction.

And before we ask a gestalt processor to recognise the whole we have built, we should make room for the possibility that hers was organised differently.

The Larger Error

There is a larger problem underneath all of this, and it is not peculiar to speech and language therapy, education, or autism. Much of what we call knowledge depends upon taking things apart. Science isolates variables. Psychology operationalises constructs. Assessment creates tasks in which one capacity can supposedly be observed without too much interference from another. We control context, reduce noise, separate signal, and make boundaries around phenomena complicated enough that we could never study them whole.

We have to do this.

I am not arguing against analysis. I am arguing against forgetting what analysis is.

Every time we isolate something, we make a cut. Language from relationship. Memory from cue. Capacity from conditions. Behaviour from threat. Speaker from listener. We hold one thing still long enough to examine it and, if we are careful, learn something real about it.

The danger comes when the boundary we created begins to look like a boundary the phenomenon already possessed.

The map becomes dangerous when we forget which boundaries were drawn by the cartographer.

Gestalt processing makes this unusually visible because the consequences of decomposition can be immediate. Remove a phrase from the relational field in which it became meaningful and we may still have all the words. Remove a skill from the larger task in which the learner understands its purpose and we may still have a perfectly valid assessment item. Remove behaviour from the threat conditions surrounding it and we may still have an accurate behavioural description.

Nothing has necessarily been falsified.

Something has been lost.

Then we study what remains.

This is where precision can become treacherous. The more accurately we describe the isolated component, the easier it becomes to believe that we have described the phenomenon itself. We measure memory under one cue condition and call the result memory. We observe language under one relational condition and call the result language. We elicit a skill in one assessment architecture and call the result capacity.

The measurement may be excellent.

The unit may be wrong.

And once we have several excellent measurements, another temptation appears. Put them together. Reconstruct the person from the scores, the observations, the language sample, the behaviour report, the diagnostic criteria, the fragments accumulated across settings.

A + B + C + D.

Surely, somewhere inside enough data, the whole must return.

But synthesis does not undo the cuts we made during analysis. Context removed at the beginning does not necessarily reappear because enough decontextualised findings have been assembled at the end.

This is not an argument for abandoning measurement in favour of intuition. It is an argument for remembering what our measurements can and cannot contain.

The whole may exceed them.

For gestalt processors, perhaps that should make us particularly cautious whenever a person seems inexplicably different across contexts. Whenever language appears somewhere and disappears somewhere else. Whenever a student demonstrates something in the whole but cannot produce its supposed component on demand. Whenever behaviour changes dramatically with relationship, familiarity, threat, timing, or mode.

Those differences may be noise.

They may also be the phenomenon.

If the whole organises the meaning of its parts, then context is not merely something surrounding the thing we are studying.

Sometimes context is inside the thing.

And once we cut it away, we cannot assume that putting everything else back together will bring it home.

Epilogue — Give Me Back My Whole

I should probably apologise for the length of this.

Both pieces, really.

The Sensual Residue piece that started this, and this one that came backwards from it into AutSide. They are long. They are recursive. They circle things already said and return to them carrying something slightly different. They spiral.

I know.

I considered breaking this whole into a series. That is often how I work when an idea becomes too large to hold comfortably inside one piece—find the natural fractures, let each section breathe, move through the architecture over several days.

I couldn’t do it here.

Or rather, I could have divided the words.

I wasn’t convinced the gestalt would survive.

There was something about the movement itself that mattered—from whole to parts, from parts to reconstruction, from reconstruction to interpretation, from interpretation to power, from power to threat, from threat back into language. Each return changes what the earlier sections mean. The recursion is not accidental. The spiral is doing some of the work.

So there are two pieces.

Both quite long.

One began in the body.

This one followed the architecture.

And after all these words, I find myself back at an extraordinarily simple moment.

Someone places the pieces in front of me.

You said this.

Yes.

Then this happened.

Yes.

And you did this.

Yes.

So this must be what you meant.

And there it is again.

The distance.

I recognise the pieces.

They are mine.

The whole is not.

Perhaps that is all I have been trying to make visible.

Not that interpretation is wrong. Not that teachers should stop analysing language, or SLTs should stop identifying components, or psychologists should abandon inference, or researchers should stop making the careful cuts that allow complicated things to become knowable.

Not that listeners should somehow receive another person’s meaning without making meaning of their own.

We cannot.

The warning is smaller.

And larger.

Remember who made the cut.

Remember who moved the pieces.

Remember that coherence can be produced by the reader as well as discovered in the text.

And when the person in front of you cannot recognise the whole you have reconstructed from her words, leave open the possibility that her difficulty is not evidence that she does not understand herself.

Especially when you have the power to decide what her difficulty means.

Because that is where misunderstanding changes form.

You heard my gestalt.

You took it apart.

You made meaning from the pieces.

Then you handed your meaning back to me and asked me to explain it.

I couldn’t.

And because you had the power to decide what my inability meant, my inability became evidence.

That is not merely misunderstanding.

That is what happens when the Wrong Reader is given authority over the text.

So yes.

This was long.

It spiralled.

It came back to the same places more than once.

I could have broken it into smaller pieces.

But perhaps, by now, you understand why I didn’t.

Sometimes the whole needs to remain whole.

Going back to school this year has made something visible to me that competence usually hides. My Learning Centre is gone. I no longer have that stable place where relationships, routines, language, and expectations can accumulate over time. Instead, I move between classrooms, pushing in for 40 minutes here, an hour there, supporting a new caseload across multiple curricula. I know the mathematics. That is not the problem. The problem is that I do not yet have the scripts for these students, these teachers, these rooms, these particular versions of the curriculum. I am tending a new script garden—and, in each new field, I am beginning much closer to the beginning than anyone watching my language from the outside might imagine.

That matters when we talk about gestalt language processing and particularly Stages 4–6. There is a seductive assumption that once someone has grammar, language has become generally available: you have language now, so use it. But language does not exist outside context. Put a student into a new classroom, beside someone they have never met, present unfamiliar mathematical language, and tell them to “turn and talk to your partner about normal distributions,” and what exactly are we measuring when they stare silently at one another? They may understand the mathematics. They may have grammar. What they do not yet have is a script for this. I recognise that because, despite everything I can apparently do with language, I experience the same thing. Stages cannot simply describe something possessed by the person. What is available is partly a property of the field.

Cathy made the connection beautifully from her own work. A child arriving at the Speech Den comes with reports, histories, diagnoses, information from parents—but none of those is the child. Knowing begins in the encounter: noticing, becoming curious, establishing routines, learning one another, planting those first seeds from which something relational can grow. And the gestalt begins long before the child crosses the threshold. It contains the morning, the journey, sleep, medication, home, yesterday, previous classrooms, previous failures, previous experiences of being misunderstood. The behaviour we encounter now is only the visible edge of something already underway. A head on a desk does not inherently mean disengagement. Silence does not inherently mean refusal. The surface does not tell us what produced it.

Yet schools are increasingly being organised as though context were expendable. Larger classes, fewer special education teachers, overloaded caseloads, disappearing Learning Centres, brief push-in services—and then we call the resulting arrangement inclusion. Full inclusion may be presented as pedagogical progress, but increasingly there is an economic reality beneath it: there simply are not enough qualified people to provide the structures we once provided. The terrible contradiction is that we are requiring children to become more context-independent at precisely the moment institutions are becoming less willing or able to provide context.

Perhaps that is also why my own Level 2 diagnosis can appear contradictory from the outside. People can hear me speak for thirty minutes, read what I write, watch me teach mathematics, and reasonably wonder where requiring substantial support for language could possibly be hiding. But perhaps they are looking for support in the wrong place. The support is not necessarily another person standing beside me helping me form a sentence. It is everything that has to become sufficiently organised for language like this to become available—the scripts, histories, searchable information, routines, relationships, technologies, and relational fields that allow the whole to resolve. The fluency is real. So is the infrastructure beneath it.

Introduction

Siegel, M., Lord, C.L., Tager-Flusberg, H. et al. Developing a consensus research definition for profound autism using a modified Delphi method. Molecular Autism 17, 28 (2026). https://doi.org/10.1186/s13229-026-00727-y

I was several pages into Developing a consensus research definition for profound autism using a modified Delphi method before the part of the paper that stayed with me stopped being the definition.

It was the geography.

The authors are attempting to solve a legitimate research problem. Autism research has disproportionately represented autistic people who can participate in conventional research protocols, complete standardised assessments, communicate in ways researchers readily recognise, and tolerate the practical demands of being studied. People with substantial intellectual, communicative, adaptive, and support needs have consequently been underrepresented in the literature intended to describe autism as a whole. The paper cites evidence that the proportion of treatment studies including severely affected autistic people has actually declined over the past two decades.

That absence matters.

If a substantial group of autistic people is systematically difficult to recruit, assess, accommodate, or retain in research, then the resulting science will become increasingly precise about the people its methods can reach whilst remaining strangely uncertain about everyone else. A research definition capable of making that population easier to identify across datasets could therefore have genuine value. I have no interest in solving the problem of exclusion by pretending that profound differences in support need do not exist.

The definition produced by Siegel and colleagues is consequently worth taking seriously. Their consensus group identified several characteristics as central: adaptive functioning substantially below age expectations, an inability to perform most activities of daily living independently, and the need for adult supervision to maintain physical and mental health, safety, and wellbeing, accompanied by severely impaired cognitive ability and/or very limited verbal communication primarily used to meet basic needs.

The strongest agreement gathered around support.

Ninety-seven per cent agreed with statements concerning support for basic needs and round-the-clock access to an adult for safety. Adaptive functioning substantially below age expectations received similarly overwhelming agreement. The authors consequently describe continuing access to an adult as a possible “core defining feature” of “profound autism.”

And this is where I began to feel the ground move beneath the definition.

Because a person requiring another human being nearby throughout the day and night certainly tells us something about that person.

It also tells us something about the world into which that requirement falls.

The Geography of Consensus

The first Delphi round included seventy respondents. Sixty-two of them—89 per cent—were from the United States. Only eight respondents represented Canada, Europe, Asia-Pacific, and the Middle East/Africa combined.

The authors acknowledge this. Geographic concentration appears appropriately among the study’s limitations, alongside variation between countries in diagnostic practices, measurement, and access to services. They explicitly recognise that cultural and contextual factors may influence how characteristics associated with “profound autism” are recognised.

I think geography reaches further into this study than a limitation on generalisability.

It reaches into the construct.

We identify the people whose lives require the most collective support from inside a society that has made that support least collective—and then call the consequences profound.

That sentence has been sitting with me.

The United States has no national health service. Access to healthcare, therapies, long-term support, respite, communication systems, supported housing, and personal assistance is distributed through a fragmented architecture of private insurance, Medicaid programmes, educational entitlements, state agencies, charitable provision, family resources, eligibility determinations, waiting lists, and administrative persistence. The precise arrangements vary enormously between states and across the lifespan. What remains remarkably consistent is the amount of labour pushed back towards individuals and families when somebody requires continuing care.

A requirement for “24/7 access to an adult” enters that system carrying an enormous material weight.

Someone has to be the adult.

Someone has to remain available overnight. Someone has to attend appointments, negotiate services, respond when a programme refuses placement, complete forms, contest denials, manage medication, arrange transport, communicate with schools, find clinicians willing and able to provide care, plan for emergencies, and eventually confront the question that shadows many families supporting disabled children into adulthood:

What happens when I cannot do this anymore?

These are consequences of support need interacting with infrastructure.

And once those consequences accumulate around a person for years, separating them becomes extraordinarily difficult.

Where Does “Profundity” Live?

This is the question I wanted the paper to ask more forcefully.

Where does “profound autism” end and “profound unsupportedness” begin?

I am cautious here because there is an easy argument available that I do not think survives contact with the lives under discussion. Social accommodation cannot make every autistic person independently safe. Publicly funded care cannot give someone capacities they do not possess. A national health service cannot guarantee independent communication, eliminate intellectual disability, or make assistance with eating, dressing, bathing, toileting, navigation, or personal safety unnecessary.

Material impairment remains material.

Yet impairment never enters the world alone.

It encounters housing, money, schools, transport, healthcare, communication access, family capacity, employment policy, social services, and other people. Those encounters accumulate. They become developmental environments. They determine which skills receive support, which opportunities become available, how much exhaustion gathers around ordinary activities, whether family relationships have room for anything beyond care, and whether adulthood expands or contracts the person’s world.

The organisation of support determines what impairment is allowed to become.

That distinction matters enormously when adaptive functioning itself becomes part of the definition. Adaptive functioning measures what someone can actually do in the demands of ordinary life. Those capacities matter. They are also necessarily expressed inside environments that provide radically different opportunities for learning, participation, assistance, autonomy, and practice.

A person does not perform adaptive functioning in a vacuum.

Neither does a family perform caregiving in one.

The American Field

Seen from this angle, the American dominance of the Delphi panel becomes intellectually fascinating.

Researchers and clinicians working in the United States encounter autistic people after those people have already passed through American systems. Caregivers describe lives shaped by those systems. Autistic participants arrive carrying their own histories of access and exclusion. The observations are real. The difficulty lies in deciding where their causes reside.

A child requiring extraordinary assistance may produce extraordinary financial strain within a family.

A parent may leave employment because continuous care is unavailable.

A family may become geographically trapped by services.

Siblings may become informal carers.

A young adult may remain at home because supported housing is unavailable, inappropriate, unaffordable, or unsafe.

A person may lose access to community activities because nobody is funded to accompany them.

Behaviour may intensify inside environments poorly equipped to understand communication, sensory distress, pain, uncertainty, or loss of control.

Each consequence can then return to the observer as evidence of the original severity.

The field disappears into the person.

This possibility becomes particularly important because the paper itself contains a moment when the neutrality of classification visibly fractures. In discussing the minimum age for applying the definition, participants expressed concern that insurers or policymakers might use a “profound-autism” designation in young children to restrict access to intensive interventions.

I find that passage remarkable.

The respondents are constructing a research category and simultaneously anticipating what may happen when that category escapes the research paper and enters the American machinery of eligibility and payment.

They understand the danger because classification already does things here.

A category can open a service.

A category can close one.

A score can authorise hours.

A score can remove them.

A diagnosis can make treatment reimbursable, and an insurer’s interpretation of that diagnosis can make the same treatment disappear.

The distance between description and allocation is very small when care is rationed through categories.

The People the Research Has Missed

None of this resolves the original problem identified by Siegel and colleagues.

That problem remains.

Autistic people with the greatest communication, cognitive, adaptive, and support needs have been badly underserved by autism research. The paper is particularly compelling when it describes the methodological machinery producing that exclusion: increased research costs, limited validated assessment tools, researcher unfamiliarity, participation barriers, and the perceived difficulty of including people with substantial intellectual and communication disabilities.

There is an uncomfortable symmetry here.

The people requiring the greatest accommodation to participate in research are excluded because research has difficulty accommodating them.

Their exclusion then produces a shortage of evidence.

The shortage of evidence makes their capacities and needs harder to understand.

That uncertainty encourages reliance upon the measures already available.

Those measures are often precisely the ones they struggle to access.

The paper encounters this problem directly with IQ. Ninety-six per cent of respondents agreed that people fitting the proposed “profound autism” category struggle to complete IQ assessments, and respondents raised concerns about the validity of those assessments for people with significant cognitive and/or language differences. No proposed IQ threshold actually reached the study’s predefined consensus level, although IQ below 50 was ultimately incorporated as one side of an “and/or” criterion.

I recognise something larger in that methodological problem.

When access to the instrument becomes difficult, psychology has repeatedly struggled to distinguish inability to demonstrate a capacity from absence of the capacity itself.

That does not mean the capacity is secretly intact.

It means the epistemic problem remains open longer than a score sometimes allows.

What a Definition Carries

Perhaps that is what troubles me most about “profound autism” as this paper attempts to stabilise it.

The category carries too much history inside too little language.

It gathers intellectual disability, communication difference, adaptive functioning, supervision, safety, dependence, and basic-needs communication into a research construct intended to make an underserved population more visible. There is value in that visibility. Research populations need boundaries if findings are to be compared across studies.

Yet the resulting boundary inevitably travels.

It will travel into clinics, advocacy organisations, funding debates, schools, insurance decisions, government reports, journalism, and family conversations. The authors explicitly emphasise that they are proposing a research definition, and that more information and guidance would be needed before applying it clinically. That distinction deserves protection.

Because categories acquire power through use.

And once “profound” becomes attached to the person, grammar begins doing political work almost invisibly.

The person has “profound autism.”

The “profoundness” belongs to her.

Her supervision needs belong to her. Her limited independence belongs to her. Her communication difficulty belongs to her. Her adaptive functioning belongs to her.

Soon the exhausted caregiver, unavailable respite worker, inaccessible community programme, years-long service waiting list, inadequate communication support, unaffordable housing, and insurance denial have disappeared from the sentence entirely.

Only the autistic person remains.

“Profound.”

The Question I Am Left With

I do think Siegel and colleagues have identified something that autism research needs to confront.

There are autistic people whose lives involve extraordinary support needs. There are families performing astonishing quantities of care. There are people whose communication remains difficult for unfamiliar partners to recognise, whose safety requires continuing assistance, whose daily living requires extensive support, and whose experiences have been largely absent from the research literature through which autism has increasingly come to understand itself.

They deserve far more research attention than they have received.

They also deserve a research language capable of keeping the world around them visible.

That may be especially important when the language is being developed in the United States, where need and provision have such an unstable relationship. A definition centred upon dependence cannot be culturally weightless in a society where dependence carries enormous economic consequences. A measure of adaptive functioning cannot be wholly detached from the opportunities through which adaptation develops. A requirement for constant supervision cannot tell us, by itself, whether that supervision is supplied by a well-supported network of paid professionals and family members or by one exhausted parent who has not slept properly in fifteen years.

The underlying autistic person may be the same.

The life produced around her can be radically different.

Perhaps, then, the next research question should extend beyond whether we can reliably identify a population called “profoundly autistic.” We also need to ask which parts of the “profundity” remain stable when healthcare changes, when income changes, when communication access changes, when housing changes, when respite becomes available, when skilled support arrives, when families are no longer required to manufacture an entire care infrastructure privately around one human being.

Because those are empirical questions too.

And until we ask them, I am wary of allowing a society’s failure to provide collective care to settle quietly into our description of the people who require it.

We identify the people whose lives require the most collective support from inside a society that has made that support least collective.

Then the consequences gather.

Financial precarity. Exhaustion. Restricted participation. Lost employment. Institutional dependence. Family crisis. Reduced autonomy. Narrowed lives.

Eventually the field becomes difficult to see.

And we call the consequences “profound.”

Mirage

During the pandemic, after my forensic science consulting business collapsed, I applied for more than two hundred jobs.

Every résumé tuned.
Every cover letter rebuilt.
Every application shaped for the machine waiting first.

I got one interview.

I did not get the job.

No.

Because?

No.

Now I watch my adult children enter a version of the same field.

Autistic gestalt processors.

Capable.

Prepared.

Trying.

And again:

No.

Because?

No.

We keep talking about what becomes possible when the field holds.

Processing time.

Explicit criteria.

Repair.

Questions in advance.

Enough relational stability for generativity to arrive without spending itself constructing the conditions of its own arrival.

All of that is possible.

That is what makes this harder.

Because possibility is not access.

The stable field becomes another horizon autistic people are told exists somewhere ahead of us—if the employer is enlightened enough, if the institution understands enough, if the right accommodation is requested in the right language at the right moment, if someone with power decides the conditions matter.

Meanwhile the applications keep going out.

The refusals keep coming back.

The field we can describe beautifully remains, much of the time, imaginary.

A mirage produced by knowing exactly what would help and having almost no power to make the person on the other side build it.

And eventually those lives become statistics.

Unemployed.

Underemployed.

Outside the labour force.

Numbers large enough to sound structural and still somehow interpreted as individual failure.

For fuck’s sake.

Perhaps the question was never simply whether she could do it.

Perhaps we should have asked what happened to her capacity when we built the conditions under which she was required to show us.

But even that question assumes someone will ask.

Most of the time, nobody does.

The decision arrives first.

No.

Because?

Still no.

Final Thoughts …

I resisted finishing this series on a high note for a reason.

There is an entire economy built around telling people that the next adjustment will make the difference. Rewrite the résumé. Optimise the keywords. Practise the interview. Learn the right scripts. Improve the eye contact. Build confidence. Find a coach who knows the secret route through the machinery. Some of that advice can be useful, certainly, but there is also a grift living very comfortably inside hope—the promise that if the person seeking work performs the correct sequence of improvements, access will eventually follow. The structure disappears again. Failure becomes evidence that one more adjustment is required.

I cannot write that ending because I lived the other one. During the pandemic, after my forensic science consulting business collapsed, I applied for more than two hundred jobs and received one interview. I had professional help tuning every résumé and cover letter to the ATS used by each employer. I knew how to work, had decades of experience, and did everything I was being told should improve my chances. It did not produce access. Now my adult children are encountering their own version of the same machinery, and I cannot offer them the comforting fiction that effort, preparation, or sufficiently sophisticated masking will reliably make the gate open.

There is a generational difference here that feels particularly bleak. When I was young, an entry-level job could begin with physically entering a place. You could walk into a shop, restaurant, office, or business, ask whether they were hiring, fill out an application, and sometimes meet the person who could say yes. The encounter might be awkward, biased, unfair, or arbitrary, but there was an encounter. My children increasingly meet an Applicant Tracking System first. Before anyone can learn their pause, their humour, their intelligence, their precision, their extraordinary capacity once a field begins to hold, they have to become legible to machinery designed to reduce a pile of applicants before anybody needs to meet them.

There is no relational field there because there is no one there to have a relationship with.

That is bleak. I think pretending otherwise would betray the argument I have spent this series trying to make. Better fields can be designed. Assessments can become more valid. Employers can change their practices. None of that means the person currently sending out applications has access to those better conditions, and I do not want to turn structural possibility into another instruction for autistic people to hope harder.

In a more private sense, this series exists because I realised I did not have the scripts to talk about any of this in my own house. I could see what was happening to my children. I recognised pieces of my own history inside it. I had the employment statistics, the language-development framework, the experience of masking, the repeated No. Because? No.—but they had not yet become one thing I could hand to them and say, this is what I think is happening.

It is still summer holiday, so I had the rare luxury of time enough to let the pieces find one another. I followed the question through double empathy, relational fracture, field foreclosure, masking, generativity, assessment, employment, and finally back into my own house.

That makes The Absent Field a script garden in the oldest sense of what those have been for me: somewhere language can grow around an experience before I need to use it.

I grew this one because I needed to talk with my kids.

I am sharing it because I suspect some of you need the script too.

Building the Field on Purpose

The obvious objection to everything I have argued so far in this series is that a stable field takes time. Usually, it does. People learn one another slowly. They discover what a pause means, what directness means, what kinds of language signal strain, what kinds signal concentration, which misunderstandings recur, which forms of repair actually hold. Shared reference accumulates through repetition until the interaction no longer has to begin from zero, and the field becomes more stable because the people inside it have history.

But duration is one way of producing stability. It is not the definition of stability.

That distinction matters because otherwise the argument collapses into something unusable: autistic gestalt processors can only show their capacities inside long-standing relationships, therefore any first encounter will necessarily mismeasure them. I do not think that follows. A field can be young and still be structured well enough to support generativity. It can lack history whilst offering predictability, explicitness, room to process, and enough repair that uncertainty does not immediately become consequence.

The question therefore changes from how do we make strangers feel familiar? to something much more practical: what does the field need in order to stop making unfamiliarity carry so much weight? Some of the answers are almost embarrassingly ordinary. Tell people what is being assessed. Give questions in advance where surprise is not itself the competency being measured. Allow processing time without treating latency as evidence. Permit clarification. Offer written or asynchronous response when rapid live speech is irrelevant to the work. Use work samples. Let people demonstrate competence through the thing they will actually be doing rather than through an adjacent performance whose relationship to the job has never been examined particularly closely.

None of this requires intimacy. It requires design.

That is the turn institutions routinely miss because accommodation is still framed as something added after the normal process has already been defined. The interview exists in its standard form. The autistic candidate arrives. Something about her makes the standard form difficult. An adjustment is therefore offered so that she can participate more successfully in the thing everyone else is presumed to access naturally. The structure remains innocent. The person carries the difference.

The methodological argument from the previous piece makes that framing harder to sustain. If hidden criteria, rapid social response, unfamiliarity, power asymmetry, and limited repair alter access to the capacities being measured, then reducing those conditions is not simply kindness toward someone who struggles with the standard test. It is an attempt to stop the test contaminating its own result.

That is assessment design.

Nothing about explicit criteria requires lowering substantive expectations. If a job requires complex reasoning, assess complex reasoning. If it requires technical competence, assess technical competence. If it requires collaboration, assess collaboration. If rapid decisions under genuine time pressure are part of the work, then time pressure may belong in the assessment too. But make the relation visible. Do not require rapid verbal improvisation merely because interviews have traditionally required rapid verbal improvisation. Do not treat eye contact as evidence of confidence when the job does not require eye contact. Do not treat conversational ease with a stranger as evidence of collegiality when the actual workplace will involve relationships accumulated over months or years. Do not convert performative enthusiasm into evidence of motivation simply because enthusiasm in a familiar social register is easier to recognise than motivation organised differently.

The question is not whether a demand feels normal. The question is whether it belongs to the construct.

Consider the ordinary behavioural interview question: Tell me about a time you had a conflict with a colleague and how you handled it. The candidate has to identify an appropriate event, decide what level of conflict counts, reconstruct enough history to make the event intelligible, select which details matter, infer what the interviewer is actually trying to assess, build a coherent answer, monitor how long the answer is taking, preserve an acceptable degree of eye contact, sound reflective without sounding rehearsed, demonstrate enough ownership without accepting blame that was not hers, and end with a lesson sufficiently polished to resemble the expected form.

Perhaps the employer wants to know how she handles conflict. The interview is measuring much more than that.

Give the question in advance and some of that noise falls away. Make the criterion explicit—we are assessing how you identify disagreement, communicate during it, and adjust afterwards—and more falls away. Allow a written response or work sample and more falls away again. The substantive demand remains. What changes is the amount of irrelevant field instability wrapped around it.

That is the design target. Not comfort, not an interview without difficulty, not a guarantee that the autistic candidate will perform well, but a cleaner relation between what the institution says it wants to know and what the assessment actually requires in order to reveal it.

This is also where repair becomes more important than it first appears. Most assessment formats treat first interpretation as final interpretation. A candidate misunderstands a question and answers the wrong thing. An interviewer interprets the answer as evidence of poor judgement. The encounter moves on, and the error remains embedded in the evaluation. A more stable field allows some degree of correction: that was not quite what I was asking; let me phrase it differently, or I think I answered a different question; can I try that again?

Those sentences are ordinary enough in real working relationships. People clarify. People revise. People discover that a first formulation did not carry what they meant. A workplace in which no one could ever repair a misunderstanding would be unbearable. Yet we often design interviews as though the capacity to avoid all misunderstanding during a first encounter is itself evidence of workplace competence.

If repair is part of competent adult communication, then an assessment that forbids repair may be removing evidence rather than preserving rigour.

The same applies to processing time. Latency is often treated as a problem because the social rhythm of an interview makes silence visible. The interviewer asks. The candidate pauses. The pause begins acquiring meaning before the answer arrives. But silence is not empty. For a gestalt processor, it may be precisely where the answer is being built. If the assessment is intended to measure the quality of that answer, there is no obvious methodological virtue in contaminating it with a second, unrelated question about whether the answer can be produced within a socially comfortable interval.

Sometimes speed matters. Often it does not. Design means knowing the difference.

This is where the argument about masking returns too. In The Manufactured Field, I described high masking as the private labour of carrying into the encounter some of the stability the encounter refuses to provide. The candidate prepares scripts, rehearses stories, calibrates expression, anticipates questions, monitors herself, and tries to construct enough predictability internally that the interaction remains traversable. Building the field on purpose changes who carries that burden. Instead of requiring the candidate to guess the criteria, disclose them. Instead of requiring her to predict every question, provide some in advance. Instead of requiring her to conceal processing latency, permit it. Instead of requiring her to manufacture a socially familiar version of confidence, specify what evidence of competence actually matters. Instead of making repair risky, design repair into the process.

The autistic person no longer has to privately subsidise the accessibility of the encounter.

That shift is larger than it sounds. So much disability accommodation is organised around individual exception. Someone identifies a need. Documentation may be required. A deviation from the standard process is approved. The system remains unchanged except where a particular person has successfully demonstrated entitlement to modification. Field design asks a different question: which features of the standard process are doing useful evaluative work, and which are simply inherited friction?

That question benefits more than autistic people. An explicit criterion helps anyone who has been excluded from the tacit cultural codes through which professional expectations are usually communicated. Advance questions can reduce irrelevant performance variance for people working in a second language, people with anxiety, people whose thinking benefits from preparation, people whose cultural norms around self-promotion differ from those embedded in the interview. Written options may reveal capacities that live badly inside rapid oral exchange. Universal benefit does not make the autistic argument less specific. It reveals how much of what we call accommodation may actually be better design.

And better design gives the field account something else it needs: a way to test itself. If I am right that field stability changes the dependable availability of generative language, then deliberately altering field conditions should change observable performance. We do not have to settle this by choosing which theoretical account feels more humane. We can compare the same substantive competency under different field conditions: conventional interview versus explicit criteria, surprise questions versus advance questions, immediate oral response versus processing time, one-shot answer versus clarification and repair, live verbal account versus work sample.

If nothing meaningful changes, the field account weakens. If performance changes substantially and systematically, then the design itself becomes part of the explanation.

That is why I resist describing these changes merely as accommodations. Accommodation asks whether the person can be helped to survive the assessment. Design asks whether the assessment deserves to survive in its current form.

The employer still evaluates. The university still assesses. The clinician still gathers evidence. The institution still has standards, obligations, thresholds, and legitimate reasons to distinguish one outcome from another. The field argument does not remove judgement. It asks judgement to become more accountable to the conditions under which its evidence was produced.

For most of my life, stability has been treated as something I am supposed to bring with me: prepare enough, regulate enough, script enough, predict enough, mask enough, become legible quickly enough that the people holding power do not have to experience uncertainty for very long. That is one way to manufacture a field. It is expensive, fragile, and extraordinarily convenient for the surrounding structure because the structure never has to change.

There is another possibility.

Build some of the stability into the encounter itself. Make expectations visible. Give language time to become. Let clarification count as communication rather than failure. Measure the thing you actually care about. Permit the field to learn, even if only a little, before interpretation becomes final.

A stable field does not have to be old. It has to be sufficiently coherent that the person inside it is not spending most of her available capacity constructing the conditions under which her capacity can appear.

That can be designed.

And once it can be designed, its absence stops looking inevitable.

It becomes a choice.

The Aggregate

There is a number that follows autistic adults almost everywhere employment is discussed.

Eighty-five per cent unemployed.

I have repeated versions of it myself. It is memorable, alarming, and sufficiently enormous that it seems to explain something simply by being stated. If anything close to eighty-five per cent of a population is unemployed, then whatever is happening is not a marginal problem.

The difficulty is that the number becomes much less stable once you ask what exactly it is counting.

Unemployment has a specific meaning in US labour statistics. It generally describes people without work who are available for work and actively seeking it. That is different from being outside the labour force, which can include people who have stopped looking, cannot currently work, are studying, are caring for others, or are otherwise not counted amongst the actively employed or unemployed. Both are different again from underemployment: working fewer hours than wanted, working intermittently, working in supported employment, or working in jobs that make very little use of a person’s education, experience, or actual capability.

Autism employment research has not always kept those categories cleanly separated.

Different studies sample different populations, use different age ranges, recruit through different systems, define employment differently, and ask different questions about what counts as successful labour-force participation. A person working six hours a week may appear as employed in one account. Someone with an advanced degree working in a position requiring no comparable qualification may also appear as employed. Someone who spent years applying for work, repeatedly failed to enter it, and eventually stopped searching may disappear from the unemployment rate altogether.

The categories matter.

So I do not think we should keep repeating eighty-five per cent unemployed as though it were a settled population statistic.

But refusing the seductive number does not make the underlying disparity disappear.

Across definitions, datasets, and populations, autistic adults experience substantial difficulty entering and remaining in employment. The exact size of that difference changes according to what we measure, but the direction does not become mysterious merely because the headline number is unstable. Autistic people are less likely to be employed, more likely to experience disrupted employment, and often work under conditions that do not reflect their qualifications or capacities.

There is something here requiring explanation.

The conventional explanations tend to begin with the autistic person.

Communication difficulties. Executive functioning. Sensory needs. Social skills. Cognitive inflexibility. Co-occurring disability. Difficulty navigating workplace expectations. Problems with interviews. Problems maintaining employment. Sometimes discrimination is added, sometimes inadequate accommodations, sometimes employer attitudes, but the basic unit of analysis remains remarkably durable: autistic adults have characteristics that make employment difficult, and when enough autistic adults possess those characteristics, the aggregate becomes an employment gap.

I am interested in what happens if we move the point of observation upstream.

Before someone can struggle to maintain employment, she has to enter it.

Before she can enter it, she usually has to survive a series of gates.

The application.

The résumé screen.

Perhaps an automated assessment.

Perhaps a personality inventory.

The telephone screen.

The interview.

The second interview.

The panel.

The selection exercise.

The reference check.

The disclosure decision running underneath all of them: Do I tell them I am autistic? If I do, when? What will they infer? If I do not, what happens when I need something later?

Then, if she succeeds, onboarding begins.

New people. New rules. New sensory conditions. New expectations, some written and many not. A probationary period in which performance is being interpreted before much shared history exists. Managers who may or may not understand autistic communication. Colleagues learning one another under conditions where the autistic worker may already know that being misread carries consequences.

Employment is not one gate.

It is a sequence of fields.

And those fields do not all ask the same thing of the person moving through them.

That distinction matters because the argument I have been developing in this series does not predict that autistic people should simply be either employable or unemployable. It predicts something more conditional: if access to particular capacities changes with field conditions, then different selection and employment environments should produce different patterns of observable performance.

That is an empirical question.

The employment statistics cannot answer it by themselves.

A large employment gap does not prove field foreclosure. It does not tell us whether an applicant was rejected because of an inaccessible interview, insufficient qualifications, discrimination, labour-market conditions, support needs, transportation, health, education, geography, or any of the other variables capable of shaping employment.

Aggregate statistics are very good at showing that something is happening.

They are much worse at telling us what happened in the room.

That is precisely why I think the room matters.

If autistic candidates are disproportionately filtered out during socially intensive, rapid, opaque selection procedures, but the disparity narrows when substantive competencies are assessed under more explicit and field-stable conditions, that would tell us something important.

If it does not narrow, that tells us something too.

The question is not whether employment statistics can be made to corroborate a theory I already prefer.

The question is where people are being lost.

That requires following the pathway rather than staring only at the outcome.

How many autistic applicants enter an application process?

How many survive initial screening?

How many are invited to interview?

How many progress after interview?

What happens when interviews are structured?

What happens when questions are provided in advance?

What happens when evaluation criteria are explicit?

What happens when candidates can respond in writing, demonstrate work directly, or clarify an answer without the clarification itself becoming evidence against them?

What happens after hiring, when the field begins acquiring history?

Those questions interest me much more than the eighty-five per cent figure because they begin locating mechanisms.

They also make room for something employment statistics handle awkwardly: the person who has stopped trying.

Labour-force non-participation is easy to read as absence.

The person is not working and is not actively seeking work, therefore employment is no longer part of the immediate picture. At population scale, she may sit outside the unemployment statistic entirely.

But the category tells us very little about how she arrived there.

Perhaps she cannot work.

Perhaps she does not want paid employment.

Perhaps other responsibilities matter more.

Perhaps the available work cannot accommodate significant support needs.

Or perhaps she applied for fifty jobs.

Then another fifty.

Perhaps she prepared for interviews, rehearsed answers, bought appropriate clothing, travelled to unfamiliar places, managed sensory environments, tried to decode what each interviewer wanted, received another polite rejection containing no usable information, changed something, tried again, and discovered that the next rejection contained no more information than the last.

At what point does stopping become evidence of incapacity?

At what point does it become evidence of learning?

I do not want to romanticise withdrawal. Being unable to obtain work can be economically devastating, socially isolating, and profoundly limiting, particularly in a society that ties income, healthcare, housing, retirement, status, and much of adult autonomy to employment. Nor do I want to convert every person outside the labour force into a secretly willing worker excluded only by bad interviews.

The category is more heterogeneous than that.

That is exactly the point.

Not seeking work describes a present relationship to the labour market.

It does not explain how that relationship formed.

And if repeated exposure to employment gatekeeping offers high effort, uncertain criteria, consequential interpretation, and almost no corrective feedback, then withdrawal deserves to be studied as an outcome of the system as well as a characteristic of the person.

The feedback problem matters here.

Most learning depends upon discovering some relation between action and consequence. I try something. Something happens. I alter what I did. The next result gives me more information. Over time, the environment becomes more predictable because the pattern begins to hold.

Opaque selection interrupts that process.

The applicant receives outcomes without enough information to identify which variables produced them. She may improve her résumé and still not know whether the résumé mattered. She may rehearse interviews and still not know whether she was rejected because she sounded rehearsed. She may mask more intensely, disclose autism, conceal autism, request accommodations, avoid requesting accommodations, and still receive the same carefully polished sentence about another candidate being a better fit.

More attempts produce more data points.

They do not necessarily produce more knowledge.

For a gestalt processor, that distinction is especially consequential because accumulation without dependable relation does not automatically become a usable whole. Fifty unexplained rejections do not necessarily teach fifty lessons. They may teach one enormous lesson instead.

This field does not become more knowable when I return to it.

That is a different interpretation of discouragement.

And it brings the aggregate back to the methodological problem from the previous piece. If assessment conditions participate in producing observable performance, then population-level employment outcomes may contain the accumulated effects of those conditions. Not all of the disparity. Perhaps not even most of it. We do not yet know.

But enough of a possibility exists that the mechanism deserves to be isolated rather than hidden inside a general category of autistic employment difficulty.

The research question is therefore not why are autistic people unemployed?

That question is too large, and it quietly invites the person to become the explanation.

I want smaller questions.

Where are autistic candidates being lost?

Under what conditions?

At which gate?

What capacities are those gates claiming to measure?

What else do the gates require in order for those capacities to become visible?

Does the pattern change when the field changes?

And after someone enters employment, what happens as the field acquires history? Does performance become more dependable as expectations become explicit, relationships accumulate shared reference, and colleagues learn what one another’s communication means? Or do new forms of foreclosure appear in probation, evaluation, promotion, discipline, and workplace social life?

Those are questions the headline number cannot answer.

They are also questions we can actually investigate.

That matters to me because the purpose of a structural account should not be to replace one certainty with another. Saying the deficit is in the system can become every bit as intellectually lazy as saying the deficit is in the autistic person if neither claim survives contact with evidence.

The field account asks for something more difficult.

Measure the person.

Measure the conditions.

Change the conditions.

Follow what changes.

Then follow where people disappear from the pathway.

The employment gap is real.

The mechanism is not yet settled.

And somewhere between the application submitted and the person eventually recorded as employed, unemployed, underemployed, or outside the labour force, there are a series of encounters we have spent remarkably little time treating as part of the phenomenon itself.

Perhaps that is where we should look next.

Not only at who gets through.

At the gates.

The Test Produces the Deficit

Up to this point, I have been making a theoretical claim. If dependable generativity is partly field-dependent, then changing the field should change what becomes observable. That is the point at which the argument stops being only descriptive and becomes answerable to evidence.

The hiring interview gives us an unusually clean place to look because it routinely introduces, all at once, many of the conditions I have been describing as destabilising: unfamiliar people, hidden expectations, asymmetric power, live social evaluation, rapid transitions, time pressure, novel questions, uncertain conversational rules, little accumulated shared reference, and almost no meaningful opportunity to repair an interpretation once it has become consequential. The candidate is asked to reason, explain, retrieve, generate, shift, improvise, persuade, regulate, and remain socially legible inside a field built from instability, then the resulting performance is treated as though it emerged from the candidate alone.

That is the part I want to examine more carefully, because assessment always makes a claim about what an observed performance stands for. An interviewer asks a novel question and believes the answer reveals flexibility. A panel watches someone explain a difficult situation and believes the performance reveals communication. A timed response becomes evidence of thinking on one’s feet. A behavioural interview becomes evidence of judgement, leadership, creativity, collaboration, resilience, self-awareness. Perhaps it does. But only if the instrument is actually measuring those things rather than some mixture of those things and the candidate’s ability to preserve access to them under the conditions the instrument itself has imposed.

This is where the language of construct validity becomes useful. The phrase sounds technical because it is, but the question underneath it is simple enough: are you measuring the thing you think you are measuring? If I want to know whether someone can reason flexibly, but I create a situation in which unfamiliarity, threat, hidden criteria, rapid response demands, and social monitoring alter the person’s access to flexible reasoning, then the answer I receive is no longer a transparent measurement of flexibility. It is flexibility-under-these-conditions.

Those are different constructs.

And yet our systems collapse them constantly. A person struggles to answer quickly in an interview and is described as poor at thinking on her feet. A student cannot retrieve language during an oral examination and is described as not knowing the material well enough. An autistic person becomes less verbally flexible in a clinical encounter and the reduced flexibility becomes evidence about her underlying communication. A worker asked to explain herself during a disciplinary meeting under threat and surveillance produces narrower language, and the narrowness becomes evidence of evasiveness, rigidity, or lack of insight.

The field enters the performance and then disappears from the interpretation.

That is the methodological problem.

We routinely treat the environment as background and the person as signal, even when the environment is actively shaping what the person can make available. That is not unique to autism. Testing has always struggled with this distinction. Noise, fatigue, familiarity, anxiety, cultural expectations, language background, threat, motivation, and context can all alter performance. Good assessment practice tries to notice those influences precisely because an observed response is never produced in a vacuum.

What gestalt processing adds, for me, is a more specific question about the relation between field and generativity. If the later stages of language are increasingly assembled in relation rather than simply retrieved from storage, then the conditions under which that assembly is demanded cannot be treated as neutral. A stable field does not merely make the same task more pleasant. It may make a different level of generativity dependable.

That gives us a consequence we should be able to test.

Take a substantive task and hold it as constant as possible, then change the field. Give the questions in advance. Make the criteria explicit. Reduce live performance pressure. Allow time for processing. Permit written or asynchronous response. Allow clarification without penalty. Create enough continuity that an interpretation can be corrected before it becomes final. Then ask whether the observable performance changes.

If it does, the original assessment was not simply measuring the person. It was measuring person-plus-field and reporting the result as though the field were irrelevant.

That is a much bigger problem than accommodation, because accommodation language still tends to preserve the assumption that the standard instrument is fundamentally sound and the autistic person needs an adjustment in order to access it. The test remains the test. The person is the variable.

I am asking whether, in some cases, the test itself may be producing the deficit it then claims to discover.

Suppose an autistic candidate gives brief, rigid, overly rehearsed answers during a conventional interview, then produces nuanced, flexible, highly generative reasoning when the same substantive questions are provided in advance and answered under explicit criteria with processing time. Which performance is the real one?

The usual framework invites us to treat the first as baseline and the second as accommodated. I think that may be backwards. Both are real. Both are performances of the same person. But they are performances produced under different fields, and if the field materially changes the availability of generativity, then neither can be interpreted without the conditions that produced it.

This is where the distinction between possible, available, and dependable becomes essential. A capacity can be possible without being available in a particular moment. It can be available sometimes without being dependable across settings. And it can become dependable only when enough of the supporting conditions are present. Those distinctions are easy to lose because assessment wants clean categories: can she do it, yes or no?

A person-in-field model asks a better question: under what conditions does this capacity become dependably available?

That question is more difficult, but it is also more honest, because if a capacity appears reliably in one set of conditions and disappears reliably in another, the variability is not noise to be discarded until we uncover the true person underneath. The variability is part of the phenomenon that needs explaining.

The field has become data.

This is why I think the conventional interview is vulnerable as an assessment instrument for autistic gestalt processors. It does not simply observe generativity. It perturbs the conditions under which generativity occurs, then treats the perturbation as neutral. The candidate is unfamiliar with the people, the norms, the criteria, the timing, the hidden expectations, and the consequences attached to each answer. She has little relational history to draw upon, little room for repair, and little ability to know whether a pause, a clarification, a revision, or an atypical prosody has already become part of the evaluator’s judgement. Then we ask for spontaneous flexibility.

That is not obviously a clean test.

It is closer to measuring how much generativity survives field instability.

And this is where the argument has to become falsifiable if it is going to matter beyond description. If field conditions are altered and performance does not change in any systematic way, then the field account needs revision. If explicit criteria, advance questions, processing time, reduced live social pressure, and opportunities for clarification produce no meaningful difference in generative performance, then I cannot simply keep insisting that the field was the hidden variable.

The claim has to be allowed to fail.

But if performance changes substantially and predictably when field conditions change, then the burden shifts. We would have to stop describing the original performance as a transparent reading of the person and recognise that the instrument was partly manufacturing the thing it claimed merely to observe.

That possibility reaches far beyond employment. Education is full of situations in which students are asked to demonstrate knowledge under conditions that may alter access to it: cold calling, timed oral response, rapid classroom discussion, oral examinations, behavioural compliance interpreted as readiness, writing produced under surveillance, assessments that reward immediate retrieval whilst treating delayed access as absence. Clinical settings do something similar when a person is asked to report feelings on demand, reconstruct a developmental history under pressure, demonstrate insight in real time, maintain social reciprocity with an unfamiliar clinician, then whatever becomes available in that encounter is converted into evidence about enduring capacity.

Disciplinary settings may be even more extreme. Someone is frightened, accused, monitored, perhaps already dysregulated, and then required to produce coherent, flexible, socially calibrated language quickly enough to be judged credible. The field tightens, language narrows, and the narrowing becomes evidence.

Once you begin seeing the pattern, it appears in many of the places where autistic people are described as inconsistent, oppositional, lacking insight, unable to generalise, poor communicators, weak under pressure, or incapable of demonstrating what everyone insists they ought to know. Sometimes those descriptions may be accurate. Sometimes the capacity genuinely is not there. The field account does not require every apparent deficit to disappear once conditions improve.

It requires something more basic: that we stop pretending conditions do not participate in the observation.

An assessment cannot remove, destabilise, or overload the conditions under which a capacity becomes dependably available and then straightforwardly interpret what remains as the capacity itself. That is the methodological claim, and it changes where scrutiny belongs: not only on the person being tested, but on the test; not only on whether she performed, but on what the assessment required her nervous system to do before her capacity could become visible.

Because sometimes the deficit we think we have measured is partly the trace left by the measurement itself.

What Stages 4 Through 6 Require

There is a question people ask autistic children constantly that I have come to find increasingly strange: Can she do it?

I understand why we ask. Education, clinical assessment, and employment all depend upon identifying capacities, and much of the machinery surrounding autistic people therefore assumes that competence is something located inside the person, relatively stable across settings, waiting for the right task to reveal whether it is present. She can read. She can speak in sentences. She can solve the equation. She can answer novel questions. She can participate in reciprocal conversation. She can think flexibly. She can generate original language. Or she cannot.

I have lived too long with capacities that change their availability across contexts to find that binary particularly convincing anymore. The knowledge does not necessarily change. The person does not become less intelligent. The language does not vanish in any ordinary sense. Yet something fluid in one setting can become extraordinarily expensive in another; something available in writing may resist live speech; something I can explain easily with one person becomes difficult to assemble with someone else. A thought may remain entirely intact whilst the route from that thought into language narrows until almost nothing dependable can pass through it.

Then the field changes, and the language moves.

I think this matters enormously for how we understand the later stages of the Natural Language Acquisition framework, because there is a qualitative change occurring as language moves from stored gestalts and their increasingly transferable parts toward genuinely generative language. Stage 4 is not simply Stage 3 with longer sentences, and Stages 5 and 6 are not merely more sophisticated versions of the same process. Earlier language can carry remarkable complexity. A stored gestalt can preserve relations much larger than the speaker is yet able to construct independently. Mitigation can introduce flexibility within that stored material. Stage 3 makes words increasingly transferable, capable of travelling beyond the situations and larger gestalts in which they first became meaningful.

Then language begins doing something different.

It assembles itself here.

The speaker combines words into original sentences not previously stored as wholes. Grammar becomes increasingly generative. Relations between meanings become more flexible. Language can be constructed for this particular thought, with this particular person, under these particular conditions, because the sentence no longer has to arrive already made. This is ordinarily described as a developmental achievement inside the person, and of course something has changed there. Capacities that were not previously dependable have become increasingly available. I am not trying to relocate language development entirely outside the speaker or turn grammar into some mystical property of relationship.

I am arguing for something narrower and, I think, more consequential: generation requires somewhere to happen.

A stored phrase can be retrieved. A rehearsed answer can be retrieved. A familiar story can be retrieved. A sentence that has never existed before must be assembled under the conditions in which it is being requested, and those conditions are not incidental. They participate in what becomes available.

I know this most clearly because I have experienced the difference from inside it. There are people with whom language becomes almost absurdly easy. I do not mean people with whom I always agree, people who never challenge me, or people around whom I feel permanently relaxed. Some of the most generative conversations of my life have been difficult, intellectually demanding, emotionally complicated, full of disagreement and correction. But the field held. I knew enough about what the other person meant, and they knew enough about how I moved through language, that a pause could remain a pause without immediately becoming evidence. A sentence could begin imperfectly because I trusted there would be enough time to discover where it was going. I could revise something halfway through without the revision becoming proof that the original thought was false. I could say, no, that isn’t quite it, and continue searching.

The language had somewhere to become.

That is what I mean by a stable field. Not comfort, agreement, familiarity in the simple sense, or the absence of stakes, but enough predictability and relational coherence that generation does not have to simultaneously defend the conditions of its own production.

When I am generating language, I am already doing something extraordinarily complex. I am taking a whole that may not have arrived propositionally, finding relations within it, selecting words, sequencing them, building grammar, monitoring whether the sentence still belongs to the thought that produced it, and adjusting as the language itself reveals distinctions I did not possess before I began speaking. Sometimes I do not know exactly what I think until the sentence exists. The sentence is not merely reporting a completed cognitive product. It participates in producing one.

Generative language therefore requires some permission to remain unfinished whilst it is becoming, and the field can either provide that permission or consume it. If I must simultaneously monitor whether my pause is too long, whether the listener thinks I am avoiding the question, whether my face is doing enough, whether my directness is being interpreted as hostility, whether I have exceeded the expected length of an answer, whether the person with institutional power is becoming impatient, whether a clarification will sound argumentative, or whether asking what the question means will be interpreted as incompetence, then the language is no longer being asked simply to generate. It is being asked to generate whilst monitoring the conditions under which generation will be judged.

That is a different task.

And this is where I think our ordinary model of competence begins to fail. We tend to imagine a capacity as something stored inside the person and then degraded by environmental difficulty: the person has generative language, but stress interferes with it; anxiety interferes with it; sensory load interferes with it; time pressure interferes with it; social uncertainty interferes with it. The underlying capacity remains conceptually untouched whilst the environment is treated as noise around the measurement.

I am no longer convinced that description goes far enough.

For a gestalt processor, generativity is not simply a possession sitting behind the interference. It is an event that has to assemble itself somewhere. The field is part of the conditions under which that assembly becomes dependable. Change those conditions enough and you have not merely made an otherwise identical task more difficult. You have changed the task itself.

The conventional hiring interview makes this especially visible. A stranger asks novel questions under partially hidden criteria, often expecting rapid answers, socially familiar affect, continuous self-presentation, and enough flexibility to move quickly between unrelated topics. The candidate has little accumulated shared reference with the people evaluating her, little opportunity to repair a misreading once it becomes consequential, and almost no way of knowing which parts of her behaviour are being converted into evidence.

This approaches the inverse of the field I described above.

The language has nowhere safe to remain unfinished.

A pause is no longer simply part of thought. It is also something being interpreted. A correction is no longer simply refinement. It may be read as uncertainty. Asking for clarification may be treated as failure to understand. Taking time to build an answer may become evidence that the answer was not readily available enough to count.

Under those conditions, stored and rehearsed language becomes disproportionately useful for an obvious reason: it asks less of the immediate field. The prepared story already has shape. The familiar introduction already has sequence. The anticipated answer does not require the same degree of live construction. The candidate can retrieve something where generating something new would require more relational stability than the encounter offers.

This is why the previous piece described masking as a manufactured field. Preparation, scripting, rehearsal, and mitigation do not merely hide autistic difference. They supply portable structure when the actual encounter offers too little of it. What looks like spontaneous flexibility from outside may therefore be the successful deployment of language prepared under very different conditions.

That does not make the language false.

It makes the route to it visible.

And once that route becomes visible, I think the conventional question—Can she do it?—starts to look badly underspecified.

Can she do it where?

With whom?

Under what temporal pressure?

With how much shared reference?

With what opportunity for clarification?

Under whose interpretation?

With what consequences attached to hesitation?

These are not accommodations added after the real capacity has been measured. They are part of the conditions under which the capacity becomes observable at all.

This is the theoretical centre of the argument I have been building across this series. Meaning can fail to cross. Meaning can cross and fail to remain. A field can be foreclosed before enough reciprocity develops for correction to matter. And when the field itself becomes unstable, the later generative capacities we often treat as context-free possessions can become correspondingly unstable in their availability.

The unit of analysis therefore cannot remain the person alone.

It has to become person-in-field.

That does not mean every difficult environment erases generativity, nor that every familiar relationship produces it, nor that autistic people are incapable of generating language with strangers. It means something more precise: dependable generativity belongs to an interaction between capacity and conditions, and we misdescribe what we are seeing when we attribute the entire result to one side of that relation.

A hiring interview does not simply reveal the candidate.

It also reveals what that candidate can make available inside the field the interview creates.

Those are not the same thing.

And if the field is unstable enough, the difference between them may be exactly where the apparent deficit appears.