Half a Paradigm Shift: Co-occurring conditions aren’t separate items. They’re what the configuration produces under load.
[VISUAL: Five-panel diagram extending the original “Person with autism / Autistic person” advocacy image.
Top row, two panels: the original image reproduced. Left panel, “Person with autism,” a gray figure holding a rainbow handbag, captioned the framing identity-first advocacy fought against. Right panel, “Autistic person,” a gray figure with a rainbow head, captioned the win identity-first advocacy secured.
Bottom row, three panels showing what comes after.
First bottom panel, “Autistic person with co-occurring conditions”: a figure with a rainbow head and a gray body, carrying a handbag, backpack, grocery bag, and a dog on a leash, each labeled with a co-occurring condition (anxiety, GI, burnout, PTSD, sleep dysregulation). Captioned the careful separation.
Second bottom panel, “Autistic person under environmental load”: a figure rendered in rainbow color from head to toe, no separate items, with six dark arrows pressing inward from all sides, labeled masking demand, sensory overload, stigma, minority stress, discrimination, trauma exposure. Strain indicators emanate from multiple points on the body. Captioned the conditions are not separate items, they are what the configuration produces under sustained environmental load.
Third bottom panel, “Neurodiversity as interconnected configurations”: five figures with distinct whole-body patterns, connected to each other by a network of fine lines. Captioned neurodiversity is everyone, health requires understanding the architecture, not removing the variation.
Beneath the entire diagram: Neurology is not an accessory. It is the whole architecture.]
Autistic anxiety. Autistic burnout. Autistic menopause.
These may share diagnostic labels with general-population presentations, but they often emerge through different neurological, sensory, interoceptive, and perceptual architectures meeting different conditions.
The diagnostic label aggregates across configurations and hides what is actually happening underneath.
Some configurations are themselves shaped by trauma — the architecture is not only autistic or ADHD; it can also be trauma-formed, with its own patterns of vulnerability and protection.
Different bio-neurotypes reach distress through different routes. They often require different interventions, in different sequences, and sometimes different medications altogether.
The same person may also move through different regulatory states over time depending on cumulative load, predictability, trauma exposure, inflammation, sleep disruption, hormonal transitions, and social buffering.
This matters because most pharmacological efficacy testing has historically been conducted primarily in populations closer to the statistical and social-regulatory average (“neurosocial”) configuration — populations that tend to manifest diagnoses more typically and respond more predictably to standardized interventions.
Emergent neurotypes often do not.
This is why the call for emergent paradigms is not philosophical. It is imperative to treat the architectures that the existing framework was not designed to model.
The image below has been circulating in autistic community spaces for years from an amazing writer, speaker, researcher and autistic advocate, Becca Lory Hector. It shows two figures side by side. The first is labeled “person with autism.” Their head is gray and empty. They are holding a small rainbow purse, the autism slung over their arm like an accessory. The second figure is labeled “autistic person.” There is no purse. The rainbow is in the head. The caption reads: Neurology is not an accessory.
The image was a triumph of advocacy. It made a point that the field had been refusing to make for decades. Autism is not something a person carries. It is who they are. The identity-first move that grew out of this insistence has been one of the most successful conceptual shifts in the history of neurodiversity advocacy. It changed the language. It changed the politics. It changed how autistic people are allowed to describe themselves in public.
It did not change the medical model. And looking at the image again recently, I realized why.
What identity-first accomplished
The identity-first fight was a fight against a specific framing: autism as an addition to a person, a condition layered onto an otherwise normal self, something that could in principle be removed without removing the person. That framing had real consequences. It supported cure-oriented research. It supported behavioral interventions designed to extinguish autistic features. It supported a clinical posture that treated the autistic part of a person as the part that needed to go.
The identity-first move shut all of that down conceptually. If autism is who someone is, you cannot cure it without erasing them. You cannot extinguish autistic features without extinguishing the person whose features they are. The neurology is not the purse. It is the head. There is no way to take it off.
This was correct. It is still correct. The advocacy that produced this shift deserves credit for changing what could be said in public about autistic people, and for protecting a generation of autistic children and adults from the worst excesses of the cure model.
The careful separation
Within the identity-first frame, thoughtful autistic writers developed a careful practice of separating identity from co-occurring conditions in their public work. I am an autistic woman. I am a woman with cancer. I am autistic. I have anxiety. John Elder Robison has done a version of this for years, holding his autistic identity steady while writing carefully about specific experiences — including the changes after TMS — as separate things he has navigated alongside it. Other thoughtful writers have done the same with chronic illness, with hormonal transitions, with mental health conditions, with whatever else has shown up alongside the autism.
The construction is consistent: autistic as identity, condition as something else, deliberately held apart. This is not casual practice. The writers who do it have thought about it, defended it, and in many cases written about why they make the distinction the way they do. It is the most rigorous response available within the framework that has been on offer.
It is also, I want to argue, structurally limited in a way the framework prevents us from seeing.
I want to be careful here, because the move I am about to make is easy to misread. I am not saying the writers who have practiced this distinction are wrong. They are not. They are doing the most precise possible work within the available framework. What I am saying is that the framework itself is too small, and the precision of their work has been bumping against the limits of the framework for as long as the framework has existed. The work has been right. The framework has been wrong. These can both be true at once.
What identity-first left intact
To make the identity-first move stick, the co-occurring conditions had to be separated out. If anxiety and gastrointestinal symptoms and seizures and sleep dysregulation stayed bundled with autism as part of a single pathological picture, then defending autism-as-identity meant also defending those conditions as identity, which they are not. Anxiety is not who someone is. GI dysfunction is not who someone is. So the conditions were peeled off, located in their own diagnostic categories, treated as separate problems that happen to occur in autistic people.
This was a tactical necessity at the time. It was also the move that left the medical model’s deepest architecture untouched.
Pathology did not go away when autism became identity. It moved. The purse came off the autistic person. It got refilled with the co-occurring conditions, and handed to the same person, who now carried the conditions instead of the autism. Anxiety became the pathology. GI symptoms became the pathology. Sleep dysregulation became the pathology. Burnout became the pathology, eventually, after enough years of having it. The deviation logic of the medical model — the assumption that there is a normal default and that everything else is deviation requiring intervention — stayed in place. The location of the deviation just shifted from one diagnostic category to several.
This is why neurodiversity, as a concept, has ended up positioned awkwardly between two boxes that already exist in clinical thinking. There is the box for normal, and there is the box for pathological, and the medical model has no third box for “different in a way that is constitutive rather than deviant.” There is no stable category in the framework for a configuration whose features are neither dysfunction nor accessory but simply what the configuration is. So neurodiversity gets squeezed, split or hold dual roles in one or both of the two available boxes — usually the pathological one, sometimes the normal one, never quite fitting either — and the squeeze is painful because the framework is two-valued and what is being described is something the framework cannot represent.
Where the careful separation breaks
The framework starts to break, visibly, when writers try to describe phenomena that do not fit either box cleanly. Cancer fits the with framing well in one important sense. Cancer is exogenous in a way that anxiety in an autistic configuration is not. A regulatory configuration does not produce cancer as part of what it does. Cancer happened to the woman; the woman did not produce it. The with relationship captures something true.
Autistic burnout does not fit the with framing in the same way. Burnout is not who someone is — no one wants to claim burnout as identity. Burnout is also not a condition that arrived from outside. It is what an autistic configuration does after years of being operated outside its sustainable range. It is the configuration’s response to chronic overload. The relationship between burnout and the autistic configuration is not the relationship between cancer and the woman who has it. Burnout is what the configuration produces under load. Cancer, in the standard rendering, is something that happened to the woman.
The community has noticed this. The literature on autistic burnout has been pushing, slowly, against the framework’s binary. The writing keeps reaching for language that says burnout is related to being autistic without quite being a feature of it, and separate from the autism without quite being a condition that has nothing to do with it. The reaching is the framework breaking in real time. There is no good available word, because the available words assume a binary the phenomenon does not respect.
The same thing happens with autistic anxiety, autistic PTSD, autistic menopause, autistic anything-that-is-not-the-general-population-version-of-the-thing. Each time the field tries to name a phenomenon specific to the autistic configuration, the available framing — autistic with — fails to capture what is being named. The phrase reads as if the autism and the condition are two separate things that happen to occur in the same person. But they often are not. The condition is, in many cases, what the configuration does under specific environmental conditions. The phrase imports a parametric assumption that the configuration cannot support.
“What appear clinically as separate conditions are often emergent expressions of shared regulatory architecture under different environmental pressures.”
Why the demand for recognition keeps failing
This is the structural reason that advocacy for recognition of autistic burnout, autistic menopause, autistic anxiety, and autistic PTSD keeps running into the same wall and same pushback. People are correctly observing that these phenomena are not the general-population versions of burnout, menopause, anxiety, and PTSD. They are something specific to autistic configurations, with different presentations, different trajectories, different treatment responses, different patterns of recovery. The advocates want these to be understood as part of the autistic configuration, because they manifestly are.
The medical model cannot grant this. Granting it would re-collapse the very separation that the identity-first advocacy fought for. If autistic anxiety is part of being autistic, then anxiety is part of autism. If anxiety is part of autism, then autism is partly anxiety. If autism is partly anxiety, then autism is partly pathological, because anxiety is in the pathology box. The whole identity-first frame starts to wobble. The advocates who have been defending autism-as-identity for decades cannot also concede that the conditions belonging to autistic people are part of what autism is, because doing so would re-pathologize the identity they have spent decades de-pathologizing.
This is the trap. The trap is not produced by either side being wrong. Both moves are correct in isolation. Autism is identity. The co-occurring conditions are part of the configuration. Both are true. And both cannot be true within the medical model, because the medical model has only two boxes, and the truths require a third.
The same trap, in two other places
This same structural trap is producing the stuck quality in two other debates the field is currently having.
The first is the debate about how to describe co-occurring conditions in lived experience. Writers like Bridgette Hamstead have been doing important work tracking what happens when autistic and ADHD configurations coexist with cumulative trauma exposure, rejection-sensitivity dynamics, and chronic illness clusters. The phenomenon she is describing is real. The lived experience is accurate. The clinical observations are sharp. But the rendering she produces — autism plus ADHD plus CPTSD plus RSD, layered on each other as discrete conditions — distorts what she is pointing at. It distorts not because she is mistaken about what she sees, but because the available vocabulary forces her to render a single regulatory configuration as a stack of separate conditions. The vocabulary supports stacking. It does not support naming the configuration the conditions are readouts of. The cross-sections are accurate. The composite reconstruction is wrong. And it is wrong in exactly the same way that the medical model is wrong — by treating what is constitutive as if it were additive, treating what is one thing as if it were several things layered together.
We are not stacks of conditions. We are configurations meeting environments.
The second is the debate over spelling-based communication methods, which surfaced again earlier this year in the exchange between Amy Lutz and Barry Prizant in the New York Times. Lutz argued that Facilitated Communication and its descendants — Rapid Prompting Method, Spelling to Communicate — are debunked methods being repackaged for new generations of nonspeaking autistic people, with real costs to families. Prizant responded that the science Lutz cites is dated, that contemporary assistive communication approaches hold genuine promise, and that nonspeaking individuals are now typing independently after starting with these methods. Both arguments are sincere. Both have real evidence. The studies showing that spellers cannot transmit information unknown to their facilitators are real. The independent typists who started in spelling and now communicate without facilitation are real. These two findings are not contradictions. They are what you would expect to find if the underlying population were architecturally heterogeneous and the methods were doing different work for different nervous systems on different developmental trajectories.
The Lutz-Prizant debate cannot be resolved by adjudicating which side has the better evidence, because both sides have real evidence. The debate is stuck in the same trap. The medical model has two boxes — methods that work and methods that don’t work — and the actual answer requires a third: methods that work for some configurations under some conditions and not for others, where the question is not whether the method works but which configuration is being engaged and what developmental work the method is doing for that configuration in particular. Recognizing heterogeneity does not eliminate the need for rigorous safeguards, independent validation, or protection against facilitator influence.
Three different debates. Same trap. The medical model is two-valued. What is being described requires a third value the model does not contain.
The way out is not picking a side
Neurology is body. Body is environment. The architecture is the relationship.
The way out of this trap is not to give up on identity-first advocacy. It is not to give up on recognition of co-occurring conditions. It is not to pick Lutz over Prizant or Prizant over Lutz. It is not to abandon the careful separation that thoughtful writers have been practicing without offering an alternative. The way out is to reframe pathology itself.
If pathology is not deviation from a default, but failure of a configuration to be supported within its sustainable operating range, then the binary collapses. There is no longer a normal box and a pathological box and a missing third box for configurations that are neither. There are configurations, with characteristic dynamics, with operating ranges, with predictable failure modes when operated outside those ranges across long developmental periods. None of this is pathology in the deviation sense. None of it is identity in the unchangeable-essence sense. It is what regulating beings do.
This kind of move is not new for the framework I work in. The Center for Adaptive Stress recently published a long essay — Complexity Evolution: Why Variation Persists, Stress Reveals Structure, and Cooperation Generates the System — engaging Robert Lustig’s claim that autism and Alzheimer’s are versions of the same disease because they share metabolic signatures. The essay does not dispute the metabolic science. The chemistry is real. What the essay argues is that shared substrate is not shared pathology. Autism and Alzheimer’s share mitochondrial infrastructure because every regulatory system in the body runs on the same energy infrastructure. Autism is a developmental emergence — a high-gain regulatory architecture that emerged as an adaptive configuration. Alzheimer’s is a degenerative collapse of regulatory capacity accumulated over decades of allostatic load. One is the architecture. The other is the architecture falling down. Same substrate. Different evolutionary events. Different clinical implications.
The structural move is the same one that needs to happen at the level of identity and condition. Just because two phenomena share substrate does not make them the same phenomenon. Just because autism and autistic anxiety share an underlying configuration does not make anxiety part of autism-the-identity, and it does not make anxiety a separate condition that arrived from outside. There is a third option. The configuration produces certain dynamics under certain conditions. Those dynamics are neither identity nor exogenous pathology. They are configurational dynamics — what the architecture does when it is operated under the conditions it is being operated in.
In this framing, autistic burnout is not pathology layered on autistic identity. It is what happens when an autistic configuration is operated outside its sustainable range across long developmental periods. Autistic anxiety is not a comorbid condition that happens to occur in autistic people. It is a regulatory pattern that emerges from how autistic configurations interact with environments built for different configurations. Autistic menopause is not menopause-with-autism. It is the specific reorganization an autistic regulatory architecture undergoes during a hormonal transition. Autistic PTSD is not PTSD plus autism. It is what trauma does to an autistic configuration, which is not the same as what trauma does to a neurotypical configuration. None of these are pathology. None of these are accessories. They are configurational dynamics — what the configuration does under specific conditions.
What the architectural reframe is not saying
I want to be careful here, because the move I have been making is easy to misread. The architectural reframe is not a project of dissolving conditions into configurations. Conditions are real. Cancer is real. Burnout is real. Anxiety is real. The reframe does not say these phenomena do not exist or do not require attention. What it says is that the question of how to address them changes.
The medical model asks how to treat the condition in the person. The architectural model asks what is producing the condition in the relationship between the architecture and the environment, and how that relationship can be changed. The clinical intervention does not go away. It is joined by an environmental one.
We do not want to be rid of autism. We do want to be rid of the environments that produce autistic burnout. We do not want to be rid of cancer in the sense of pretending cancer does not happen. We do want to change the environments that are making cancer increasingly common in architectures that meet them. Different kinds of intervention work together. The architectural reframe widens the lens. It does not narrow it.
What the autism-cancer epidemiology shows
The autism-cancer literature is one of the cleanest empirical illustrations of why the architectural framework is needed and why the parametric framework cannot account for what the data actually show.
The headline finding comes from Darbro and colleagues, in a 2016 paper in PLOS ONE. Autistic people carry significantly more mutations in cancer-promoting oncogenes than non-autistic people, yet show lower overall cancer rates — particularly in childhood and young adulthood, where the protective effect is strongest. The odds ratio for cancer in the youngest autistic cohort, ages 0 to 14, was 0.06 compared with controls. The protective effect attenuates with age but remains visible into adulthood. This is the paradox that the parametric framework cannot resolve. If cancer were a parametric phenomenon — more mutations producing more cancer in a linear way — the autistic population should have higher cancer rates. The actual finding is the opposite. Something about the autistic regulatory architecture is metabolizing those mutations differently than the typical configuration does.
The broader picture, across multiple studies, is more complex and more interesting. The 2022 Nordic cohort study of 2.3 million individuals found a modest increased risk of any cancer in autistic populations overall (OR 1.3), but that increased risk was driven entirely by autistic people with co-occurring intellectual disability or birth defects. Autism alone, without those co-occurring conditions, was not associated with increased cancer risk. A Taiwanese cohort study found elevated risk for specific cancers — especially genitourinary and ovarian. A 2025 Frontiers in Oncology review summarized the field as showing a context-dependent relationship: protective for some subpopulations, elevated for others, varying by age, by specific cancer type, and by the conditions co-occurring with the autism.
The conceptual interpretation that holds this pattern together comes from work by Crespi, by Crawley, Heyer, and LaSalle, and by others tracking the shared genetic architecture between autism and cancer. The shared pathways are real — PTEN, MAPK, PI3K/AKT, mTOR — and they operate in both conditions. But they operate differently. In autism, these pathways show high signaling in differentiation. In cancer, the same pathways show high signaling in proliferation. Same substrate. Different regulatory direction. Different outcome. The autistic configuration is not a cancer-prone configuration that happens to dodge cancer. It is a configuration that uses the same molecular infrastructure as cancer to do something else — to drive neural differentiation rather than tumor proliferation. The evidence suggests that identical molecular substrates can participate in different regulatory programs depending on developmental architecture and environmental context.
This is what architectural variation looks like at the cellular level. Same genes. Same pathways. Same molecular substrate. Different regulatory architecture metabolizing the substrate differently, producing different patterns of vulnerability and protection. The autistic configuration is not less cancer-prone or more cancer-prone in any simple sense. It is differently cancer-prone, with patterns of protection and susceptibility that follow from the architecture’s particular way of using the shared pathways. These findings suggest differences in how developmental regulatory systems utilize shared signaling pathways.
What this implies for intervention
The architectural framework changes what intervention can mean.
If autistic regulatory architecture metabolizes carcinogenic substrate differently — protecting against some cancers, contributing to others — then the cancers that do appear in autistic populations are not random events. They are configurational outcomes. They emerge from the meeting of a particular architecture with a particular environment. The cancer that develops in an autistic person is the product of how that architecture handled the carcinogenic load it encountered.
The clinical implication is that treating the cancer in the person is necessary but insufficient. The architectural implication is that reducing cancer in autistic populations — and in non-autistic populations — requires changing the environments that are producing carcinogenic load in the first place. This is not blame. The woman did not choose her cancer. The architecture did not fail her. What happened is that the environment delivered more carcinogenic load than the architecture could metabolize without consequence, and one of the consequences was cancer.
The same logic extends across the configurational dynamics the framework describes. Autistic burnout is real, and it requires attention. The intervention is not to medicate the burnout out of the person. It is to change the environment producing the chronic overload that the configuration is metabolizing into burnout. Autistic anxiety is real, and it requires attention. The intervention is not to dissolve the anxiety into identity or to pathologize it as a separate condition. It is to recognize that the configuration is producing anxiety as a regulatory response to environments it was not built for, and to ask what those environments are doing to architectures generally.
This is the move the medical model cannot make on its own. The medical model intervenes on the person. The architectural model intervenes on the relationship — the meeting between architecture and environment where the conditions are emerging. Both kinds of intervention are necessary. The architectural one is the half of the picture that has been missing.
We do not want to be rid of autism. We want environments that do not produce autistic burnout. We do not want to be rid of cancer. We want environments that do not produce carcinogenic load at the rates contemporary environments do. We do not want to be rid of any of the conditions the configurational framework names. We want the relationship between architectures and environments to be sustainable, so that the conditions emerging from the relationship are not the inevitable cost of being a regulating being in a world that was not built for the architecture you happen to have.
What this opens
This is the move that finishes what identity-first started. It does not undo the identity-first claim. It dissolves the binary that has been forcing identity-first advocates to choose between defending autism-as-identity and acknowledging that the co-occurring conditions are part of the configuration. The choice was always a false one, produced by a medical model that could not represent what the configuration actually is. Once pathology is reframed, the choice disappears.
It is also the move that makes the careful separation more sustainable rather than less. The thoughtful writers who have been holding the line on autistic woman with cancer and autistic with anxiety have been doing precise work within a framework too small for the precision they were attempting. Once a third option exists, they no longer have to render every co-occurring phenomenon as either identity or as a foreign condition. They can describe the cancer with the with framing that fits cancer cleanly at the personal level, and they can also recognize that even cancer is shaped by the meeting of architecture and environment at the population level. They can describe the anxiety as a configurational dynamic that emerges from how their architecture meets the environment they are in. Different kinds of phenomena get different kinds of description, because not all phenomena that co-occur with autism have the same relationship to it. The framework expands to fit what was always there.
It is the move that makes the stacking model unnecessary. If there is a vocabulary for naming configurations, then advocates and lived-experience writers no longer have to render their experience as a pile of diagnoses. They can name the configuration they operate with, describe its dynamics, locate the conditions under which it functions well and the conditions under which it does not, without having to translate every observation into the cross-section vocabulary the diagnostic system imposed on them.
And it is the move that makes the Lutz-Prizant debate productive instead of stuck. If methods are evaluated by which configurations they engage and what developmental work they do for those configurations, then the question is no longer whether spelling-based methods “work” in the abstract. The question becomes empirical and architectural: which configurations are these methods doing work for, what kind of work, on what trajectory, under what conditions, with what failure modes. That is a research question. It is a question the field can answer. It is the question that the binary debate has been preventing the field from asking.
Closing the half-shift
The image of the autistic figure with the rainbow head is still correct. Neurology is not an accessory. The fight that produced that image was a real fight, and the people who fought it were right. What I am saying is that the move was half a paradigm shift, and the field has been stuck in the half-shifted state for long enough that the cost of staying there has become visible.
The cost is that thoughtful autistic writers have to keep performing a careful separation of identity from co-occurring conditions because the framework offers no third option, even when the separation strains against the phenomena they are trying to describe. The cost is that lived-experience writers have to keep stacking diagnoses to describe what they are because the available vocabulary forces it. The cost is that debates about methods stay polarized because the binary doesn’t have room for architectural heterogeneity. The cost is that intervention stays focused on the person rather than on the relationship, because the framework cannot see the relationship as the thing producing the conditions.
The other half of the paradigm shift is the move from a deviation-based framework to a configurational one. From a model that asks how much someone differs from a default, to a model that asks how their regulatory architecture is built, what it does well, where its operating range is, what conditions it needs to function. From a vocabulary of categories to a vocabulary of configurations. From a clinical apparatus organized around fixing deviations to one organized around supporting architectures meeting environments.
Neurodiversity is not a category. It is what regulating beings look like when you stop pretending there is a default.
This is the work that finishes the move that identity-first started. It honors the lineage of careful writers whose work made the next layer possible. It does not undo the careful separation; it expands the framework so that the separation does not have to do all the work. It is the work that makes the lived-experience writers’ observations legible as observations rather than as complaints. It is the work that turns method debates into research questions. It is the work that closes the gap between what advocates have known for years and what the field has been able to recognize. It is the work that lets us ask not only what to do about the conditions in the person but what to do about the environments producing the conditions across architectures.
The neurology is not an accessory. It is also not a pathology. It is a configuration, meeting an environment, producing dynamics. The framework that can hold all four of those statements at once is the one that has been waiting on the other side of the half-shift the field has been stuck inside.
That is the direction the work is going. It is closer than it looks. It has been waiting for us to notice that the purse came off but the pathology stayed, and that the next move is not about what we carry but about how the architecture meets the world.
We are not what we carry. We are how we regulate.
Lori Hogenkamp — Center for Adaptive Stress
A note. The Evolutionary Stress Framework is a conceptual lens, not clinical guidance. It is offered as a way of thinking, not as a treatment protocol or a substitute for medical, psychological, or therapeutic care.
This piece is written from an autistic perspective, by an autistic independent scholar, as part of the ongoing development of the framework. The architecture being described here is still being built — through papers under review, through collaborations across disciplines, through the lived expertise of the people whose configurations the field has been misreading. None of it is finished.
Responses, pushback, and conversation are welcome. The framework gets stronger when it meets the people it is trying to describe.
Leave a Reply