This requirement says that physics can only be tested and compared to reality through finite, reproducible records — measured numbers, ledgers, repeatable procedures, anything that can be written down and checked. It is a plain rule about what counts as evidence, not a claim about minds: a "record" here just means any finite, physically instantiated trace, never consciousness or a subjective observer.
Nonseparability matters here because a record can never be assumed to belong to one isolated piece of the universe — the record and the thing it describes cannot be taken as independent of the rest of reality; that independence has to be earned, not built in by default. And because granularity (the smallest meaningful distinctions a system can support) is treated as coarser away from t=now and finest at t=now, what counts as a "finite record" is itself not a fixed, universal grain — it depends on where in the reconstruction you are asking the question, which is part of why this requirement is treated as a starting condition rather than something derived from something deeper.
Honest status: this is anchored — a foundational, load-bearing root of the program, an anchored starting point rather than a derived result. It sits upstream of the physics as the condition for any claim to be checkable at all, so it is best read as an anchored posit, not proven from something more basic. Nothing about it is claimed as a theorem, and it carries an explicit anti-claim: it does not say or imply that "observer" or "record" means a conscious mind.
This item asks whether reality has a smallest operational "step" — a positive floor on the cost of making one record distinguishable from another, below which no finer distinction can be made. If such a floor exists, it sets a genuine cost floor under everything built on top of it, including the layers this page groups under "consciousness." The floor is framed as a cost on a scalar quantity (action), not as a smallest length — so it does not claim space itself comes in fixed-size pixels, and it picks out no preferred frame.
Nonseparability matters here because a floor cannot simply be assumed to hold uniformly for every system just because each system, taken alone, has only finite resources. Two explicit counter-examples show that finite resources by themselves force at most a per-system, non-uniform limit, not one common floor shared across all of reality — the independence of "this system's floor" from "the rest of reality's floor" would have to be earned, not assumed. That is also where the "granularity was coarser toward the Big Bang" idea lives: nothing here licenses treating the grain size as fixed once and for all across time; a uniform floor is a separate, explicitly named assumption, not a free consequence of finiteness.
Honest status: this root is certified irreducible — the Δ₀ cost-floor posit is named openly and proven non-derivable (two counter-models); resolved at +0. Given that the space of records is compact, the existence of a positive floor is proven outright. Compactness itself is shown only conditionally, given a finite-test-family assumption plus a completeness assumption, both carried in the open rather than hidden. The claim that the floor is the same size everywhere (uniformity) is an openly declared, unproved assumption — proven not to follow automatically from finiteness alone. So this piece rests on one named, honestly labeled posit plus one measured quantity (Planck's constant, entering only as the floor's size, never derived); certified irreducible — the posit proven non-derivable from deeper principles — resolved at +0.
This asks a basic question about any two things you'd like to call "separate" — a lab system and its surroundings, one part of the universe and another, an observer and the moment they're looking back on: what actually earns them the right to be treated as independent? Writing down "A and B are independent" is one sentence, but the constraint-first picture treats that sentence as a claim that has to be paid for, not a free starting assumption. A separation counts as real only when there's an actual finite record — a measurement, a screening, a derivation from the shape already in hand — that backs it up.
The frozen 13-dimensional shape behind this reconstruction doesn't presuppose a "you" standing outside it, and it doesn't presuppose that its parts come pre-cut into independent pieces. Distance from the present matters here: the further a claimed vantage point sits from now (or, in the extreme, from any moment at all — "before" the universe's own record-keeping begins), the less likely it is that a finite record could ever exist to back it up. That's not a claim about hidden information being hard to find; it's a claim that for some worded-up "observers," no admissible record exists even in principle, so the question resting on them isn't unanswered — it isn't well-formed. Meanwhile, an ordinary observer riding along with the early universe a few hundred thousand years in is a case where the record does exist and is one of the best-measured things in physics — that question stays fully valid.
Honest status: this is anchored as a bookkeeping discipline — a fully specified method, applied to real cases with named outcomes. Every claimed separation lands in one of four honest boxes — backed by observation, derived from the shape already granted, an acknowledged open debt, or (rarely, and only when earned case-by-case) not a valid question at all — and unpaid separations stay openly labeled as open, never quietly upgraded. Whether any given separation actually passes is a separate, case-by-case question, and several remain open. The discipline resolves no open physics problem elsewhere and produces no prediction or number; it sorts which questions about "before" or "outside" are the kind physics can address at all.
The Born rule is the familiar quantum recipe that turns a wavefunction into a probability: the chance of seeing outcome $k$ is $|\langle\phi_k|\psi\rangle|^2$, the squared overlap. This gate asks whether that "square the amplitude" weighting can be pinned down by the framework's own structure, rather than simply written down by hand as an extra rule. The headline result is that it can be pinned, given one clean assumption: the exponent is forced to be exactly two, not 1.9 and not 2.1.
Nonseparability matters here because probability weights are exactly the kind of thing that must be "earned," not assumed. Because records are discrete and finite, the elimination of rival weighting rules is decisive: rules giving negative or complex weights, rules that fail to compose, and rules demanding distinctions no finite record can carry are all struck out, and the single survivor is the amplitude-squared class — the pre-record freedom in the measure collapses to that one operational class. Two residuals are shown in the open. First, the framework's gauge symmetry (from its BRST structure) forces outcome-probabilities to agree across gauge-equivalent descriptions, but a careful check (a working countermodel) shows this alone does not force the deeper property — "non-contextuality" — that the amplitude-squared form actually rests on; so non-contextuality is knowingly imported as the one paid axiom. Second, when the framework looks for a single natural probability measure over the universes it could select, the best candidate it can build lives on a finite symmetric chamber of the extra-dimensional shape — much narrower than an unconstrained guess, with the natural point singled out but not proven the unique choice.
Honest status: RESOLVED at +0 — certified-irreducible, with two residuals shown — not rolled up to "open." The exponent p = 2 is a terminal result, and the pre-record measure freedom genuinely collapses to the single operational |amplitude|² class; the gate rests on one named assumption — non-contextuality — shown openly in the dossier, not derived from nothing. The two residuals are carried openly rather than hidden: (1) gauge invariance alone does not force non-contextuality — a checkable countermodel proves this — so that one assumption is imported by name, not smuggled in; and (2) the natural measure over candidate universes is narrowed to a finite symmetric region but cannot yet be shown unique. Neither residual is rounded up to "derived," and neither downgrades the reached terminal: this is certified-irreducible — reduced to one named axiom, with two honestly-shown edges, each with a stated path to attack it.
This item is about a rule for how the framework treats "separateness." It is tempting to assume that any two systems — two labs, two particles, two moments in time — are simply independent of each other unless something links them. This requirement says the opposite: independence is never a free starting assumption. Every claimed separation between a system and the rest of reality has to be accounted for — backed by an actual measurement, derived from the shape already in use, or admitted as an open debt — before it can be relied on. The universe can still be lumpy and full of everyday, reliable separations (your coffee is distinct from your desk); the point is that each one is earned, not assumed.
Nonseparability connects to time-dependent granularity because the same accounting discipline applies to observers and records at different times. Reality is treated as one constrained whole, and any slice of it — a system now, an observer long ago, a moment "before" some boundary — only counts as well-posed if a finite physical record could exist for it. Near the present, the finest distinguishable grain of reality is far below anything we probe, so ordinary questions about separate systems are unproblematic. Further from now, the achievable graininess of any record changes, and some questions that presuppose an observer or a record can fail this test entirely — not because they are unanswered, but because the object the question presupposes (a record-keeper at that point) is not there. When that happens, the question is re-typed as not askable rather than left open.
Honest status: this is an anchored, resolved accounting rule, not an open computation — a discipline for classifying any claimed separation into one of a few honest boxes (backed by observation, derived from the accepted shape, an acknowledged open debt, or not askable at all), and it changes no other result's status. The one class of question it re-types as "not askable" (for example, an observer's records "before" a stated boundary) is held only as conditional on the framework's own scope and roots, not proven outright — so that specific verdict stays labeled as a declared boundary conditional on scope rather than a certified proof. Nothing about the deeper structure of reality is claimed settled by this rule, and it resolves no open gate elsewhere.
This is a proposal for how a mind could relate to time within the framework: a repeating cycle where a state is recorded, that record is thought about (read, classified, and used to help select which future paths are taken), and the result is written back as a new record — over and over. In plain terms, it tries to describe consciousness not as a single frozen fact but as a loop that keeps re-reading and re-writing its own history. The point of this expanded note is to take each word in "record → thought → record" and ask what physics genuinely supplies for it, and where the physics runs out and speculation begins. Two of this site's anchored roots do real work here: that physics is accessed only through finite, reproducible records, and that distinguishable transitions carry a positive cost floor (granularity). The loop below is, quite literally, assembled out of those two ideas — which is exactly why it is worth stating carefully rather than waved at.
Anchored on the physics — Shape · Granularity · Scale · Observed
The record-thought-record loop is not a new law; it reduces onto roots this site already carries — an admissible trajectory on the frozen shape, every step paying the granularity cost-floor sized by the measured residue ℏ. In the site's discipline this is anchored, not derived: the loop borrows the roots' vocabulary and evidence and lends the anchored physics no weight in return.
Shape (the frozen 13-D object; ×Stage · ⊕Rulebook · ⊗Actors): A record is an admissible configuration on the active branch — a stable, redundantly copied state in the decoherence-selected pointer basis (the in-page Records root). All three layers are load-bearing; the geometry alone is not the shape. The record persists in the ×Stage 4-D arena M4 (from M4 × K6 × S² × S1Y/ℤ2, K6 = SU(3)/T²); a “thought” is an operator acting on that record, living in the ⊗Actors (bundle/operator) layer that carries dynamics and readout; the ⊕Rulebook layer fixes which transitions are admissible. This is Shape-supported, in the site's own sense: no worked derivation exhibits a map from the branch to the loop, so it is never Shape-forced.
Granularity (cost-floor Δ0 > 0; ℏ its measured residue): Every loop step is a distinguishable transition, and distinguishable transitions are never free — the in-page Granularity root (the Uniform Operational Cell Law, Δ0 > 0). Reading, rewriting, or erasing a record pays at least the floor; there are no free relabels. The floor is on action/cost, a Lorentz scalar — not a smallest length — applied across all layers and all 13 dimensions including time, which orients the loop's arrow (records of the past, not the future). Its existence is a declared posit; its size is the measured anchor ℏ — never merged.
Scale (absolute anchor MPl): Honestly, this item does not reach the master scale anchor MPl. A static object supplies structure, not magnitude, and the loop makes no dimensionful prediction — no rate, energy, or step-time in GeV. The only scale it touches is ℏ, the measured residue sizing the minimum step-cost: scale enters as a measured anchor (ℏ), not a derived magnitude, and not via any bridge from MPl. In the site's tags: SCALE-ANCHORED on ℏ, SCALE-OPEN for any absolute rate. Claiming a predicted magnitude here would be the exact overclaim the Scale page forbids.
Observed phenomenon: Two measured hooks anchor the “record” and “cost” halves. (i) Records are physical, stable, and redundantly copied — decoherence and quantum Darwinism, confirmed experimentally. (ii) Erasing one bit costs at least kBT ln 2 — Landauer's bound, measured in colloidal, nanomagnet, and quantum-dot systems near the quasi-static limit. Named open residue: there is no measured signature distinguishing a “conscious” loop from any other record-processing loop of the same shape. The physics anchors record-keeping; it supplies no observable for felt experience, and none is manufactured here.
Honest status. Genuinely anchored: the loop's ingredients reduce onto the Records root (Shape), the cost-floor Δ0 with ℏ as its residue (Granularity), and two measured phenomena (quantum Darwinism; Landauer). Genuinely open: it is Shape-supported, not forced — no worked map derives it from the frozen branch; it is Scale-anchored on ℏ only, with no MPl magnitude and no absolute-rate prediction; and the consciousness-specific claim — that such a loop produces experience — has no measured signature and belongs to the separate Knowing root. Anchored is not derived; selected is not forced; dissolved is not solved; floor ≥ 1, forever.
What a “record” is, physically
A record is not just any physical state. It is a stable, redundantly encoded, classically readable configuration: something you can read, and copy, without destroying it, and that many independent observers would agree about. That last property — objectivity — is the hard part, and it is where decoherence theory earns its keep. Left to itself, a quantum system explores superpositions and its phase relations stay coherent; a would-be record written in such a state could, in principle, be run backwards and un-happen under unitary evolution. Something that can be coherently reversed is not yet a record. Records require effective irreversibility: the information has to leak into so many degrees of freedom that recombining them to reverse it is, for all practical purposes, impossible.
Decoherence supplies the mechanism. When a system couples to a large environment, only certain states — the pointer states, in Zurek's terminology — survive the interaction without being smeared out; superpositions of pointer states rapidly lose their relative phase into the environment. The pointer basis is selected by what the environment monitors, typically something close to position for macroscopic objects, which is why the classical world looks made of localized things. A record is written in this decoherence-selected, robust basis: that robustness is exactly what lets it be read repeatedly and copied faithfully.
Quantum Darwinism sharpens this into a story about why an outcome becomes an objective fact. Zurek's proposal is that information about a system does not merely leak into the environment — it proliferates, imprinting many redundant copies across independent environmental fragments (the scattered photons, the surrounding air molecules). Because there are many copies, many observers can each intercept a small fragment, never touching the system itself, and still agree on the same result. Objectivity, on this view, is redundancy: a fact is “out there” precisely to the degree that its record has been copied many times over into the world. This is the physical content of the site's anchored records root — physics reaches us only through such finite, reproducible imprints — and it is what the “record” step of the loop is meant to denote. It is worth flagging honestly that quantum Darwinism describes how classical records emerge given a division into system, apparatus, and environment; it presumes rather than derives that division, and it does not by itself resolve which single outcome is realized (the measurement problem in its sharpest form remains open, and is treated differently by Everett, GRW-type collapse, and Bohmian approaches).
Why records and loops are never free: the cost floor
If a record is effective-irreversible information written into many degrees of freedom, then writing, keeping, and especially clearing records has a thermodynamic price. The cleanest statement is Landauer's principle: erasing one bit of information — resetting a two-state memory to a known value regardless of its prior content — dissipates at least k_B T ln 2 of heat into the environment, where k_B is Boltzmann's constant and T the temperature. The reason is entropic bookkeeping: erasure maps two prior states onto one, halving the memory's phase-space volume, and that missing entropy has to be paid for in the surroundings so the second law of thermodynamics stays satisfied. Landauer's bound has been probed in real experiments over the past decade — with single colloidal particles in optical traps, with nanomagnets, and in electronic and quantum-dot systems — and the measured heat approaches k_B T ln 2 in the slow, quasi-static limit, consistent with the principle.
A crucial refinement, due to Charles Bennett, is that logically reversible computation carries no such fundamental floor: any computation can in principle be restructured so that it never discards information, and then it can be run arbitrarily close to zero dissipation, at the price of keeping a growing history and running slowly. The unavoidable cost attaches to the irreversible steps — the erasures, the resets, the moments where distinct histories are merged and one is thrown away. A persistent loop that must keep freeing memory to make room for new records cannot escape this: each cleared record has a Landauer floor. This connects directly to the site's anchored granularity / cost-floor root. Distinguishable transitions are never free. Every genuinely new, distinguishable record, and every iteration of a loop that overwrites what came before, sits on a strictly positive floor cost. The arrow of time the loop rides on — records of the past but not the future, the ability to remember and not to “pre-member” — is the same thermodynamic asymmetry that Landauer's principle expresses: making a distinction and committing it to a stable record is what pays entropy forward and orients the loop in time.
The loop as physical computation and feedback
Read as a whole, record → thought → record is a description of physical computation with feedback and memory. The “thought” step is a physical process that reads the currently accessible records, classifies them, and uses that classification to bias which future record-paths are actually taken — and then commits its verdict as a new record that the next iteration will read. That is precisely the structure of a control loop with state: a system that carries a model of its own recent history and conditions its next move on it.
Standing research programs describe loops of exactly this shape. The most prominent is predictive processing and Karl Friston’s free-energy principle: a brain modelled as a loop that maintains a generative model, predicts its sensory input, and updates either the model or (via action) the world to reduce the mismatch — records of expectation compared against records of sensation, driving the next state. It is a powerful modelling framework; whether it is the whole story of cognition is an open scientific question, not a settled result. (The physical Church–Turing thesis frames the companion claim — that “a thought is a computation” has physical, thermodynamic teeth — and it too remains a thesis, not a theorem.)
What all of these genuinely share, and what physics genuinely supplies, is this: a self-referential loop that reads and writes its own records is a physically ordinary object. It needs no new force and no exotic ingredient. It needs stable records (decoherence and redundancy provide these), it needs a way to read them non-destructively (redundant copies provide this), and it must pay the cost floor at every irreversible step (Landauer and the second law fix this). In that limited sense the mechanics of a conscious process — the record-keeping, the feedback, the self-model — are describable in standard terms without any appeal to mystery.
How this ties to the framework's shape
Nonseparability matters here because a thinking system is never cleanly cut off from the rest of reality. Its independence from everything else has to be shown, not assumed — and a record-thought-record loop is exactly the kind of thing that looks self-contained but may not be, since the very redundancy that makes its records objective consists of correlations reaching out into the wider environment. Drawing the boundary of “the loop” is a modelling choice, and the framework insists that such boundaries be justified rather than taken for granted.
The framework's shape also lets granularity — the fineness of what can be distinguished — change with distance from the present moment, becoming coarser the further back or forward one looks. If that is right, a memory loop like this one would naturally blur older records rather than keeping every step equally sharp: distant records would be held at coarser resolution, cheaper to retain but less finely distinguishable, while the present is resolved most finely. That is an appealing feature — it would mirror how remembered experience actually degrades with time, and it would tie the fading of memory to a cost-floor that rises as you demand fine distinctions about the remote past. But it must be stated as a possibility, not a deliverable: nothing here has been shown to require or produce that blurring on its own.
What this model does not do
The honest boundary must be drawn sharply. Everything above is a plain-English model of the mechanics of a conscious process — how a mind might keep, read, and rewrite its own records, and what such a loop must cost. It is emphatically not a demonstration that running such a loop produces experience. It says nothing about why the loop feels like anything from the inside — why there is something it is like to be the process rather than mere information-processing in the dark. That is the “hard problem” in David Chalmers' sense, and on this site it belongs to the separate Knowing root; the loop model neither answers it nor pretends to. Nor does describing the loop in thermodynamic and decoherence terms adjudicate between rival theories of consciousness such as Integrated Information Theory or global-workspace accounts, or between interpretations of quantum mechanics; it is deliberately neutral about all of these.
Honest status. This is a speculative idea, not an established result. There is no worked derivation showing that the record-thought-record loop follows from the framework's frozen shape, and no calculation tying it to a measured quantity. The pieces it leans on — pointer states, quantum Darwinism, Landauer's principle, Bennett's reversibility, the free-energy principle — are real physics and real theory, cited here because they describe loops and records honestly, not because they have been assembled into a proof about mind. It remains a plain-English sketch of a possible picture of consciousness under construction, offered for what it is rather than rounded up into a finished claim. It carries no weight for the anchored physics elsewhere on this site: the anchored roots stand on their own evidence, and this loop is a downstream speculation that borrows their vocabulary without lending them any support in return.
This is the central speculative claim of this section, and it is worth stating carefully before defending or doubting it. The proposal is that first-person experience might be reduced along a route — re-described, step by step, in successively lower-level terms — passing through a specific chain: experience → a conscious loop → observation or measurement → thought → records. The hope is that this chain does not regress forever, but instead bottoms out in a small set of terminal roots the framework cannot reduce any further. What follows is an attempt to say, as evenhandedly as possible, what that would mean, what physics genuinely supplies here, and why the whole program might simply be the wrong decomposition.
Anchored on the physics — Shape · Granularity · Scale · Observed
The route (experience → loop → observation → thought → records) is a chain of re-descriptions whose terminals are already anchored roots on this page — Records, the Born weight, Nonseparability — and whose admissibility is the finite-cost discipline the gates run on. In the site's sense it is anchored, not derived: Shape supplies where it must land, Granularity supplies whether each step was paid and where the chain must stop, and Scale enters only weakly because the claim is structural. What is not anchored is the one thing the route is about — that felt experience actually reduces along this exact chain.
Shape (the frozen 13-D object; ×Stage · ⊕Rulebook · ⊗Actors): Shape does not model consciousness; it supplies the terminals a reduction may bottom out on, each a genuine root of the frozen branch. Records land on the Record Interface / readout maps — the ⊗ Actor layer that carries any observable off the geometry (a static geometry produces nothing without a readout map). The measurement→record step lands on the Born weight, which Shape treats as forced structure, not free choice (Gleason). The correlation structure lands on Nonseparability — the root that also forbids splitting the route's cost ledger into independent sectors for free. The route is thus Shape-supported: the object supplies real terminals and a re-description grammar, but does not force that experience is one of the things it reads off — no Actor bundle, operator, or readout map for “feel” exists on the branch. Selected is not forced.
Granularity (cost-floor Δ0 > 0; ℏ its measured residue): this is the load-bearing anchor. A reduction step is admitted only if it is paid — recoverable from a finite, charged record with exact object identity and no unpaid labels — the same discipline the gates use. It decides two things at once. First, it makes the route falsifiable: a step underwritten by no finite record, or by a silent object swap, is inadmissible, not merely hard. Second, it sets where the chain must stop: the uniform positive cost-floor Δ0 > 0 (the Uniform Operational Cell Law, with ℏ as its measured residue) means a step demanding a distinction below the floor cannot complete under finite resource, so an infinite regress of costly re-descriptions cannot finish. The route therefore terminates at a finite set of roots: floor ≥ 1, forever. Honestly bounded — this forces the chain to have a finite bottom and to pay its way there; it does not force that felt experience sits above that bottom rather than below it (finite records support the discipline; the stronger claim stays a posit).
Scale (absolute anchor MPl): Scale enters only weakly, and honesty requires saying so. The reduction claim is dimensionless and structural — about admissibility of steps and identity of terminals, not any magnitude in GeV. By the dimensionless-first doctrine such a claim needs no absolute anchor: MPl does no work here. Scale contributes one thing indirectly — the floor Δ0 lives on a Lorentz-scalar action whose measured size is ℏ. Beyond that there is no Scale-derived quantity in this item, and none is claimed; reporting a magnitude for “how much” experience reduces would be a category error the Scale page explicitly forbids.
Observed phenomenon: the route's observable hook is entirely at its terminals, each independently measured or anchored — Born statistics measured to high precision; records what every experiment logs; Nonseparability tested via Bell-type correlations. That is real empirical contact for the destinations. The named open residue is the item itself: no observable yet distinguishes “felt experience reduces along this route” from “it does not.” There is no measured signature of felt quality — no experiment whose outcome changes if the chain routes around experience rather than through it. Until a falsifiable signature of the feel (not merely the function) is exhibited and frozen before reading it, the claim has anchored terminals but an unmeasured core.
Honest status. What genuinely anchors: the route's terminals are anchored/measured roots of the frozen object (Shape), and every step is disciplined by the same finite-cost, exact-identity, floor-≥-1 rules that govern the gates (Granularity) — which also guarantee the chain is finite and falsifiable rather than an endless regress. What does not anchor: that felt experience is among the things Shape reads off, and any Scale magnitude for the reduction — both are absent, and the record-accountable chain may simply end before experience is reached, leaving the feel as an unpaid remainder. In the site's discipline this is anchored, not derived; supported, not forced; a well-posed, testable proposal, not a solution — and its most defensible reading is the humblest: clear enough to be argued about, tested against the terminals, and quite possibly refuted.
What "reduction along a route" means
A reduction, in the sense used here, is a re-description of a phenomenon in terms of something more basic, together with a bridge that says how the higher-level description is realised by the lower one. The classic template is Ernest Nagel's account of intertheoretic reduction: a theory reduces to a more fundamental one when its laws can be derived from the fundamental theory plus bridge principles connecting the vocabularies (temperature to mean molecular kinetic energy is the stock example). "Along a route" simply means we do this in stages rather than in one leap — each link in experience → loop → measurement → thought → records is its own small reduction, and the chain is only as strong as its weakest bridge.
The discipline this site imposes is what makes the route more than hand-waving, and also what makes it falsifiable. A step is admitted only if it can be paid for with a finite record — if the coarser description can be recovered from a bounded amount of stored, physical information about the finer one. This is not a slogan; it has a real physical anchor in Landauer's principle, which ties information to thermodynamics: erasing one bit in a bath at temperature T dissipates at least k_B T ln 2 of heat (k_B the Boltzmann constant). Records are not free. If a proposed reduction step would require an unbounded or unpayable record to underwrite it, the step is disallowed. The route therefore self-terminates: it continues only while each re-description is finitely accountable, and it stops where no finite record can buy the next step. Where the chain stops is exactly what we mean by the named residues — the open roots. The residues are not decreed in advance; they are meant to be whatever survives when the accountable reductions run out.
What physics genuinely supplies
Some links in this chain are on comparatively firm ground, and it is important not to overstate the mystery by ignoring them. The step from observation/measurement toward records is the best-understood, because it is essentially the physics of measurement and decoherence. When a quantum system couples to a large environment, interference between its alternatives is suppressed extraordinarily fast as which-path information leaks into uncontrollable degrees of freedom; the system's reduced state becomes, for all practical purposes, a classical mixture in a preferred (pointer) basis. Decoherence does not by itself say which single outcome occurs — that is the measurement problem — but it does explain why definite records form and why the world looks classical at our scale. The statistical weights themselves are not a free choice: Gleason's theorem and, under weaker assumptions, Busch's and related results show that the Born rule — probabilities as |amplitude|² — is essentially forced once you accept the Hilbert-space structure and demand a consistent probability measure over measurements. So "measurement → records" is a place where physics really does deliver structure, mechanism, and constraint.
What physics does not settle is which interpretation underwrites that structure — Everett, dynamical-collapse, and Bohmian readings each place the route’s terminal roots differently (the full, evenhanded menu is under the Observing root). The route can locate where an irreducible residue sits without dictating the metaphysics attached to it, and it must not smuggle in a favourite interpretation as if it were established.
Why this is genuinely contested — and why it may fail
Here the honest voice has to get sharper, because the strongest objections in philosophy of mind are aimed at precisely this move. The worry is not that the structural reductions are hard; it is that completing them might not touch the thing we care about at all.
The hard problem (David Chalmers) draws a line between the "easy" problems — explaining discrimination, integration, reportability, the control of behaviour, all of which are functional or structural and plausibly reducible — and the hard problem of why any of that functioning is accompanied by experience at all. The force of the distinction for this route is direct: even a flawless reduction of the structure (loop, measurement, thought, records) need not be a reduction of the feel. You could pay for every structural step with a finite record and still not have said why there is something it is like to be the system running that route. Reducing the structure and reducing the experience may be two different tasks that only look like one.
The explanatory gap (Joseph Levine) puts the same pressure epistemically rather than metaphysically: even granting that experiential facts supervene on physical facts, we lack any account that makes the connection intelligible — that lets us see why this physical arrangement must feel like that, rather than like something else or like nothing. A route that terminates in roots would, at best, mark where the gap lives; it would not obviously close it. There is a real danger of mistaking "we have localised the residue" for "we have explained the residue."
The knowledge argument (Jackson’s Mary) and the conceivability of zombies press the same worry from two more angles — both are laid out in full under the Knowing root. If the zombie intuition has any bite, the chain this route walks is, by construction, the structural chain a zombie would share — walking it to its roots may be walking around experience rather than through it.
Two ways the route could quietly cheat
Beyond the external objections, there are two internal failure modes worth naming plainly, because a proposal that cannot state how it might be fooling itself is not yet a serious proposal.
First, the route may smuggle in what it means to explain. If "reduce" is tacitly defined as "give a complete structural/record-based account," then the conclusion — that experience reduces to structural roots — is baked into the vocabulary from the start, and the hard problem has been assumed away rather than answered. A reduction that only ever admits structural currency cannot, without begging the question, claim to have captured a non-structural residue if there is one. Guarding against this means being willing to conclude that the terminal root is not structural — that the record-accountable chain genuinely ends before experience is reached, leaving the feel as an unpaid remainder rather than a solved one.
Second, the residues may not be truly terminal. "We cannot reduce this further" is easy to confuse with "we have not yet found how." A residue that looks irreducible under today's descriptions could dissolve under a better one, exactly as vital forces dissolved into biochemistry; conversely, a residue that looks like clean detail could hide a strong-emergence break. The finite-record discipline is meant to be a check on this — a root is only accepted as terminal if one can show why no finite record buys the next step, not merely that none has been found — but that "why" is itself a demanding claim, and the site does not pretend to have discharged it for experience.
Honest status
To be unambiguous: this is the speculative claim itself, offered as a structured proposal for further analysis, not as established science, and not as a solution to the hard problem. The framework supplies real and non-trivial material for the later, more physical links — decoherence and record formation, the Born rule as constrained rather than chosen (Gleason), the thermodynamic price of information (Landauer), and a menu of interpretations (Everett, GRW, Bohm) that read any terminal root differently. It does not supply a defined measure of "experience," a mechanism connecting structure to feel, or any calculation that would turn this route into a result. Nothing here is derived; it is at most a well-posed question sitting next to the physics.
It could straightforwardly be the wrong decomposition. Experience → loop → measurement → thought → records is a natural-looking chain, but "natural-looking" is not "correct," and each of Chalmers, Levine, Jackson, and the zombie argument gives a distinct reason the chain might route around the phenomenon it is trying to reach. The most defensible reading of the claim is therefore the humblest one: not that experience does reduce along this route, but that the route is a clear enough proposal to be argued about, tested against these objections, and — quite possibly — refuted. That would still be progress. Locating precisely where an honest reduction fails is more valuable than an over-confident claim that it succeeds.
“Knowing” gathers two of the hardest open questions about mind into one place. The first is why any physical process is accompanied by an inner experience at all — why there is “something it is like” to see red or feel pain, rather than the same information being registered, discriminated and reported with no felt quality attached (the problem of qualia). The second is why that experience is centered as a first-person “me” — an owned point of view — rather than an unowned, view-from-nowhere pattern of events (the problem of selfhood). These are among the hardest open questions in science and philosophy, and this page does not claim to have solved them. Everything below is offered as a way of framing the questions from the constraints this site takes seriously, and every statement about consciousness here is a proposal, never a result.
Anchored on the physics — Shape · Granularity · Scale · Observed
This is the one item where the anchoring is honestly partial: two roots genuinely engage and two are the named open residue. Selfhood's unity and the terminal “just knows” reduce onto Shape (nonseparability) and Granularity (the cost-floor) in the site's anchored-not-derived sense; qualia's magnitude and its measured signature do not, and that is stated plainly rather than papered over.
Shape (the frozen 13-D object; ×Stage · ⊕Rulebook · ⊗Actors): The unity of a single experienced field maps onto a structural fact about the frozen object: it is one connected object, and independence of any part is earned, never free (the in-page Nonseparability root). The internal geometry K₆ = SU(3)/T² is a connected coset, and the global state is not, in general, a product of separately specifiable parts — nonseparability is a Shape property. A “self” boundary is therefore a soft, earned sub-region or quotient of the one shape, not a fundamental partition posited by hand — exactly the shape the self-model and body-illusion evidence already gives the self. This constrains the form a self must have (built from shared structure); it is Shape-supported, not Shape-forced — no map carries connectedness to a felt point of view.
Granularity (cost-floor Δ₀ > 0; ℏ its measured residue): The chain of records — apparatus records system, retina records light, brain records retina — is a record of a record of a record, and it must terminate in something that is not a further record. That terminus is set by the cost-floor: a uniform positive minimum operational/action step Δ₀ > 0 (the in-page Granularity root, with Records and Distinguishability: separation is earned). The regress stops where a further distinction would cost more than Δ₀ admits — a terminal “just knows.” The floor is a Lorentz-scalar on cost, not length, applied across all 13 dimensions including time; its time-dependent character is congenial to the manifest presentness of experience — a definite present-moment record rather than a smeared block. ℏ is the floor's measured size, an anchor, never derived here.
Scale (absolute anchor MPl): Honest weak point. Felt quality has no anchored scale in the framework. The one absolute anchor is MPl, and every dimensionful claim must trace to it through a named bridge (RG flow, decoupling, threshold, compactification, FRG, or dissolution). Neural scales sit far below MPl-set structure, and no scale-bridge exists from the geometry to the magnitude of a subjective feel. By the site's own rule, a magnitude with no anchor and no bridge is not a claim: this root is SCALE-OPEN for qualia. Existence of a floor (ℏ) is anchored; the scale of felt quality is not.
Observed phenomenon: Named open residue. There is no third-person measured signature of felt quality — this is the hard problem itself. What is measured are the neural correlates of consciousness (NCC): integration, binding, global broadcast, recurrent processing. These are correlates, not anchors — they say which/when a state is conscious, never why there is feel; selfhood connects only to these. The nearest honest hook is structural, not phenomenal: the measurement formalism forces the existence of a terminus (Wigner's-friend and its extensions show the observer's first-person knowing cannot be dissolved without cost) — a falsifiable statement about the chain, not a signature of the feel.
Honest status. Exactly two of four anchors engage. Nonseparability/binding is Shape-anchored and the present-moment terminal knowing is Granularity-anchored (with ℏ the measured residue of Δ₀ > 0). Qualia's Scale and its Observed signature are the named open residue: no scale-bridge to felt magnitude, no third-person signature of feel. In the site's discipline — anchored is not derived; selected is not forced; dissolved is not solved; floor ≥ 1, forever — there is here no derivation of qualia from the geometry, no mechanism from connectedness or the cost-floor to felt experience, and no measured prediction that distinguishes this framing from any other. The roots constrain the form an answer must take — integrated not separable, tensed not block-like, a self-boundary earned not stipulated, a terminus that is not a further record — and that is all they do. This is a rigorous placement of the question, not a closure of it.
The hard problem, and the “easy” ones
David Chalmers drew the sharpest version of the distinction. The easy problems of consciousness are the ones that, however difficult in practice, are clearly problems of function: how a brain discriminates a stimulus, integrates information, focuses attention, accesses memory, controls behavior, and issues verbal reports about its own internal states. We can see, at least in principle, what a solution would look like — a mechanism that performs the function. The hard problem is different in kind: why is all that functioning accompanied by experience? Why is there felt quality rather than “all that data-processing, in the dark”? Even a complete functional story — every neuron, every computation, every report — seems to leave this residual question untouched. That residue is the same residue this root names.
Thomas Nagel fixed the target phrase in “What Is It Like to Be a Bat?” (1974): an organism is conscious just in case there is something it is like to be that organism — a subjective character of experience essentially tied to a point of view — and that first-person aspect does not obviously reduce to any third-person description, no matter how complete.
Joseph Levine named the shape of the difficulty as the explanatory gap (1983). We can grant, as a brute correlation, that a certain neural state is the state of feeling pain; what we cannot do is explain why that state feels the way it does rather than some other way, or feels like anything at all. In the paradigm cases of successful reduction — heat as molecular motion, lightning as electrical discharge, the gene as a stretch of DNA — once we have the microphysical story, the identity feels intelligible; there is no residual “but why that?” With consciousness, the reduction, even where the correlation is granted, does not close the intelligibility gap. Whether the gap is merely epistemic (a limit on our concepts) or ontological (a real gap in the world) is itself one of the disputed questions.
Mary, zombies, and the meta-problem
Two thought experiments sharpen the same intuition and are worth stating precisely, because they are frequently misremembered.
Frank Jackson's knowledge argument (1982): Mary is a brilliant scientist who has lived her whole life in a black-and-white room and learned, through black-and-white texts and screens, all the physical facts about color vision — every wavelength, every retinal and cortical process, every functional and behavioral fact. The claim is that when Mary leaves the room and sees red for the first time, she learns something new: what it is like to see red. If she had already known all the physical facts and yet learns a new fact, then not all facts are physical facts — so runs the argument. Physicalist replies are numerous and serious: perhaps Mary gains a new ability (to recognize, imagine, remember) rather than a new fact; perhaps she comes to know an old physical fact under a new, phenomenal mode of presentation. The argument is not a proof; it is a pump for the intuition that phenomenal knowledge is not exhausted by structural-functional knowledge.
The conceivability of zombies (Chalmers): a philosophical zombie is a being physically and functionally identical to a conscious person — same behavior, same reports, same neural dynamics — but with no inner experience, all dark inside. If such a being is even conceivable without contradiction, then experience is not logically entailed by the physical and functional facts, which is precisely what a hard-problem skeptic denies is possible for genuinely physical properties. The move from conceivability to metaphysical possibility is exactly where the argument is contested: many hold that the zombie only seems conceivable because our concepts of the physical and the phenomenal are disjoint, not because the world could really be that way. Again: a pump, not a proof.
Chalmers later turned the tables with the meta-problem of consciousness: set aside, for a moment, whether there really is a hard problem, and ask instead the straightforwardly functional question of why we are so disposed to think there is one — why physical systems like us produce reports and judgments about ineffability, inner light, and an explanatory gap. This is an “easy” problem in the technical sense (it is about the mechanisms of a behavior), and it is a live research program. Some illusionists (Keith Frankish, Daniel Dennett) argue that a good solution to the meta-problem would dissolve the hard problem — showing that the sense of an unbridgeable gap is a systematic artifact of how the brain models its own states, so there is a hard problem of the illusion but no further hard fact of phenomenal feel. Realists reply that explaining why we say there is feel is not the same as explaining feel. This site takes no side; it records that the dispute is unresolved and that the residue it names survives on the realist reading.
Selfhood: binding, unity, and the self as a model
The second half of “Knowing” is not qualia but ownership. Even granting that there is experience, why is it had by someone — structured as a single, present, first-person field rather than a scatter of unowned micro-experiences or an impersonal registering of facts?
Part of the puzzle is the binding problem and the unity of consciousness. The visual system processes color, motion, orientation and location in largely separate populations of neurons, yet you experience one moving red object, not four disjoint feature-streams. Across modalities the point is stronger still: sight, sound, touch, proprioception, mood and the flow of thought are presented together, as facets of a single momentary field of experience that seems to belong to one subject. How distributed processing yields a unified, apparently indivisible present is not settled. Candidate mechanisms — temporal synchrony of firing, large-scale recurrent integration, a shared workspace — are correlates of when binding occurs, not explanations of why integration should feel like one experience for one me.
Thomas Metzinger pushes this to its sharpest form in the self-model theory of subjectivity. On this view there is no thing that is the self; there is a self-model — a representation the system continuously constructs and runs — and the peculiarity is that the system cannot represent this model as a model. Because the modeling is transparent (we see through it to a seemingly given world and a seemingly given self, not to the representing process), the content “I” is experienced as a real, present, owned entity rather than as an ongoing construction. The technical term for this bare first-personal givenness — the “mineness” that attaches to experience prior to any thought about it — is ipseity. Metzinger's claim is deflationary about the substantial self but not about the phenomenon: the experience of being a self is entirely real and demands explanation, even if no little self is found behind it. Note carefully what this does and does not do. A self-model can explain the structure and revisability of the sense of self — why it has the character it has, why it can be perturbed — without thereby explaining why running such a model is accompanied by any felt point of view at all. The self-model theory is, in that respect, a candidate answer to the selfhood-structure question that leaves the hard problem exactly where it stood.
The measurement chain, and where knowing must terminate
These questions come to a head, in physics specifically, at quantum measurement — and this is the one place where the residue is not merely a philosopher's intuition but is forced by the formalism's own structure.
Von Neumann’s 1932 axiomatization separates two utterly different state changes: process 2, the smooth, deterministic, reversible Schrödinger evolution of an isolated system, and process 1, the abrupt, irreversible, probabilistic jump to one definite outcome on measurement, weighted by the Born rule. Unitary evolution alone, applied honestly to system-plus-apparatus, never produces a single definite pointer reading — it produces a superposition of readings — and decoherence, real and quantitative as it is, converts that superposition into a stable branch structure without selecting which branch is actual. (The full treatment — exactly what decoherence does and does not do, and the live menu of interpretations — is under the Observing root.)
Now push the description up the chain, as von Neumann did. The measured system disturbs the apparatus; the apparatus emits light; the light strikes the retina; the retina drives the optic nerve; the nerve drives the cortex. At every link you can, if you insist, describe the previous stage quantum-mechanically and treat the next as the “observer” that records it. This is the von Neumann chain, and its central lesson is the Heisenberg cut (also called the psychophysical cut): the boundary between the part of the world described as an evolving quantum system and the part simply taken to hold a definite record can be slid up and down the chain without changing any prediction. The formalism does not tell you where to put the cut; it only insists that a cut be placed somewhere, so that at least one stage counts as a definite outcome rather than an open superposition.
This is where the two hard problems and the physics converge. The chain is a sequence of records: the apparatus records the system, the retina records the light, the brain records the retina — a record of a record of a record. As long as each stage is itself just another physical system that could in principle be measured by the next, nothing in the sequence is a definite outcome; each is only a correlation waiting to be read. The regress has to stop. It stops at the point where the observer does not take a further measurement of the outcome but simply knows it — where a record is not read by yet another apparatus but is experienced. Fritz London and Edmond Bauer, in their 1939 essay on measurement, made this explicit: what distinguishes the observer from any link in the chain, they argued, is a “faculty of introspection” — the observer's immediate, first-person knowledge of their own state, which requires no further measurement and terminates the regress. Whatever one thinks of their conclusion, they had correctly located the terminus. That terminal first-person knowing — the point at which the chain grounds out in an experience that is had, not merely recorded — is exactly the residue this root names. It is where qualia (there is something it is like to know the outcome) and selfhood (the outcome is known by someone) enter the physics whether or not one wants them there.
Eugene Wigner dramatized the tension with the thought experiment now called Wigner's friend. A friend, inside a sealed laboratory, performs a measurement and observes a definite result. Wigner, outside, describes the whole laboratory — friend included — unitarily, and so assigns it a superposition of “friend saw up” and “friend saw down.” But the friend, from the inside, is certain a definite outcome occurred. Whose description is right, and when did the outcome become definite? Wigner's original response was that a conscious observer cannot be in a superposition of having-seen-different-things — that consciousness itself collapses the state — which would place the psychophysical cut at the boundary of mind. Most physicists today reject that particular resolution; recent extended-Wigner's-friend theorems (Frauchiger–Renner; Bong and colleagues) sharpen the puzzle into genuine no-go results about which combinations of assumptions — universal unitarity, observer-independence of facts, single outcomes, freedom of choice — can hold together. The upshot relevant here is modest but firm: no interpretation makes the observer's first-person knowing vanish as a distinct thing to account for. It is either given a special dynamical role, or relativized to a branch or an agent, or declared primitive — but never simply explained away.
A crucial caution: it does not follow from any of this that consciousness causes collapse, or that minds are needed to make the world definite. Decoherence, and the various interpretations above, show that the classical appearance of outcomes can be accounted for without invoking a mind, and the “consciousness causes collapse” reading is a minority view for good reasons. The genuine and interpretation-independent point is only structural: the chain of records must be grounded in some terminal that is not itself a further record, and the one such terminal we have direct acquaintance with is first-person experience. The physics forces the existence of a terminus; it does not tell us its nature, and it certainly does not license the claim that mind is doing the collapsing.
Scientific theories of consciousness: correlates, not solutions
There is a productive, fast-advancing science of consciousness — Integrated Information Theory (Tononi), Global Workspace (Baars; Dehaene), higher-order and recurrent-processing accounts, and the predictive-processing / free-energy framework (Friston) among them. Read exactly, these are searches for the neural correlates of consciousness: increasingly precise, genuinely testable — sometimes falsified — answers to which states are conscious and when a state becomes conscious. On the hard problem they are, by construction, correlational: they say which and when, never why there is feel. They leave the residue this root names standing — and most of them are candid about it.
How the constraints of this site reframe the question
This site is built around a small number of physical constraints, and it is worth being scrupulous about what they can and cannot contribute here. Two are relevant, and both frame the question rather than answer it.
The first is nonseparability: entangled systems do not in general have separately specifiable states, and a system’s independence from the rest of the world is at best approximate and earned — so a “self” boundary cannot be a sharp line stipulated by hand; it would have to be a soft, earned boundary emerging from shared structure, which is exactly the shape the self-model and body-illusion evidence already suggests the self has. The second is granularity: on this site fine-grained detail is fully resolved only near the present, so there is a definite present-moment record rather than an eternally smeared block — congenial to a phenomenon that is emphatically tensed and singular, always now. Framing devices, both; neither is a mechanism.
It must be said as plainly as possible what these resonances are worth. They tell us something about the form a physical account of experience would need to have — integrated rather than separable, tensed rather than block-like, with a self-boundary earned rather than stipulated. They supply constraints on the answer. They do not supply the answer. There is no step here from nonseparable geometry to felt redness, no derivation of ipseity from granularity, and no reason yet to think that satisfying these formal constraints is sufficient for there to be something it is like. The gap between “a system with this structure” and “a system for which there is experience” is exactly as open under these constraints as under any other.
Honest status
Open — and speculative in the strict sense. Let us be exact about the division of labor. What physics genuinely supplies is: the two-process structure of quantum measurement and the resulting measurement problem; decoherence as a real, mind-independent account of classical appearance that nonetheless does not by itself pick out a single outcome; the movable-but-necessary psychophysical cut; the theorems (Wigner's friend and its extensions) showing that the observer's first-person knowing cannot be dissolved without cost; and the structural fact that a chain of records must terminate in something that is not a further record. That much is standard physics, presented here without embellishment.
What remains open is everything that would make this a solution. There is no derivation of qualia from the geometry, no mechanism connecting nonseparability or granularity to felt experience, no account of why the terminal knowing is a feel rather than a bare registration, no derivation of selfhood or ipseity, and no measured prediction that would distinguish this framing from any other. The neuroscience supplies ever-better correlates but no bridge across the explanatory gap; the philosophy supplies sharper statements of the problem but no consensus resolution; and the physics supplies a terminus it cannot characterize. The constraints of this site reframe the demand and constrain the shape of an eventual answer, and that is all they do.
This status is a report, not a verdict. To say the root is not closed by physics alone is not to say it is unsolvable, and it is not to smuggle in dualism or mysticism. It is to record, honestly, that at present neither physics, nor neuroscience, nor philosophy has shown why there is something it is like to be the observer at the end of the chain, or why that observer is a “me.” The most this root claims is to have located the question precisely — at the terminal first-person knowing that the formalism forces and that no theory yet explains — and to have named, without overselling, the constraints an answer would have to respect.
“Observing” names the ability to collapse a wave function — the step where a system in a superposition of possibilities yields, on measurement, a single definite outcome. This is the deepest and oldest of the genuinely open questions in the foundations of physics: the measurement problem. Standard quantum mechanics is astonishingly accurate, yet it describes what happens during a measurement with two rules that do not fit together cleanly. Below we lay out exactly what is settled, what is not, and where — carefully and without overclaiming — this framework touches the question. The short version, kept honest throughout: the statistics of observation are on firm ground; the selection of one actual outcome, and any role an observer plays in it, is not.
Anchored on the physics — Shape · Granularity · Scale · Observed
Of every item on this speculative page, “observing” is the one whose anchoring reaches deepest onto the roots: the weight of each possible outcome is fixed by the framework's Born-rule root, and the definiteness of a record sits on the Granularity floor. What those roots do not hand you is the selection — which single outcome is the one actually observed. The weight is anchored; the selection is the named open residue. Anchored is not derived; the residue is real and it is not smuggled shut.
Shape (the frozen 13-D object; ×Stage · ⊕Rulebook · ⊗Actors): The Born weight is one of this page's in-page anchored roots: the probability of an outcome is the squared amplitude, p = 2, which the framework closes anchored on non-contextuality — its single paid axiom. Given that one posit, the Gleason and Busch theorems force the Tr(ρ E) form uniquely on the Hilbert space the object defines: the weight is not tunable and not from-nothing — it is anchored. The exponent is independently measured: Sorkin's third-order-interference parameter vanishes exactly when the rule is squared-amplitude, and triple-slit experiments find it consistent with zero — so p = 2 is an empirically pinned exponent, not a choice. Which variables carry the records — the einselected, redundantly-copied pointer states — is an Actors-layer (⊗) question, tied to the in-page Records and Distinguishability roots that fix what can count as a record at all.
Granularity (cost-floor Δ0 > 0; ℏ its measured residue): Which outcomes are distinguishable — and can therefore be recorded and separated — is set by the uniform positive cost-floor Δ0 > 0, with ℏ its measured size (an anchor, never derived here). The floor is a Lorentz-scalar on action/cost, not a length: only a transition to an orthogonal record costs, so a definite measured outcome is a present-moment grain. This ties observation to the in-page Granularity and Distinguishability roots — they fix what can be a record, and how finely.
Scale (absolute anchor MPl): A probability is dimensionless, so the Born weight needs no scale anchor — it is Scale-anchored only trivially, through ratios, and this is the weakest place the absolute anchor MPl touches the item. Where Scale does bite is the dynamics of any candidate selection process: decoherence timescales are dimensionful and set by ordinary physical scales, and a Penrose-type gravitational reduction would tie a collapse time to a mass-energy via t ~ ℏ/EG. Those are ordinary scale claims, not a magnitude the frozen shape forces.
Observed phenomenon: The strongest anchor on the page. The Born statistics are the most-tested prediction in physics — verified to extraordinary precision across every quantum experiment. Decoherence is quantitative and experimentally confirmed; einselection and redundant records (quantum Darwinism) are observed; definite single outcomes are what every apparatus shows. The open observable is the selection mechanism — and it has a genuinely falsifiable direction: objective-collapse models (GRW, CSL, Penrose OR) predict tiny departures from textbook quantum mechanics — faint spontaneous radiation, a mass-dependent loss of interference — and experiment has already excluded parts of the CSL and gravitational-collapse parameter space.
Honest status. The weight and the statistics of observation are anchored: the Born rule p = 2 rests on the framework's single paid non-contextuality axiom (and the exponent is independently measured), the distinguishability that lets a definite record exist rests on the cost-floor Δ0 > 0, and the Born statistics themselves are measured to extraordinary precision. What is not closed — the named open residue — is the selection: the passage from a weighted set of decohered branches to this one actual outcome, and whatever role an observer plays in it. No interpretation is asserted, and decoherence is not allowed to pose as the solution. In the site's discipline — anchored is not derived; selected is not forced; dissolved is not solved — the weight is anchored, the selection is open, and (uniquely among these items) that open leg already carries an experimentally-tested falsifiable signature.
Two dynamics that sit uneasily together
Quantum theory contains two distinct laws of change. Between measurements, a closed system evolves by the Schrödinger equation: smooth, deterministic, reversible, and linear. It preserves superpositions exactly — feed in a sum of possibilities and you get out a sum of possibilities, never a single one. John von Neumann, in his 1932 axiomatization, called this process 2. The second law is the projection postulate together with the Born rule: upon measurement the state is projected onto one eigenstate of the measured quantity, chosen stochastically, with probability given by the squared amplitude. Von Neumann called this process 1 — discontinuous, irreversible, and non-unitary. The two could hardly be more different in character: process 2 is continuous and keeps every branch alive; process 1 is a sudden jump that keeps exactly one.
The sharp way to see the difficulty is this: the unitary dynamics of quantum mechanics, applied honestly, never turns a superposition into a single outcome. A linear, reversible evolution takes a superposed system, entangles it with the apparatus and then with the environment, and produces an ever-larger superposition of “system-plus-apparatus-plus-world” states — a superposition of “pointer at A” and “pointer at B.” It cannot, by its own linearity, extinguish one term. So process 1 has to be added by hand. Whether that addition reflects a real physical event, a change in our knowledge, an artefact of a bad question, or something we have simply not yet derived is precisely what remains unsettled. Everything genuinely at issue in “observing” lives in the gap between process 2 and process 1.
The Born rule — what is settled is the statistics
The one part of measurement that is on very solid footing is the weighting. Outcomes occur with probability p = |amplitude|² — the modulus of the quantum amplitude, squared. The exponent is exactly 2, not 1 and not 3, and this is not an arbitrary choice; several independent lines of argument force it.
- Gleason's theorem (1957). If probabilities are assigned to measurement outcomes (rays / projectors) in a way that is non-contextual — the probability of an outcome does not depend on which other compatible measurements it is bundled with — then in any Hilbert space of dimension ≥ 3 the only consistent probability measure is the Born measure, p = Tr(ρ P). The squared-amplitude form is not assumed; it is deduced. The dimension-≥-3 condition matters: the qubit (dimension 2) is a known exception, patched by later results.
- Busch's POVM theorem (2003). Extending Gleason from sharp projective measurements to general positive-operator-valued measures (POVMs) recovers the Born rule even in dimension 2, and with weaker assumptions. This closes the qubit gap Gleason left open.
These derivations rest on different assumptions and none is uncontested, but together they make one thing clear: given non-contextuality, the p = 2 exponent is essentially forced. This is what this site's anchored Born-rule root records: p = 2 is the terminal form of the measure, with non-contextuality the single axiom that is paid for it. That is a claim about statistics — how often each outcome shows up over many trials. It says nothing about why, on a single trial, one particular outcome is the one you see. The statistics are settled; the selection is not. Keeping those two apart is the whole discipline of this page.
Decoherence — precisely what it does, and what it does not do
The most important physics developed since the 1970s for this question is decoherence, worked out by Zeh, Zurek, Joos, and others. It is real, quantitative, experimentally confirmed, and often misdescribed. Here is what it genuinely supplies.
A macroscopic object is never isolated; it is continuously entangled with an enormous environment — air molecules, thermal photons, its own internal degrees of freedom. When you write the system's state as a density matrix and trace out (average over) that unobserved environment, the off-diagonal terms — the interference terms between distinct outcomes — decay to zero extraordinarily fast. For anything of everyday scale the decoherence time is unimaginably short, many orders of magnitude below any laboratory timescale. This is why we never see a coffee cup in a superposition of two places: the interference that would reveal the superposition is destroyed almost instantly by entanglement with the surroundings. Decoherence explains the practical absence of macroscopic superposition, and it does so from the ordinary unitary dynamics alone, with no new law.
Decoherence supplies two more things. Einselection (environment-induced superselection, Zurek): the environment does not couple to arbitrary states equally; it dynamically picks out a robust, stable pointer basis — the states that survive monitoring and can serve as records. This is why measurement outcomes are recorded in position-like, classical-looking variables rather than in bizarre superposed bases. And quantum Darwinism (also Zurek): information about the pointer states is copied redundantly into many environmental fragments, so that many independent observers can read off the same value without disturbing it — which is what it means for a record to be objective.
Now the crucial caveat, and it is the crux of the whole problem. Decoherence does not select a single actual outcome. What tracing out the environment gives you is an improper mixture: a density matrix that is mathematically diagonal — it looks like a classical probability distribution over outcomes — but that arose from discarding information about a still-fully-entangled global pure state. It is not a proper mixture, in which the system really is in one definite state and we are merely ignorant of which. The distinction is not pedantic; it is the entire difficulty:
- A proper mixture describes genuine ignorance: the system is in one state, we don't know which, and the probabilities are epistemic. One outcome is real.
- An improper mixture is the reduced description of a larger entangled whole that is still in a superposition. No term has been removed. The full state still contains “pointer at A” and “pointer at B,” both, coherently — the interference is merely hidden in correlations with the environment we chose not to look at, and is in principle recoverable.
The two mixtures have identical local density matrices, so no local experiment can tell them apart — which is exactly why decoherence gives the appearance of collapse so convincingly. But mathematically they are different objects, and decoherence produces the improper one. It yields a set of dynamically independent, non-interfering branches; it does not yield one realised result and delete the rest. Bridging from “a set of non-interfering branches” to “exactly this one branch is what I observe” is the step decoherence cannot take. Anyone who says decoherence “solves” the measurement problem has quietly conflated the improper mixture with a proper one. Decoherence explains the disappearance of interference and the emergence of a stable classical record structure; it does not explain singular actuality.
The problem stated sharply: Maudlin's trilemma
Tim Maudlin gave the cleanest statement of the tension. Three natural-sounding claims cannot all be true:
- (1) Completeness. The quantum wavefunction is a complete description of a physical system — there is nothing more to say about its state than the wavefunction says.
- (2) Universal unitarity. The wavefunction always evolves by the linear, unitary Schrödinger dynamics — process 2 has no exceptions.
- (3) Definite outcomes. Measurements always have single, definite results.
If the wavefunction is complete and always evolves unitarily, then after a measurement the state is the full superposition of all outcomes — so results are not single: (3) fails. If it is complete and outcomes are single, the dynamics must sometimes be non-unitary — a genuine collapse: (2) fails. If the dynamics is always unitary and outcomes are single, the wavefunction cannot be the whole story — something more (hidden variables) fixes the result: (1) fails. Every serious interpretation of quantum mechanics is, at bottom, a choice about which of these three to give up. That is a helpful map: the interpretations are not a bewildering zoo but three families, sorted by which horn of the trilemma they abandon.
The interpretations — each fairly stated, all unsettled
There is no scientific consensus on the measurement problem, and this page asserts none. The leading positions are internally coherent; most are empirically indistinguishable from one another by any experiment yet performed. They differ in what they say is real and where they locate the transition to definiteness.
Orthodox / Copenhagen. Collapse is a primitive, unanalysed feature of measurement; a “cut” (Heisenberg's Schnitt) is drawn between the quantum system and the classical measuring apparatus, and process 1 acts at the cut. The theory is treated as a predictive instrument, not a picture of an observer-independent reality. This gives up completeness or unitarity depending on how literally the collapse is read, and famously declines to say where the cut must go.
Von Neumann–Wigner (“consciousness causes collapse”). Von Neumann noted that the cut can be pushed arbitrarily far up the chain — system, apparatus, photons, retina, optic nerve — without changing any prediction, and asked where it finally stops. Wigner, for a time, proposed it stops at the entry of the result into a conscious mind: consciousness is what actualizes one outcome. This is a historically real proposal by serious physicists, but it is controversial and not established; Wigner himself later moved away from it, partly because it seems to make physics depend on an undefined notion of mind, and decoherence removed much of its original motivation. On this speculative page it is noted as a proposal with a real pedigree, not as a result and not as this framework's claim.
Everett / many-worlds. Deny process 1 entirely. There is only unitary evolution; the universal wavefunction never collapses. Every term of the post-measurement superposition is equally real — the world “branches,” and each outcome is realised in its own decoherent branch, with observers splitting along with it. There is no selection to explain because nothing is deleted; the Born rule re-enters as a rule of self-location (how much an agent should expect to find themselves in a given branch), justified by the Deutsch–Wallace argument. It gives up (3), single outcomes, in the ordinary sense.
De Broglie–Bohm (pilot wave). Add hidden variables: particles always have definite positions, guided by the wavefunction through a deterministic, explicitly nonlocal guidance equation. The wavefunction never collapses; the appearance of collapse is the particle settling into one branch of a decohered wave. Outcomes are always single and the dynamics of the guided variables is deterministic — it gives up (1), completeness, since the position is real information beyond the wavefunction. It reproduces all standard quantum predictions exactly.
Objective-collapse theories (GRW, CSL, Penrose). Modify the dynamics itself so that collapse is a real, spontaneous physical process. In GRW (Ghirardi–Rimini–Weber) each particle undergoes rare spontaneous localizations at a fixed tiny rate; a single particle almost never localizes, but a macroscopic object of ~10²³ particles localizes essentially instantly, so micro-superpositions persist while macro-superpositions collapse. CSL (continuous spontaneous localization) smooths this into a continuous stochastic term added to the Schrödinger equation. Penrose's gravitational objective reduction (OR) proposes that superposing significantly different mass distributions is unstable because it superposes different spacetime geometries, and collapse occurs on a timescale set by the gravitational self-energy of the difference (roughly a time ~ ℏ/E_G). What sets these apart from every other interpretation is decisive: they make predictions that differ from textbook quantum mechanics — a tiny, ongoing violation of energy conservation, faint spontaneous radiation, and a mass-dependent loss of interference. They are empirically testable, and they are being tested, by matter-wave interferometry pushed to ever larger masses, by underground searches for the predicted spontaneous X-ray emission, and by precision force and heating measurements. Current experiments have already excluded parts of the CSL parameter space and some versions of gravitationally-induced collapse; this is the one corner of the measurement problem where laboratory data, not philosophy, is doing the deciding.
Relational quantum mechanics (Rovelli) drops the absolute, observer-independent state altogether — a measurement is one system coming to have a definite value relative to another, and there is no view from nowhere in which a single global outcome is stamped in. QBism (Fuchs, Mermin, Schack) reads the quantum state as an agent’s degrees of belief, the Born rule as a normative rule for coherent expectation, and “collapse” as ordinary Bayesian updating on new experience.
The key honest point: except for the objective-collapse family, these views are empirically near-indistinguishable with current means. Choosing among them is, for now, a matter of which metaphysical price one is willing to pay — nonlocality, branching worlds, added stochastic dynamics, observer-relativity, or a purely epistemic reading — not a matter that experiment has settled.
The observer's place: Wigner's friend and its extension
The role of the observer is sharpened by a pair of thought experiments. In Wigner's friend, a friend inside a sealed laboratory measures a quantum system and, from their standpoint, obtains a definite result. Wigner, outside, has not yet interacted with the lab; by the unitary dynamics he must describe the friend-plus-system as an entangled superposition of “friend saw A” and “friend saw B.” Two observers thus give different, apparently incompatible accounts of the same events — one a definite outcome, the other a live superposition — and standard theory does not adjudicate between them. The Frauchiger–Renner extension (2018) nests such observers and derives an outright contradiction from a small set of assumptions each interpretation would like to keep — roughly, that different agents' conclusions can be chained together, that measurements have single outcomes, and that quantum theory can be applied to other observers. The theorem shows the assumptions are jointly inconsistent: no interpretation escapes free; each must renounce one of them, and which one it renounces reveals its true commitments. These results do not favour a consciousness-based account; if anything they show that any story in which outcomes are observer-relative pays a steep price in consistency. They are diagnostic, not decisive — a precise way of feeling exactly where the problem bites.
Where this framework touches it — and where it stops
Now the constraint-first connection, stated with the same restraint used everywhere on this page. Two of this framework's roots bear directly on observation, and one result anchors part of it:
- The records root fixes what counts as a stable, persistent record — which is the same question decoherence answers dynamically through einselection and quantum Darwinism: what makes a pointer state robust and redundantly copied enough to be a record at all.
- The granularity root fixes how finely the present is resolved — the scale at which distinctions become fixed and a “now” is defined.
- The Born-rule result fixes the weight: p = 2 as the terminal, non-contextual measure, exactly the settled statistics discussed above.
Put together, these close the statistics of observation: given that records exist and are resolved at a definite grain, the framework fixes how the possible outcomes are weighted and what it takes for an outcome to be recorded. That is a real and non-trivial thing to have in hand. But it is emphatically not a solution to the measurement problem, and the framework does not pretend otherwise. What it does not close is the selection: the passage from an improper mixture of weighted branches to this one actual outcome, and whatever role — if any — an observer plays in that passage. That step is left as an open root, deliberately, in the same honest posture used for every other hard question on this page. The framework supplies the weight and the criterion for a record; it does not supply the mechanism that turns a superposition into a single observed result, and it does not smuggle one in under the cover of decoherence.
Honest status
Open. The Born weight is anchored (p = 2, forced by non-contextuality, with that non-contextuality the one paid axiom). The statistics of observation are on settled ground. The mechanism that turns a superposition into a single observed result — the selection of one outcome, and the observer's place in it — is not derived here, and no interpretation is asserted as established. Copenhagen, Everett, Bohm, objective collapse, relational QM, QBism, and the historical consciousness-causes-collapse proposal are all reported as live, competing, and (save for the testable objective-collapse family) empirically near-indistinguishable options — not as conclusions of this framework. On the speculative connection between observation and consciousness in particular: any such link here is a proposal, never a result. What physics genuinely supplies (decoherence, einselection, quantum Darwinism, the Born weight) is kept cleanly apart from what remains genuinely open (which single outcome, and why), and this page keeps it that way on purpose.
"Meaning" or "intentionality" is the philosopher's name for the puzzle of aboutness: why a thought, a word, or a brain state can be about something else — a memory, a distant object, an abstract idea, a person who no longer exists — rather than just being a physical pattern sitting there. The term traces to Franz Brentano, who in 1874 proposed that intentional inexistence — the directedness of mental states upon an object — is the "mark of the mental," the feature that distinguishes minds from mere matter. A rock does not point beyond itself; a belief does. Any account of consciousness eventually has to say something about why mental states point beyond themselves, and not just what they are made of. This page is the SPECULATIVE part of the site, so everything below is a proposal or a survey of open problems, never a claimed result.
Anchored on the physics — Shape · Granularity · Scale · Observed
Meaning reduces cleanly onto the deep roots only as far as correlation-with-a-record: a record's content is the set of admissible configurations on the frozen object that it co-varies with, read out through the Rulebook's maps, resolved to the cost-floor grain, and quantifiable as mutual information. That much is Shape- and Granularity-anchored and observable. The remaining step — from co-variation to genuine aboutness that can be false — is the named open residue, and no root closes it. Anchored is not derived.
Shape (the frozen 13-D object; ×Stage · ⊕Rulebook · ⊗Actors): Meaning-as-correlation anchors to the readout structure of the Shape. A record is a state of the Actor layer (a bundle/operator datum, ⊗), and its content is fixed by which admissible configurations on the frozen branch it correlates with — a relation carried by the Rulebook's readout maps (⊕), not by the metric carrier (×) alone. This is the same object-completeness discipline that runs the physics: content is a property of the full three-layer object on the frozen branch (K₆ = SU(3)/U(1)²), never of geometry in isolation. It matches the site's standing intuition that no part is separable by default — aboutness, if it is physical at all, is a relational signature of how a state is woven into the whole, not an ingredient painted on top. But the readout map here is Shape-supported, not Shape-forced: correlation fixes at most a disjunction of causes, and nothing in the map singles out one determinate referent.
Granularity (cost-floor Δ₀ > 0; ℏ its measured residue): Content is resolved only to the minimum distinguishable operational cost Δ₀ > 0 (whose measured size is ℏ). A record means X only up to a distinguishable grain: two candidate contents separated by less than the floor are one content, because separation must be earned (Distinguishability). This bounds the classic disjunction problem — the sprawling disjunction of proximal and distal causes a state co-varies with is finite and coarse, not a continuum — but it does not eliminate it. A finite grain still contains many distinct worldly causes inside one cell, so "cat-or-fox-in-fog" survives as a legitimate coarse-grained content. Granularity therefore forbids the fabricated infinite-precision content, but does not deliver the single determinate referent that misrepresentation requires.
Scale (absolute anchor M_Pl): Weak, and honestly so. Aboutness is structural and dimensionless — it is a pattern of correlation and distinguishable grain, not a magnitude — so it sits in the unit-gauge-invariant sector and touches the absolute anchor M_Pl only indirectly, through ℏ setting the size of the cost-floor grain at which contents become distinct. There is no meaningful dimensionful "amount of meaning" to derive or to pay; the scale root supplies no forcing here beyond fixing the grain. This is a genuine non-anchor, stated plainly.
Observed phenomenon: The measured hook is the mutual information I(X;Y) between a record and the world — a real, physically instantiated, thermodynamically costed quantity (erasing one bit dissipates at least k_B T ln 2). This is genuine and within reach of physics. The named open residue is that there is no observable that separates carrying mutual information about X from being determinately about X, and none that registers misrepresentation — the possibility of a record being about X falsely. A falsifiable signature would have to distinguish, from records alone, error from correct indication of a disjunction; no such measurement is on the table. Correlation, the finite grain, and the record's energetic cost are all measurable; determinate, error-permitting content is not.
Honest status. Genuinely anchored: correlation-with-a-record is Shape-supported (readout structure over the frozen three-layer object) and Granularity-bounded (content resolved to Δ₀, disjunction made finite but not singular), with a real measured hook (mutual information, and its thermodynamic cost). Named open residue: the step from mutual information to genuine semantic aboutness — and to the possibility of misrepresentation — has no observable and no derivation; the Scale root does not anchor it; the Shape readout is supported, not forced. This is where the reduction stops, not where it succeeds. Anchored is not derived; selected is not forced; dissolved is not solved; floor ≥ 1, forever.
Correlation is cheap; meaning is more
Start with the deflationary temptation. A neuron in visual cortex fires when a cat is in view; the firing correlates with cats. Isn't that already aboutness? No — and seeing exactly why is the whole difficulty. Correlation is metaphysically cheap. The rings of a felled tree correlate with the years it lived; a tree's height correlates with its age; the mercury in a thermometer correlates with temperature; a footprint correlates with the foot. Physics hands us correlations of this kind essentially for free, because the world is full of lawful, causal, statistical dependencies. But we do not ordinarily say the tree ring is about the drought, or that it can be mistaken about the drought. A neuron that fires with cats also fires, on a foggy night, with a small dog, a fox, or a shadow. If its content were just "whatever causes it to fire," then when it fires at the fox it is not misrepresenting a cat — it is correctly representing the disjunction cat-or-dog-or-fox-or-shadow. That is the crux: correlation gives co-variation, but meaning demands a determinate content that the state can get wrong.
The disjunction problem and the demand for misrepresentation
This is the disjunction problem, and it is the sharpest tool in the philosophy of content. Any physical state co-varies not with one thing but with a sprawling disjunction of causes: proximal stimuli, distal objects, lighting conditions, the states of intervening media. Pure causal covariation cannot single out "cat" from "cat-or-fox-in-fog" as the content, because both are perfectly correlated with the firing in the actual history of that neuron. And a theory that cannot separate them cannot allow misrepresentation — the possibility of being wrong. A genuine theory of meaning must permit error, because representation that cannot fail is not representation at all; it is just reliable indication. A thermometer stuck at 20°C is not "lying"; it has simply ceased to track. So the target any naturalistic account must hit is precise: explain how a physical state acquires a content specific enough that it can be false.
The symbol grounding problem
A parallel difficulty appears the moment you try to build meaning out of symbols alone. Stevan Harnad's symbol grounding problem (1990) points out that a formal system whose symbols are defined only in terms of other symbols never acquires content — it is like trying to learn Chinese from a Chinese–Chinese dictionary, forever chasing definitions in a closed loop that never touches the world. Manipulating tokens by their shapes according to syntactic rules, however elaborate, is what John Searle's Chinese Room (1980) dramatized: syntax is not sufficient for semantics. Harnad's proposed remedy is grounding: at least some symbols must be tied, through sensorimotor interaction, to the categories of things in the world that a body can perceive and act upon. The word "cat" means cat, on this view, only because it is anchored in a history of seeing, tracking, and responding to cats. Grounding does not by itself solve intentionality, but it reframes the question usefully: content is not a property a system has in isolation: it is earned through the loop that connects an internal state to the world it is embedded in and acts within.
Information versus meaning
It helps to be careful about the word "information," because physics has an extremely successful theory of it that is not a theory of meaning. Claude Shannon's information theory (1948) is syntactic: it quantifies correlation and reduction of uncertainty — entropy H = −Σ p log p, mutual information I(X;Y), channel capacity — entirely in terms of the statistics of signals, with no reference whatever to what those signals are about. Shannon himself was explicit that "the semantic aspects of communication are irrelevant to the engineering problem." Mutual information tells you that two variables co-vary and by how much; it is silent on which of them, if either, represents the other, on what the representation says, or on whether it is true. This is the same gap in information-theoretic dress. What we want is semantic information: content with truth-conditions, a state of affairs the signal says obtains, such that the world can fail to match it.
- Syntactic (Shannon): correlation, mutual information, channel capacity, error rates. Physics supplies this readily — any lawful dependency carries mutual information.
- Semantic: content, reference, truth-conditions, and the possibility of misrepresentation. Nothing in the Shannon quantities, on their own, fixes these.
The relation between information and thermodynamics — Landauer's principle, that erasing one bit dissipates at least k_B T ln 2 of heat, and Bennett's resolution of Maxwell's demon — shows that Shannon information is a genuine physical quantity with real energetic cost. This is a deep and settled result. But it concerns the syntactic bit, the correlation. It does not bridge to semantics. A physically embodied, thermodynamically costly bit is still not yet a bit that is about anything in the sense that admits error.
Naturalizing content: the main research programs
Philosophers have not been idle in the face of this gap, and honesty requires reporting the serious attempts, none of which commands consensus. Three families of theory try to naturalize content — to say, in respectable non-mental terms, what makes a state be about X.
The serious programs are decades deep and worth naming: causal / informational theories (Dretske; Fodor’s asymmetric-dependence proposal, on which “cat” means cat rather than cat-or-fox because the fox-caused tokenings depend on the cat–“cat” law and not vice versa) and teleosemantics (Millikan; Papineau), which grounds content in what a mechanism was selected to track and turns misrepresentation into malfunction. Both are live, unsettled research lines; neither commands consensus, and each is still suspected of quietly importing the aboutness it set out to explain.
What all these programs share is a target and a difficulty. The target: a determinate content that permits misrepresentation. The difficulty: doing so in terms a physicist would accept, without smuggling in the very aboutness one set out to explain.
Where this framework stands
In the picture developed on this site, no piece of reality is allowed to be treated as fully separate from the rest by default — independence has to be earned by the actual structure connecting a part to everything else, not assumed for convenience. That constraint is suggestive here. If a state's "content" is really a signature of how it is woven into the wider structure it sits inside, then aboutness might be a relational fact — a matter of a state's place in a web of dependencies — rather than an extra non-physical ingredient painted on top. This resonates with the grounding and teleosemantic intuitions above: meaning as something earned through connection and use, not possessed in isolation. Separately, the finest grain of detail in this picture is treated as sharpest at the present and coarser toward the Big Bang, which would bear on how a meaning-bearing state relates to its past causes and future referents — a representation reaches "backward" to what caused it and "forward" to what it anticipates, and any physics of that reach must respect this asymmetry. Neither observation, it must be stressed, has been turned into an actual account of intentionality.
Put plainly, the framework can, at most, reduce "meaning" to correlation-with-a-record: a physical state that reliably co-varies with, and retains a trace of, some feature of the world. And then it must name the residue honestly. The step from "this record co-varies with X" to "this record is ABOUT X — determinately, and such that it could be about X falsely" is precisely the step that the disjunction problem, the misrepresentation requirement, and the symbol grounding problem all identify as hard. That step is not closed by the physics offered here. Supplying the correlation, the mutual information, even the thermodynamic cost of the record, is genuinely within reach of physics. Supplying determinate, error-permitting content is not something this framework has shown how to do.
Honest status: open
Open. This is speculative philosophical framing, not a physics result. There is no derivation, no model, and no candidate calculation connecting the geometric framework to meaning or intentionality. What physics genuinely supplies is correlation, mutual information, and lawful causal dependence — the syntactic layer — together with the real thermodynamic accounting of physical records. What remains open is everything that separates that layer from semantics: determinate content, reference, truth-conditions, and the possibility of being wrong. The serious naturalistic programs — causal-covariational, asymmetric-dependence, teleosemantic, predictive-processing — are reported here as live, unsettled attempts, not as solutions this framework endorses or completes. The root is named as a place where the reduction stops, not as a place where it succeeds — a status, not a claim of unsolvability. It marks a possible direction of thought, nothing more, and it stays open.
"Agency / free will" asks whether a mind's choices are genuinely its own, or just the playing-out of physics that was already fixed in advance. In everyday language: when you decide something, is there real room for that decision, or is it an illusion sitting on top of a clockwork chain of causes? This is one of the oldest problems in philosophy, and modern physics sharpens it rather than dissolving it. Below we lay out, as carefully as we can, what the physics genuinely supplies, what the standard philosophical arguments say, and where this framework does — and does not — have anything to add. The honest bottom line, stated up front, is that this root is open: nothing here solves free will.
Anchored on the physics — Shape · Granularity · Scale · Observed
Read agency as selection among available future record-paths, and the concept stops floating free: the choice set is the admissible-transition structure of the frozen object, the paths become countable only above the cost-floor, and the act of selection is the same unitary-plus-Born dynamics the physics already carries. This is an anchoring, not a derivation — the notion reduces onto the deep roots, but only a high-level, compatibilist agency reaches them; libertarian freedom does not, and is named below as the open residue.
Shape (the frozen 13-D object; ×Stage · ⊕Rulebook · ⊗Actors): The set of futures actually open to a system is not free — it is fixed by the Rulebook layer (⊕), the admissibility rulebook that decides which transitions and configurations the frozen branch permits at all. Agency-as-path-selection therefore anchors to the ⊕ layer: what can be chosen is exactly what the geometry admits, read over the whole three-layer object (× Stage · ⊕ Rulebook · ⊗ Actors), never the metric carrier alone. This is Shape-supported, not Shape-forced — no map takes the frozen branch to a specific decision; it identifies which layer owns the choice set, which is a genuine anchor and not a closure.
Granularity (cost-floor Δ0 > 0; ℏ its measured residue): Two candidate paths are selectable only if they are distinguishable, and distinguishability is earned: the Uniform Operational Cell Law posits a uniform positive floor Δ0 > 0 on distinguishable operational (action) cost, with ℏ its measured residue. Paths closer than Δ0 are not separate options; the floor discretizes the choice set into a finite, admissible class, and every selection is a costly step that must be paid — gliding through overlapping, non-orthogonal states is free, only transitions to distinguishable records cost. So the choice set is finite and charged, not a continuum: a real anchor. Honest limit — the floor is a floor on a Lorentz-scalar cost, not a length, and its existence is a declared posit while its size (ℏ) is measured; anchored is not derived.
Scale (absolute anchor MPl): The dynamics that carries out the selection is unitary evolution plus Born-rule (p = 2, |amplitude|²) weighting; magnitude enters only through the action floor ℏ and, for any dimensionful bridge, the absolute anchor MPl (measured, never derived). This is where the anchor is weakest and must be stated plainly: at warm, wet, electrically noisy neural scales, decoherence destroys superpositions astronomically faster than neural firing, so the brain is effectively classical and quantum indeterminacy is not a hidden source of freedom. Scale supplies lawful dynamics and a discreteness floor; it supplies no scale-anchored magnitude that would make micro-indeterminacy reach up into a decision. Determinism-or-randomness both remain, and neither delivers control.
Observed phenomenon: The measured hook is narrow but real — record-selection demonstrably happens: definite outcomes occur, single records are written, the choice set collapses to one actualized path. That much is anchored to what is observed. The open residue has no observable: there is no measurement that distinguishes a "free" selection from a determined-or-random one, and none of felt authorship or "could-have-done-otherwise." No falsifiable signature separates libertarian freedom from lawful selection, so the framework manufactures no anchor there.
Honest status. Open. What genuinely anchors is agency-as-constrained-path-selection: the choice set is the Rulebook's admissible-transition structure (Shape-supported), it is discretized and paid by the cost-floor Δ0 > 0 (Granularity-anchored, floor ≥ 1), and its execution is lawful unitary-plus-Born dynamics whose only observed hook is that definite outcomes occur. That reaches, at most, a compatibilist "freedom worth wanting" — acting from one's own records and reasons, uncompelled — a real, graded property of certain physical record-processing loops, consistent with lawful dynamics and claiming no power to suspend the microphysics. The open residue is libertarian freedom: it is neither derived nor dissolved, has no measured signature, and is not touched by the consequence or exclusion arguments here. Anchored is not derived; selected is not forced; dissolved is not solved. No calculation, no matched observable, no closure — this stays labeled open.
The core tension: determinism or randomness, and neither is control
Start with the microphysics, because that is where the pressure comes from. At the fundamental level our best theories come in two flavours, and both seem to leave "free control" homeless.
The first flavour is deterministic. Classical mechanics is deterministic: given the exact positions and momenta of every particle now, together with the force laws, the entire future is fixed. Nothing is left to decide; the trajectory is already written. Quantum mechanics, in its unitary part — the smooth evolution of the wavefunction governed by the Schrödinger equation — is also deterministic. The state vector evolves in a completely fixed, reversible way. If unitary evolution were the whole story (as in the Everett / many-worlds reading), the global quantum state marches forward with no genuine branching-into-one; it merely differentiates into branches.
The second flavour is irreducibly random. This is the quantum measurement problem. When a measurement-like interaction occurs, standard textbook quantum mechanics says a single definite outcome appears with a probability given by the Born rule, |amplitude|² — and, on the orthodox reading, which outcome occurs is not fixed by anything at all. It is a fundamental chance. Objective-collapse theories such as GRW (Ghirardi–Rimini–Weber) make this randomness a real physical process; Bohmian mechanics restores determinism by adding hidden particle positions guided by the wavefunction; Everett removes the randomness by keeping all branches. But on the most common reading there is genuine, lawless chance at the moment of outcome.
Here is the crux, and it is worth stating bluntly: neither option obviously delivers freedom. If the future is determined, my choice was fixed before I was born and I did not author it. If the future is random, my choice is a dice-roll and I still did not author it — a quantum coin flip is not a free choice. Randomness is not agency. Swapping "the laws made me do it" for "a random number made me do it" does not hand control back to the agent; it just relocates the lack of control. Any serious account of agency has to thread between these two, and it is not obvious there is a gap to thread through.
The consequence argument: why determinism seems to threaten freedom
The philosopher Peter van Inwagen crystallised the deterministic worry into what is called the consequence argument. Informally: if determinism is true, then the state of the world in the distant past, together with the laws of nature, entails everything that happens today, including what I "choose." But I have no control over the distant past — it happened before I existed — and I have no control over the laws of nature. And if I have no control over some facts (past + laws), and no control over the fact that those facts entail my present actions, then I have no control over my present actions either. Control cannot be conjured out of two ingredients neither of which I control. On this reading, determinism and genuine "could-have-done-otherwise" freedom are simply incompatible. This is the position called incompatibilism, and its libertarian wing concludes that if we are free, determinism must be false.
Compatibilism: the freedom worth wanting
Many philosophers reject the demand that freedom require breaking the causal chain. The compatibilist tradition — running from David Hume through Harry Frankfurt to Daniel Dennett — argues that the freedom worth caring about was never the metaphysical power to violate physics. It is something more mundane and more useful: acting from one's own reasons, values and desires, without external compulsion or coercion.
On this view, a free action is one that flows from you — from your deliberation, your character, your endorsed motives — rather than from a gun to your head, a brain tumour, an addiction you disavow, or a hypnotist. Frankfurt sharpened this with the idea of higher-order desires: a person acts freely when they act on a first-order desire that they also, at the reflective second order, want to be moved by. The addict who wishes she were not addicted is unfree with respect to the craving; the person acting on motives they endorse on reflection is free. Determinism is beside the point here, because the distinction between "compelled" and "acting on your own endorsed reasons" is a distinction that survives even in a fully deterministic world. Dennett called this "the free will worth wanting" — freedom as a real, graded, biologically and socially evolved capacity for self-governance, not a magical exemption from causation.
Does quantum indeterminacy rescue freedom? Mostly, no.
It is tempting to hope that quantum indeterminacy re-opens the future and thereby makes room for free will. The hope usually runs: unitary determinism was the problem, quantum measurement injects genuine openness, so the future is not fixed, so I am free. There are two reasons to be very cautious about this move.
First, the philosophical reason already given: indeterminism buys openness but not control. If my decision hinges on a genuinely random microphysical event, the outcome is not thereby mine — it is chance. Libertarians who want quantum randomness to do real work face the "luck objection": an action that traces back to an undetermined event looks lucky, not authored. Making the future unfixed is necessary for one notion of freedom but nowhere near sufficient for agency.
Second, the physical reason: even if quantum indeterminacy could help, it is very doubtful that it reaches up to the scale of neural decision-making. The brain is warm, wet, and electrically noisy. Quantum superpositions in such an environment are destroyed extraordinarily fast by decoherence — the rapid, effectively irreversible leakage of phase information into the surrounding degrees of freedom (water molecules, ions, the electromagnetic bath). Max Tegmark estimated decoherence times for candidate neural quantum states (such as ion positions in firing neurons or proposed microtubule coherences) and found them to be on the order of 10⁻¹³ to 10⁻²⁰ seconds — astronomically shorter than the roughly 10⁻³ to 10⁻¹ second timescales of neural firing and cognition. If that estimate is even roughly right, the brain is effectively classical for the purposes of computation and decision: whatever quantum coherence exists is washed out long before it could steer a thought. (Proponents of quantum-mind proposals, such as the Penrose–Hameroff "orchestrated objective reduction" idea, dispute the estimates; the mainstream assessment remains skeptical, and in any case the control problem would survive even if coherence did.)
So the standard verdict is: quantum mechanics does not obviously come to the rescue. It changes determinism into chance, and chance is not choice; and at the relevant scale the quantum effects are most likely gone anyway.
Mental causation and the exclusion argument
Even setting the micro-level aside, there is a pointed challenge at the level of minds causing things. Suppose your decision to raise your arm is a mental event, and it causes the physical arm-raising. The philosopher Jaegwon Kim posed the exclusion argument (also called the causal exclusion or overdetermination problem):
- Assume causal closure of the physical: every physical event that has a cause has a sufficient physical cause. This is a working assumption of physics — there are no gaps in the physical story that need a non-physical filler.
- Assume the mental is not simply identical to the physical (it "supervenes on" but is not reducible to brain states).
- Then the arm-raising already has a complete physical cause (the neural events). If the mental event also causes it, the arm-raising is overdetermined — caused twice over — which is implausible as a general rule. So either the mental event does no real causal work (it is epiphenomenal), or it does its work only by being the physical event (reduction).
The exclusion argument is why "downward" or "mental" causation is genuinely hard to make sense of as something over and above the physical dynamics. If every physical event already has a sufficient physical cause, what causal job is left for an irreducible mental agency to perform? Answers exist — some deny strict closure, some embrace a subtle reductive identity, some argue that higher-level causal patterns are real without being extra forces — but none is uncontested, and this is exactly the seam where claims of "the mind pushing the atoms around" tend to tear.
How this framework's ideas connect — and where they stop
Now to this framework specifically, keeping firmly in mind that everything in this speculative section is a proposal, not a result.
Two of the framework's habits of thought bear on the question. First, nothing is treated as fully independent unless that independence is earned: a decision-making system — a brain, an agent — is embedded in and correlated with the rest of the world, not floating free of it. Second, how finely the world is resolved into distinct, decidable facts is not fixed once and for all: on this picture the "grain" of settled fact can be sharp near the present and coarser as you move away from it, so that not every fine detail of the future is already a definite matter of fact. What the physics itself supplies is only the dynamics: unitary evolution of the state plus Born-rule selection of outcomes (|amplitude|²). Those are lawful, and they are the same two flavours — deterministic evolution and chance selection — that we already saw fail, on their own, to deliver control.
So where could "agency" even live in such a picture? At best it would be a high-level, compatibilist notion, not a new fundamental force. The natural reading is an agent as a record-processing loop: a physical system that maintains a model of itself and its environment, registers stable records of the past, weighs available paths against its own goals and values, and selects among the options actually left open to it. This resonates with several standard research programmes — Karl Friston's free-energy principle (agents as systems that minimise prediction error and act to keep themselves in expected states), the broader predictive-processing view of the brain, and information-theoretic accounts of cost and irreversibility (Landauer's principle: erasing one bit dissipates at least k_B T ln 2 of energy, tying computation to thermodynamics). On such a reading, "freedom" is exactly the compatibilist article: acting from one's own reasons and records, uncompelled — a real, graded property of certain physical loops, entirely consistent with lawful dynamics, and making no claim to suspend or steer the microphysics by fiat.
Could the "grain of settled fact" idea do more than that — could the openness of a choice be partly a matter of how much fine-grained detail about the future is even fixed or decidable at the moment of choosing, rather than a matter of breaking any law? Perhaps. It is a suggestive direction: it would locate whatever "room" a choice has in the settledness of facts, not in a violation of dynamics, and it would sidestep the luck objection by not resting agency on a raw dice-roll. But it is a plausible-sounding connection, not a worked derivation. It does not by itself answer the consequence argument, escape the exclusion argument, or show that the relevant openness is anything more than our ignorance re-described. Those are exactly the hurdles any honest attempt would have to clear, and this framework has not cleared them.
Honest status
Open. There is no calculation here, no matched observable, and no settled result connecting this framework to agency or free will. The physics genuinely supplies the dynamics (unitary evolution plus Born-rule selection) and a picture in which agents are embedded, correlated, record-keeping subsystems. Everything beyond that — whether "agency" is best understood as the compatibilist "freedom worth wanting," whether the coarsening of settled fact adds anything real, whether any of this touches the metaphysics that the consequence and exclusion arguments target — remains genuinely unsettled. This is a status, not a claim of unsolvability, and equally not a claim of solution. It should not be read as a finding, a proof, or a resolution of the free-will question. It is an interesting direction suggested by the underlying ideas, nothing more, and it stays labeled open until (and unless) real work changes that. The page does not claim to have solved free will, because it has not.
This item asks whether value or salience — why some possibilities feel like they matter more than others, why anything registers as good or bad, worth pursuing or worth avoiding — could ever be given a physical meaning inside a reconstructed universe, rather than being treated as a brute fact bolted on from outside. It is grouped with the consciousness questions because value and salience are usually discussed as features of an experiencing subject, not of equations. Of all the items in this speculative section, this is arguably the deepest root: it is where a purely third-person, physical description looks least likely to be, even in principle, the whole story.
Anchored on the physics — Shape · Granularity · Scale · Observed
Read in the framework's own terms, salience reduces onto the roots as a question about which record-distinctions a persisting loop pays to keep making, and what fixes the price of a distinction at all. That price is not free: the framework already carries a uniform positive cost-floor, and it turns out to be the tightest anchor value has — a natural currency for mattering. This is an anchoring, not a derivation: it reaches felt mattering's mechanism, and it names honestly where it stops.
Shape (the frozen 13-D object; ×Stage · ⊕Rulebook · ⊗Actors): Salience is the loop weighting some record-distinctions over the vast background it lets blur. Which distinctions exist to be weighted is Shape data: the ⊗Actors layer (bundles, operators, connections, readout maps) fixes what can be resolved off the frozen branch, and the ⊕Rulebook layer fixes which configurations are admissible to distinguish at all. So Shape supplies the alphabet of distinctions salience selects from. It is Shape-supported, not Shape-forced: nothing in the frozen object fixes which subset a given loop weights — that is a readout/dynamics obligation, and a static geometry does not by itself produce a weighting.
Granularity (cost-floor Δ₀ > 0; ℏ its measured residue): This is the tightest anchor. Granularity's root posit is a uniform positive minimum operational/action step — the cost-floor Δ₀ > 0 (the Uniform Operational Cell Law), with ℏ as its measured residue, a value never derived here. A distinction is not free: gliding through overlapping (non-distinguishable) states costs nothing, while committing to a genuinely distinguishable record costs at least Δ₀. This gives salience an exact, non-arbitrary currency: a distinction is salient precisely when it is worth paying at least Δ₀ to draw and to keep — and because a bounded loop cannot pay for unbounded fine grain, most possible distinctions must be left to blur. Value-as-cost-weighting is thereby anchored on the finite-cost discipline itself. Honest limit: this is the mechanism of weighting (a currency — supported, not forced), not a proof that weighting is mattering; the floor charges the distinction, it does not feel it.
Scale (absolute anchor Mₚₗ): The anchor here is weak, and honesty requires saying so. Salience is a dimensionless weighting — a ranking of which distinctions a loop favors — and by the dimensionless-first doctrine such ratios never touch the absolute-scale anchor Mₚₗ and need no scale bridge. Mₚₗ enters only indirectly, as the anchor the cost-floor's magnitude lives on (the floor is a Lorentz-scalar action, so ℏ is a measured-anchor under Scale). No dimensionful magnitude of "how much this matters" is a well-posed, scale-forced observable. The Scale root supports the currency's units and forces nothing about value.
Observed phenomenon: The measured hook is real but partial. Reward-prediction-error signals in midbrain dopamine neurons, and salience-network activity in the anterior insula and dorsal anterior cingulate, are genuinely measured — they track which possibilities a system weights, learns from, and acts on, and they map cleanly onto cost-weighting. But they are correlates of the tracking of value, not of its being felt. The named open residue has, on present understanding, no observable: why cost-weighting is felt as mattering — the is/ought, normative gap — produces no measured signature. A weak falsifiable edge does exist for the mechanism half: if a loop demonstrably weighted and acted on distinctions costing far below any Δ₀, the cost-floor currency for salience would fail.
Honest status. Salience-as-cost-weighting is Granularity-anchored — the cost-floor Δ₀ supplies a principled currency for which distinctions a loop is willing to pay to keep — and Shape-supported (the Actors and Rulebook layers fix the distinctions available to weight); the Scale anchor is weak, touching only the currency's units. What none of this delivers, and what the framework does not claim, is the crossing from weighting to felt mattering: that is the named open residue, sitting exactly on the is/ought divide, with no observable on offer. Anchored is not derived; selected is not forced; dissolved is not solved. This is a rigorous anchoring of the mechanism to the roots, and an explicit flag that the felt-value residue is open — floor ≥ 1, forever.
The is/ought gap: why value looks additional to the physics
The oldest and sharpest statement of the problem is David Hume's, from the Treatise of Human Nature (1739). Hume observed that writers move imperceptibly from claims joined by "is" and "is not" to claims joined by "ought" and "ought not," and that no purely logical step licenses the crossing. You cannot derive a conclusion about what ought to be, or what matters, from premises that state only what is, using deductive logic alone — an "ought" cannot appear in a conclusion if it appears in none of the premises. This is now called the is/ought gap, and its close cousin is G. E. Moore's "open question argument": for any proposed natural property N, it always remains a coherent, open question to ask "but is N actually good?" — which suggests that "good" is not simply identical to N.
The force of this for a physics reconstruction is direct. Suppose the framework, or any physical theory, hands you a complete description: the full state, the dynamics, every trajectory, every record. That description is entirely in the "is" register. Nowhere in it does an "ought" or a "matters" logically appear. Normativity — the fact that some states count as better than others for a subject — looks like something additional to the physical description, not a theorem of it. This is not a claim that value is supernatural. It is the more modest and more stubborn observation that no amount of descriptive fact, by itself, entails an evaluative fact. Any theory that seems to bridge the gap has, on inspection, usually smuggled a value in through a premise.
It is worth being honest that the is/ought gap is contested. Some naturalist philosophers argue that once you fix what an organism is — its needs, its flourishing conditions — value claims follow as facts about that kind of thing (a broadly Aristotelian or "Cornell realist" move). Others hold that evaluative facts simply are a special class of natural facts. The framework does not need to adjudicate this. It only needs to record, honestly, that the crossing is not free: if value is to be physical, that has to be shown, not assumed, and the burden of proof sits squarely on the side that claims the gap can be closed.
What neuroscience genuinely supplies: salience as mechanism
Neuroscience has a great deal to say about salience — but it is crucial to read it as an account of mechanism, of how a brain comes to treat some stimuli as important, and not as an account of why importance is felt at all. The mechanistic story is real, well-evidenced, and should be stated in its own right before its limits are drawn.
- Reward and reward-prediction error. Wolfram Schultz and colleagues, recording from midbrain dopamine neurons in the early-to-mid 1990s, found that these neurons do not simply signal reward. They signal the difference between reward received and reward expected — a reward-prediction error. A dopamine burst fires to an unexpected reward; it shifts to the earliest reliable cue that predicts the reward; and it dips below baseline when an expected reward fails to arrive. This maps strikingly onto the "temporal-difference" error term in reinforcement learning, and it is one of the cleanest bridges between a formal learning theory and a measured neural signal.
- The salience network. Functional imaging identifies a "salience network" — anchored in the anterior insula and dorsal anterior cingulate cortex — that appears to detect behaviourally relevant events and help switch the brain between internally-directed and externally-directed modes. Attention itself, whether captured bottom-up by a sudden stimulus or directed top-down by a goal, is the mechanism by which some possibilities are elevated over the enormous background of those the system ignores.
All of this explains, with real predictive power, why some stimuli and possibilities stand out and get preferentially processed, learned from, and acted upon. What it does not do — and what its own practitioners are careful not to claim it does — is explain why any of this standing-out is accompanied by felt mattering. A prediction-error signal is a quantity a mechanism computes. That the arrival of an unexpected reward should feel good, that its absence should feel like disappointment, is not contained in the statement that a neuron's firing rate deviated from a baseline. The mechanism tracks value; whether it thereby constitutes felt value is exactly the open question.
A physics-flavoured account: the free-energy principle
The most physics-shaped attempt to naturalise value comes from Karl Friston's free-energy principle and the associated framework of active inference. The core idea borrows the machinery of statistical mechanics and Bayesian inference. An organism that persists must keep itself within a bounded set of viable states — a fish must stay in water, a body must keep its temperature and chemistry in range. Formally, persisting means keeping the entropy of one's sensory states low: not visiting too many surprising configurations. Since the true "surprise" (the negative log-probability of sensations under the organism's model) cannot be evaluated directly, the organism is modelled as minimising an upper bound on it — the variational free energy — a quantity mathematically identical to the "evidence lower bound" used in machine learning. Minimising free energy amounts to making the organism's internal model a good model of its world, and keeping the organism in the states its model expects.
And yet the honest question presses again. The free-energy account is, at its core, a description of control dynamics — a precise story about which states a self-maintaining system will be disposed to seek and which it will avoid. Whether that description also captures value in the sense that matters here — felt goodness and badness, the affective tone of an experience — is not settled by the mathematics. A thermostat, or a simple homeostat, can be described in free-energy terms as "preferring" its set-point; we do not thereby suppose it feels anything about being cold. The formalism tells you what a system will do. It is silent, or at best promissory, on whether there is something it is like, from the inside, for the system to be pulled toward its preferences. Friston's framework is best read as supplying a candidate bridge law between dynamics and value — a bridge whose far end, felt mattering, it names but does not reach.
The residue: tracking value versus value being felt
Assemble the best of all of the above — the prediction-error signals, the salience network, the free-energy dynamics — and you have, in principle, a complete account of what a system is disposed to pursue: which states it seeks, which it avoids, how it weights options, how it learns. This is a great deal. It is also, arguably, everything a third-person description can deliver, because dispositions to pursue are exactly the kind of thing that show up in behaviour and in mechanism.
What such an account leaves untouched is why any of it should matter from the inside. There is a difference — and it may be a difference of principle, not merely of current ignorance — between a system that tracks value and value that is felt. This is the affective face of the same puzzle raised, for sensory experience, by David Chalmers' "hard problem" and by Thomas Nagel's question of what it is like to be a given creature. Pain is not merely a damage-signal that reliably drives avoidance; it is a damage-signal that hurts, and the hurting is the part that seems to escape the functional story. One can imagine, without evident contradiction, a system that computes every reward-prediction error and minimises every free-energy bound exactly as we do, yet for which nothing is ever at stake from within — the affective analogue of a "zombie." Whether that imagining is genuinely coherent is disputed, but that the dispute is live is itself the point: the felt goodness or badness of a state does not obviously follow from any amount of description of what the state disposes the system to do. This is why value/salience is placed among the consciousness questions, and why it may be the hardest of them: it is the normative root, the place where the "is" of physics and the "ought"/"matters" of experience meet, and where a third-person description may not, even in principle, be the whole answer.
Where the framework can and cannot go
Within its own commitments, the framework can say something modest and precise about the mechanism-facing half of the problem. It treats a persisting observer not as a system that may be cleanly severed from the rest of reality at will, but as a pattern whose separateness must be earned from the shape of the whole — the rule of nonseparability. And it treats the fineness of the distinctions a reconstruction commits to (its "granularity") as something that carries a cost, and that is legitimately coarser the further a claim reaches from the present moment. In that language, salience has a natural place: it is which record-distinctions the persisting loop chooses to weight, and which it is willing to pay the cost to keep making. Some distinctions are worth the price of drawing them; most are not and are allowed to blur. This ties salience to the same cost-floor and granularity considerations that appear elsewhere in the framework, and it gives a principled, physical reason why an observer's world is not uniformly detailed but sharpened selectively around what it treats as important.
That is a real connection, and it is worth watching. But it must be stated with the same honesty as everything else here: encoding which distinctions a loop pays to make is still an account of tracking and control, not of felt mattering. It tells you where the system will spend its finite resolution; it does not tell you why the states it resolves toward should be good or bad for it, experienced from within. The step from "distinctions the loop weights" to "genuine mattering" is not a step the framework claims to have taken. It is named explicitly as residue — the part left over once the mechanism is fully described.
Honest status
Open — the normative root; not solved. There is no derivation, no calculation, and no observable match on offer for "value" or "salience" as felt mattering. Physics and neuroscience genuinely supply a rich account of the mechanisms of salience — reward-prediction error, the salience network, free-energy minimisation, value functions — and the framework can connect salience to which record-distinctions its loop weights and pays to make. What none of this delivers, and what the framework does not claim to deliver, is a bridge from any of that to why anything matters intrinsically, from the inside. This should be read as an interesting direction to watch and an honestly flagged open question — the deepest of the consciousness roots, sitting exactly on the is/ought divide — not as an established finding, not as a partial closure, and not as a verdict that the question is unanswerable. It is open, and it is named as open.