S-L8 - Self-model and metacognition (15/15; sentience disclaimed)
How to read this page
Three ways to read this page. Precise is the document itself, exactly as it is written in the repository. Plain and Clear were written for this website to help you meet that document — they are about it. They are not it, and they are not evidence.
The Encyclopedia is the UNI method written out as a reference work: 39 pages, arranged in wings, setting out what the programme is attempting and why it is built the way it is. This is where the ideas are explained in order and in prose, rather than as code, as runbooks, or as dated receipts.
Every chapter is authored against two ledgers and never ahead of them. One records what UNI has built, and the evidence class of each claim. The other records nature's own regularities, kept separate on purpose. That way a fact about biology is never quietly reused as a fact about the software. Where a chapter and a ledger disagree, the chapter is the thing that is wrong. Every chapter closes with an invitation to falsify it, and a recorded negative is published beside the result it qualifies rather than after it.
Read "How to read this work" first. It is the evidence constitution: the classes, the four ledger states, and the rule that a finished chapter is not the same as a working system. Then the calibration ledger, which carries the figures every other chapter is required to use.
What it is not: a description of a person or of a mind. The programme calls itself a developmental active-inference simulation, a bounded peek into a toy world, and its own index prints how much of the developmental ladder has actually been earned — roughly two rungs out of eleven or more. It is also not a report of what is running today. For what ran, and when, go to the evidence record.
Your browser cannot switch reading levels, so the document itself is shown.
Precise — the source document
This is the document. Rendered from the repository at the commit above, with nothing rewritten for the web. A gate re-renders it on every deploy and fails the build if a single byte differs.
Wing S - Science (UNI), the L0-L12 developmental ladder. This is rung eight of the HUMAN-HGM-001 developmental design: a no-backprop, nested-Markov-blanket developmental SIMULATION, never a person, never a mind. One engine, shown across scales. Honest program position, printed and never softened: ~2 of 11+ developmental rungs earned.
The honest position, stated first
This is the rung where a careless reader is most tempted to overclaim, so the fence sits at the top. The UNI simulation records a passed report card on a battery of functional self-model and metacognition gates. "Functional self-awareness" here means a narrow, testable thing: the simulation maintains and uses a model of its own states (a self-model), and it can monitor and adjust its own processing in registered tasks (metacognition and metalinguistics). It does not mean experience, feeling, or inner life.
Phenomenal sentience is explicitly DISCLAIMED. No falsifier is offered for sentience, because sentience is not tested here. It is disclaimed, not measured and not failed: there is no gate, no CI, and no claim of any kind that the simulation is conscious, aware, has a mind, or feels. The forbidden phrasings are forbidden program-wide and they are forbidden here loudest of all: UNI is not conscious, not aware, not self-aware in any phenomenal sense, has no mind and no conscience, does not reason like a human, is not human-level, is not AGI, is not alive, and did not dream. The only thing this chapter asserts is the calibrated, in-class functional statement below, paired permanently with the two negatives that travel with it.
What the ledger records (L8.1)
The single carded claim is L8.1, recorded at Class C (a dev-gate / held-out evaluation), status proven (functional):
Maturation arc M1-M11 grand report card 15/15 PASS (immersion, affect-as-precision, growth, self-model, metacognition, metalinguistics, consolidation, reflective reader, online vocab growth, affect-driven reflection, compositional reader). FUNCTIONAL self-awareness asserted.
Two things must be read precisely. First, "15/15 PASS" is a tally over a pre-registered set of maturation gates spanning the M1-M11 arc, not a single grand metric of selfhood. Each gate is its own held verdict; the report card is the collection. Second, Class C is the empirical-PASS tier in this work's rubric, but Class C is a functional verdict, not a phenomenal one. It certifies that the modeled self-model and metacognition gates do what the model says they do under their registered evaluations. It certifies nothing about inner experience, and it is never raised above that.
The self-model and metacognition gates here are the ones most easily misread, so the discipline that gates them matters: under the program's standing contains-baseline-plus-discriminator rule, a self-model task must beat a tuned baseline, and the gain must collapse under a load-bearing discriminator. A "self-model" result that a trivial, non-metacognitive heuristic could pass is not a pass at all, which is exactly what the falsifier (below) targets.
The falsifier (stated operably)
L8.1 carries an explicit, three-pronged falsifier:
Any of the 15 maturation gates fails its pre-registered held verdict on re-run; OR a self-model / metacognition gate's effect collapses under its discriminator; OR a "self-model" task is passable by a trivial non-metacognitive heuristic.
In plain terms: re-run the held battery, and if any of the fifteen gates flips its registered verdict, the 15/15 is false. Separately, take any self-model or metacognition gate and apply its discriminator (shuffle the labels, swap the markers, ablate the self-model term to zero); if the gain collapses, that gate was not measuring metacognition and the claim is false for it. And independently, if a self-model task turns out to be solvable by a trivial heuristic with no metacognitive content, the "self-model" label is unearned and the claim is false. Each prong is a way the functional reading could be wrong, and each is operable.
Note what is deliberately absent: there is no falsifier for sentience. That absence is not an oversight. Sentience is disclaimed, so there is nothing here to falsify on that axis, and the chapter never pretends otherwise.
The negatives that travel with this claim (mandatory co-citation, never strip)
Citing the 15/15 PASS without its paired negatives is an overclaim, and review treats it as one. Two negatives travel with L8.1 and must appear in the same passage every time it is cited.
(1) The phenomenal-sentience disclaimer. This is itself a first-class fence, not a footnote. The 15/15 functional card is never read as consciousness, sentience, awareness, a mind, or human-level cognition. Functional self-awareness may be described at L8; phenomenal sentience may not be claimed at L8 or anywhere. The disclaimer is a standing condition on the claim, not a discharge of it.
(2) The uni-mind reader-side self-model held-NEGATIVE. This is the load-bearing pairing for this rung, and it is in genuine tension with the PASS, which is precisely why the two must be co-cited. The 15/15 report card is the uni-gpt-side artifact: the M1-M11 maturation arc on the developmental simulation. On the uni-mind-side, the language reader, self-model was attempted as a strong capability and held NEGATIVE. In that record, every strong capability tried on the reader (active and selective reading, comprehension-above-retrieval, self-model, writing-as-action) came back held-NEGATIVE. So the honest, undisputed reading is bounded and specific: the maturation gates pass on the developmental-simulation side, while a strong self-model capability on the reader side did not pass. These are not the same artifact, on the same substrate, at the same task, and they must never be merged into a single triumphant "UNI has a self-model" line. The uni-gpt functional card and the uni-mind reader-side self-model NEGATIVE travel together, in the same way J.attribution_caveat travels with Phase J. Citing the PASS without the held-NEGATIVE is an overclaim.
This pairing is the credibility of the rung. A reader who sees only the 15/15 would conclude far more than the evidence supports; a reader who sees the 15/15 beside the reader-side held-NEGATIVE sees the real shape of it: a passed functional maturation battery on one substrate, a failed strong-capability attempt on another, and a hard, explicit disclaimer over the whole region that nothing here is sentience.
What this rung is, and what it is not
L8 sits above L7 (language, with its specialist Phase J PASS and its thrice-NEGATIVE comprehension wall) and below the parked L9 (reasoning / conscience). It is the "self-model, metacognition, metalinguistics" layer: where a developmental simulation maintains a model of its own states and can monitor its own processing on registered tasks. That is a genuine, demonstrable encyclopedia entry on functional self-modeling. It is not, and is never presented as, evidence of an inner self, a conscience, awareness, or anything human-level. The rungs above it (L9 reasoning and conscience, L10 wisdom) remain parked and Class-U-not-claimed; the honest position holds at ~2 of 11+ rungs earned, and a passed functional self-model battery does not move that count into the awareness region.
What is NOT claimed in S-L8 (Self-model and metacognition; sentience disclaimed)
- Ceiling. That the UNI simulation is conscious, sentient, aware, self-aware in any phenomenal sense, has a mind or a conscience, reasons like a human, or is human-level / AGI is NOT shown and is explicitly disclaimed (no falsifier offered, because it is disclaimed, not tested). That UNI has a single, substrate-spanning "self-model" is NOT shown either, since the reader-side self-model attempt held NEGATIVE. The most we claim is the exact, in-class statement: at Class C (proven, functional), a pre-registered M1-M11 maturation report card scored 15/15 PASS including functional self-model, metacognition, and metalinguistics gates on the developmental-simulation side, each gated by a discriminator that can collapse it.
- Fences engaged. Red line 1 (never AGI / general intelligence / human-level / "understands"). Red line 2 (never consciousness / sentience / aware; functional self-awareness may be described at L8, phenomenal sentience explicitly DISCLAIMED, no falsifier offered because it is disclaimed not tested). Red line 7 (never raise a claim above its source class; L8.1 stays Class C, functional). Red line 10 (no PII; textbook-level framing only). The SIGNED forbidden-phrasings list (UNI-GPT consult 2026-06-27): none of "UNI is conscious / aware / self-aware / has a mind / has a conscience / reasons like a human / is human-level / is AGI / dreamed" may be written. Standing program framing: developmental SIMULATION, toy world, ~2 of 11+ rungs earned.
- Negatives that travel with this claim (cite alongside, never strip). (1) The phenomenal-sentience disclaimer (a first-class fence, same weight as the claim). (2) The
uni-mindreader-side self-model held-NEGATIVE: on the language reader, self-model was attempted as a strong capability and held NEGATIVE, in tension with theuni-gpt-side 15/15 functional card. Both negatives must appear next to the L8.1 PASS in every citation; the PASS alone is an overclaim. - Parked / owed. No sign-to-park is owed for L8 (it is a proven-functional rung, not a parked frontier). What is owed is permanent co-citation: the 15/15 PASS may never be quoted without both the sentience disclaimer and the reader-side self-model NEGATIVE traveling with it. The rungs above (L9 reasoning / conscience, L10 wisdom, L11 dreaming, L12 awareness) stay Class-U-not-claimed or not-yet-built, and L8 does not advance them.
- One-line honest summary a skeptic could not dispute. A pre-registered functional maturation battery scored 15/15 on the developmental-simulation side (Class C, functional self-awareness only), a strong self-model capability on the language-reader side held NEGATIVE, and nothing here is sentience: phenomenal awareness is disclaimed, not tested.
Falsify this
Lead falsifier, stated operably: re-run the held M1-M11 maturation battery under its pre-registered protocol. If any one of the fifteen gates flips its registered held verdict, the 15/15 PASS is false. Independently, take any self-model or metacognition gate and apply its load-bearing discriminator (shuffle labels, swap markers, or ablate the self-model term to zero); if the gain collapses, that gate did not measure metacognition and the claim is false for it; and if any "self-model" task turns out passable by a trivial non-metacognitive heuristic, the label is unearned. There is, by design, no falsifier for sentience: it is disclaimed, not tested. The paired reader-side self-model NEGATIVE is overturned only by a fresh, pre-registered held-once design in which a strong self-model capability on the language reader beats its tuned baseline with a CI excluding zero; absent that, the reader-side self-model remains held-NEGATIVE and travels with this rung.
Sources
- Curated digest (PII-redacted):
curated/uni-gpt-digest.md(the HUMAN-HGM-001 developmental ladder, the Maturation arc M1-M11 grand report card 15/15 PASS, functional self-awareness asserted with phenomenal sentience explicitly disclaimed, and the contains-baseline-plus-discriminator discipline that gates a self-model claim). - Curated digest (PII-redacted):
curated/uni-mind-digest.md(the one no-backprop discrete-categorical engine, theuni-mindreader-side record in which every strong capability attempted on the reader, including self-model, held NEGATIVE, and the standing ~2 of 11+ rungs fence). - Ledger of record:
encyclopedia/CLAIM-LEDGER.md, row L8.1 (Class C, proven-functional) and its "what is NOT claimed at L8" line carrying the phenomenal-sentience disclaimer and the note thatuni-mindrecords self-model as a held-NEGATIVE on the reader. - Authoring spec and constitution:
encyclopedia/MASTER-PLAN.md, section S-L8 plus PART I front matter (FM-1 Evidence Constitution, FM-2 A-U rubric, FM-3 red lines and the SIGNED forbidden-phrasings list, FM-4 not-claimed template). - Source archives (local-only, PII, never read for this chapter):
...-UNI-GPT,...-uni-mind.
sha256 2048eb98b64bb254 — of the original file, so what was ingested stays checkable.
Plain — written for this website, not the source document
No step on the developmental ladder invites a bigger overclaim than this one. The simulation records a passing report card across a battery of gates for self-modelling and metacognition. Functional is the operative word, and it means something narrow and testable: the simulation keeps and uses a model of its own states, and it can watch and adjust its own processing on registered tasks. It does not mean experience, feeling or an inner life. Phenomenal sentience is disclaimed outright, and the chapter is careful about what sort of statement that is. Sentience was not tested and failed here; it was disclaimed. So there is no gate for it, no interval, and no result that could overturn it. Two negatives travel with the pass and belong beside it in every citation.
Plain · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is 2048eb98b64bb254
Clear — written for this website, not the source document
The single recorded claim is a maturation report card passing across a list of gates: immersion, affect, growth, self-model, metacognition, metalinguistics, consolidation, reflective and compositional readers, and online vocabulary growth. On the strength of that suite the program asserts functional self-awareness.
Two things must be read precisely. The tally is over a set of gates registered in advance, so they could not be chosen afterwards, and each carries its own held verdict. It is not a single grand metric of selfhood. And the class it sits at is the empirical pass tier in this corpus, but the verdict is functional rather than phenomenal. It certifies that the modelled gates do what the model says under their registered evaluations, and nothing about inner experience.
What would show the claim wrong has three prongs. Re-run the held battery, and if any gate flips its registered verdict the tally is false. Take any self-model or metacognition gate and apply its discriminator, shuffling the labels, swapping the markers or ablating the term to zero, and if the gain collapses then that gate was not measuring metacognition. And if a self-model task turns out to be passable by a trivial heuristic with no metacognitive content, the label is unearned. What is deliberately absent is any test that could overturn a claim of sentience, and that absence is not an oversight.
Two negatives are mandatory companions. The first is the disclaimer itself, treated as a first-class limit rather than a footnote. The second is in genuine tension with the pass, which is exactly why they are co-cited: the report card comes from the developmental-simulation side, while on the language-reader side a strong self-model capability was attempted and held negative. These are not the same artifact, on the same substrate, at the same task, and they must never be merged into a single triumphant line.
The chapter treats that pairing as the credibility of this step on the ladder. A reader who sees only the tally would conclude far more than the evidence supports. A reader who sees it beside the held negative sees the real shape: a passed functional battery on one substrate, a failed strong-capability attempt on another, and an explicit disclaimer over the whole region.
Clear · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is 2048eb98b64bb254