Quantified and Contained: Autism, the DSM, and the Eugenic Roots of Statistical Authority
How Eugenics, Statistics, and the DSM Shaped—and Still Constrain—Autistic Lives.
This piece traces autism’s path through the DSM, exposing its eugenic roots and the statistical tools used to contain neurodivergent lives. It calls for narrative sovereignty, relational care, and an end to data-driven erasure.
Introduction: The Numbers Were Never Neutral
In the early 1900s, an entire field of mathematics was developed to legitimise eugenics. It did not emerge in the abstract, nor in pursuit of neutral scientific discovery, but rather out of a deliberate effort to quantify human value along racialised, classed, and ableist lines. Figures such as Francis Galton and Karl Pearson did not merely influence the development of modern statistics—they designed it explicitly to uphold selective breeding, population control, and the elimination of so-called “undesirable traits.” The statistical concepts we now take for granted—regression, correlation, standard deviation—were forged in this crucible of control, not in the service of understanding, but in the service of power. These tools, alongside the emerging diagnostic categories of psychiatry, were not neutral measures. They were, and often remain, technologies of enclosure: apparatuses for sorting, ranking, and regulating lives deemed divergent from a socially constructed norm.
This entanglement between statistics and psychiatry is nowhere more evident than in the history of autism’s inclusion in the Diagnostic and Statistical Manual of Mental Disorders (DSM). Although the DSM presents itself as a clinical reference—an evolving text guided by research and consensus—it is also a social artefact, reflecting the ideological assumptions and political imperatives of its time. The pathologisation of autistic experience, from its early framing as a form of childhood schizophrenia to the later consolidation of the autism spectrum, cannot be disentangled from the eugenic anxieties that shaped 20th-century psychiatry. Nor can we ignore how statistical reasoning—what gets counted, who gets diagnosed, which traits are deemed deviant—continues to constrain the way autistic lives are understood, supported, or denied care. This piece traces that historical arc: how autism came to be defined within the DSM, how eugenic thought shaped those definitions, and how the statistical tools used to track, classify, and treat autism today carry forward a legacy of control masked as objectivity.
Eugenics and the Birth of Modern Statistics
Modern statistics did not arise in a vacuum. It was developed in direct service of a project designed to measure, rank, and ultimately control human populations. At the centre of this development was Sir Francis Galton, often credited as a pioneer of statistical thinking, whose driving obsession was white supremacist social engineering. Galton coined the term eugenics and introduced concepts like correlation and regression to the mean—not to better understand human variation, but to lend scientific authority to the idea that certain races, classes, and disabled people were biologically inferior. These tools were developed to support policies aimed at eliminating “undesirable” traits from the population through forced sterilisation, institutionalisation, and the prevention of reproduction among those deemed unfit. They were not neutral instruments; they were designed to quantify human worth in order to justify exclusion, confinement, and eradication.
Galton’s protégé, Karl Pearson, took these ideas further and institutionalised them. As the founder of the first academic department of statistics at University College London, Pearson helped embed eugenics within the very architecture of modern data science. He believed that statistical knowledge should serve the state in managing its population, and he explicitly argued that the poor, the disabled, and the racialised should reproduce less for the good of society. Under his guidance, statistics became a mechanism of surveillance and control—a means of transforming social prejudices into seemingly objective facts.
Ronald Fisher, often named the third founding father of statistics, continued this legacy deep into the 20th century. Whilst he is widely known for contributions to evolutionary biology and the development of statistical significance testing, Fisher also vigorously defended eugenics programmes, including forced sterilisation. He argued that Britain’s declining birth rate was the fault of the lower classes and that the nation was at risk because those deemed genetically “unfit” were having too many children. For Fisher, data was a means of protecting racial and class hierarchies under the guise of scientific rationality.
The infrastructure they built—conceptual, academic, institutional—still shapes how we collect, analyse, and interpret human data. Journals like Biometrika, co-founded by Pearson and Galton, and the Annals of Eugenics (now published under the name Annals of Human Genetics), provided platforms for embedding these ideologies into scientific practice. These publications did not simply reflect the prejudices of their time; they helped codify them into policy. The statistical methods we continue to use today were not developed for justice, equity, or care—they were created to exclude, to rank, and to decide who got to live freely and who did not.
The DSM and Autism: A Eugenic Framing
The Diagnostic and Statistical Manual of Mental Disorders—known as the DSM—has long been upheld as the gold standard for psychiatric diagnosis. Yet beneath its clinical exterior lies a system deeply rooted in eugenic logic and structured around normative ideals of utility, productivity, and conformity. Far from being a neutral or universally representative tool, the DSM was developed through a lens that privileged certain bodies and minds—chiefly cisgender, white, middle-class boys—whilst pathologising those who deviated from this constructed norm. Its criteria have consistently been shaped not only by scientific trends, but by economic imperatives and political pressures, often at the expense of those it purports to describe.
In its first two editions—DSM-I (1952) and DSM-II (1968)—autism did not appear as a distinct category. Children who would now be identified as autistic were typically diagnosed with childhood schizophrenia or labelled psychotic. These early frameworks mirrored broader eugenic attitudes toward mental difference, viewing such children as defective or dangerous, destined for institutionalisation. The goal was not understanding or support, but removal from public life—a sanitising of the population through segregation. This approach was grounded in the belief that these children would never be “useful” to society, and that their presence posed a burden on familial and state resources.
DSM-III (1980) marked autism’s formal entrance into psychiatric discourse as “infantile autism,” now separated from schizophrenia. Whilst this shift may seem like progress, the accompanying criteria reflected a rigid, deficit-oriented framing. Communication delays, social withdrawal, and repetitive behaviours were pathologised without context, stripped of meaning, and positioned as evidence of disorder rather than difference. These criteria were normed almost exclusively on cisgender, white boys, a methodological choice that would go on to marginalise girls, non-binary people, people of colour, and those whose presentation did not conform to this narrow template. The publication of DSM-III coincided with the rise of Applied Behaviour Analysis (ABA), a practice rooted in the belief that autistic traits should be extinguished in pursuit of a more “normal” presentation. ABA did not aim to support autistic flourishing—it sought to suppress, erase, and retrain, often through coercive methods. Autism, in this framework, was a problem to be solved, not a neurotype to be understood.
DSM-IV (1994) expanded the diagnostic umbrella to include Asperger syndrome and Pervasive Developmental Disorder Not Otherwise Specified (PDD-NOS). This broadening led to a dramatic increase in autism diagnoses, not because autism itself became more prevalent, but because more people—especially those previously excluded—were now eligible to be recognised. This expansion came with both increased access to support and a rapid rise in behavioural intervention programmes. Once again, the underlying logic was not acceptance but correction. The expansion functioned as a form of enclosure: autistic children were no longer warehoused in institutions, but they were now subject to highly controlled educational environments and treatment regimens aimed at behavioural conformity. The visibility of autism increased, but so too did its medicalisation and commodification.
DSM-5 (2013) collapsed the previously distinct subcategories into a single diagnosis: Autism Spectrum Disorder (ASD). This reorganisation was framed as a move toward diagnostic consistency, but it carried significant consequences. Many individuals who had previously received diagnoses under the DSM-IV criteria—particularly those with Asperger syndrome or PDD-NOS—no longer met the stricter thresholds, resulting in the loss of services, support, and in some cases, even legal protections. The removal of Asperger syndrome also prompted profound disruption within autistic communities, especially for those who had built identity and belonging around that label. The spectrum model, whilst seemingly more inclusive, continued to prioritise observable deficits over lived experience, and retained its reliance on normed behavioural checklists rooted in a narrow demographic base.
The most recent revision, DSM-5-TR (Text Revision, 2022), underscores how these decisions are often driven less by clinical insight than by economic and political interest. One of the most consequential changes was the quiet tightening of diagnostic criteria: where DSM-5 had previously allowed for “two or more” criteria in a particular diagnostic domain, DSM-5-TR now requires that all listed criteria be met. This revision was not driven by new evidence or improved understanding—it was implemented through backdoor contracting arrangements that prioritised the needs of insurance companies over those of autistic people. By narrowing eligibility, insurers could deny coverage and limit liability, effectively weaponising diagnostic precision against those seeking support. This change disproportionately harms multiply marginalised people—those whose traits are masked, inconsistent, or misunderstood within the rigid expectations of clinical gatekeeping.
Throughout its history, the DSM has functioned less as a mirror of autistic experience and more as a tool of containment—defining who counts, who qualifies, and who gets to be supported. It has pathologised difference in the language of disorder, and reduced richly varied ways of being into checklists normed on the very populations most likely to benefit from their own inclusion. Its revisions have never been neutral; they have been shaped by ideologies of utility, filtered through systems of whiteness, ableism, and economic expedience. Autistic people deserve better than a diagnostic framework built on our exclusion. We deserve language, structures, and systems that see us clearly and treat us justly.
Autism and Statistical Containment
Diagnosis today functions less as a route to understanding and more as a system of statistical enclosure—where legitimacy, care, and support are rationed based on whether a person meets a set of narrowly defined thresholds. Screening tools like the RAADS‑R, AQ‑10, or CAT‑Q, widely used in both clinical and informal settings, claim to identify autistic traits using fixed numerical cut-offs. But these thresholds are based on statistical averages drawn from research samples that overrepresent cisgender, white, and male populations. As a result, those who do not match the expected presentation—particularly women, trans people, people of colour, and those who mask or use non-standard communication—frequently fall short of the “required” score. The diagnosis hinges not just on what one experiences, but how well one’s traits align with the statistical profile of past diagnoses. Regression and correlation, the statistical tools first developed to justify human hierarchies, are now embedded in psychometrics: how closely your responses correlate with pre-defined autistic patterns determines whether you’re seen as valid. Scores that hover near—but not beyond—a diagnostic threshold are treated as “subclinical,” a term that conveniently removes the obligation to support whilst still signalling deviance.
The logic extends into interventions as well. In educational contexts, statistically-based methods like Functional Behavioural Assessments (FBAs) purport to identify the root causes of behaviours by collecting and analysing frequency data. These assessments rely heavily on behaviourist principles: counting how often a behaviour occurs, what precedes it, and what follows. The data is usually compiled into frequency tables or ABC charts (Antecedent, Behaviour, Consequence), which are interpreted to determine function. But these tools often prioritise surface patterns over context, treating correlation as causation. A child flaps their hands before maths class; the behaviour is recorded, tallied, and judged to be “escape motivated,” with no inquiry into whether the classroom is overwhelming, the instruction inaccessible, or the demand itself harmful. Bias enters at every level—from what gets counted as “challenging,” to how behaviour is labelled, to what outcomes are considered desirable. The child’s internal state is rarely considered; instead, meaning is imposed through statistical proxies and interpreted through normative expectations.
This data-centric approach underpins interventions like ABA, which markets itself as the gold standard for autism support by emphasising its “evidence base.” But what ABA measures is not growth, connection, or autonomy—it measures compliance. Progress is defined by reduction in so-called problematic behaviours, tracked through repeated trials and charts. The goal is to make the autistic person appear more like those of the dominant culture, regardless of whether the change is distressing, unwanted, or coerced. Consent is often absent entirely, particularly for children, and distress is reinterpreted as noncompliance. Even success, within this model, is suspect: if a behaviour disappears, the data declares improvement—even if the child has simply shut down, masked more effectively, or withdrawn entirely.
In this framework, data does not serve to understand the person; it serves to justify controlling them. The more quantifiable we are, the more governable we become. Diagnoses, eligibility decisions, behavioural plans—all hinge on numerical thresholds and statistical norms that were never designed with our liberation in mind. What they offer is not care, but containment.
Major Players: Architects of Enclosure
The standard narrative in Western psychiatry often positions the Global North as the rational counterpoint to the coercive, politically weaponised practices of the Soviet era. We are quick to condemn the diagnosis of “sluggish schizophrenia” as a tool for silencing dissent, or to cite the political abuses of Pavlovian conditioning and psychiatric internment. But we rarely examine our own systems with the same lens. And yet, the development of autism diagnosis in the West—its criteria, its enforcers, its consequences—was no less shaped by ideology, no less driven by a desire to control behaviour, and no less willing to sacrifice human dignity in the name of statistical order. The difference is that our enclosures were built with academic credentials, not prison bars. Our behavioural reprogramming was rendered palatable by data tables, not just dogma.
Leo Kanner, long credited with “discovering” autism in 1943, did not merely describe a condition—he constructed a problem. His early case studies pathologised traits like sensory sensitivity, deep focus, and relational difference, casting them as signs of innate emotional deficiency. He characterised autistic children as isolated, aloof, and inaccessible, reinforcing the idea that their humanity was somehow incomplete. His work gave rise to a parent-blaming framework that dovetailed neatly with eugenic fears of degeneration: if the autistic child was unfeeling or unreachable, perhaps the defect lay not just in them, but in their heredity. Kanner’s diagnostic legacy was less about care and more about classification. He helped formalise autism as something that needed to be managed, separated, and ultimately, rendered productive—or invisible.
Hans Asperger, working in Nazi-era Vienna, occupies a darker space still. Whilst often lauded for identifying “high-functioning” autism, Asperger was complicit in a eugenic regime that sorted children by perceived social utility. Those he deemed capable of becoming useful citizens—often boys with academic strengths—were protected, at times even praised. But others, particularly those with higher support needs, were funnelled toward clinics like Spiegelgrund, where they were experimented on, neglected, or killed. Asperger’s clinical gaze was not neutral; it was a filter through which the value of a life was determined. And although he likely had access to the pioneering work of Grunya Sukhareva—who had described similar profiles with far more nuance and compassion nearly two decades earlier—he did not cite her. Whether by design or by erasure, her inclusion of girls, her resistance to moral judgment, and her refusal to pathologise difference had no place in the fascist medical order. And in many ways, no place in ours either.
Then came Ivar Lovaas, whose work in the United States laid the behavioural foundation for what would become ABA. Lovaas did not conceal his motives. He described autistic children as “not people in the psychological sense” and proposed that through systematic reinforcement and punishment, one might build a person where before there was none. His early work involved electric shocks and other aversives. Later, his methods would be sanitised, datafied, and mainstreamed—but the aim remained: to make autistic children indistinguishable from their peers. Lovaas also co-founded the UCLA Feminine Boy Project, which used similar techniques to “correct” gender non-conforming behaviour. In both cases, his methodology was rooted in the same logic: deviation from the norm was not to be understood—it was to be eradicated.
These men did not act alone. They were part of a broader psychiatric-industrial complex that sought to reshape human behaviour to better align with normative standards of productivity, obedience, and legibility. The DSM gave their work structure; statistics gave it authority. The screening tools, the behavioural interventions, the frequency tables—all of it functioned to convert difference into deviance, and deviance into something treatable, correctable, or disposable. This system, for all its claims to clinical neutrality, operated no differently than the Soviet one it so often condemned. It pathologised nonconformity. It medicalised dissent. It enforced compliance. It did so not through political slogans, but through standard deviations and significance levels.
What might have been possible had we followed Sukhareva instead—had the field honoured her early, empathetic descriptions of autistic children not as defective, but as different? Her work recognised sensory and motor differences, emotional complexity, and the adaptive potential of special interests. She did not seek to fix these children, but to understand them. That path was available to us. We simply chose another.
And we are still living inside the choice.
What We Inherit: Current Implementation Challenges
What we inherit from this lineage is not simply a set of outdated ideas, but an active infrastructure of misunderstanding. The systems that diagnose, educate, fund, and supposedly support autistic people continue to be shaped by statistical tools that were never meant to see us fully. These tools, derived from the same eugenic frameworks and behavioural algorithms detailed above, have not only survived—they’ve become embedded in policy, practice, and public consciousness. Their language is still one of thresholds and burdens, and their impact is most visible in the ways autistic people are misrecognised, mistreated, or missed altogether.
Diagnostic ambiguity remains one of the most pervasive and damaging challenges. The criteria used to determine autism are narrow, fixed, and still largely normed on studies of cisgender, white, male children. Autistic people who present differently—those who mask, who use alternative forms of communication, or whose distress is internal rather than disruptive—are often excluded. High-masking autistic girls and gender-diverse people, in particular, are routinely misdiagnosed with anxiety, borderline personality disorder, or left undiagnosed entirely. The growing use of Zoom and telehealth for assessments has only sharpened this problem. The flattened screen environment rewards surface-level performance and punishes sensory regulation strategies—eye contact, tone, posture, and language fluency are all misread as evidence of neurotypicality. Those who’ve spent their lives learning to blend in are penalised by the very success of their adaptation.
Training gaps compound the issue. Clinicians are often taught to rely on standardised screening tools—tools that generate scores, flags, and cut-offs—but receive little training in recognising the full range of autistic presentation. Diagnosis becomes a matter of arithmetic, not understanding. Professionals tally up symptoms whilst clients wait in the hallway, their humanity reduced to a checklist. There is no time for context, no attention to developmental history or trauma, no curiosity about what the traits mean to the person. Instead, difference is filtered through a deficit lens, and anything that cannot be quantified is ignored. Those who don't fit the template are told they don’t qualify, even when they are drowning.
These diagnostic decisions have real-world consequences. Schools, insurance companies, and government services often treat DSM criteria as fixed gospel, failing to adapt as science, community insight, and lived experience evolve. Support plans are denied because the scores are “too low.” Access to therapy is cut off because the person is “too high-functioning.” Students are pushed out of classrooms because their behaviours don’t map neatly onto bureaucratic definitions of need. These systems were designed not to accommodate complexity but to filter it out.
All of this unfolds under a broader cultural backdrop where statistics still drive stigma. Public conversations about autism are dominated by numbers—“one in thirty-six,” “higher risk of aggression,” “increased healthcare burden.” The language of prevalence and cost is rarely accompanied by discussions of systemic violence, ableism, or unmet needs. Instead, autism is presented as a problem to be managed, a threat to be minimised, a deviation from the mean. We are rendered legible only when we become legible as risk.
What we inherit, then, is a map of care designed to keep us out. It is calibrated not to see autistic life in all its richness, but to categorise it, limit it, and—where possible—erase it. We are measured constantly, and yet rarely understood.
Reclaiming Autistic Narrative Sovereignty
Against the weight of this history, a countercurrent has emerged—rooted not in pathology, but in lived experience; not in statistical norming, but in narrative sovereignty. Autistic people are no longer waiting to be studied, scored, or defined by others. We are writing, theorising, organising, and remembering ourselves into visibility, reclaiming the language and frameworks that have long been used to contain us. This movement is not simply a call for better treatment within existing systems. It is a refusal to be positioned as research subjects whilst others extract profit, power, and legitimacy from our pain.
The neurodiversity paradigm, driven largely by autistic thinkers, writers, and researchers, fundamentally rejects the deficit model that has shaped autism discourse for decades. It challenges the idea that autistic traits are problems to be solved and instead frames them as natural variations in human neurology—variations that carry both strengths and struggles, but that require understanding, not correction. Neurodiversity-affirming scholarship has exposed the harms of behaviourist interventions, questioned the validity of current diagnostic criteria, and demanded an overhaul of clinical, educational, and research practices. This work does not simply critique—it reconstructs.
Yet, even as this movement grows, the machinery of the diagnostic-industrial complex continues to operate, and often co-opts the language of inclusion whilst doubling down on control. Nowhere is this clearer than in the legislative and research frameworks built around the Autism CARES Act in the United States. Despite its name, the Act has disproportionately funnelled funding into biomedical and genetic research, surveillance initiatives, and applied behaviour interventions. Large-scale data collection efforts—such as those conducted by the CDC and private research firms—rely on statistical models that frame autistic people as public health risks or financial burdens. The numbers generated are then used to justify further funding for “early intervention” and “cost-reducing therapies,” whilst autistic-led services, community supports, and access to peer-directed care remain grossly underfunded. The autism research economy is vast, and it thrives on autistic people being seen as problems to be managed, not as thinkers in our own right.
But we are not passive data points. We are analysts of our own experience. Autistic writers, researchers, educators, and activists are reframing what it means to understand autism—not as a disorder to be diagnosed, but as a way of being in the world that carries its own logic, temporality, and relationality. We are building new frameworks grounded in ecological validity and relational ethics. We are asking different questions: not “How can we make autistic people behave more normally?” but “What conditions help autistic people thrive?” Not “How do we reduce challenging behaviours?” but “What unmet needs or systemic violences are those behaviours responding to?”
To reclaim autistic narrative sovereignty is to centre our interpretations of our own lives. It is to name the violence of systems that treat our difference as deviation. It is to build knowledge not through comparison to the norm, but through dialogue with each other. This is not just academic or therapeutic—it is political. Because if we define ourselves, we can redefine the systems that surround us. And if we tell the story differently, perhaps we can begin to live differently too.
Final thoughts …
Autism has never existed outside politics. The language we use to describe it—clinical, statistical, diagnostic—was not born of neutral observation, but forged in a crucible of control. Psychiatry and statistics were co-constructed in the service of eugenics, bound together by the belief that human variation could and should be measured, ranked, and corrected. From the early biometric models of Galton and Pearson to the rigid diagnostic thresholds of the DSM, autistic lives have been shaped by systems that were never designed to see us, let alone support us. That legacy persists—not just in outdated terminology or historical footnotes, but in the very criteria, tools, and assumptions that still govern how autism is recognised and responded to today.
To name this history is not to suggest that tools of measurement have no place. It is to insist that their origins matter, that their uses must be interrogated, and that their authority is not immune to critique. Data can serve liberation—but only if it is reclaimed, repurposed, and made answerable to the communities it purports to represent. Diagnosis can be a gateway to understanding—but only if it is wielded with humility, context, and consent. The problem is not that we measure, but that we have allowed measurement to dictate meaning, to overwrite narrative, to stand in for care.
What might it look like to build systems of support that do not require our compliance with someone else’s bell curve? What could we design if we started not from deficit, but from relationality—if we treated autistic people not as outliers to be corrected, but as co-constructors of knowledge, culture, and kinship? These questions are not rhetorical. They are urgent. Because as long as care is conditioned on conformity, and legitimacy on legibility, autistic people will remain inside a quantified cage.
It’s time to open the door.
sea goats