The Making of Masking

Masking is arguably one of the most influential additions to the modern autism diagnostic framework.

And a little-known part of the history of ‘autistic masking’ is that the construct entered the DSM-5 not through a programme of empirical research, but through sustained political lobbying.

An account by Kapp and Ne’eman, published in 2020, describes a 3-year campaign by members of the Autistic Self Advocacy movement to influence the wording of the DSM-5 autism criteria. At the time, the Neurodevelopmental Disorders Work Group was deliberating over a consequential and contentious change: the proposal to merge Autistic Disorder, Asperger’s Disorder and PDD-NOS into a single Autism Spectrum Disorder diagnosis.

The advocacy group, led by individuals with Asperger’s Syndrome, was concerned that some people with existing diagnoses might no longer qualify under the proposed criteria. They also wished to ensure continuity of support and services across the lifespan, arguing against the possibility that one may ‘recover’ (or learn to adapt) from autism - and suggested that the DSM-IV had made it difficult for autistic people to obtain, or retain, a diagnosis due to masking or passing skills they had developed. According to their own account, members of the advocacy group actively sought opportunities to meet DSM committee members, and build relationships in order to influence the revision process.

The paper, published seven years later, provides a detailed first-person account of the advocacy process and the origin of the masking wording.

One of the results of this lobbying was the insertion of a short clause into Criterion C:

“Symptoms must be present in the early developmental period (but may not become fully manifest until social demands exceed limited capacity, or may be masked by learned strategies in later life).”

From this brief addition, a much broader explanatory narrative gradually emerged - one involving hidden autism, lifelong masking, and the possibility that autistic traits had remained undetected across development. The wording itself did not establish these propositions, but it created conceptual space in which they would flourish.

The Significance of Criterion C

The significance of Criterion C lies in the conjunction of its two clauses. The first allows neurodevelopmental symptoms to become apparent “when social demands exceed capacity”. The second allows those symptoms to have been “masked by learned strategies”.

Together, they create the possibility that autism can be retrospectively diagnosed at any stage of life, in the absence of evidence that signs existed in childhood – as these may have been suppressed.

Masking was introduced as part of a much broader revision of the autism criteria. DSM-5 removed the separate Communication domain, broadened the definition of social communication difficulties, reduced emphasis on atypical language production that had previously served as a clear developmental marker, and introduced a more dimensional understanding of symptom emergence. For readers unfamiliar with DSM-IV, these changes substantially altered what clinicians understood by the term "communication difficulties".

Additionally, a diagnosis of ‘Autism’ (or ASD) now frequently occurred at precisely the moment that decades of developmental research had shown adolescents and young adults - particularly girls - were entering the age range associated with rising rates of anxiety, emotional distress and social difficulties.

The masking construct had arrived on the diagnostic stage in the middle of a developmental traffic jam.

The Scientific Response

How do we know that masking exists? What evidence supports it?

At the time it entered the DSM-5, the construct of masking (or camouflage, as it was also known) had not been clearly defined, operationalised or empirically validated. It’s impact had not been field tested as part of the DSM process (McPartland et al, 2012). The DSM-5 committee had not established what, precisely, qualified as the autistic behaviours that were being concealed. What behaviours were being masked?

Limited guidance was provided to clinicians. Nor had developmental boundaries been established. For example, how plausible was it that a child could successfully conceal neurodevelopmental features of autism throughout childhood? Under what circumstances might this occur? How common would it be? How does masking relate to those formerly diagnosed with Autistic Disorder?

These questions were not explored, nevertheless masking had now entered the diagnostic manual, giving it immediate visibility within both clinical and research settings.

The scientific effort to define and operationalise masking did not begin in earnest until several years after its appearance in the DSM-5.

Beginning around 2016, researchers such as Hull and Bargiela began interviewing later-diagnosed adults - mostly women - who had received a Asperger diagnoses under DSM-4 or autism diagnoses under the broadened DSM-5 criteria. Participants were invited to look back over their lives through the newly emerging idea of “camouflaging” or “masking”, reflecting on whether they had hidden, suppressed or compensated for autistic traits during childhood and adulthood.

The accounts were compelling and quickly gained traction. But they also relied on retrospective interpretation: revisiting childhood experiences through a concept that had only recently entered the diagnostic conversation. These questions were being asked of people who also reported significant anxiety, and in one sample more than 30% reported a history of sexual assault. Diagnoses were not always independently verified, co-occurring conditions were not consistently controlled for, and participants were largely recruited through online autism communities.

This does not invalidate their experiences. But it does make it difficult to know exactly what experience was being captured.

The themes identified in these interviews informed the development of the Camouflaging Autistic Traits Questionnaire (CAT-Q), which rapidly became the dominant measure of masking. The questionnaire was then tested on similar samples of late-diagnosed autistic adults and shown to be internally consistent.

To summarise the process:

1. Introduce a concept

2. Build a questionnaire from retrospective accounts

3. Test the questionnaire on a similar group

4. Interpret endorsement as evidence that the construct has been captured

This is methodologically problematic, as the sequence can create problems of circularity, particularly when the same population contributes to concept generation, instrument development and validation.

Over the following decade, hundreds of studies used the CAT-Q as a proxy for autistic masking (Arnold et al, 2026). Most were cross-sectional. Many were correlational. The literature expanded rapidly. Autistic Burnout, mental health problems, suicide - were attributed to masking.

Inferences became headlines.

But one basic question remained unanswered:

What exactly was the questionnaire measuring?

Social Transmission

All of the above was unfolding during what might be called the Age of Social Transmission.

Ideas no longer spread only through journals and conferences. They moved through podcasts, social media, advocacy networks, online communities and algorithms. Concepts that were still awaiting empirical validation could nonetheless achieve enormous cultural traction if they resonated with people’s lived experience.

This does not imply that the idea was false, nor that those who recognised themselves in it were mistaken. It simply reflects the way contemporary psychological concepts now diffuse through society.

The philosopher of science Ian Hacking had described this process years earlier. Human categories, he argued, are not passive labels. Once people adopt them, identify with them and organise their lives around them, the categories themselves begin to change. The classification loops back and alters the people being classified.

Masking was a perfect candidate for such a looping effect.

The idea was immediately recognisable. It spoke to familiar experiences of adaptation, exhaustion and self-monitoring. That recognisability helped it travel.

But What Is Masking?

At its heart, discriminant validity asks a deceptively simple question:

Is there actually a distinct thing called masking?

Or are we describing a collection of familiar human behaviours—social adaptation, impression management, self-monitoring, anxiety management, attachment strategies, trauma responses - and giving them a new name?

Was masking actually specific to autism?

Or was it capturing something broader and more familiar?

Professor Fombonne raised this concern in 2020, noting that many descriptions of masking appeared to overlap with impression management - a sophisticated social skill that sits somewhat awkwardly alongside a condition defined by difficulties in social communication.

Williams later observed that the CAT-Q bore a striking resemblance to measures of social anxiety and self-monitoring. Only a small number of items referred to behaviours commonly regarded as specifically autistic, such as suppressing stimming behaviours, or reduced eye contact.

Yet the literature continued to grow. Correlations accumulated. Associations were reported. Interpretations multiplied.

Masking as a theoretical proposition

In making their case around the introduction of masking, Kapp, Ne’eman and colleagues drew on the clinical observations of Hans Asperger, Lorna Wing and Tony Attwood. These clinicians had long described individuals with Asperger Syndrome learning social rules, copying gestures, rehearsing conversations and developing scripts as ways of navigating the social world.

Individuals with Asperger Syndrome were often recognisably different because of atypical prosody, pedantic or overly formal language, pragmatic language difficulties, one-sided conversation, unusual gestures, literalness, reduced awareness of non-verbal communication and an idiosyncratic interpersonal style. These were the characteristics that distinguished the Asperger phenotype and were arguably the behaviours that individuals were attempting to conceal through masking.

But there was an important caveat.

This group might learn explicit social rules. They might imitate others. They might become increasingly socially skilled. But they generally remained, in Wing’s memorable phrase, “observably odd” (Wing, 1981) - language that many autistic people today may understandably experience as confronting, despite its use as a descriptive clinical observation at the time.

Compensation was not invisibility.

Fast forward to 2026.

Since 2013, masking has increasingly become an explanatory framework through which many people have recognised their own experiences of social adaptation, self-monitoring and exhaustion, and come to understand these as evidence of a previously unrecognised neurodevelopmental condition.

Over time, the construct itself has evolved. Accomplished professionals—including clinicians and other mental health practitioners, skilled communicators and confident public speakers - have increasingly described their longstanding social adaptation through the lens of hidden autism or ADHD, although masking does not form part of the ADHD diagnostic criteria.

One of the most visible developments has been the emergence of “high masking” autism, particularly among professional women. Many describe the cumulative demands of balancing careers, parenting, caregiving, and the emotional labour of sustaining families and relationships while maintaining an outwardly composed and capable exterior.

While not wishing to invalidate the many people who connect with this description, it represents a very different picture from the one originally described by Lorna Wing.

The “science” behind Masking

As Flake and Fried (2020) have argued, a measure can demonstrate impressive internal consistency while still failing to measure the thing it claims to measure. That possibility was precisely what critics such as Fombonne and Williams were highlighting.

In 2024, several of the field’s leading research teams examined whether masking, as operationalised within the CAT-Q, could reliably distinguish autistic from non-autistic individuals.

Studies by Ai and colleagues and van der Putten and colleagues found substantial overlap with impression management, emotion regulation and anxiety-related processes. Masking was not unique to autism, nor was it confined to autistic individuals.

Cross-sectional studies can identify associations, but they cannot determine whether masking predicts later psychological distress. Longitudinal studies are better suited to testing that question.

If masking contributed to autistic burnout and poorer mental health, this prediction should be evident over time.

In 2025, van der Putten and colleagues provided the first opportunity to examine this directly, following autistic adults longitudinally.

The findings painted a more nuanced picture than had emerged from the cross-sectional literature. Depending on the individual and the context, masking appeared to be adaptive, burdensome, or have little apparent effect. Overall, its association with later mental health outcomes was weak and, if anything, slightly positive.

Rather than supporting a simple pathway from masking to psychological distress, the authors concluded that future research should focus on individual differences - traits such as conscientiousness, neuroticism and negative affect - that might explain why masking proves helpful for some people and burdensome for others.

These findings shifted attention back toward a familiar question in psychology: whether similar behaviours might arise from different developmental pathways, with personality, anxiety, attachment, temperament and autism each contributing in different ways.

Very recently, some of the most prominent researchers associated with the construct of masking have cautioned against the use of the CAT-Q in assessment or diagnosis (Hannon, Mandy, Hull et al., 2026).

And this is significant, as the CAT-Q had effectively become a proxy for “autistic masking” in public discourse.

However, rather than measuring the concealment of autism-specific behaviours, the construct appears to have drifted towards measuring the subjective emotional experience of social adaptation.

Where to from here?

The behaviours now grouped under the term masking are real. Social adaptation, self-monitoring, impression management, role performance and compensatory strategies have been described for decades by psychologists, sociologists and developmental researchers. Some are conscious. Some are not. All are part of being human.

Masking-like behaviour can arise through personality (perfectionism, conscientiousness, agreeableness), temperament, emotional neglect in childhood, shyness, introversion, anxiety, trauma, autism, professional codes of conduct, and socially performative behaviour.

A lifetime of adaptation can be exhausting, particularly when juggling work, home, children, ageing parents, friendships and modern life.

The difficulty is that similar subjective experiences can arise from very different developmental backgrounds and life experiences. Masking entered the diagnostic framework before it had been established as a valid psychological construct, and much of the subsequent literature focused on elaborating the idea rather than testing whether it represented a distinct phenomenon. That matters.

Because the existence of a subjective experience does not automatically validate the explanation that is attached to it.

The question is therefore not whether people mask, but how those experiences are best understood, what evidence supports one explanation over another, and how we distinguish among competing interpretations.

This is particularly important because, among late-diagnosed people, masking is increasingly used to explain the absence of observable developmental features in childhood. In doing so, the construct may change not only who receives a diagnosis, but also how people come to understand their own developmental histories and life experiences.

This paper reflects my interpretation of the current literature. I welcome thoughtful discussion and alternative interpretations in the comments.

Kapp & Ne’eman (2020) Lobbying Autism’s Diagnostic Revision in the DSM-5. In: Kapp, S. (eds) Autistic Community and the Neurodiversity Movement. Palgrave Macmillan, Singapore