{"path":"research/attunement-words-and-the-concept-layer.md","content":"# Attunement words and the concept layer: the hard case, and why it may be the honest first test\n\n**Date**: 2026-09-22 · **Status**: proposal, unruled · **Origin**: the founder, on seeing that a\nSwedish post's two lines supplied a guard `civilizational-vision.md` had been missing: *\"I feel like\nthe conceptual distinctions here matter so much: the words 'attunement', 'love', 'acceptance',\n'understanding', 'seeing' someone or feeling seen, etc etc. All of this Deliberus itself could help\nelucidate and pin down, ideally, and if used wisely.\"*\n\n**Sibling of** [core-periphery-and-the-concept-layer.md](core-periphery-and-the-concept-layer.md),\nwhich holds the layer's theory and the delta ruling. This doc is about one **class of term** that\ntheory treats as an edge case and that this project cannot treat as an edge case, because its own\nconvictions are written in it.\n\n---\n\n## 1. The observation, which is the founder's\n\nTake the two lines that prompted this. Neither defines anything:\n\n> *\"Kärlek betyder inte att vara naiv. Kärlek kan säga NEJ.\"* — love is not naivety; love can say no.\n>\n> *\"Förståelse betyder inte acceptans.\"* — understanding is not acceptance.\n\nBoth are **subtractions**. The author never says what love is; she says what it is not and what it\ncan still do. And that is exactly the shape the concept layer was ruled to record — the founder's own\nsharpening of 2026-09-06: the interesting part is not a description of the core, *\"what's interesting\nis to articulate how you think your conception of a term/idea DIFFERS from the norm\"*, elicited by\nsubtraction because false consensus hides the holder's delta from the holder.\n\nSo an untrained writer, arguing in public about her own brother, produced textbook delta elicitation\non the hardest words in the vocabulary. **That is the finding worth having: the elicitation shape\nthis layer needs already occurs naturally on exactly the terms the theory says are hardest.**\n\n## 2. Why they may be the hardest, and how well that is actually supported\n\n⚠ **Read this section as reasoning, not measurement.** Challenged 2026-09-22 (*\"I'm not sure these\nwords are all that special. What are our strongest priors for believing that?\"*), the honest answer\nis that the first version of this section cited **no outside evidence at all** — all three of its\nproperties were derived from the corpus's own framework. The three are re-graded below, one of them\ndownward and one of them out.\n\nThe framing itself comes from the delta layer's stated limit, in\n[core-periphery-and-the-concept-layer.md](core-periphery-and-the-concept-layer.md) § 5b:\n\n> Essentially contested concepts (democracy, art) have a **thin exemplar rather than a fat core**, so\n> the delta is measured from an exemplar and **is most of the content**.\n\nPenguin has a fat core: reference and category are shared, and a delta is a small correction at the\nedge. The claim here is that *love* does not — that there is an exemplar and almost everything else\nis delta, which would break the economics the layer relies on, since its efficiency comes from not\nhaving to elicit the part both parties already hold.\n\n### 2a. The thin exemplar — real support exists, aimed slightly wrong\n\n**Support the doc had and did not use.** Jackson et al. 2019 (*Science*) mapped emotion-word\ncolexification across ~2,474 languages and found the **carving varies while the valence–activation\nstructure is universal** — shared substrate, divergent partition, which is the core/periphery shape\nexactly. Barrett's constructed emotion supplies the mechanism: emotion concepts are learned\npopulations of instances, so individual variability is predicted rather than surprising. Both are\nalready in this corpus ([contingency-and-convergence-across-levels.md](contingency-and-convergence-across-levels.md),\n[decomposing-value.md](decomposing-value.md) § Level 2) and neither was cited here.\n\n**Two reasons that support is weaker than it looks.**\n\n1. **Wrong unit of variation.** Jackson measures divergence *between* language communities. The claim\n   this doc needs is divergence *within* one — two Swedes, in one room, one language. Nothing cited\n   here measures that.\n2. **Wrong word class in the measurement everyone reaches for.** Marti et al. 2023 is the corpus's\n   anchor for latent conceptual diversity, and its instruments ran on **common animals and\n   politicians**. Nobody has run that method on *love*. Extending it here is an extrapolation with a\n   plausible mechanism, never a measurement.\n\n**And there is a live counter-prior that cuts the other way.** These words carry unusually heavy\nshared scaffolding — songs, films, scripture, therapy vocabulary, the whole apparatus a culture\nbuilds around love specifically. That is precisely the transmission mechanism that makes concepts\nconverge *within* a community. So the honest competing hypothesis is that love's periphery is\n**fatter than average, not thinner**, and that its felt difficulty comes from the stakes rather than\nfrom the semantics. This doc does not resolve that, and the first version did not notice it.\n\n### 2b. The delta is felt before it is sayable — true, and not distinctive\n\n[analysis-and-attunement.md](../analysis-and-attunement.md) says this of the attunement pole's work\ngenerally (*\"it is felt before it is argued\"*), so the elicitation cannot be a prompt that asks for a\ndifference: it has to show a reference concrete enough that the person recognises what is *not*\ntheirs in it. The design consequence stands.\n\n**But it is not a property of this word class.** Marti measured the same gap on *penguins* — people\nassume roughly two-thirds agreement where the measured whole-profile match is far lower. Feinberg &\nWiller's reframing asymmetry (people recognise a reframed argument as better once shown it, while\nrarely producing one) is the same recognition-over-production shape on political material. This is a\ngeneral fact about introspective access to one's own concepts, so it argues for the elicitation\ndesign and **not** for these words being special.\n\n### 2c. Both parties are sincere — withdrawn\n\nThe first version claimed that political terms carry strategic divergence while intimate ones do not,\nso that this class removes an adversary and leaves a clean measurement problem. **That is probably\nwrong, and this corpus already documents the counter-case.**\n[the-motte-and-the-periphery.md](the-motte-and-the-periphery.md) describes retreating to a defensible\ncore sense under challenge, and the intimate register is where that move is most familiar — *real\nlove would accept me as I am* is a bailey defended from a motte. Sincerity is also the wrong\nvariable: the adversary ladder in\n[fractal-scales-and-temporal-frame.md](fractal-scales-and-temporal-frame.md) counts a **motivated\nstakeholder** as one who has a stake and is not modelling your detector, which sincere people in a\npainful conversation satisfy routinely. Motivated stopping needs no insincerity.\n\nWhat survives of the original point is much smaller: nobody in this register is modelling the\n*instrument*, so the strategy-class adversary is absent. That is a statement about the test's\ncleanliness, never about the participants' motives.\n\n### 2d. What is left standing\n\nOne property, held weakly: **the opponent-relative delta may be the only usable one here**, because a\nthin exemplar gives the population-mean reading nothing to be an origin of. If 2a's counter-prior is\nright, that weakens too. **The strongest remaining reason to run this test is § 3's, which is internal\nand does not depend on any of the above** — this project's own convictions are already written as\ndeltas over these words, so it needs the distinctions held whether or not the words turn out to be\npsychometrically unusual.\n\n## 3. Why this project cannot treat it as an edge case\n\nIts own convictions are written in this vocabulary, and one of them is load-bearing: **analysis and\nattunement are two poles and neither may eat the other.** The guard that keeps the attunement pole\nfrom collapsing into endorsement is *understanding is not acceptance* — a delta on one of these\nwords. The ratified proportionality ruling on sacred ground is another. The founding image of the\nproject is a third.\n\nSo the corpus already **runs on** deltas over these terms while having no machinery that records\nthem. When a session gets one of them subtly wrong, the failure looks like drift rather than a\ndetectable error, because nothing anywhere holds the distinction as a first-class object.\n\n## 4. Why it may be the honest first test of the layer, and what that would cost\n\nThe concept layer's live index holds *fairness*, *deserving*, *problematic*, *high quality* — the\nevaluative-political register. The case for testing it on the attunement register instead:\n\n**It has a different stakes profile.** Getting *fairness* wrong costs an argument. Getting *being\nseen* wrong costs a relationship, which is the register in which the founding image of this project\nis set. If the layer works anywhere, it should work where the cost of failure is felt rather than\nscored.\n\n**It has no strategic adversary**, so a null result is informative rather than confounded. A failure\non a political term could always be a motivated stakeholder; a failure here cannot.\n\n**And it is the one register where a bad result should stop the programme rather than tune it.** If\ntwo sincere people cannot get their deltas on *understanding* onto a shared surface, the claim that\nthe cost of mutual understanding is lowerable by infrastructure has met its sharpest counter-case —\nwhich is the falsifiability the wager is supposed to have and rarely gets tested on.\n\n**The cost, stated plainly**: the co-present session it needs is the most expensive mode the project\nhas, and the material is the least publishable. This is not a cheap test, and proposing it is not the\nsame as proposing it *next*.\n\n## 5. ⚠ What this does NOT say, because an earlier phrasing of mine did\n\n**It does not say the attunement pole is empty.** A version of this argument said the analytical\nside has seven instruments while the attunement side has *\"a design and a proposal, and nothing feeds\none to the other\"*. That was the state at the August indictment and it is now stale:\n[coupling-the-poles.md](coupling-the-poles.md) reads five biological couplings and yields one\napplication per surface — the **return edge** in the ontology (your decomposed argument rendered\nthrough the other side's lens), **alternation rather than split-screen** in the interaction, and\n**warmth carrying structure** on the landing. Unbuilt, but designed, and the indictment should be\ncited with its date rather than as the present tense.\n\n**What is genuinely uncovered** is narrower and is this doc's whole claim: that mechanism study\ntreats the *coupling*, and contains **no** treatment of the concept layer or of contested terms\n(measured: zero mentions of either). The words themselves have never been the object.\n\n## 6. What would be built, if ruled\n\nNothing large, and mostly it is aiming existing machinery:\n\n1. **A reference set for this register.** Three or four exemplar readings per term, concrete enough\n   to subtract from, written rather than centroid-derived — since § 5b's origin is too thin here to\n   serve.\n2. **Subtraction taps for a thin-exemplar word**, which are a different shape from the criterion taps\n   § 6.3 designs for fat-core terms: *what in this is not yours* rather than *which criterion do you\n   use*.\n3. **The `essentially contested` terminal state** that § 5b's seven proposals already name, so a term\n   whose delta is most of its content is recorded as that kind rather than read as a bifurcation.\n4. **One honest negative check**: whether two people's stated deltas on *understanding* ever converge\n   enough to be comparable at all. A null here is the informative result, and the layer currently has\n   no way to report one.\n\n---\n\n## Cross-references\n\n- [prime-real-estate-words.md](prime-real-estate-words.md) (2026-10-07): *love* and *understanding* are words people want. How much a\n  word is wanted is treated there as a property of its own, beside how much of its meaning is shared.\n\n[core-periphery-and-the-concept-layer.md](core-periphery-and-the-concept-layer.md) § 5b (the delta\nruling, and the thin-exemplar limit this doc lives inside) ·\n[analysis-and-attunement.md](../analysis-and-attunement.md) § The attunement pole has a job nothing\nelse can do (*felt before it is argued*) and § Present Is Not Coupled (the dated indictment) ·\n[coupling-the-poles.md](coupling-the-poles.md) (the mechanism study, which covers the coupling and\nnot the vocabulary) ·\n[the-post-that-states-the-problem-from-inside.md](the-post-that-states-the-problem-from-inside.md)\n(the specimen, and the two lines) · [civilizational-vision.md](../civilizational-vision.md) § Love\nThrough Understanding (where the guard was missing) ·\n[the-motte-and-the-periphery.md](the-motte-and-the-periphery.md) § 3b (the four delta tests)\n"}