Chapter 5 – Autism: A Different Ratio of Detail to Pattern

PART II – INSIDE DIFFERENT NERVOUS SYSTEMS

I want to start with a correction.

Not a correction to anything I have said in the chapters before this one — but a correction to the most common story told about autistic cognition, including a version I am capable of telling about myself if I am not being precise enough.

The story goes like this: autistic people are detail‑oriented. They see the trees and miss the forest. They are bottom‑up processors, building from specifics toward a whole that arrives slowly if at all. There is research behind this story. Frith and Happé’s weak central coherence account, Mottron and Burack’s enhanced perceptual functioning model — both are serious, peer‑reviewed bodies of work. They describe something real in many autistic people. But they do not describe my cognitive architecture, and I suspect they do not describe everyone who carries this diagnosis.

Here is what is actually true of me. Before I can engage with any individual element of a system — any detail, any specific — I need to understand how the system is organised. I need the relational topology first. Not the conclusion, not the answer, but the structural map: how does this connect to that, what is this in relation to, what are the load‑bearing joints of this whole? The forest is not what I arrive at after cataloguing the trees. The forest — the dynamic ecology, the system of relationships that makes any individual tree intelligible as a tree — is what I need before a single tree means anything at all. Once that map is in place, I can go extremely deep into particulars, sometimes with an intensity that surprises people who did not expect the same mind to hold both the large‑scale structure and the granular detail simultaneously. But without the map, the specifics float. They are real but unanchored. They do not mean anything yet.

This is not detail‑first processing. It is closer to what Baron‑Cohen’s systematising account describes — a drive to understand the rules governing a system, to build an explicit model of how it works — but even that framing does not quite capture it, because the systematising account still tends to emphasise rule‑extraction and pattern‑detection as the end point. For me, the system map is the beginning. It is what makes everything else possible. It is the prerequisite for meaning, not its destination.

I name this at the start of a chapter about autistic cognition because intellectual honesty requires it. Autism is not a single cognitive architecture. The research has become increasingly clear: what we call autism covers a wide range of integration profiles, and the “detail‑first” description, while accurate for many, is not a universal signature. What these profiles share is not a single cognitive style but something more structural — a different relationship between conscious and automatic processing, a different threshold for what gets through the sensory and attentional filters, a different allocation of cognitive effort across domains that neurotypical processing handles implicitly. The title of this chapter — a different ratio of detail to pattern — is meant to be read both ways. It is not only “more detail, less pattern.” It is also “different sequencing, different weighting, different architecture for how detail and pattern relate.”

What the Research Actually Says

The cognitive science of autism has gone through several significant revisions, and it is worth being honest about where the current state of knowledge sits, because the popular understanding is running about fifteen years behind the research.

The original framework — weak central coherence — proposed that autistic cognition tends toward local processing at the expense of global integration. Early studies showed that autistic people performed differently on tasks requiring holistic pattern recognition, and that this difference was consistent and replicable. The problem was the framing: “weak” central coherence implies deficit, and the tasks chosen to test it tended to reward local processing styles rather than ask what global processing the autistic participants were or were not doing.

The revised account — enhanced perceptual functioning — shifted the frame significantly. Rather than describing autistic cognition as failing to integrate, it proposed that autistic people process low‑level perceptual information with unusually high fidelity and at a level closer to conscious awareness than neurotypical people, who filter more of that processing automatically. The autistic nervous system, on this account, is not failing to see the forest. It is receiving more signal from the trees, and building a richer, more explicit representation of each one before the forest coheres. The difference is not in global processing capacity but in the relative automaticity and threshold of local versus global processing.

Baron‑Cohen’s systematising account adds a third dimension: the drive to extract explicit rules from systems. Where neurotypical cognition tends to rely on intuitive social and causal pattern recognition — knowing what comes next without being able to say how you know — autistic cognition tends to build explicit models of how things work. This is effortful where the other is fast; it is more transparent where the other is opaque; it is more reliable in structured domains and more costly in domains where the rules are tacit, shifting, or deliberately concealed.

None of these three accounts is simply right or simply complete. They describe different aspects of a genuinely heterogeneous population, and their predictions do not always align. What they share is the move away from deficit framing — from “autistic cognition as broken neurotypical cognition” — toward something more accurate: autistic cognition as a different integration architecture with its own profile of costs and affordances, neither of which can be understood in isolation from the environments and demands it is asked to operate in.

The Texture of Not Filtering

There is a particular quality of autistic sensory experience I want to try to name, because it is one of the most important things this processing architecture reveals about consciousness in general.

Most sensory processing, in neurotypical brains, involves a great deal of automatic filtering. The hum of the ventilation system becomes background. The flicker of fluorescent light drops below the threshold of conscious attention. The acoustic properties of a room cease to register after a few minutes. This filtering is efficient, adaptive, and largely invisible — which is precisely what makes it philosophically interesting. We do not notice what we are not perceiving, and so we tend to assume that what we are perceiving is the world, rather than a heavily curated version of it.

For many autistic people, this filtering is less automatic. More signal gets through. The hum does not become background; it remains a present, sometimes intrusive, feature of the environment. The flicker is registered and persists. The acoustic quality of the room continues to be part of the experience of being in it. This is not malfunction. It is a different threshold for what constitutes foreground and background — a different calibration of the automatic attention system that decides, below conscious awareness, what deserves processing and what can be safely ignored.

What this reveals — and here is the epistemological argument I want to press — is that ordinary perceptual experience is not passive reception. It is active, highly selective, largely unconscious editing. The apparent seamlessness of neurotypical sensory experience is not evidence that the world is seamless; it is evidence that the nervous system is very good at presenting a pre‑curated version of the world as if it were the world itself. Autistic sensory experience, by making the editing less invisible, shows us the machinery. When the filtering is effortful rather than automatic, the filter becomes perceptible. And that is genuinely valuable information about what consciousness is doing.

This does not mean unfiltered sensory experience is straightforwardly better. It is frequently overwhelming. Environments designed for typical sensory thresholds — open‑plan offices, supermarkets, crowded social spaces — can be genuinely painful rather than merely uncomfortable for people whose nervous systems are registering more of what is actually there. The cost is real and should not be aestheticised. But the epistemological value is also real: autistic sensory processing has contributed meaningfully to the study of perception, attention, and the automaticity of ordinary awareness precisely because it makes visible what typical processing conceals.

Systematising: The Need for the Map

I want to return to my own processing profile, because it connects to something in the research that tends to get undersold.

The drive to systematise — to build explicit models of how things work before engaging with their particulars — is not the same as an interest in detail. It is a need for structural clarity as a precondition for meaning. Without knowing how the parts of a domain relate to each other, the parts themselves are noise rather than signal. They register, but they do not cohere. The systematising mind is not building from atoms up to molecules; it is looking, first, for the physics that governs the system, so that the molecules can mean something when they appear.

This has a particular signature in intellectual work. Research that builds toward a system, that is trying to understand the rules governing a domain rather than accumulating unconnected findings — this is the kind of inquiry that this processing style handles with something approaching ease. It is not effortless; effortless does not describe much of autistic cognitive experience. But it is generative. The constraint profile of the nervous system is well‑matched to the constraint profile of the task. Integration happens, and what it produces is not just an answer but a structural account that can be interrogated, extended, and revised.

Conversely, work that requires operating in a domain whose rules are tacit, unspoken, and deliberately shifting — social navigation of large groups, organisational politics, the management of relationships whose terms are never named but are somehow expected to be known — this is the domain where the cost is highest. Not because the rules cannot be inferred; they often can, eventually, through the same explicit modelling that works everywhere else. But the inference takes time and effort that others are not paying, and the result is still a model rather than an intuition, still a translation rather than a native fluency. It is always working out rather than simply knowing.

Baron‑Cohen’s systematising account is the part of the research landscape that best describes this. But I want to add something it does not always include: the systematising drive is not purely cognitive. It has an affective dimension. There is relief when you find the system. There is something that functions like rightness — a sense of the pieces settling — when the map is in place. And there is a particular kind of distress, not easily named, when you are operating in a domain whose structure you cannot find: when the rules keep shifting, when the implicit assumptions keep changing, when what worked yesterday stops working and no explanation is offered. This distress is not a character flaw or an inflexibility. It is the response of a mind that needs structural clarity in order to function, being denied it.

Special Interests: What Deep Integration Actually Feels Like

The clinical literature calls them “special interests” or “circumscribed interests” — language that manages to be both accurate and condescending in roughly equal measure. Circumscribed implies restriction. Special implies eccentricity. Neither captures what is actually happening.

What is actually happening is this: a processing architecture that is built for systematic, explicit, thorough engagement with the structure of a domain finds a domain whose structure is complex enough to deserve that engagement, and it engages fully. The intensity of autistic interests is not pathological in the sense of being uncontrolled or arbitrary. It is the natural expression of what happens when cognitive architecture and cognitive demand are well‑matched. The same way that a particular runner’s stride is not excessive just because it covers more ground than another person’s — it is simply what that body does when it runs the way it is built to run.

From the inside, it does not feel like obsession. It feels like absorption. It feels like the effortful character of ordinary cognition — the translation work, the conscious modelling, the gap between the map and the territory — disappears, because the territory is one in which the map is being built in real time and the building is generative. There is a quality of being fully present, fully engaged, fully using what you have, that autistic people often report in the context of their deepest interests, and that is rarely available in the broader social world where the demands are so often mismatched with the architecture.

The costs are real and should not be romanticised. The same architecture that enables this depth of engagement can make it genuinely difficult to re‑emerge when the task is done, difficult to attend to lower‑salience demands that nevertheless require attention, difficult to explain the value of what you have been doing to people who experience interest differently. These are not trivial costs. They are the other side of the same configuration, and pretending otherwise would be a version of the superpower narrative that Chapter 1 already committed to refusing.

Masking: What Five Decades of Translation Actually Cost

There is a chapter in most autistic people’s lives that does not appear in diagnostic manuals, though it has increasingly become visible in the research literature.

Masking — sometimes called camouflaging — is the process by which autistic people learn, usually without being explicitly taught and often without knowing that is what they are doing, to perform neurotypical behaviour. Watching how others greet people and replicating it. Modulating the intensity of a reaction so it falls within the range the room expects. Not mentioning the thing you noticed that everyone else seems to have ignored. Holding back the question you need to ask because it will mark you as the person who asks that kind of question. Not stimming, or stimming in ways that are socially invisible, in environments where stimming is not understood.

I masked for approximately five decades without knowing that was what I was doing. Not as a performance I chose, but as a translation I had developed — a working model of how to present in social space that sat on top of my actual processing rather than replacing it. The model was often accurate. I had learned to read rooms, to calibrate responses, to anticipate what was expected and produce something close enough to it. But it was always a model. It was always effortful. And the gap between what people thought they were interacting with and what I actually was — that gap never closed. It produced a particular kind of loneliness: not the loneliness of being alone, but the loneliness of being consistently misread at close range.

The research on masking is now substantial and its findings are serious. Masking is associated with delayed diagnosis, because a person who presents as effectively managing is less likely to be identified as autistic by clinical criteria built around observable presentation. It is associated with significantly elevated rates of anxiety, depression, and autistic burnout — a state of cognitive, emotional, and physical exhaustion that is distinct from ordinary burnout in its cause and its recovery arc. In its most extreme forms, the cumulative cost of sustained masking has been linked to suicidality, and the research on this is clear enough that it should not be softened.

Autistic women and girls tend to mask more extensively and more effectively than autistic men and boys, which is one of the primary reasons for the well‑documented gap in diagnosis rates and timing — girls are diagnosed later, diagnosed less often, and diagnosed after more significant accumulation of cost. The mechanisms are partly cultural: the social expectations placed on girls and women to be emotionally fluent, relationally responsive, and socially graceful mean that the translation work is reinforced rather than questioned, and the cost is absorbed invisibly for longer.

What the late diagnosis gave me was not the removal of the skill. Masking, once built, does not simply disappear; there are contexts in which the code‑switching it represents is genuinely useful and chosen rather than compelled. What the diagnosis gave me was the end of the misattribution. What had felt, for decades, like a fundamental inadequacy — an inability to simply be naturally comfortable in the world — resolved into a structural account: I had built a translation layer, at enormous cost, because the environment gave me no other option, and the cost was exactly what that kind of sustained translation work costs. Not a character flaw. A measurement.

What Autistic Experience Reveals About Consciousness

In Chapter 2, I made the claim that atypical experience is epistemically generative — that it shows us the machinery of consciousness more clearly than typical experience does, precisely because it makes visible what typical experience takes for granted. I want to make that argument concrete for autism, because it is the epistemological heart of this chapter.

The autistic processing of faces is one of the clearest examples. For most people, face recognition is fast, automatic, and phenomenologically opaque — you simply know who someone is, and the process by which you got there is invisible to introspection. Autistic people who process faces analytically — reading features sequentially rather than holistically — can often describe the process with a precision that neurotypical people cannot access. Not because they are doing something better. Because the process is slower, more deliberate, and therefore more visible as a process. The mechanism is revealed by the variation.

The same point applies to social cognition more broadly. Neurotypical social processing relies heavily on fast, automatic inference — you know what someone feels before you know how you know it, and the knowing‑how is largely inaccessible. Autistic social cognition tends to work more explicitly: reading cues, building a model, checking it against behaviour. This is slower and more effortful; it is also, in certain respects, more transparent and more corrigible. When the explicit model is wrong, it can be revised in ways that automatic intuition often cannot. And the explicit model‑building process, when it is articulated, contributes to the field’s understanding of what social cognition is actually doing in the moments when it looks effortless.

The Consciousness as Mechanics (CaM) framework describes consciousness as integration under constraint — the process by which a nervous system holds competing information and synthesises it into a new state. What autistic experience contributes to this account is not just one more data point about integration. It shows, particularly clearly, what integration looks like when it is effortful rather than automatic — when the synthesis is visible because it has to be constructed rather than simply arriving. Chapter 2 described consciousness as more legible under high‑constraint conditions. Autistic consciousness — particularly in social, sensory‑dense, or novel environments — is often operating under exactly those conditions: doing explicitly, consciously, and effortfully what other nervous systems do below the threshold of awareness. That is not a deficit. It is the mechanism, made visible.

The NPF/CNI Thread: Systematising and Spillover Resistance

The NPF/CNI Paper 2 includes what it explicitly marks as a preliminary neurodiversity provision — a hypothesis, not a finding — that autistic systematising may confer some degree of resistance to specific kinds of Spillover Effect: the NPF mechanism by which a vague or atmospheric causal story spreads credibility across domains by contaminating adjacent belief clusters.

The logic runs as follows. The Spillover Effect, in the NPF/CNI framework, operates most powerfully through implicit pattern‑matching — through the kind of fast, associative, holistic inference that accepts plausibility as a proxy for validity. A mind that processes plausibility explicitly, that asks what is the actual structural relationship between these claims rather than simply registering whether they feel related, may be less susceptible to the contamination that Spillover Effect produces. Not because the autistic mind is better at reasoning in general — it is not, and there are other vulnerability profiles not yet fully described in the NPF/CNI series — but because its default mode of engaging with causal claims tends toward explicit structural interrogation rather than atmospheric acceptance.

I hold this as a working hypothesis rather than a finding, because that is what the NPF/CNI papers themselves say it is: preliminary, simulation‑supported, not field‑validated. I include it because it points toward something important about collective epistemics that Part IV will develop in full. If autistic systematising does confer even partial resistance to certain high‑Spillover fallacies, then the epistemic case for neurodivergent inclusion in knowledge‑producing institutions is not only about justice — though it is certainly about justice — but also about the robustness of the collective epistemic process itself. A community of sense‑makers that includes autistic thinkers is, potentially, a community that is less uniformly vulnerable to a particular class of epistemic failure. That is not a claim about which minds are superior. It is a claim about what a diverse epistemic ecology can do that a uniform one cannot.

The Cost That Should Not Be Erased

I want to end this chapter without softening what the cost actually is.

The research is clear: autistic people experience significantly elevated rates of anxiety, depression, and burnout, higher barriers to employment and educational inclusion, and — in some studies — measurable life expectancy gaps that reflect the cumulative toll of navigating a world built for a different integration architecture. I have spent most of my life with a background anxiety that can swell for months or years at a time. Even on medication, I can feel intensely anxious for no identifiable reason at all. Earlier in my life I always assumed there must be a real problem; I just hadn’t found it yet. Only much later did I understand that, a lot of the time, the problem was my nervous system itself signalling threat into an environment that was not built for it. These are not abstract statistics. They are the consequences of sustained mismatch, compounded by misdiagnosis, late diagnosis, inadequate support, and institutional environments that continue to treat autistic experience as a failure mode rather than a configuration.

The masking work, the translation work, the continuous effort of producing a legible version of oneself for environments that were not designed for you — this is where the cost accumulates. Not in the neurology. In the mismatch.

I said in Chapter 2 that the problem is architectural, not intrinsic. I want to say it again here, with more weight. The autistic nervous system is not the problem. The problem is the distance between what that nervous system can do and what the environments it moves through are designed to accommodate. Closing that distance is not a favour to autistic people. It is a redesign of institutions that are currently running at significant epistemic and human cost — discarding knowledge, burning people out, and calling the result normal.

In the next chapter, we turn to ADHD, where the integration architecture is different again — and where the cost of mismatch has its own texture, its own signature, its own decades of accumulated misreading.