Composite Conditions and the Weight of Repetition
- Troy Lowndes
- 11 hours ago
- 6 min read
Why diagnostic categories outlive their evidence, and what that costs the people inside them
By Troy Lowndes
I was assessed for ADHD in 2023 (aged 48)
The category that finally described me had been quietly rewriting itself for decades before I walked into that room. The clinician who named it was working from a field that had spent a generation un-bundling phenomena that used to travel under one label. Without that un-bundling, I would not have been legible. With it, I was. The difference between those two states is roughly forty years of research that most of the people who lived through the old framing never got to benefit from.
This piece is about that gap. Not the personal one. The structural one. The gap between when a field notices it has been wrong about the shape of a category and when the correction reaches the people the category is used on.
The pattern
Psychiatric diagnostic categories accumulate co-occurring phenomena as if they were features. A condition gets identified by its most observable surface traits. Everything that tends to show up alongside those traits drifts into the diagnostic gestalt. Research builds on the composite for a generation. Clinical training absorbs the composite. Public understanding crystallises around the composite. Then, quietly, someone does the statistical separation work that shows the composite was never one thing.
The un-bundling begins. It takes another generation to reach the clinic. Longer to reach the public. Meanwhile, the people being assessed under the composite framing continue to be assessed under the composite framing, because the composite framing is what their clinicians were trained on, and because the accumulated weight of thirty years of papers, courses, handbooks, and shorthand has gravity.
Two case studies, both still in mid-correction.
Case one: ADHD and the subtype that wasn't
Before DSM-5, ADHD was structured around three subtypes. Predominantly inattentive. Predominantly hyperactive-impulsive. Combined. The subtypes were taught as quasi-distinct presentations of the same condition, and a generation of clinicians learned to triage along those lines.
The problem was that the subtypes didn't behave like types. A hyperactive seven-year-old routinely became an inattentive twenty-seven-year-old. Same person, same neurobiology, different surface. The categorical model couldn't account for that, because types aren't supposed to morph into each other across a lifespan.
DSM-5 reframed the subtypes as presentations, acknowledging that the same person shifts across them over time. The underlying construct was recognised as a dimensional executive-function and self-regulation difference, with inattention and hyperactivity being the expressions, not the components.
That change was formalised in 2013. It is now 2026. A meaningful proportion of clinicians still think in subtypes. A larger proportion of the public thinks in subtypes. "ADHD means hyperactive boys" is a framing you can still find in the wild, operating on evidence that was dismantled before the children currently being misread were born.
The un-bundling happened. The lag is still running.
Case two: autism, alexithymia, and the work of Geoff Bird
For decades, the autism literature treated emotion-recognition difficulty as a core feature of the condition. If you met an autistic person who struggled to read facial expressions, the explanation was right there in the diagnosis. The broken mirror theory, which proposed a dysfunction of the mirror neuron system as the underlying mechanism, hardened that framing through the 2000s and travelled far beyond the scientific community.
Geoff Bird, working across King's College London and University of Oxford through the 2000s and 2010s, is one of the people who refused to let the composite stand.
His work with Richard Cook, Rebecca Brewer, Caroline Catmur and others did the patient statistical separation that eventually showed what the composite had been hiding. Roughly half of autistic people meet alexithymia criteria, a condition defined by difficulty identifying and describing one's own emotional states. Roughly half do not. When the two are disentangled and analysed separately, much of what had been attributed to autism in the emotion-reading literature actually tracks alexithymia. Autistic people without alexithymia perform typically on emotion recognition. The 2013 paper carried the title that should have been the headline of the decade: Alexithymia, not autism, predicts poor recognition of emotional facial expressions.
Alongside that, Bird and colleagues spent years challenging the broken mirror theory directly. The 2016 paper with Sowden and Dziobek was titled, with characteristic directness, Intact Automatic Imitation and Typical Spatial Compatibility in Autism Spectrum Disorder: Challenging the Broken Mirror Theory. The automatic imitation that the broken mirror theory predicted should be impaired turned out to be intact. The composite that had carried the field for twenty years was built on a mechanism that, on closer inspection, wasn't there.
Bird's corpus is one of the clearest examples in contemporary psychiatric research of long-running crusader work. Not a single breakthrough paper. A steady, methodical, decade-plus dismantling of a composite. Every year, another piece. Every year, the field's gravitational centre shifts a little.
And yet. If you ask a general clinician today whether alexithymia is a symptom of autism, you will still often be told yes. The un-bundling is in the literature. It is not yet in the room.
The weight of repetition
There is a question sitting underneath both case studies, and it is the question that the lag hinges on.
Why does correction take so long?
Not because the evidence is unclear. By the time DSM-5 changed its framing of ADHD, the evidence had been accumulating for a decade. By the time Bird and colleagues published the alexithymia-not-autism paper, the statistical case was strong. The lag isn't epistemic. It's gravitational.
This is where I find myself reaching for the instrument I've spent the last few years building. SpectralBinary is a framework I've developed for modelling communication and signal across multiple axes. Its fifth axis, added earlier this year, measures temporal persistence and signal gravity over time. I won't unpack the mechanics here. What matters for this piece is the shape of the observation.
Ideas accumulate mass by repetition. A framing that has been published, taught, clinically applied, internalised, and passed on for thirty years has discursive gravity. New evidence doesn't replace it by arriving. It has to overcome the accumulated weight of the old framing, in minds, in curricula, in handbooks, in institutional training programs, in the language clinicians use to describe what they see. The un-bundling work can be complete in the literature and still take a generation to reach the clinic, because the clinic is running on the gravity of what it was taught.
This is not a failure of individuals. It is a property of how knowledge moves through large systems. The cost, though, is paid by individuals. By the child assessed in 2026 under a framing the evidence moved past in 2013. By the adult assessed at forty-eight who lived a whole working life under the wrong self-model. By the clinician still teaching composites their field has un-bundled, because nobody has yet written the training program that reflects the correction.
What to do with the observation
The practical use of noticing the lag is not to assign blame. It is to hold diagnostic categories with appropriate epistemic humility, especially the ones we currently feel confident about.
Every diagnostic composite that has ever been un-bundled was, at some point, a composite nobody questioned. The categories we are most confident about today are, by the structure of the pattern, the ones most likely to be the composites of tomorrow. This doesn't mean we throw them out. It means we carry them with a lighter grip than the confidence of their presentation tends to invite.
Three questions worth asking, of any category that feels settled:
What phenomena are we bundling into this that might be co-occurring rather than constituent? Whose work is quietly doing the separation that hasn't yet reached the clinic? And how much of our confidence is evidence, and how much is accumulated weight?
The un-bundling is not an event. It is an orientation. The clinicians and researchers doing it well are the ones who treat every current category as provisional, every composite as a question rather than an answer, and every assessment as a piece of work that will be judged, eventually, against a framing that hasn't been written yet.
That is the work Bird has been doing for two decades. It is the work Barkley and others have been doing on ADHD. It is the work, quietly, that keeps the field honest.
And for those of us who have been legible only because the un-bundling reached our room in time, it is also the work that gave us back a self.
If you work in this space, particularly on composite conditions you suspect are still awaiting their separation, I'd be glad to hear from you.
ToneThread's interest in this sits at the intersection of neurodivergent communication, diagnostic framing, and the modelling of how signals accumulate weight over time. The fifth axis of SpectralBinary is proving to be a useful instrument for naming the gravity of repetition. I suspect it has more to say about diagnostic lag than I've surfaced here.
Bonus Easter Egg - for those drafting the next DSM: True to form, I got sick of waiting and had a crack at drafting it myself. Not peer-reviewed. Not clinical guidance. A manifesto-style reframing built from the inside of the composite, aimed at where the un-bundling might go next.
Written roughly a year ago, when SpectralBinary had four axes rather than five. The fifth axis, Resonance, would change how some of these profiles read - however I've left the document in its original four-axis form because the trajectory matters.
Comments