Cognitive architecture
Variant slug: cognitive-architecture
Version: v1.0
State: Canonical
Reviewed against: derivation-protocol v1.0
Author: Prayas Abhinav
Scale: The cognitive scale — SDC operating inside a single practitioner's act of thinking, rather than inside a pipeline, a curation, or a public artwork.
Stake / associated work: The Museum of Vestigial Desire archive (2012–2021) as nine-year practice ground; design-studio teaching across NID, IIT Gandhinagar, CEPT, Srishti, Karnavati, and Anant; the Dynamics of Moral Perception working paper as empirical sister-register.
Opening¶
This variant is for those who already find themselves bothered by the fusion of language and judgement in their own thinking. The suspicion that a sentence you have just produced has the rhythm of conviction without the substance of it. The recognition that a position you defended in conversation last week has been said many times before by you, with little fresh judgement under it. The sense that two arguments which look identical across fields are actually doing structurally different work, or that two arguments which look different are doing the same work in different clothes.
If none of these moments are familiar, the variant has no purchase on you. It does not argue them into existence. It addresses those who already have them, and it describes the architectural move that lets them become reliable perceptions rather than passing intuitions.
The cognitive variant, briefly stated¶
Split-Domain Cognition operates at three nested scales. Inside an AI tool, it is the pipeline architecture that keeps language work (qualification and narration) apart from judgement work (rules). Inside a medium — a studio, a curation, an essay, a public artwork — it is the discipline that keeps generation apart from evaluation. Inside a single practitioner's cognition, it is the perceptual capacity to distinguish, at any moment of thought, what is linguistic form from what is judgement-shape.
The first two scales have been described elsewhere in the canon. This variant is about the third. The cognitive scale. The version of the architectural move that operates not in software, not in a public artwork, but in the practitioner's own act of thinking — the place where the move either is performed silently, dozens of times a day, or is not performed, and the practitioner's thinking is colonised by its own language without their noticing.
Five moments¶
The variant proceeds by showing. The reader who already has the question will recognise the moments described below. The reader who does not will read past them. Neither response is wrong. The variant cannot argue recognition into being; it can only describe the moments in which the recognition arrives.
1. The writer's moment.
You are writing. The sentence is forming and it has rhythm. There is something pleasing about how it is landing — a cadence you have produced before, in pieces you were proud of. The sentence finishes and stands on the page. You almost go on. Then a small thing in you stops. The sentence has the rhythm of conviction, you notice; but you have not actually thought what it says. The language did the work. The judgement was supposed to have done the work, and instead the language ran ahead of the judgement and produced the appearance of having thought. You delete the sentence. You write a different one — uglier, slower, the rhythm is gone — and this one is what you actually meant. The fusion would have let the first sentence stand because the saying would have been the thinking. The separation is what lets you catch what the language did without the judgement having joined it.
2. Same words, different judgement-shape.
Two people say we have to keep our work alive. The first is an artist whose teacher is dying and whose tradition has no formal institutional home. The second is a startup founder whose pivot has just failed and is using the line to keep investors from withdrawing. The sentence is the same. The fusion-collapsed reader hears the same words and either reaches for the same response twice, or contradicts themselves trying to manage the dissonance. The reader who holds the distinction hears the words and perceives that the judgement-shape underneath is not the same — one is about the survival of a form of life across generations; the other is about the management of a balance sheet for one more quarter. The language has converged. The judgement-shape has not.
3. Two students, same judgement-shape under different language.
A design teacher who has read deeply in one tradition encounters two students. The first says: I want my work to live in the moment before you fall asleep. The second says: I am interested in liminal thresholds and transitional states. The teacher who fuses language and judgement responds only to the vocabulary they recognise — to the second student with feedback that lands; to the first, with confusion or with a request to clarify. The teacher who holds the distinction perceives that both students are working in the same judgement-shape, that this judgement-shape has a name in design discourse but does not require the name to be working, and that feedback can be given to both. No translation overhead, because the perception is already operating below the linguistic surface.
4. The relabelling.
You read a news article. The phrase is enhanced interrogation. You feel one way about what is described. The next day you read another article describing the same actions and the phrase is torture. Nothing has happened in the world between the two readings; only the linguistic form has shifted. The fusion-collapsed cognition follows the linguistic form because the linguistic form is the judgement. The cognition that holds the distinction catches the substitution as it happens — registers that the language is asking the judgement-shape to follow, declines to follow, holds the judgement-shape steady regardless of which linguistic form is presenting it. Most of what passes for persuasion in public discourse is the substitution of linguistic forms across stable judgement-shapes, hoping the reader's cognition will follow the language because the distinction has not been held.
5. The everyday self-catching moment.
You are about to say something. You pause. The thing about to be said has been said before, by you, many times. The pause produces a small piece of perception: I am about to say this because I have always said it, not because I have just thought it. The language is moving on its own momentum; the judgement has not been done freshly. The fusion-collapsed cognition cannot catch this, because the saying is the thinking and there is no gap to perceive. The cognition that holds the distinction catches it because there is a small clearing between the linguistic form and the judgement-shape, and the clearing is the place where the perception lives. Whether you go on to say it anyway, or revise it, or stay silent — that is a different decision. What the distinction makes available is the noticing.
The multi-domain precondition¶
The perceptual capacity described in those five moments does not arrive by intention. You cannot wake up tomorrow and decide that you will, from now on, distinguish your language from your judgement in your own cognition. The capacity arrives — when it arrives — through sustained practice across multiple, structurally distinct domains.
A single-domain practitioner can have extraordinary judgement within their field. Cézanne's eye. Bach's ear. The master craftsperson's hands. The monastic's depth of prayer. The surgeon's intuition for tissue. The poet's ear for the line. The variant does not contest any of this. What single-domain practice does not naturally produce is the cognitive separation of language from judgement at the level of perception itself. The practitioner who has lived inside one judgement-space has rarely had reason to articulate their judgement-shape as distinct from the linguistic form it has taken, because the linguistic form has been adequate to the work. Painter speaks painter; the painter's judgement and the painter's language have grown together until they are inseparable in the painter's own cognition. The painter does not need the separation to paint. The painter may even paint worse if the separation is forced.
What multi-domain practice produces is something else. Working across structurally distinct fields forces the practitioner to perceive that the same concern can wear different linguistic clothes in different domains — and, more sharply, that what appears to be the same concern under shared linguistic clothes is sometimes structurally distinct. After enough crossings of this kind, the practitioner stops mistaking the linguistic form for the judgement-shape because the experience of the distinction has been repeated often enough to become perceptual rather than inferential. The separation moves below the threshold of conscious method. It becomes how the practitioner sees.
A consequence follows. The same concern in one field can be an entirely different concern in another. This is not the trivial claim that disciplines have different vocabularies. It is the sharper structural claim that what appears as the same disagreement across fields is sometimes not the same judgement at all but two structurally distinct judgements whose linguistic forms happen to converge. Conversely — and more important — what appears as different disagreements across fields can be the same judgement wearing different linguistic clothes. The multi-domain practitioner learns to make this distinction not as a method but as a perception: the same-or-different question is answered by seeing the judgement-shape directly, not by translating one vocabulary into another.
A second consequence follows. Once this distinction is reliably perceived, attachment to positions shifts. A position is a linguistic form with judgement-criteria attached. Treating positions as if they were judgement is the naive move. Treating positions as language is the multi-domain move. The practitioner sees that any position they hold is one linguistic form their underlying judgement has taken, and that the underlying judgement may take other forms in other fields without becoming a different judgement. This is not relativism. It is the opposite: the judgement is held more firmly because it is no longer collapsed into any particular linguistic form. The form can be released — for translation, for revision, for occasion — without the judgement being released with it.
What this is and is not¶
This is not relativism. A practitioner who can perceive the language-judgement distinction in their own thinking is not therefore a position-less sophist. They are a person who knows what their judgement is underneath the language they happen to be using to hold it. They can revise the language without abandoning the judgement; they can refuse the language without losing the judgement; they can translate the language across fields without distorting the judgement. The instability that relativism produces — no position is more defensible than any other — is the opposite of what the distinction produces. The distinction lets positions be servants of judgement rather than disguises for it.
This is not a hierarchy. Deep single-domain practice produces goods that multi-domain practice cannot. Cézanne could not have produced Cézanne's paintings as a polymath. The variant is not arguing that multi-domain cognition is superior to deep single-domain cognition. It is arguing that multi-domain cognition produces a specific capacity — the perceptual separation of language from judgement — that does not arrive by other routes except rarely. The single-domain master and the multi-domain practitioner are doing different work, in different cognitive shapes, producing different goods. Neither is the right shape for the other's task.
Some single-domain practitioners arrive at the separation by other routes. Certain mystics. Certain contemplative practitioners who interrogate language and judgement within a single practice. Certain writers whose body of writing is itself a multi-domain laboratory — Gandhi, Wittgenstein, Iris Murdoch, Simone Weil, those who moved between practices serious enough to function as distinct domains. Certain teachers who, by transmitting across many minds over decades, become aware of how the same judgement-shape recurs across different student vocabularies. The multi-domain route is more reliable. It is not exclusive.
This is not prescriptive. The variant does not argue that anyone should become multi-domain. It addresses those who already find themselves troubled by the fusion. It describes one route to the separation. The route is real, the perceptual capacity is real, the consequences are real. Whether to pursue the route is a different question and is not the variant's concern.
Relationship to the other variants and to Koher¶
Koher applies this principle in AI tooling — the three-layer architecture is SDC at the scale of software. The other variants apply the principle to specific external domains: animal rights and veganism, curation, urban planning, data work in India, design research, psychiatric diagnosis, the valuation of talent. Each is the same architectural move performed in a different medium with different stakes.
This variant applies the principle to the cognitive substrate that makes the others possible. The cognitive variant is not the cause of the others — Koher's AI architecture came before the term Split-Domain Cognition existed, and the curatorial practice of the Museum of Vestigial Desire precedes both — but it is the cognitive precondition for noticing that the architectural move is the same move across the others. Without the cognitive distinction, the AI variant and the curatorial variant and the animal-rights variant look like separate methods for separate problems. With it, they are recognisable as the same move performed in different domains, and the recognition itself is the practitioner's evidence that the architecture is real.
The empirical sister-register for this variant is The Dynamics of Moral Perception — a working paper in the SDC project that treats cross-domain transfer of moral perception as a dynamical system with measurable parameters: perceptual readiness P(t), a transfer rate τ, two attractors separated by a critical threshold. The paper makes the empirical case in the cognitive-psychology register; this variant makes the practitioner case in the philosophical register. The two documents address the same scale of claim at the same architectural move. They are not redundant. They speak in different registers to readers who arrive at the question through different routes.
The variant also bears on the Level 1 / Level 2 position drafted in the SDC repository. Level 1 (the lived practice) and Level 2 (the shared practice) are themselves a language–judgement split at the scale of a life — generative on one side, evaluative on the other, the practitioner's task being to keep them from collapsing. The cognitive variant is what makes the Level 1 / Level 2 distinction perceptible from the inside. A practitioner who cannot perceive the language–judgement distinction in their own thinking will, predictably, collapse Level 1 and Level 2 — measuring their lived practice by Level 2 outputs without noticing the collapse. The cognitive variant is the precondition for the larger architectural move at the scale of a life.
Philosophical grounding¶
Felt-sense lineage. Eugene Gendlin's Focusing (Bantam, 1981) names the bodily-implicit precision against which a formulation either carries forward or fails to. Sondra Perl's "Understanding Composing" (College Composition and Communication 31:4, 1980, 363–369) imports Gendlin's felt-sense work into writing phenomenology specifically. Together they describe the layer in which the writer's moment in this variant — the catching of the sentence whose rhythm landed but whose judgement did not — operates.
Fluency-as-truth empirical floor. Reber, Schwarz & Winkielman, "Processing fluency and aesthetic pleasure" (Personality and Social Psychology Review 8:4, 2004, 364–382), and Oppenheimer, "The secret life of fluency" (Trends in Cognitive Sciences 12:6, 2008, 237–241), establish empirically that processing fluency — including rhythmic and phonological ease — increases perceived truth and conviction. The empirical floor under the writer's moment's claim that the rhythm of conviction can produce the sensation of having-thought without the thinking having been done.
Tacit-knowing background. Michael Polanyi, Personal Knowledge (Routledge, 1958) and The Tacit Dimension (Routledge, 1966). Polanyi's distinction between articulate and tacit knowing is adjacent to the language–judgement distinction this variant develops; Polanyi's emphasis is on tacit knowing as the ground of articulate knowing, where this variant's emphasis is on the moment of detecting the misalignment between them. The lineages are sister, not identical.
Practice ground. The variant draws on the Museum of Vestigial Desire archive (2012–2021), in which the nine-year sustained refusal to collapse description into evaluation was the practice; on the design-studio teaching that produced the Coherence Diagnostic and the Play Shape Diagnostic; on the SDC pipeline architecture itself; and on the daily practice of writing in which the distinction is either held or lost, sentence by sentence.
Closing — what the variant makes possible¶
The cognitive variant is the inward face of every other variant in this canon. It does not propose a method. It does not prescribe a discipline. It describes a perceptual capacity, names the route by which the capacity arrives, and locates that capacity as the precondition for the architectural move at every other scale.
A practitioner who has the perception can write a variant in a new domain not because they have learned a technique but because they can see, in that domain, the shape that has been visible to them in other domains. A practitioner who lacks the perception can apply the protocol mechanically and produce a derivation that has the form but not the bite. The four validation tests — recognition, specificity, cost, stake — are designed in part to catch this difference, but no test substitutes for the perception itself.
The variant exists to name what the canon's other documents assume. The principle is articulated in ../sdc.md. The position is defended in ../principle-not-pattern.md. The methodology is operationalised in ../process.md. The seven derivation questions and the four validation tests are scaffolded in README.md. All of these are external instruments. This variant names what has to be true inside the practitioner for those external instruments to do anything more than reproduce themselves with the labels swapped.
What the variant makes possible is the readership for itself. The reader who recognises the five moments is already inside the perception the variant describes. The reader who does not has not been argued out of anything; they have simply read a description that did not address them this time. The architecture continues either way. The variant is a description of the inside of the move, written for those who will recognise it.
Version 1.0 — 12 May 2026. The variant operates at the cognitive scale of SDC, distinct from the medium scale (curation, public art) and the software scale (AI tooling). It is offered for readers who have already noticed the fusion in their own thinking and are looking for a description of the architectural move that operates inside cognition. Pair-references: ../sdc.md, ../principle-not-pattern.md, ../process.md, README.md. Reviewed against derivation-protocol v1.0.
How to follow this variant¶
- Reference — cite the variant in framing. See Adoption — Reference.
- Alignment — your artefact performs the separation. See Adoption — Alignment.
- Derivation is for new domains; this slug is canonical, so derivation does not apply here. See Adoption — Derivation.
The badge URL pattern for this variant:
https://splitdomaincognition.org/badge/cognitive-architecture-reference.svg
https://splitdomaincognition.org/badge/cognitive-architecture-reference-dark.svg
https://splitdomaincognition.org/badge/cognitive-architecture-alignment.svg
https://splitdomaincognition.org/badge/cognitive-architecture-alignment-dark.svg
To declare an adoption of this variant, use the chooser on the Declare page.