Umwelten

The ethics of intervention

12 min read

Summary

Every intervention in a neurodivergent life rests on an assumption about what that life should look like. The medical model assumes the goal is to reduce symptoms, to bring the person closer to a neurotypical norm. The neurodiversity paradigm asks a different question: does this intervention expand what the person can do and be, or does it narrow their life to fit someone else’s expectations?

This page does not argue that all intervention is wrong. It argues that intervention without ethical scrutiny is dangerous, and that the history of autism intervention provides ample evidence of why. The questions are not complicated: Who benefits? Who decided? Was the person asked? Could the environment be changed instead? These questions have not always been asked, and when they have been, the answers have not always been comfortable.

The normalisation problem

The foundational ethical tension in autism intervention is the distinction between reducing suffering and reducing visible difference.

An autistic child who experiences genuine distress from sensory overload in a school dining hall is suffering, and helping that child is straightforwardly justified. But the help could take two forms. One modifies the environment: acoustic panels, a quieter seating area, permission to eat elsewhere. The other modifies the child: sensory desensitisation, compliance training, or simply the expectation that they will learn to tolerate it. The first addresses the mismatch. The second addresses the child.

When the goal of intervention is “indistinguishability from peers,” as Lovaas (1987) explicitly stated it was (see Lovaas and the origins of ABA), then the benchmark for success is not the person’s wellbeing. It is their ability to pass as neurotypical. Jaswal and Akhtar (2019) put the problem precisely: much of autism intervention conflates being socially engaged with appearing socially engaged. A child taught to make eye contact has learned a performance. Whether that performance reflects, or costs, their actual social experience is a different question.

The Autistic Self Advocacy Network’s intervention ethics paper (2021) documents where this logic leads. When the goal is to eliminate visible difference, behaviours that serve the autistic person (stimming as regulation, gaze aversion as sensory management, scripts as social navigation) become targets for reduction. The intervention succeeds by the clinician’s metric while failing by the person’s. Rapp and Vollmer (2024) confirmed this empirically: when autistic adults were asked which intervention goals they found acceptable, they endorsed quality of life, safety, and autonomy. They rejected normalisation benchmarks.

Chapman (2023) places this in a broader political frame. Empire of Normality argues that the drive to normalise disabled people is not a neutral scientific programme. It is a product of industrial capitalism’s need for standardised labour, dressed in clinical language. The “empire of normality” defines which minds are productive, which bodies are compliant, and which deviations require correction. Intervention that targets visible autistic behaviour without asking whether the behaviour serves the person is, on this account, not healthcare. It is social conformity enforcement.

This is a strong claim, and reasonable people disagree about its scope. But the core insight is hard to dismiss: the history of autism intervention has repeatedly prioritised the comfort of the observer over the wellbeing of the observed. Recognising this does not require abandoning intervention. It requires changing what intervention is for.

The ABA question

Applied Behaviour Analysis occupies a contested position in autism intervention. It has the largest evidence base of any autism intervention for behaviour change. It also has the most sustained ethical critique from the community it claims to serve.

The wiki’s treatment of ABA’s history and its relationship to sensory processing lives on dedicated pages (see Lovaas and the origins of ABA and ABA and sensory processing). What belongs here is the ethical architecture of the debate.

Kupferstein (2018) surveyed 460 autistic adults and found that 46% of those exposed to ABA met the diagnostic threshold for PTSD. The study has been methodologically contested (self-selected sample, retrospective self-report, no control for other trauma sources), and the ABA field has responded with both critique and attempted reform (Wilkenfeld and McCarthy 2020; Melton et al. 2025). But the ethical significance of the finding does not rest on whether the precise number is correct. It rests on the pattern: a substantial proportion of the people subjected to this intervention report it as traumatic, and the field spent decades not asking them.

That silence is itself an ethical failure, and it connects directly to the epistemic injustice described on the epistemic justice page. If autistic people’s testimony about their own experience of intervention had been treated as credible evidence from the beginning, the debate would look different. The ethical critique of ABA is not primarily about technique. It is about who counts as a reliable reporter of their own experience.

Contemporary ABA practitioners increasingly describe their work as neurodiversity-affirming, focused on functional skills rather than normalisation. Whether this represents a genuine transformation or a rebrand is actively debated within and outside the field. The ethical test remains the same: does this intervention expand what the person can do and be? Or does it teach them to perform in ways that serve someone else’s definition of success?

The most fundamental ethical question in intervention is also the simplest: did the person agree to this?

For autistic adults without intellectual disability, the answer should be straightforward. Adults have the right to accept or refuse treatment. In practice, the picture is muddied by power imbalances in clinical relationships, by the pressure to “try everything,” and by the social consequences of refusing intervention in contexts (employment, education, social services) where compliance is rewarded and refusal is punished.

For children, the question is harder. Parents consent on behalf of children for all medical interventions, and autism intervention is not unique in this respect. But it is unusual in one way: the intervention often targets behaviours that the child may, as an adult, come to value. A parent who consents to a programme that eliminates stimming is making a decision about a behaviour whose significance the child cannot yet articulate and the parent may not understand. Bottema-Beutel, Malay, and Gates (2023) argued that early intervention goals should be evaluated against what the child’s future autistic self might want, not only against what the parent or clinician wants now.

For autistic people with intellectual disability, the problem deepens. Proxy consent (a guardian or family member consenting on the person’s behalf) is the legal standard in most jurisdictions. Wasserman and Asch (2016) showed that the two standard models for proxy consent both fail in this context. “Substituted judgement” (deciding as the person would decide if they could) requires knowledge of the person’s preferences that may not exist. “Best interests” (deciding what a reasonable person would want) defaults to neurotypical assumptions about what a reasonable person would want, which may not reflect what the autistic person with ID would choose.

The alternative is not to abandon proxy consent but to transform it. Beemer and Sterkenburg (2024) argued for a relational autonomy framework: rather than asking “what would this person choose?” (which may be unknowable), ask “how can we maximise this person’s participation in the decision, given their communicative capacities?” This means accessible information, adapted consent processes, and a presumption of competence that can be adjusted but should not be abandoned.

The capabilities test

The capabilities approach (see The capabilities approach) offers a framework for evaluating intervention that does not depend on the medical model’s deficit framing or require the full political apparatus of the neurodiversity paradigm.

Goering (2022) applied Martha Nussbaum’s ten central capabilities to autistic adulthood and proposed a simple test: does this intervention expand or contract the person’s capability set? An intervention that enables an autistic person to communicate their needs (expanding capability) is ethically different from one that teaches them to suppress stimming in public (contracting a self-regulatory capability to satisfy a social norm). Both might be described as “treating autism.” They are not ethically equivalent.

The capabilities framework is useful because it avoids two traps. It does not require you to deny that autism can be disabling (the capabilities approach takes seriously that some people cannot do things they want to do). And it does not require you to accept that all intervention is justified (the test is whether the person’s capabilities are expanded, not whether the clinician’s goals are met).

Nicolaidis, Milton, and colleagues (2022) found empirically that autistic people define flourishing in terms that map well onto capabilities: autonomy, meaningful work, community belonging, the ability to live according to their own values. Symptom reduction barely featured. If intervention aims to support flourishing, it should aim at what autistic people actually say flourishing means.

The environment question

The most ethically straightforward intervention is often the one that does not target the person at all.

If an autistic employee is overwhelmed by the office, the options include: sensory integration therapy to increase their tolerance, coaching to help them mask the overwhelm, or noise-cancelling headphones and a quieter workspace. The first two modify the person. The third modifies the environment. The ethics of the first two require justification (why is the person being changed rather than the setting?). The ethics of the third barely need stating.

This is not to say that person-directed intervention is always wrong. An autistic person who wants to develop strategies for navigating noisy environments is exercising autonomy. The ethical problem arises when environment modification is not considered, when the default assumption is that the person must adapt to the environment rather than the other way around. The evidence problem in sensory interventions (see The evidence problem in sensory interventions) compounds this: many person-directed sensory interventions have weak evidence bases, while environmental modifications (acoustic treatment, lighting changes, reduced demand) often have clearer and more immediate effects.

The social model of disability (see Models of disability) makes this the default ethical position: disability is created by the environment’s failure to accommodate, so the first obligation is to fix the environment. The capabilities approach reaches a similar conclusion by a different route: if the goal is to expand what a person can do, changing the environment to enable functioning is often more effective and less intrusive than changing the person.

The timing problem

Early intervention in autism carries a particular ethical weight because it intervenes before the person can express preferences about their own life.

The evidence for early intervention is strongest for communication support and weakest for behaviour modification (Bottema-Beutel et al. 2023). Fletcher-Watson, Pain, and Hammond (2024) interviewed parents of young children at elevated likelihood of autism and found a tension that the research literature rarely acknowledges: parents valued early identification but worried about the pressure to intervene in a child who might not need or want the specific intervention being offered. The “earlier is better” narrative leaves little room for watchful waiting, for allowing the child’s own trajectory to become visible before deciding what, if anything, to change.

The ethics sharpen further with pre-symptomatic intervention, the emerging possibility of predicting autism from biomarkers before behavioural signs appear (see Autism screening, surveillance, and genetic testing). If a child can be identified as likely autistic at six months, what does ethical intervention look like? The answer depends entirely on what you believe autism is. If it is a disorder to be prevented, early intervention is urgent. If it is a neurological architecture to be supported, early intervention means preparing the environment, not pre-empting the child.

Open questions

How should intervention ethics apply to autistic people with severe intellectual disability and high support needs, for whom environmental modification alone may not be sufficient and whose preferences may be difficult to determine? The capabilities framework offers a starting point (expand what the person can do and be), but its application in practice requires more work than the philosophy has yet done.

Can ABA be reformed into something ethically defensible, or is the normalisation logic structural to the approach? This question divides not only the autistic community and the ABA field but also disability scholars, and the answer may depend on distinguishing between ABA-the-science (behaviour analysis applied to any goal) and ABA-the-industry (a clinical and commercial system with its own institutional incentives).

What would a truly neurodiversity-affirming intervention framework look like in practice? Wagland et al. (2025) have begun mapping this for children, but the field is young and the gap between principle and implementation is wide.

Key sources

  • Autistic Self Advocacy Network. (2021). For whose benefit? Evidence, ethics, and effectiveness of autism interventions. Washington, DC: ASAN.
  • Bottema-Beutel, K., Malay, R. and Gates, J.A. (2023). At a crossroads: reconsidering the goals of autism early behavioural intervention from a neurodiversity perspective. Frontiers in Psychology, 14, 1106154. doi: 10.3389/fpsyg.2023.1106154
  • Chapman, R. (2023). Empire of Normality: Neurodiversity and Capitalism. London: Pluto Press.
  • Fletcher-Watson, S., Pain, H. and Hammond, S. (2024). The earlier, the better? An in-depth interview study on the ethics of early detection with parents of children at an elevated likelihood for autism. Journal of Autism and Developmental Disorders, 54(5), 1847-1862.
  • Goering, S. (2022). A capabilities approach to understanding and supporting autistic adulthood. Nature Reviews Psychology, 1(10), 620-631.
  • Jaswal, V.K. and Akhtar, N. (2019). Being versus appearing socially uninterested. Behavioral and Brain Sciences, 42, e82. doi: 10.1017/S0140525X18002091
  • Kupferstein, H. (2018). Evidence of increased PTSD symptoms in autistics exposed to applied behavior analysis. Advances in Autism, 4(1), 19-29. doi: 10.1108/AIA-08-2017-0016
  • Rapp, J.T. and Vollmer, T.R. (2024). The social validity of behavioral interventions: seeking input from autistic adults. Journal of Autism and Developmental Disorders, 54(6), 2341-2360.
  • Wagland, Z., Sterman, J., Scott-Cole, L., Spassiani, N. and Njelesani, J. (2025). Promoting neurodiversity-affirming care for autistic children: a scoping review. AJOB Neuroscience, 16(1), 42-64.
  • Wasserman, D. and Asch, A. (2016). Autism, intellectual disability, and a challenge to our understanding of proxy consent. Medicine, Health Care and Philosophy, 19(2), 161-176. doi: 10.1007/s11019-016-9745-y

Nearby

Current page
Visited
Not yet visited
Tag

Backlinks