In certain situations when the symptoms are too few, too mild, too many, too severe, or too spread out across many domains, the diagnostic possibilities begin to all blur together. The presentation takes on an undifferentiated quality, a blob of psychopathology that you could reasonably describe using various names for differentiated things if you had to, but anything you call it is too ill-fitting, too artificially precise.
It’s like a low-resolution, pixelated photo.
A blurry photo can be enhanced into a variety of possible faces, and we can arrive at a very plausible but untrue end result.
In the case of a blurry photo, there is usually a truth of the matter. There is or was a face that has lost its features and we can’t reconstruct it anymore with accuracy because of information degradation.
There are other situations, however, where there is no truth of the matter, like generating a low-resolution blur of a face from scratch disconnected from any particular face. There is no truth of the matter regarding what face it really is.
And there are situations where the relevant concepts simply break down or lose their meaning. For example, in the fraction of a second after the Big Bang, the temperature was so high that electromagnetism and the weak nuclear force did not exist as separate forces. There was a unified electroweak state in which neither electromagnetism nor weak nuclear force was a meaningfully separate category. As the universe cooled, the two forces emerged as distinct. It is not that we cannot tell the two forces apart at those high temperatures; rather, there aren’t two forces to tell apart.
Similarly, I feel that there is a point where autism gets too pixelated and hard to differentiate from its cousins; the uncertainty becomes too high. And then there is also a point where it no longer makes sense to talk about “autism” at all because what we are dealing with is just a blob of psychopathology.
In my physics analogy, the dial is energy or temperature. In the case of autism, we have multiple dials to consider: there is developmental history, severity, comorbidities, and the explanatory level of description.
There is too much uncertainty in situations like:
Diagnosis of autism in adults where developmental history is unknown or unreliable, and years of compensatory behaviors have made assessment difficult.
Differentiating autism from severe social anxiety, schizoid personality traits, etc.
Differentiating from “broader autism phenotype” or subclinical autistic traits, in situations where impairments are hard to assess.
Like the photo “enhancement” example above, a clinician can fill in the missing pixels based on educated guesses and priors, but there is no guarantee that the autism reconstructed is the correct reconstruction.
Developmental time is perhaps the closest parallel to the early universe situation. Take a 6-month-old baby; it isn’t that the baby is already autistic, and we just can’t detect it yet. Arguably, there isn’t yet an autism to detect. What exists, probably, is a kind of undifferentiated neurodevelopmental liability that can manifest in different ways later in development, including autism.
At the level of genetics also, there is no such thing as autism, differentiated from other neurodevelopmental and psychiatric conditions.
And then in cases of dense psychiatric multimorbidity, when a person has elevated ratings on many or nearly all psychopathology dimensions, there may be no distinct autism to separate from the rest of psychopathology.
In physics we can distinguish between measurement breakdown and the actual breakdown of phenomena, but in our current science of psychopathology, we cannot empirically distinguish between “measurement is no longer possible due to a high margin of error” and “the concept no longer applies.”
The usual commentary on autism overdiagnosis these days, e.g. Uta Frith and company, leaves me fairly frustrated and dissatisfied. The impression I get is that the diagnostic concept of autism and the composition of people being diagnosed, and those self-diagnosing, has expanded and changed so substantially since the 1980s that folks like Frith, who were used to seeing autism comorbid with intellectual disability or significant speech impairments, struggle to make sense of it. Their instinct is to be skeptical (rather than curious) and to try to preserve some old idea of autism as more real and more genuine. But both the scientific understanding and the clinical landscape of autism have genuinely changed, and the sure ground they hope for doesn’t exist and never existed. This instinct leads them to focus on trying to exclude this or that subgroup—the mildly impaired, those diagnosed late, those who mask, etc., etc.—from the autistic camp. RETVRN!
My interpretation of the situation is different. There are definitely differences between early-diagnosed and late-diagnosed individuals. There are different patterns of severity and comorbidity that lead some people to come to attention fairly early in life and that lead some people to come to attention later in life, but as long as they share the clinical features we call autism, they are all autistic. Heterogeneity is the price we pay for syndromic diagnoses. There are also differences, for instance, between those with severe intellectual disability and mild intellectual disability, or adolescent depression and geriatric depression. Variation in correlates are reasons for specifiers and subtypes, not exclusion from the syndrome.
What makes autism “autism” rather than a different diagnosis? The field has settled on social communication impairments and restricted, repetitive behaviors as the clinical core, and as far as these clinical features are concerned, it is rather clear that they exist on a continuum, with heterogeneity existing within the continuum. There are certainly autistic folks who also have intellectual disabilities or speech impairments, but neither intellectual disability nor speech impairments are autism. To insist that only those with intellectual disabilities and speech impairments are people with “real” autism is to profoundly misunderstand what makes autism autism.
Frith thinks that autism should be reserved for people with severe communication challenges, that people who are fluent in conversation cannot be autistic. When Holden Thorp (Editor-in-Chief of the Science family of journals and diagnosed autistic) asked her that if people like him are not autistic, then what are they, she didn’t quite know what to say:
“That’s the main place where Frith and I diverge. Even someone like me who has most of life’s possible advantages has experienced difficulty from autistic traits imposed by society and the environment — and their preferences for a neurotypical affect. (I told Frith that I had a meltdown at the hotel before I came over; it had to do with technology, as it so frequently does.) Now imagine how that plays out for someone with fewer resources who is also excluded based on their other identities.
Finally, I asked her if people agreed with her, what would she call the rest of us who are no longer autistic. She had noticed I was staring out the window and agreed that was an autistic trait. When I explain why I do that to someone, I just say it’s because I’m autistic. What should I say instead?
She didn’t know. She suggested “autistic-like.” Then she tried “autistic temperament.”
I didn’t think these would fly, and she reluctantly agreed.”
Autistic-like and autistic temperament… woah, it almost sounds like autism exists on a… spectrum?!
We should obviously recognize and better characterize this heterogeneity, and we shouldn’t accept misdiagnosis, but that doesn’t mean we can chop off the spectrum for arbitrary reasons. Frith raises skeptical questions that are fine to raise but is unable to offer a better account of what autism is.
Where I do agree with Frith is that the reliance on self-report, appeals to an overly broad notion of masking (but not masking per se1), social-media-fueled self-diagnosis, and inadequate consideration of other diagnostic possibilities have made diagnosis of autism in adolescents and adults, especially those with only mild impairments and high comorbidity, more uncertain and fallible, and we collectively have not yet come to terms with that fallibility.
My grievance with contemporary autism is neither that people with mild impairments are getting diagnosed, nor that we now recognize and account for masking as a phenomenon. My worry is that at a certain point the confidence interval around diagnostic judgments becomes so wide that we can no longer say with meaningful confidence what we are dealing with.
We need a new clinical vocabulary to talk about this diagnostic blurriness.
In addition to already recognized constructs such as “broader autism phenotype” (no clinical impairments) and social communication disorder (social communication impairments without restricted, repetitive behaviors), there are moments as a clinician when I find myself reaching for descriptions like:
Autistic features with indeterminate developmental history
Autistic features with context-dependent impairment
Autistic features with self-reported impairment, not clinically observed
Autistic features with psychiatric multimorbidity and uncertain differentiation
I am less concerned with trying to preserve some old idea of autism or restrict it to people with severe impairments. I think people with mild impairments or people who mask also merit clinical recognition. I am more concerned with the margin of error, with epistemic uncertainty. I am worried that when developmental history is uncertain, when impairment is only self-reported, when psychopathology is spread out over multiple spectra, a diagnosis of “autism” conveys a certainty that just doesn’t exist.
See also:
See Jesse Meadows’s review of the scientific literature on masking: The Scientific Basis for Autistic Masking (So Far)
I'm probably one of the people they think have "fake" autism because I'm socially warm. What that hides is that I had intense sensory and emotional overload starting from early childhood— so much so that my mom took me to see a child psychiatrist when I was 3 in 1968. (no one did that then!) Unbeknownst to me at the time, I was diagnosed with ADHD and "hyperesthesia"— known to me I was diagnosed as "gifted" because I was reading at 3 also. I didn't like to be touched, had repetitive compulsive behaviors that I sought to hide and didn't like change. I took everything literally; before puberty, I was focused almost exclusively on ideas and reading and my particular obsessions (scifi and opera) rather than people. Was bullied, excluded etc until I realized that being obsessed with drugs was an obsession other people my age did share. What would you call that if not autism spectrum disorder? Should the spectrum be divided by IQ, even though it's really difficult to actually test in severe cases even if it may actually be high? I don't know. All I know is that being diagnosed with OCD, ADHD, depression, various addictions and even once bipolar made far less sense to me than having an underlying condition of autism via which oversensitivity led to self medication via repetitive behavior and obsessive interests and later, drugs. (And thanks Awais for your contributions to the piece I wrote about this for the New York Times, which can be found here: https://www.nytimes.com/2026/10/05/opinion/mental-health-diagnosis-labels.html?unlocked_article_code=1.GVE.80Ex.FmHG28JTFoLI&smid=url-share )
If the ground keeps moving beneath our feet, maybe we're not actually standing on "ground." What is it like to be Holden Thorp or Uta Frith? We really have no clear idea of that, and until we gain the ability to train subjects in how to bring disciplined, high-resolution, precise observation to their first-person subjective experience in a way that enables us to actually discern patterns that stand firm, we have no good maps. Our current diagnostic categories are not good maps.