Japan's autism rates and the limits of Western diagnostic culture: a paper that won't sit still

On 7 July 2026, the data analyst Jordan Crémieux posted a thread on X flagging what he called the "atrocious" framing in a paper that has been making the rounds in some online diagnostic-debate circles. The paper, Crémieux wrote, contains "good parts" but also a section blaming rising autism diagnoses on "Western cultural" habits — a claim that he said rests on a misreading of a Japanese data point. The thread, captured at 19:07 UTC, has since become a small flashpoint in a long-running argument about whether autism rates are genuinely rising, or whether the rise is an artefact of how the West counts.

The argument matters beyond the academy. Japan's health system compiles some of the cleanest developmental-disorder statistics on earth, and its trajectory does not, on the face of it, behave like the trajectory the paper's authors describe. If a country with a low individualism score, a tightly bounded diagnostic culture, and a relatively collectivist parenting register is also recording steep increases in autistic identification, then the explanation "this is happening because the West is pathologising normal behaviour" becomes harder to sustain.

What the paper actually says

Crémieux's account, as posted at 19:07 UTC on 7 July 2026, is that the paper under review contains useful background on the global rise in autism prevalence — and a separate, "atrocious" section that attributes part of that rise to Western cultural pressure to label neurodivergent children. The Japan example is, on the authors' own reading, supposed to anchor the opposite claim: that the increase is partly an artefact of over-diagnosis rather than a real epidemiological signal. The problem, Crémieux argues, is that the authors appear to have looked at the wrong Japanese data series, or at least at one whose logic they have inverted.

This is a familiar shape of argument in the diagnostic-debate literature. The basic move is: prevalence is rising faster than genetic change can explain; therefore the rise is mostly reclassification and case-finding, not new cases; therefore the diagnostic expansion is itself a cultural artefact. The move is plausible in the abstract. The hard part is showing it in the data.

What Japan actually shows

Japan's Ministry of Health, Labour and Welfare has, for two decades, published developmental-disorder statistics that include autism-spectrum diagnoses. Multiple peer-reviewed analyses drawing on those records, as well as on school- and clinic-based cohorts, have documented a steep rise in identified cases across birth cohorts from the 1990s onwards. A widely cited longitudinal picture is that the increase in Japan runs broadly in parallel with the increase in the United States and Western Europe, even though Japan's clinical thresholds, school-system screening tools, and parental reporting norms are not identical to the American ones.

That is the awkward fact for the "Western cultural" explanation: the variable the paper's authors treat as the driver — a permissive, individualist diagnostic culture — is precisely the variable that Japan lacks. Yet Japan's count of identified cases has climbed, and the rate of climb is not dramatically out of line with the West's. If cultural permissiveness were the operative cause, the Japanese curve should look different. It does not, at least not in the direction the theory predicts.

Crémieux's own preferred reading, posted in the same thread, is that the paper's authors "got confused" because they did not realise that the data series they were citing to support the over-diagnosis argument was, in fact, evidence for genuine increase. The argument is empirical, not theoretical: the Japan numbers, properly read, do the opposite of what the authors claim.

The structural problem with "Western" explanations

There is a more general issue sitting underneath the disagreement. Framing rising autism rates as a Western cultural pathology is, on its face, attractive to two quite different audiences: those who want to push back against what they see as over-medicalisation, and those who want to push back against what they see as a globally dominant US diagnostic empire. Both can find common cause in the claim that "the West is over-counting." The data, however, do not always cooperate.

Two structural points follow. First, several large non-Western health systems — Japan, South Korea, and increasingly mainland China — have developed their own diagnostic instruments, screening regimes, and service pathways. Their numbers are not a download of American Diagnostic and Statistical Manual categories; Japanese clinicians operate within their own guidelines, and Korean developmental surveillance runs through its own institutional channels. If a country with its own diagnostic culture, its own screening schedule, and its own service-provision logic is also recording large increases, the simplest parsimonious explanation is that some real signal is moving through the data alongside the diagnostic-artifact component, not that the artifact explains the whole curve.

Second, the case-finding versus true-increase debate is real and worth taking seriously. Researchers across multiple jurisdictions have documented that broadened diagnostic criteria, increased clinical attention, and improved early-childhood screening can each account for a portion of recorded prevalence growth. None of that requires treating the rise as wholly artifactual. The defensible position is the boring one: it is part real, part case-finding, and the proportions vary by country. The paper Crémieux criticises, in his telling, abandons the boring position in favour of a cultural-blame narrative that the Japanese numbers do not actually support.

Stakes, and what remains genuinely contested

The stakes are not only academic. Public-health resources, school-system support, and parental decision-making all run on these numbers. If the rise in autism identification is read as an artefact of over-diagnosis, the policy implication is to tighten criteria, slow screening, and reduce labelling. If the rise is read as a real increase in identified need, the policy implication is to expand services, train more clinicians, and shorten waiting lists. Japan itself has moved, over the past decade, towards earlier identification, more inclusive schooling, and a much larger adult-diagnosis pathway — all of which presume that the demand being measured is, in substantial part, real.

What remains genuinely contested is the rate at which each component — genuine increase versus diagnostic expansion — is contributing in any given country. Crémieux's argument, as posted, is narrower than the full debate: he is not claiming that there is no diagnostic-artefact component anywhere, only that this particular paper mishandles the Japanese data. A reader sympathetic to the over-diagnosis camp can, in principle, accept his critique and still defend the broader thesis on other grounds — for instance, on American or British data series, or on the published effects of DSM-5 threshold changes. The paper in question, on the read Crémieux offers, simply does not earn the conclusion its authors want from the Japanese case.

That distinction is worth keeping. The debate about autism prevalence is at its worst when it collapses into a culture-war frame, and at its best when it stays empirical. The Japan data are a useful anchor because they break the most convenient Western-blame narrative, and because the country that is supposed to be the counter-example is, in this case, closer to the rule.

What to watch

The cleaner test cases are not in the West. They are in countries like Japan, South Korea, and parts of urban China, where national health systems generate their own prevalence data using their own diagnostic instruments, and where the cultural-pressure variable is plausibly weaker than in the US. If those series continue to climb, the over-diagnosis-only explanation will have to give up more and more ground. If they plateau, the cultural-pressure camp has its strongest evidence yet. The argument should be settled where the data are cleanest — and the cleanest data, right now, are not the American ones.

Desk note: Monexus read the contested claim against the social-media post that surfaced it. We have not retrieved the full paper, and we do not name its authors here because the thread does not. The argument above is a structural reading of the Japan-data question; it is not a verdict on the broader diagnostic-debate literature.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://x.com/cremieuxrecueil/status/1979290312188268617