A label that no longer carves a kind from the world
Luigi Croce and Irene Fusaro published a long narrative review in Frontiers in Psychiatry in July 2026 under a deliberately provocative title: “Diagnostic inflation in autism spectrum disorder.” Their case is that the reported prevalence of autism has risen almost fourfold in two decades — from roughly 1 in 150 children in 2000 to 1 in 36 in the United States by 2023 — and that the autism category has become, in their phrase, semantically unstable: broadened at the edges by DSM-5, flooded with self-report tools in adult services, and now aggregating what they argue are neurobiologically distinct populations under one label.
Their central device is a Type I / Type II partition. Type I is the prototypical, early-onset, often syndromic presentation with high genetic load — the historical Kanner-Asperger picture. Type II is later-recognised, milder, polygenically driven, and overlaps substantially with general psychopathology: mood, anxiety, ADHD. It is, in their account, the population behind much of the recent increase — adults and women without intellectual disability, the group for whom camouflaging/masking is the operative concept. From this they build a “dual challenge”: genuine under-diagnosis in historically missed groups on one side, and over-diagnosis or misdiagnosis on the other, where presentations that overlap with borderline personality disorder, complex PTSD, ADHD with affective dysregulation, or trauma get read as autism.
To their credit, and to get this out of the way as early as possible: the paper is not a crude gatekeeping tract, and it should not be read as one. It disavows denial of care in every section. It engages the neurodiversity objection directly and grants the recognition of the female phenotype as a real clinical achievement. It frames its own project as protecting the neurodiversity paradigm rather than undoing it — the argument being that a label applied with low specificity stops identifying a population for whom particular support works, and so loses value for the people who hold it legitimately.
There is a genuine point inside this that my coherence-first view cannot wave away: if the diagnostic category really does gather distinct profiles with different needs, then supporting someone through the wrong frame is a real harm, and caring about the right response to the actual person means actually caring about that.
So their core claim — that the diagnostic category has become incoherent, and semantically unstable — is worth taking seriously. Where the paper goes next is where I part company with it — not on whether there is inflation, but on what is actually doing the inflating.
The error of treating a constitutive incoherence as a failure of measurement
The paper’s remedy is methodological. It proposes six minimum evidentiary standards for adult assessment: structured developmental history from at least two independent informants, multi-context behavioural observation, neuropsychological profiling, granular sensory assessment by modality and direction, systematic evaluation of alternative diagnoses, and a longitudinal formulation held open to revision. The whole apparatus is a better sorting machine — more informants, more instruments, more rigour at the boundary, so that the people who “really” belong in the category are separated more cleanly from those who belong in some other.
Notice what that remedy assumes, though. It assumes the problem is measurement error — that there is a correct box for each person and the inflation comes from sorting them into it badly, or too sloppily. So: if you just tighten the criteria, add the second informant, and run the granular sensory assessment, the overflow resolves back to its proper categories; bosh, job done. Inflation, on this reading, is a failure of the sorting instrument, fixable by better fine-tuning.
But here’s the thing: you cannot fix — nor even touch — an identity incentive with a better ADOS (Autism diagnostic observation schedule, the standardised assessment) protocol. If part of what is driving people toward a diagnosis in the first place is that the diagnosis itself has become something people want — something that confers meaning, community, and an account of one’s difficulties — then no amount of methodological rigour at the clinical gate touches the driver, because the driver is upstream of the clinic, simply feeding down to it.
It is operating in the culture and existing active in the substrate of the collective before anyone books an assessment. The paper’s own evidence half-sees this: it notes the eleven-billion-daily-view scale of autism content on social media, the resource incentives attached to the label, and the self-diagnosis pipeline. But it files these as contamination to be screened out at the door, rather than as actual factors in the engine, sitting several rooms away from the door entirely.
And here the paper’s own findings quietly undercut its remedy. Its strongest material — the Type I / Type II split, the demonstration that the sensory criterion is non-specific, the pervasive overlap with “general psychopathology” — is evidence that the sorting apparatus was never tracking natural kinds to begin with. If the category genuinely aggregates distinct phenotypes shading continuously into ordinary variation and into other conditions, then “sort people into it correctly” presupposes joints in nature the paper has just spent thirty pages showing are not even there. It documents that the boundary is constructed, and then sets out to construct a better boundary (Einstein’s “we cannot solve our problems with the same thinking we used when we created them” comes to mind, here). Ultimately, the instability the authors found is real, and I’ve been circling that myself in my own corpus for years now. The remedy, though, misreads it as a problem of precision when it is actually a problem of what the category is.
The valence inversion — and the other it called into being
If the driver is upstream, as I say, then it can be dated. I’d locate it at the very moment the progressive movement coined the dyad of the constructs “neurodivergent” and “neurotypical“. It was at this moment, from this moment, that the valence of divergence was flipped.
Before that inversion, to diverge was to deviate: the word sat on the deficit side of the ledger, and the clinical gaze looked outward at the patient as the thing that had gone wrong. The coining did something genuinely clever and, in its first-order intent, humane — it flipped divergent from deviance to distinction, and in the same motion made typical the faintly diminished, condescended pole: bland, conformist, and unremarkable; herdlike, even.
This was the affirming move, and it did real good. But it had a structural consequence its authors did not price in, as they could not. From the moment the valence flipped, identifying as not-typical became, generally speaking, both more attractive and more incentivised — and, crucially, the coining produced a construct that could be blamed and separated from. “Neurotypical,” “the neurotypical world,” “the system that designs for the other and has not priced me in”: a ready-made other to hold responsible for one’s sufferings and difficulties, and a ready-made self to step into that arrived pre-loaded with meaning and a grievance system already structured.
There is a deep irony here that the affirming movement has not fully metabolised, and it is the reflexive sting of the whole piece. I’ve argued across my work that “neurotypical” is itself a cybernetic construct — a norm the system produces and then measures everyone against, not a natural category of person. The valence inversion I am analysing here took that construct and made it the thing to identify against. The movement, in flipping the sign, built the very other it now defines itself in opposition to. “The neurotypical” was always a fiction of the psychiatric apparatus that revolves around statistical normativity and the industrial standardisation and institutional education that requires conformity to that myth of normality; the neurodiversity-affirming turn made it, funnily enough (though unfortunately it’s not that comedic), a load-bearing fiction — the pole you push off from to become someone.
This has to be said with care, because it is the exact point the tightening literature reaches for to cast doubt on genuine recognition — and that is not what I am doing. The valence inversion did not invent autistic people, or ADHD, or the late-diagnosed adult who spent thirty years without an account of themselves. Those are real, and the recognition of them is the achievement the progressive, affirming movement should be proud of. What the inversion did though, however incidentally, was open a third door alongside the two real ones — and it is that third door, not the first two, that the inflation curve is almost exclusively walking through.
Inflation as possession — the archetype worn in place of the earned self
The word the paper chose is doing more work than the paper knows. “Inflation” is already a Jungian technical term, and it names something the clinical sense only points at from outside. Psychic inflation, according to Jung, is what happens when the ego identifies with an archetype, without integration — the conscious self swelling beyond its own borders by claiming, as personal, the numinous authority of a collective image it has not done the inner work to earn. The result of this phenomenology is what Jung called a mana-personality: the self enlarged by appropriating the mana (the “charge”) of an archetype it has merely put on rather than integrated.
That is the third door, described precisely. There are at least three circles of neurodivergent identification, which you could also see as doors to diagnostic recognition. The first is real substrate difference — the constitutional profile, present from the start. The second is the cybernetically-formatted developmental response (clinically recognised as “virtual autism” and “virtual ADHD“) — real difference produced by a mismatched environment (between cybernetic-external and human-internal) during plastic windows and throughout the lifespan, no less genuine for being induced. Both of these deserve recognition and support without qualification. The third circle is the mana-personality: the ready-made identity claimed for the meaning and the belonging and the grievance-structure it supplies, without the individual doing any of the integrative work — the self one steps into rather than becomes. Jung’s warning was precisely against this: the archetype claimed pre-packaged, worn as identity in place of the labour of individuation.
So the paper’s diagnostic inflation and Jung’s psychic inflation are the same phenomenon seen at two levels. At the epidemiological level it appears as a prevalence curve bending upward. At the psychological level it is the ego swelling into a ready-made archetypal identity it never earned. The clinicians count the curve, and reach for better instruments to flatten it. What they are measuring, without the vocabulary to name it, is a possession — a mana-personality made culturally available the day “divergent” changed its sign, and taken up by people in both distress and need reaching for a self that comes with meaning attached. This is what I’d call mana personality inflation, and it is the engine the sorting machine cannot reach, because it does not live in the assessment room.
Which is why naming it matters, and matters especially from an affirming position rather than a gatekeeping one. Left unnamed, this third circle gets weaponised — folded back by papers like this one into doubt about the first two, so that the late-diagnosed woman and the adult who finally found their account are quietly re-sorted into borderline personality disorder or trauma and told their recognition was a measurement error. Naming the mana-personality precisely, with open eyes, is what stops that. It lets you say clearly that the inflation is real, that its engine is an identity made available by a valence inversion rather than a failure of clinical rigour, and that acknowledging the third circle is not a concession to the gatekeepers but the thing that protects the first two from them. The paper found the curve and reached for the wrong tool. The tool it needed was not a better protocol. It was hidden in Jung’s Collected Works, Volume Seven.
Citations
Croce, L. & Fusaro, I. (2026) — Diagnostic inflation in autism spectrum disorder: an epistemological and methodological reappraisal — Frontiers in Psychiatry
Litman, A., Sauerwald, N., Green Snyder, L. A. et al. (2025) — Decomposition of phenotypic heterogeneity in autism reveals underlying genetic programs — Nature Genetics
Pellicano, E. & den Houting, J. (2022) — Annual Research Review: Shifting from ‘normal science’ to neurodiversity in autism science — Journal of Child Psychology and Psychiatry
Jung, C. G. (1966) — Two Essays on Analytical Psychology (Collected Works Vol. 7) — Princeton University Press
