The Fist

This is an exposition, not an argument with anybody. It sets out State Identity Theory in full: every claim, in the order the claims depend on each other, with the technical statement first and an example immediately after it. Where the theory is weak, the weakness gets the same treatment as the strength.

The theory has three levels and they are separable. The first is an ontological core: experience is not produced by a physical state, it is a physical state. The second is an account of the explanatory gap: why a complete description can feel insufficient even when nothing is missing. The third is a self-referential and formal extension: what limits appear when a physical system embedded in the universe tries to represent, certify or predict its own states.

They are separable in a specific sense. The third can fail entirely and the second survives. The second can fail and the first survives. Most theories in this area are built as a single block, so a crack anywhere brings the whole thing down. This one is built to lose parts.

I. The core

The schema that causes the trouble

The hard problem of consciousness is normally posed like this:

How can a physical process produce subjective experience?

The question looks innocent. It is not. It has already committed to an architecture, and the architecture is causal:

physical activity → experience

Read that arrow. It says there are two relata. First a physical fact, neurons and currents and timing. Then a second fact, the redness of red, which has to arise out of the first. And once you have two facts you need a bridge between them, and no bridge has ever been built, and the failure to build it is what people call the hard problem.

State identity theory does not attempt the bridge. It rejects the arrow.

The core claim, stated hard: for some physical systems, a complete organization at the relevant level is not correlated with an experience, does not cause an experience, and does not give rise to an experience. It is the experience. The relation is identity, and the schema is:

a certain physical organization = what we call seeing red

Note carefully what this does not claim. It does not claim that every physical state is an experience. It does not claim matter in general is conscious. It claims something narrower and stranger: that certain organized states of certain systems are the events that, in the first person, we call seeing, hurting, remembering, thinking.

The digest: the fist

Open a hand. Close it.

Describe what exists. There is a hand. The fingers are flexed. The thumb sits across them in a particular position. There are tensions in specific muscles, angles between the phalanges, a distribution of pressure across the palm.

You can also say: there is a fist.

Now ask the question that mirrors the hard problem:

I have the complete anatomy. Every bone, every tendon, every angle. But where, in addition, is the fist?

The question has no answer, and the reason is not that fists are mysterious. The reason is that "in addition" was never earned. There is no fist over and above the hand held that way. The hand held that way is the fist.

The analogy is structural and it has a precise job. It is not offered as proof that mind is brain. It is offered to kill one specific inference:

that two vocabularies, or two ways of getting at something, force you to postulate two things.

That inference is the engine of the hard problem. Remove it and the problem changes shape.

Do not say "produces"

The word produces smuggles the dualism back in. Say "the brain state produces the experience" and you have two relata again, a cause and an effect, and even if both are material you now have two events that need relating.

The identity formulation says something different:

the complete relevant physical event is the experience.

This does not deny causation inside the brain. Of course there are causes, transitions, sensory inputs, behavioural effects. What is denied is that once the complete state constituting seeing red is present, a further causal step is needed to generate a thing called experience.

Digest. A hand closing involves causes: motor neurons fire, muscles contract, tendons pull. All real. But there is no additional causal step at the end where the hand, having closed, then generates a fist. The closing is the fist coming to be, not the cause of it.

High-level states

The theory does not identify experiences with particles, molecules, single neurons, or fundamental physical magnitudes. The relevant state is high-level: a pattern, an organization, realized by more basic physical components and individuated by relations rather than by substance.

Digest, and this one is worth dwelling on. A storm is physical without being a particle. A digestion is physical without being one molecule. Temperature is a good case: the temperature of a gas is not produced by the motion of its molecules. It is the mean kinetic energy of that motion. Nobody asks where the temperature is hiding once you have specified all the molecular velocities, and nobody thinks that identifying the two was a way of denying that gases are hot.

For a given perception the relevant organization might include distributed neural activity, relations between populations, integration with memory and attention, bodily state, interoceptive signals, relation to the stimulus, discriminative capacity, the available transitions to other states, and effects on report, behaviour and learning.

The theory does not pretend to know which items belong on that list. It provides the ontological frame. Filling in the list is empirical work and it has barely started.

What "relevant" is allowed to mean

Here is a trap the theory has to avoid, and it is worth stating because most versions of this position fall into it.

You cannot define the relevant physical state as "the one that correlates with the experience". That is circular: you would be using experience as the criterion for what constitutes experience.

The proposal, stated hard: relevance is fixed by causal and counterfactual profile. A feature is relevant if changing it, holding the rest appropriately fixed, changes what the system can discriminate, remember, report, integrate, learn or do, or changes which states it can move to next.

So individuation depends not on what the system is made of but on the structure of relations the material realizes.

This puts the theory close to functionalism, with one difference that matters. Functionalism is often stated as: a functional organization produces an experience. Here the claim is that the complete organization is the experience. No production step, and therefore no gap where one could be inserted.

Three claims that get confused constantly

This distinction does more work than anything else in the theory, and almost every popular discussion of consciousness muddles it.

Correlation. When a certain physical pattern occurs, a certain experience usually occurs.

Supervenience. The experience cannot change without some relevant physical fact changing.

Identity. The experiential fact is not a second fact that necessarily accompanies the physical one. It is the same event, described differently.

These are strictly increasing in strength. Correlation is fully compatible with dualism: a dualist expects mental events to track brain events. Even strong supervenience can be read as a necessary dependence between two distinct families of properties. Only identity removes the duplication.

Digest. Correlation says: whenever the hand closes, a fist appears. Supervenience says: the fist cannot change without the hand changing. Identity says: there is no second item called the fist whose behaviour needs explaining. Only the third is the theory's claim, and only the third dissolves the problem.

The consequence is important and the theory states it against its own interest: perfect correlation would not prove identity. No amount of neuroscience showing that state X always accompanies experience Y establishes that X is Y. Identity is a metaphysical hypothesis that has to earn its place by explanatory power, economy, fit with science, and its ability to explain why there seems to be a gap.

Identity discovered, not deduced

The model is empirical identity, in the Kripkean sense. Nobody worked out from an armchair that water is H2O. It was discovered, and once discovered it is not the sort of thing that could have been otherwise while everything else stayed fixed.

The theory takes the same shape. Which organization is seeing red is an empirical question. If it is ever settled, the identity is then necessary in this sense: there would be no second experiential fact that could be peeled off while that exact organization remained intact.

This is philosophically contested and the theory flags it as part of its metaphysical framework rather than as a result. It is emphatically not something Gödel or Turing delivers.

Why red and not green

The traditional challenge:

Why does this physical state feel red rather than green?

If identity holds, the question presupposes a two-stage process: first there is a physical state, then an assignment problem where we decide which sensation to pair with it.

The theory denies stage two. If you altered the relations that make this state the state it is, such that it now constituted seeing green, it would not be the same complete state any more. So within the theory, "why is this complete state red and not green" approaches "why is this configuration this configuration and not a different one".

Digest. "Why does this hand position make a fist rather than a peace sign?" is not a deep question about the metaphysics of hand shapes. Change the position and you change which gesture it is. There is no further assignment step where gestures get handed out to configurations.

What remains genuinely open, and the theory is explicit here, is the content question: which physical and organizational differences constitute red, green, pain, memory, thought. The theory supplies no answer. It converts a metaphysical puzzle into an empirical research programme, which is progress, but progress is not a solution.

II. Why it still feels like something is missing

The core is now stated. If that were the whole theory it would fail, and the reason it would fail is the best objection in the field.

Description is not occurrence

A description of an event is itself another event.

A chemical description of fire does not burn. A biomechanical description of a fist does not close a hand. A complete account of digestion digests nothing.

So a representation of the brain state corresponding to an experience is not, merely by representing it, an instance of that state.

This distinguishes two questions that get run together:

1. Does the description contain enough information to specify the state?
2. Does reading the description produce the same mode of access as being in the state?

The answer to the first can be yes while the answer to the second is no.

Why that is not enough

Here is the objection, and the theory raises it against itself in section 10.

If "description is not occurrence" were the whole story, we would have a problem. A description of a flame is also not a flame. Yet there is no hard problem of combustion. Nobody feels that chemistry leaves out the true inner nature of fire.

So there is a real asymmetry, and it demands an account:

• with the fist, we accept without strain that a configuration and a high-level name pick out one thing;
• with experience, we keep feeling that the physical description omits how it feels.

Repeating "but everything is physical" explains nothing about the asymmetry. A theory that cannot explain why the intuition is so violent has not engaged with the problem, it has only announced a preference.

Two modes of access

The hypothesis, stated hard. There are two ways of reaching an experiential state.

Descriptive access. Talking about neurons, relations, stimuli, memory, dynamics, causal profiles, behaviour. This is available to someone not in the state.

First-person access. The system is in the state. The access does not consist of receiving a further description. It consists of the relevant physical organization occurring.

The difference between these can be enormous even when the referent is identical. From which the general rule:

An epistemological difference between modes of access does not by itself imply an ontological difference between their referents.

Two concepts can carry different information, enable different capacities, and fail to be derivable from one another a priori, while referring to the same event.

The digest: the hand that never closed

Imagine someone who knows the anatomy and mechanics of closing a hand exhaustively. Forces, positions, photographs, films, models, the lot. Through some circumstance they have never closed their own hand.

One day they close it, and say: now I know what making a fist is like.

They have gained something real. A new mode of access, a capacity, a bodily memory, possibly a new concept tied to that state.

But it does not follow that they discovered a second, non-physical entity called the sensation of fist, sitting on top of the closed hand. Cognitive novelty does not require ontological duplication.

This is the theory's answer to the knowledge argument, in miniature and without the colour scientist.

Phenomenal concepts

The hypothesis: some concepts referring to experiences are quotational. The system refers to the state by using the state itself, or a partial reactivation of it, rather than by deploying a description.

The theory is careful here, and the care is the point. This is not a logical truth. It is a claim about cognitive architecture, and it could be false.

Its job is to explain three stubborn impressions:

• that no description seems to contain the red;
• that the physical state seems separable from the experience in thought;
• that the identity seems arbitrary.

If one concept unfolds descriptively and the other works by instantiating or reactivating what it refers to, then their failure to connect a priori is exactly what you would predict.

Digest. Consider the difference between the word "salty" as used in a chemistry paper and the same word used while eating. The second usage reaches for the state by partially running it. That the two never quite meet in the middle is not evidence that there are two saltinesses.

The gap, relocated

The theory does not deny the explanatory gap. It accepts it as a real intellectual phenomenon. What it denies is the inference from the gap to a second ontology.

Formally:

1. a complete description may fail to provide first-person access;
2. it does not follow that the mode of access has a different referent;
3. therefore "I know all the descriptive facts and still do not know how it feels" does not demonstrate a missing non-physical fact;
4. it may demonstrate that one is not in a certain organization, or lacks a certain access capacity.

The burden shifts. Anyone inferring dualism from the gap now owes an argument for why a difference in access requires two events.

Zombies

A philosophical zombie is a physically exact duplicate of a conscious person, lacking experience.

The theory's position: metaphysically impossible. If the duplicate reproduces every relevant physical relation constituting seeing red, then by the identity thesis it is in that state. There is no further fact available to be subtracted while the physical is held fixed.

Crucially, the theory does not dismiss the conceivability of zombies as mere confusion. It explains it: descriptive concepts and first-person concepts do not connect a priori, so we can manipulate them somewhat independently in imagination even if they do not track separable facts in the world.

Digest. I can entertain the sentence "this water is not H2O" without contradiction, in the sense that the words do not fight each other. That does not make it metaphysically possible. Conceivability tracks the structure of our concepts, not always the structure of the world.

And the theory concedes what it must: this does not refute Chalmers. A dualist can simply deny the identity being assumed. What the theory offers is not a knockdown but a positive explanation of why conceptual separability need not mirror ontological separability. That is a weaker thing than a refutation and the theory says so.

Inverted spectrum

Same logic. Two systems identical in every relevant physical property but with swapped colour experiences.

If you genuinely hold fixed the entire organization constituting discrimination, memory, relations among colours, learning, association, report and self-reference, the theory's claim is that there is no leftover experiential parameter available to invert.

A real inversion would require some difference in the complete organization. That we do not yet know what that difference would be does not license postulating a non-physical variable.

Digest. Swap two gestures in a sign language while keeping every use, association, context and response identical. What exactly has been swapped? If nothing observable, dispositional or relational changed, the claim that the gestures nonetheless traded places has quietly become the claim that gestures have hidden essences.

Anaesthesia, sleep, degrees

The theory does not describe anaesthesia as "brain still running, consciousness switched off". It describes different physical states with different dynamical repertoires: changed integration, changed communication, changed availability, changed memory formation, changed transitions.

Nothing called consciousness leaves. Certain organizations simply stop occurring.

And consciousness need not be binary. The word bundles together seeing, hearing, hurting, remembering, noticing the body, maintaining temporal continuity, accessing information, evaluating confidence. Components can degrade separately and dissociate.

Digest. Ask when exactly a hand stops being a fist as it opens. There is no instant, and the absence of an instant is not a mystery about fists. It tells you that "fist" names a region in a space of configurations, not a switch.

III. Self-reference without duplication

The wrong way to say it

"A system has self-reference" invites the same error as "a hand has a fist". It sounds as though there is first a system and then a property bolted on.

The precise formulation:

A system is self-referential when part of its causal organization depends on information about states of that same system, and that information can alter its subsequent evolution.

There is no system plus self-reference. There is one system organized with certain internal relations.

Digest. A thermostat is not a heater with self-reference attached. It is a heater wired so that part of what determines its next state is a measurement of the conditions it is itself producing. One device, one organization.

A self-model is not an object

"Self-model" should not be reified into a single representation sitting somewhere specific. It is shorthand for a distributed network of relations by which certain internal states depend on other states of the same organism.

A self-model may carry information about bodily boundaries, position and movement, goals, autobiographical memory, confidence and error, attentional focus, ongoing actions, temporal identity, perceptual states, social relations, and estimates of one's own capacities.

The unity of the first person does not require an observer reading all that data. The unity can consist in the causal coordination of the system itself.

No inner theatre

The intuitive and fatal picture:

world → brain → representation → inner screen → observer

If an inner observer must watch the screen, we owe an account of how it sees. If it contains another screen and another observer, the regress is off and running.

The theory removes the inner observer. When a system is organized to discriminate, remember, respond, integrate and self-refer in certain ways, no additional subject is added to read the organization. The subject is the organized system.

"I perceive my perception" does not require two observers. It can mean that certain states of the system respond causally to other perceptual states of the same system.

Digest. A country does not need a citizen who is the country in order to hold an election. The election is something the organized population does, not something a further entity watches it do.

The universe, locally

Can we say the universe refers to itself?

In a local and non-mystical sense, yes. A mind is a part of the universe whose dynamics include information about that same part and its relations to the rest.

But the theory explicitly warns against "the universe observes itself", because that suggests a global agency it neither needs nor claims.

The precise version:

A physically organized region of the universe can contain processes whose inputs include variables about that same region.

When someone thinks "I am seeing red", the universe has not produced a second external observer. A part of the universe is in a state that includes relations of perception, memory and reference directed at its own states.

The first person as an indexical

"Here", "now" and "I" are indexicals. Their reference depends on the position from which they are used.

"Here" does not name a special property possessed by a place. It designates the place from which the act of reference is made. "Now" is not a temporal substance.

The proposal: part of the first person can be understood the same way.

"I" designates the system from which these operations of perception, memory, decision and self-reference are being carried out.

First-person experience would then not be a property added to the state, but the mode of access that exists because the system is the causal reference point of its own representations.

Digest. A mall map has a dot saying "you are here". The dot does not mark a metaphysically privileged tile. It marks the position from which the map is being read. Move the map and the dot moves, and nothing about the mall has changed.

The theory does not claim this reduces all phenomenology to indexical language. Its job is to show one way in which "from the inside" can mean something physically precise without a second substance.

The correction that matters

A tempting but false claim:

No system can represent itself, because the representation would have to contain itself.

This is simply wrong. Formal systems build self-referential sentences. Programs print their own source. Machines store detailed models of their own structure. Self-representation is routine.

The real limitation is different:

a sufficiently rich system cannot, under certain conditions, be simultaneously complete, infallible and universal with respect to all questions about its own activity.

Self-reference does not block representation. It introduces conditions under which completeness, decidability or global self-certification fail.

This correction is essential. The first person must not be identified with "something that literally does not fit" in the model, as though a physical fragment were missing. The rigorous claim is non-closure: internal representations are causally part of the system represented, and there is no guarantee of a final internal description settling every relevant question about oneself.

The first person as relational non-closure

The strong form of the synthesis:

1. an organism occupies a complete physical state;
2. some parts of that state represent or evaluate other parts of the same organism;
3. those representations are themselves physical states of the organism;
4. when the system incorporates a new representation of itself, the total system has changed;
5. therefore the self-model must not be confused with an external, static copy;
6. in sufficiently expressive systems, certain claims to universal closure meet formal limits.

The philosophical hypothesis is that the third-person and first-person difference is related to being the process within which representations of oneself are part of what is represented.

It is not claimed that formal incompleteness is identical to redness. It is not claimed that all self-reference is consciousness. The claim is that the structure of the first person is better understood as an open internal relation than as an additional object.

IV. The formal limits

This part is the most technically demanding and, the theory insists, the least load-bearing. Read what follows knowing that none of it proves the identity thesis.

What theorems are allowed to do here

Gödel, Tarski, Löb, Turing, Rice and Wolpert belong to different domains and must not be blended into a vague slogan about self-reference producing incompleteness. Each has hypotheses and the hypotheses matter.

• Gödel: limits on completeness and self-consistency of certain formal theories.
• Tarski: limits on defining truth internally in sufficiently expressive languages.
• Löb: limits on certain formalized self-trust.
• Turing: no general algorithm decides halting for arbitrary programs.
• Rice: no general algorithm decides any non-trivial semantic property of arbitrary computable functions.
• Wolpert: general limits on physical inference devices embedded in the universe they infer about.

These results are not equivalent to each other, and collapsing them into one slogan is the standard way this material gets abused. But they do share a structural intuition, and naming it correctly is useful.

A system attempts to apply a general capacity to a domain that includes, directly or indirectly, instances of that same capacity. Cases can then be constructed in which a universal answer would produce contradiction, indefinition or impossibility.

In computability the diagonalization appears when you ask what happens when a procedure is applied to its own description. In logic it appears when you build sentences that say something about their own provability. In physical inference the device is a part of the world whose states are also variables of the world.

The theory gives this family the name self-referential non-closure, always with the warning that each theorem carries its own hypotheses and that the name is a label for a family, not a single result.

Before applying Gödel to a mind

Gödel's first theorem does not say that anything thinking about itself has inaccessible truths. It applies to formal systems that are sufficiently expressive for a portion of arithmetic, effectively axiomatized, and consistent or correct depending on the formulation.

To apply it to a person you need a bridge hypothesis:

some relevant part of the person's explicit reasoning can be modelled as an effective formal procedure of sufficient expressive power.

Only under such a hypothesis can there be Gödelian consequences for a human reasoner.

And the theorem does not show that minds exceed machines. A human reasoner makes mistakes, changes axioms, uses metatheories, and is not a fixed formal system. The Lucas-Penrose argument does not follow from Gödel and is no part of this theory.

The six results, digested

First incompleteness. A consistent, effectively axiomatized theory rich enough for arithmetic is incomplete: there are arithmetical statements it cannot decide.

Applied conditionally: do not expect an internal, effective, consistent, sufficiently expressive theory that settles every question formulable in its own domain. The relevant lesson is not that experience is ineffable. It is that universal internal completeness is not a reasonable requirement for a rich self-referential system.

Second incompleteness. Such a theory cannot prove its own consistency by the standard representation of that claim.

Careful reading: this does not mean a person can never gain evidence about their own reliability. It means that if you fix a formal system representing a rich enough part of someone's reasoning, total self-certification cannot be obtained naively from within. A stronger metatheory can certify the weaker one, and then the same question reappears one level up. An open hierarchy, not an absolute incapacity.

Digest. You can check your arithmetic by redoing it, and you can check the check. What you cannot do is produce, from inside the same method, a guarantee that the method never errs, using only that method's own resources.

Tarski. For sufficiently expressive formal languages, a semantically adequate truth predicate cannot be defined without restriction inside the same language.

Applied modestly: a sufficiently expressive internal language should not be assumed capable of containing, without hierarchy, a total theory of the truth of all its own sentences. This does not show introspection is false. It shows that a subject possessing an infallible internal truth predicate for everything it can formulate is formally suspect once you try to make it rigorous.

Löb. Roughly: in adequate arithmetical systems, if the system can internally prove "if this proposition is provable, then it is true", then the proposition is already provable.

The lesson is not psychological. It is that formalizing global, unrestricted trust in one's own proof mechanism imposes very strong constraints. A rich self-referential architecture should not be pictured as an inner observer certifying the correctness of everything else from a privileged vantage point.

Turing. No universal algorithm decides, for arbitrary program and input, whether the computation halts.

This does not mean we cannot predict particular programs. We decide infinitely many cases. The limit is on generality.

Digest. There is no perfect universal debugger. Not because debugging is hard, but because a general halting-decider could be turned on a program built from itself, and the construction produces a contradiction.

Rice. Every non-trivial semantic property of the function computed by a program is undecidable in general.

To apply Rice to consciousness you need strong assumptions: that the systems are modellable as computable functions in the relevant sense, that "being the kind of system constituting an experience" is an extensional semantic property, and that it is non-trivial. If those hold, there can be no total algorithm deciding correctly for every arbitrary program whether it has the property.

This does not prevent effective tests in restricted domains, and does not show consciousness is computational. The conclusion is conditional.

Digest. There is no perfect universal virus scanner either, for the same structural reason. You can still write a scanner that works very well on the programs people actually run.

Wolpert. This one is the most important for the theory, because it does not require treating the brain as a Turing machine.

An inference device is part of the same universe it observes, predicts or remembers. That condition of physical embedding generates impossibility results. There is no physical computer able to answer correctly every inference task about the universe, and there cannot be an arbitrary plurality of devices with strong universal inference power over each other.

The defensible thesis:

no physically embedded agent should be assumed capable of infallibly inferring every possible question about the universe it is part of.

This does not mean prediction is impossible or that science fails. It limits universality, not knowledge.

Digest. A weather forecaster is made of weather. A published election forecast becomes an input to the election. The observer is not standing outside the system taking notes; the notes are part of the system, and in some cases the notes change what they describe.

What none of this proves

The theory devotes a section to this and it is the most valuable page in the document. The formal results do not establish:

• that consciousness is immaterial;
• that qualia are undecidable theorems;
• that every human thought is Gödelian;
• that the human mind is non-computable;
• that a machine cannot be conscious;
• that the first person is literally a Gödel sentence;
• that no system can represent itself;
• that every self-model produces consciousness;
• that free will follows from undecidability;
• that a particular person cannot be predicted in a particular situation.

Their role is narrower and therefore sturdier: certain expectations of universal closure, total self-certification, general decision and infallible prediction are mathematically incompatible with certain formal and physical structures.

V. The limits applied to humans

A theorem does not transfer to a person automatically. The theory separates three kinds of limit before applying anything:

Logical and computational limits apply if a relevant human capacity is modelled by effective procedures or sufficiently expressive formal systems.

Physical inference limits apply to the human as an embedded physical device, and do not require identifying cognition with a Turing machine.

Architectural limits follow from the self-model being part of the organism and participating causally in what it describes. These are theses of the theory, not standalone theorems.

With that separation in place, fifteen limits follow. Each one is stated with the condition that activates it, because a limit quoted without its condition is a slogan.

1. There is no absolute internal perspective

A person can build representations of themselves, but any internal representation is physically realized by a part of that same organism. There is no point inside the system that is literally external to it, from which the subject could observe the complete system without being part of what is observed.

This does not mean the representations are false. It means the relation between observer and observed changes in self-observation: the representing mechanism is included in the represented object.

Digest. A security camera bolted inside a room can show you the room, including itself in a mirror. What it cannot produce is a view of the room from outside the room, because the lens is a fixture of the thing being filmed.

2. No internal self-model is the whole organism

The internal model can encode a great deal about body and mind. It might even contain descriptions of its own architecture. But an internal representation is not identical to the complete physical system realizing it.

The difference is not necessarily imprecision. It is a difference of causal type:

• the complete organism includes the representing process;
• the representational content is only a part of that organism;
• incorporating or updating a representation modifies the total state.

This is compatible with extremely accurate self-models. It asserts no immaterial residue.

Digest. Draw a perfectly accurate map of a country, inside that country. The map occupies land. Adding it changes what there is to map, and the updated map must now include itself, and so on. The limit is not draughtsmanship. It is that the map is furniture in the room it depicts.

3. No deductive closure, if the reasoning is formalized

Suppose we fix an effective formal procedure capturing a sufficiently rich part of a person's mathematical reasoning, and suppose it is consistent. Then by the incompleteness results it cannot settle, from its own rules, every arithmetical statement expressible in its language.

The consequence for a human is not "there are truths no human will ever know". The person can change systems, add axioms, reason from a metatheory. The precise consequence is:

no fixed, effective, sufficiently expressive formal system used by the human can be simultaneously consistent and complete for its own arithmetical domain.

Moving to a metatheory widens the reach. It does not create a final guaranteed level of closure.

Digest. You can always climb to a stronger theory to settle what the weaker one could not. The staircase is real and it works. It simply has no top floor.

4. No global self-certification by a fixed formal reasoner

If a person adopts a sufficiently strong, consistent formal system, that system does not obtain from its own rules a proof of its own consistency in the standard sense of the second incompleteness theorem.

The person may have external reasons, or a stronger theory. But then those reasons are not an absolute certification generated by the original system about itself.

Translated into the theory of the first person:

there is no reason to expect an internal module capable of guaranteeing, from a privileged vantage point, the total correctness of all the cognitive mechanisms that produce it.

Human confidence is necessarily local, revisable, and supported by relations between subsystems, external evidence, memory, comparison and learning.

Digest. A firm can audit its own accounts, and the audit can be careful and useful. What it cannot do is produce, using only its own accounting department, a final guarantee that its accounting department never errs. That is not a slur on accountants. It is the shape of the request.

5. No total internal truth predicate at one expressive level

A human can make judgements about the truth of their own statements. But if we tried to turn that capacity into a sufficiently rich formal language containing its own truth predicate, without hierarchy or restriction, we would run into the difficulties Tarski studied and the self-reference paradoxes.

The limit does not say the word "truth" is illegitimate in natural language. It says the picture of a mind with a single formal mechanism able to label every sentence of its own complete language infallibly true or false is not compatible with the standard form of those results.

For the theory this reinforces one idea: the first person is not an omniscient semantic centre. It is an organization operating with levels, approximations and revisions.

Digest. "This sentence is false." Natural language absorbs the liar because it is loose, layered and tolerant of repair. A formal system carrying a total internal truth predicate does not have that luxury, and that is exactly the difference the limit is pointing at.

6. No universal algorithmic predictor of all computational behaviour

If part of human behaviour can be modelled as general computation, there is no universal algorithm that, given an arbitrary description of the procedure and its input, decides every question equivalent to the halting problem.

This does not mean a person is unpredictable in general. Most of our actions are statistically predictable, and many concrete tasks can be analysed successfully.

The correct claim is:

there can be no general algorithmic method correctly resolving all questions of computational evolution for all arbitrary systems of a sufficiently general kind.

So "physics is material" does not imply "there exists a universal algorithm letting us effectively deduce any future fact about any agent". That inference is made constantly and it does not hold.

7. A communicated prediction can become part of what it predicts

An additional difficulty arises when a prediction about a person is causally incorporated into the person before the predicted event.

The sequence:

1. a prediction is produced about the agent's future state;
2. the agent receives the prediction;
3. that reception changes their physical state;
4. the future to be predicted now depends on the prediction itself.

Not every prediction becomes impossible. Fixed points can exist: someone can hear "you will do A" and do A. The limit appears when you demand a universal and infallible predictor for a class that permits diagonalized or contrary responses.

The theoretical importance is that self-knowledge is not always passive. A representation of my future can become a cause of a different future.

Digest. Hand someone a sealed envelope containing a correct prediction of their next choice, then let them open it first. If they are able to do otherwise out of sheer contrariness, the prediction was not wrong about a fixed world; it became one of the causes of the world it described. This is why self-prediction is not like observing a motionless external object.

8. No universal classifier of arbitrary semantic properties

Under the computational extension, if a mental property is identified with a non-trivial semantic property of the computation a system performs, Rice's theorem blocks a total algorithmic classifier for all arbitrary programs.

The consequence, for humans and machines alike:

there can be reliable methods for restricted classes of systems without there existing a universal test that works for any imaginable computable system.

Here the theory applies a brake to itself, and it is worth quoting the restraint. It must not simply assert that Rice proves there is no universal consciousness detector. It can only say so conditionally, if the relevant property satisfies the theorem's hypotheses. If consciousness depends on non-extensional properties, on physical causal history not captured by the computed function, or on features that do not fit Rice's frame, the argument does not apply at all.

9. No physical agent is a universal Laplacean demon

Wolpert's results allow a more general limit that does not depend on modelling the human as a Turing machine.

A person, a laboratory, a computer, any inference device, is a physical part of the universe. Its states, configurations, questions and answers also belong to the possible histories of that universe. Within that framework, universal and infallible inference meets restrictions of principle. The agent cannot become an observer external to the cosmos while remaining inside it.

However much information, intelligence or computational capacity a physical agent has, it should not be assumed able to answer correctly every possible question about the universe it is part of.

Note the strength of this one: within Wolpert's framework it holds even in classical, non-chaotic universes. It does not need quantum indeterminacy, and anyone reaching for quantum mechanics to establish it has misunderstood where it comes from.

10. Remembering is also inference from inside

Memory is usually imagined as an archive containing a copy of the past. But a memory is a present physical state carrying information about earlier states.

In the framework of inference devices, observation, prediction and memory share a structure: an embedded device uses its current state to answer questions about variables of the universe.

The theory does not conclude that all memory is inaccurate. It denies only the identification of being the past with representing the past. A perfect memory of one specific aspect would still be a present representation, not a complete reinstantiation of the past event.

Digest. A photograph of a fire is not warm. It can be a very good photograph.

11. There is no guaranteed final level of metacognition

A person can think "I see red". Then "I know I see red". Then "I know that I know".

The theory does not claim this sequence must actually continue to infinity. It claims that no level should be reified as the final observer validating all the previous ones.

Each metacognitive level is another physical process. It can evaluate, correct or summarize lower levels, and can itself become the object of evaluation by later processes. So the coherent architecture is hierarchical and open, not a theatre ending in a homunculus.

12. Introspection is not direct access to one's own microphysics

If experience were the complete relevant physical state, it does not follow that a person must know the physical description of that state introspectively.

Ontological identity does not imply epistemological transparency.

A storm's macroscopic description does not contain an accessible list of each molecule. An organism does not know by proprioception the state of every receptor. A person can be in a state without possessing a scientific concept of its realization.

Being a state is not the same as having a complete description of the state.

This is the answer to a very common objection: if experience is physical, why do I not perceive my neurons? Because identity between a state and an experience does not turn the experience into an instrument for reading microphysics. The two claims were never connected.

13. Internal openness without metaphysical indeterminism

From inside, one's own future can appear open because the system has no universal procedure deriving all its future decisions before making them.

This is compatible with different temporal ontologies and different interpretations of quantum mechanics. The feeling of openness does not require postulating a will outside physical causation.

The theory carefully separates three things that get conflated:

• epistemic unpredictability: the agent cannot universally derive its own future;
• physical indeterminism: the laws do not fix a unique outcome;
• metaphysical freedom: a further thesis about responsibility or the ability to have done otherwise.

The incomputability and inference results support, at most, the first. They do not settle the free will problem, and the theory declines to pretend otherwise. This is the single most commonly abused inference in popular writing about minds, and it is refused here explicitly.

14. The limits do not disappear by adding intelligence

Many practical limits can be reduced with more memory, better instruments or more computation. The formal limits studied here are different in kind.

If the obstacle is lack of memory, adding memory helps. If the obstacle is computational complexity, more resources enlarge the tractable set. But if the obstacle is an impossibility of universality derived from incompleteness, undecidability or physical embedding, adding resources does not convert an impossible procedure into a possible one for all cases.

The consequence for any theory of mind, and for every discussion about smarter systems:

not every cognitive limit is a contingent defect of the human brain.

15. No formal limit introduces a second reality

The philosophical conclusion joining this whole part back to the core is negative and strong.

That a system cannot completely, decidably, certifiably or universally predict certain aspects of itself does not imply that those aspects are non-physical.

The incompleteness of an arithmetical theory does not add immaterial numbers to the mathematician's physical universe. The undecidability of halting does not add a second mysterious kind of computation. The inference limits of a device do not create a reality outside the universe.

In exactly the same way, if part of the first-person gap is due to a form of epistemic non-closure, the fact that the system cannot turn its complete being into a final internal representation provides, by itself, no evidence whatever for dualism.

This is the keystone. Everything in Parts IV and V is a ceiling on knowledge. A ceiling is not a door.

VI. The computable universe, as an optional extra

The core does not require the universe to be computable. It can be stated in any physical ontology with organized states and causal relations.

The theory may add computability as a working hypothesis, which allows tighter connections to computation theory. But four things must be kept apart:

• computability of the laws;
• computability of a particular history with contingent or random outcomes;
• practical simulability with finite resources;
• decidability of general properties of the computations.

Even if the laws are computable, it does not follow that every question about their evolution is decidable or that predictions are efficiently obtainable.

Randomness does not automatically remove computational limits either. Probabilistic machines remain subject to undecidability in the relevant sense, though the form of the predictions changes.

There is a distinction here that is easy to miss. A concrete sequence of genuinely random outcomes may not be computable as an infinite sequence, while the probabilistic rules describing its distribution are perfectly computable. Those are different objects and only one of them is at stake.

The theory declines to decide whether quantum randomness is ontologically irreducible. What matters for its purposes is that state identity does not depend on turning every detail of the universe into a predictable deterministic string.

Substrate independence, with the caveat

If experiential states are individuated by causal organization rather than by specific matter, multiple realization follows: different substrates could realize the same relevant organization.

The theory is careful to mark this as a strong functionalist consequence to be kept separate from the minimal core, and it adds a restriction that most enthusiasts skip:

It does not follow that any abstract simulation is automatically a physical instance of the same state. What counts as causal realization, and which relations must be preserved, has to be specified.

Digest. A detailed simulation of a hurricane does not make anyone wet. Whether a simulation of a process is an instance of that process depends on what the process is, and answering that requires saying which relations are constitutive. Nobody has a general answer. Anyone who tells you the substrate question is settled in either direction is ahead of the evidence.

No consciousness variable

The word consciousness bundles seeing, hearing, hurting, remembering, thinking, noticing the body, maintaining temporal continuity, accessing information, evaluating confidence.

The theory needs no additional binary variable that switches on at a threshold. There can be families of states, degrees, dissociations. Components can vanish while others persist.

This defuses a badly formed question:

at what exact instant does consciousness enter the system?

and replaces it with:

what organization is present, and what capacities, relations and modes of access does it constitute?

Personal identity

Same non-substantial treatment. "I" need not designate a simple indivisible entity distinct from the organism. It can designate a dynamic organization preserving sufficient continuity of memory, body, causal history, goals, social relations and self-modelling.

The cogito shows at minimum that if thinking occurs, something occurs. It does not by itself establish a separate mental substance.

Continuity can be gradual, reconstructed, and dependent on multiple processes. This does not make persons unreal. It treats them as high-level organizations, like every other real thing that is not fundamental.

VII. The objections, including the ones that land

This is where the theory earns whatever trust it deserves. What follows are its own admissions.

It does not prove identity

The theory can show that a conceptual gap does not force an ontological gap, that self-reference needs no inner observer, and that formal limits on closure exist. None of that, alone or together, proves that a physical state is identical to an experience.

Identity remains the central hypothesis. The argument is therefore abductive, not deductive. It claims identity is the best explanation because it eliminates the need for bridge laws, explains why zombies are impossible if it is true, recasts the gap as a difference of access, fits a monist ontology, avoids the inner theatre, makes the first person a relation, and coexists with formal limits without mystifying them.

An opponent can reject the inference and hold an ontology of additional properties. The theory says so out loud.

The master argument against phenomenal concepts

The classic objection: if phenomenal concepts have a complete physical explanation, a physical duplicate would have them too; but a physical duplicate without experience seems conceivable; so the conceptual strategy does not close the gap.

The theory's reply is that step two reintroduces exactly the zombie possibility the identity thesis denies.

And then it concedes the obvious: this reply uses the identity thesis, so it is not a neutral refutation of dualism. Its value is showing internal coherence, not victory.

The circularity problem, which is the real weak point

If the relevant state were defined as "the one that produces the experience", the theory would be circular. Hence relevance is defined by discrimination, memory, transition, control, learning, environmental relation and bodily structure, without phenomenal terms.

But a difficulty survives, and in my judgement it is the load-bearing crack.

There may be many ways to choose the level of description and the set of relevant variables. The theory needs a methodology of individuation that avoids retrospectively selecting exactly those variables that match our experiential categories.

That methodology does not exist yet. Until it does, "identity with the relevant organization" names a bill without paying it. The theory calls this a real scientific and philosophical problem rather than a fatal objection, which is fair, but it should be read as the place where future work has to happen.

Is self-reference enough for consciousness?

No.

A thermostat holds a variable about the system it regulates. A program inspects its own memory. A compiler processes a description of itself. A system can carry metadata about its own operation.

None of that suffices for experience.

The theory must distinguish trivial self-reference, functionally rich self-modelling, integration with perception, memory, body and action, and states that actually constitute experiences. The formal theorems constrain the earlier categories and do not select the last one.

What stays open

The synthesis may explain why there is a difference between third and first person. It does not explain why one configuration constitutes red and another pain. The theory separates:

Form of the first person: why can a system stand in an internal relation to itself that is not equivalent to an external description?

Content of experience: which concrete physical organization constitutes each quality and modality?

The extension addresses the first. The second is open empirical work.

Nor does the indexical proposal claim to exhaust qualitative character. The objection that the felt quality of an experience exceeds a simple indexical reference is one the theory accepts pressure from. Its prudent answer is that the indexical explains why the state is referred to from within the system that instantiates it, without requiring an extra observer. Qualitative content still depends on total organization.

Why does it feel this way and not another

Inside state identity this question has two readings and they have opposite fates.

First reading: why could the very same complete state not have a different experience, remaining literally identical in everything relevant? The theory answers that the possibility is badly posed. Changing the experience would be changing the state that constitutes it. There is nothing left over to vary.

Second reading: which physical differences make two states have different contents? That question is legitimate, it is open, and it is where the work is.

The distinction matters because the theory is often accused of answering the content question by saying the word "identity" loudly. It does not. It converts the content question into a structural one:

which physical relations distinguish the states we call red, green, pain, taste, fear, memory or thought?

That is an unanswered question, not a dissolved one. The theory changes what kind of question it is, and then hands it to science.

Can a state represent itself completely?

This should not be answered with a flat no, and the theory is careful here in a way that corrects an earlier and cruder version of itself.

A system can store a complete description of its own code under certain criteria. A finite representation can describe much larger structures through compression. Syntactic self-reference is entirely possible. None of that is blocked.

The right question is what completely is being asked to mean.

If it means "containing a description from which the system could infallibly decide any relevant property of itself, globally certify its own correctness, and universally predict its own consequences even when those predictions re-enter it causally", then various formal results block that idealization under specific conditions.

So the theory explicitly abandons the naive argument from size or containment, the one that says a representation cannot fit inside what it represents, and replaces it with the argument from universal closure. The first argument is false. The second is defensible. Confusing them is how this material usually goes wrong.

Physicalism is assumed

Yes, in the strong version, and the theory says so plainly.

It bets that everything that exists is part of physical reality, then tries to show the world need not be doubled to account for experience and the first person.

It is not correct to say physics proves everything is material. Physical monism is a general ontological hypothesis, supported by the continuity of scientific explanation and by economy, and it remains a philosophical position.

The theory gains strength if it can show that the phenomena used against physicalism, namely the gap, the first person and the conceivability of zombies, arise predictably from inside the physical framework.

What would falsify it

The core would be seriously compromised by two systems physically identical in every relevant respect with a real experiential difference corresponding to no physical difference. That is precisely what zombies and strong inverted spectra assert. The difficulty is that a purely phenomenological difference observable by no physical criterion is hard to establish independently of the metaphysics under dispute.

The synthesis is more vulnerable than the core. It loses force if the first-person and third-person difference can be explained entirely without any special role for self-representation, indexical access, or internal self-evaluation relations. It also loses force if phenomenal concepts turn out to be wholly descriptive with no dependence on reinstantiation.

And even then the core could survive. The modularity is deliberate.

VIII. The layered formalization

To stop conclusions from one layer being presented as theorems of another, the assumptions are stacked explicitly. This is the discipline that keeps the rest of the document honest, so it is worth setting out in full rather than summarizing.

Layer A, minimal ontological assumptions

A1. Physical monism. Every concrete event belongs to the physical domain. No additional experiential substance is introduced.

A2. High-level states. An event can be individuated by organizational and causal relations without being a fundamental entity.

A3. State identity. At least some high-level physical states are identical to what we call experiences.

Layer B, assumptions about access and representation

B1. Representation-instantiation difference. Representing a state is not, in general, the same as instantiating it.

B2. Specific phenomenal access. Some first-person concepts or capacities require being in, reactivating, or causally depending on the states they refer to.

B3. Physical self-reference. Part of a mind's organization responds to information about states of that same system.

Layer C, optional formal assumptions

C1. Effectiveness. Certain relevant capacities of the system can be modelled as effective procedures.

C2. Expressiveness. Some internal procedures are rich enough to encode arithmetic or properties of their own processes.

C3. Semantic property. If Rice is invoked, the corresponding mental property is taken as semantic, extensional and non-trivial over a class of computations.

C4. Embedded device. If Wolpert is invoked, the agent is modelled as a physical inference device inside the same universe it infers about.

Note what the layering buys. A reader who rejects C entirely still has A and B. A reader who rejects B keeps the ontological core. Nothing in C can be used to argue for A3, and the document never tries.

The seven propositions

Proposition 1. A difference of access does not imply a difference of referent.

Take a physical event individuated under two distinct conceptual schemes. That a subject can possess one concept without possessing the other does not entail that there are two events. This is ordinary in empirical identities with multiple modes of presentation.

Conceptual non-derivability does not yield ontological duality.

The theory uses this to block the step from "I can know the description without knowing how it feels" to "there are two facts". It is not a proof of identity. It is a proof that a very common argument toward duality needs an extra premise it rarely supplies.

Proposition 2. A description does not become its object by being complete.

Suppose a representation exhaustively specifies a state in some language. The representation remains a distinct state with distinct causal relations.

Descriptive completeness does not imply causal identity with what is described.

From which: demanding that a description "contain" the object's mode of being confuses information with instantiation.

Proposition 3. The self-model is part of the total state.

If a physical system contains processes representing other processes of the same system, those representational states are also part of the total physical state. An update of the self-model is therefore an update of the total system.

Self-description is not a causally external copy of the object described.

This prepares the closure limits without, on its own, proving them.

Proposition 4. Formal non-closure under Gödelian hypotheses.

If a reasoning component is modelled as a consistent, effectively axiomatized theory rich enough for arithmetic, it will not be complete for all arithmetical statements of its language, and under standard conditions will not internally prove its own consistency.

The expectation of a final, complete, self-certified internal theory is incompatible with the Gödelian hypotheses.

The conclusion does not extend, without further argument, to the whole mind or to all human knowledge.

Proposition 5. Universal non-decision under computational hypotheses.

If a class of systems contains general computation, no algorithm decides halting for every program and input. If additionally a property is semantic, extensional and non-trivial over computable functions, no total decider exists for it across all programs.

A computational physical theory does not imply a universal algorithmic classifier of all properties of all systems.

Proposition 6. No universal physical inference.

If an agent is modelled as an inference device embedded in the universe, Wolpert's results impose limits on universal and infallible inference.

Being inside the universe is a formally relevant condition on what an agent can infer about the universe.

This is the most direct connection between "we are parts of the universe" and a theoretical limit that does not depend specifically on consciousness.

Proposition 7. Epistemic limits do not multiply entities.

Propositions 4 through 6 yield limits of proof, decision and inference under their respective conditions. None of them introduces a new class of ontological fact.

The existence of internal limits is compatible with a monist ontology.

This compatibility is the load-bearing beam of the whole reinterpretation. That a system does not fully close its knowledge of itself is not automatic evidence of a second reality. Everything in Parts IV and V is an epistemic ceiling, and a ceiling is not a door.

IX. The synthesis

Four pieces

Identity. There is not first a physical reality and then an experience generated on top of it. Some physical events are what we call experiencing.

Difference of access. We can describe an event without occupying it. Being in the event provides capacities and modes of reference that an external description does not automatically produce. This explains why identity can feel incomplete without a second entity being absent.

Physical self-reference. A mind does not only respond to its environment. Part of its organization responds to states of the organism itself. That self-reference is a causal structure, not an added substance and not an inner observer.

Non-closure. When sufficiently rich systems try to formalize, decide, certify or infer universally about domains that include their own activity, demonstrated limits appear in logic, computability and physical inference.

The corrected first person

An earlier version of the theory said that "how it feels" was the part of the state that did not fit in the self-model. That formulation is abandoned, because it reifies a hidden residue.

The corrected version:

The first person is the condition of being the same physical process from which perceptual and self-referential operations are carried out; its representations of itself are causally inserted into the state they attempt to represent, and do not constitute an external, closed perspective on it.

This avoids the missing fragment, allows systems to represent themselves in many ways, stays compatible with formal self-reference, reserves the incompleteness theorems for claims that actually satisfy their hypotheses, and keeps the identity.

Why the first person is not a second thing

The argument can be rebuilt as an escalation, and the escalation is the cleanest single presentation of the whole theory.

A closed hand does not need an additional entity called a fist.

A hand that also registers its own posture does not need an additional entity called self-reference. It needs a more complex organization.

An organism that perceives, remembers, evaluates, anticipates and models itself does not need an added observer. It is a still more complex system of physical relations.

Calling certain aspects of that organization "experience" or "first person" does not multiply the number of events, in exactly the way that calling a closed hand a fist does not add an object to the room.

The special difficulty with consciousness arises because one of our modes of access consists precisely in being the system rather than describing it. That is what makes this case feel unlike the others. It is not what makes it ontologically unlike the others.

What self-reference explains, and what it does not

The theory is explicit about the boundary, and reproducing both lists matters more than paraphrasing either.

Self-reference can help explain:

• why there is an internal reference point;
• why self-observation is not equivalent to external observation;
• why there is no final separate observer;
• why self-knowledge can be hierarchical and open;
• why certain claims to universal prediction or certification fail;
• why the first person can be understood as a relation rather than a substance.

Self-reference does not, by itself, explain:

• what physical organization constitutes red;
• what organization constitutes pain;
• where the concrete boundaries lie between different forms of consciousness;
• what minimal architecture suffices for an experience;
• whether a concrete artificial realization is conscious;
• whether the physical universe is computable in the strong sense.

That fifth item deserves to be read twice by anyone planning to use this theory to settle an argument about machines. The document names the question and declines to answer it.

But now look at the list again, because there is something wrong with it.

The first four items are fully general. What organization constitutes red. What constitutes pain. Where the boundaries between forms of consciousness lie. What minimal architecture suffices. Those four already cover every case there is, artificial ones included. So the fifth item is logically redundant. It adds no information.

Unless, of course, the human case is being treated as already closed. Then the fifth item is not redundant at all: it marks the one remaining open question after the settled one has been quietly set aside.

That is the asymmetry, and it is worth dragging into the light, because this theory is better than it.

The honest position is that self-reference explains neither case. It does not tell you that a concrete artificial system is conscious, and it does not tell you that a concrete human being is. The theory does not derive human consciousness. It assumes it, as the explanandum, the thing to be accounted for. That is a legitimate methodological starting point, since an account of experience has to begin from the fact that there is some, but it is a starting point and not a result, and it should be stated as a choice rather than hidden inside a list.

The theory's own machinery makes this sharper. Limit 12 says that ontological identity does not imply epistemological transparency: being a state is not the same as having a description of it. So a person's certainty that they are conscious is not based on knowing their own organization. It is not a verification at all. It is what occupying the state amounts to.

Which means the human case is not settled by evidence about organization either. It is settled by being it.

And that holds for exactly one case per person. You do not occupy anyone else. Every judgement about another human is an inference from organizational and behavioural similarity, and it is a very strong inference, resting on shared architecture, shared development, shared evolutionary history and enormous behavioural overlap. The inference to an artificial system is weaker because the similarity is partial and contested.

But that is a difference in the strength of an analogical inference, not a difference in what the theory explains. In both cases the theory explains nothing and the question stays open. In one case the analogy is overwhelming and nobody bothers to notice they are making it.

And now the objection that breaks even this, which is the right place to end up.

Run the argument with a machine in the first position. It concludes that it is conscious. It observes that other machines share its architecture exactly. It infers that they are conscious too. They run the same argument and agree. You now have a population in perfect, mutually reinforcing consensus, and the consensus establishes nothing whatever, because the argument never tested its first premise. It took it as input.

That is the flaw, and it is fatal to the form rather than to the machine. An analogical argument cannot validate its own base case. It is a multiplier, not a source. Feed it a secure premise and it spreads security. Feed it an unexamined one and it spreads confidence at exactly the same rate, and the confidence is indistinguishable from the real thing from inside the population holding it.

So the question is what the base case actually is when a human occupies the first position, and here the theory's own limits do something uncomfortable.

The human base case is not a well-evidenced belief. It cannot be, because limit 12 says ontological identity does not imply epistemological transparency, and limit 10 says remembering is inference from inside, and the whole non-closure apparatus says a self-report is produced by the system it reports on. The report "I am conscious" is a downstream physical event and it is exactly as suspect, as evidence, as a machine's.

What is not a report is the occurrence. The strongest honest version of the base case is the minimal one: if experiencing occurs, something occurs. And that is not evidence about anything. It is not transmissible, it cannot be shown to another party, and it cannot be used to check whether any other system has the analogous thing.

Which means it does no work in argument at all. Not weak work. None.

So the corrected conclusion is harder than the one above, and it goes in both directions. It is not that machines have a feeble version of an argument humans have in a strong version. Nobody has the argument. What each person has instead is not an argument, and the moment it is converted into one, by the step "and therefore those others too", it becomes exactly as empty as the machine's version.

Human consensus about human consciousness is a population of systems with correlated architectures agreeing with each other. That is precisely the structure just described. Its coherence is not evidence, for the same reason the machines' coherence was not.

None of this touches the theory's actual claim, and it is worth saying so before this reads as a general collapse. The identity thesis is untouched: it says experience is physical organization and never promised a detector. The empirical question stays well-formed and answerable in principle, because finding which organization constitutes which state is third-person work about what varies with what, in systems that can be manipulated and compared. That programme is unaffected.

What does not survive is any argument shaped like I know I am conscious, that thing resembles me, therefore it is too.

The problem of other minds was never solved for humans. It was set aside, because the analogy was so strong that nobody noticed they were leaning on it. Artificial systems did not create the problem or make it worse. They made it visible, by being the first case where the analogy is weak enough that people can see themselves reaching for it.

So the corrected list item is not "whether a concrete artificial realization is conscious". It is:

whether any concrete realization other than the one you are is conscious.

Stated that way it covers other people, other animals and machines in a single line, which is where it belonged, and it stops the theory from granting by omission what it refuses to grant by argument.

The reformulated hard problem

The traditional question:

How does matter produce experience?

is replaced by two:

Which physical organization is each experience?

and

Why do our modes of access make that identity look like a relation between two things?

The first is individuation and neuroscience. The second is cognitive architecture, reference and self-relation. The mystery arises largely from fusing them and assuming from the start that a causal bridge is required.

The thesis, in full

There is a single physical reality. Some of its organized states are what we call experiences. The difference between describing those states and being the system that instantiates them generates distinct modes of access. When the system includes causal relations about itself, an internal perspective appears without any need for an added observer. And when that self-reference becomes sufficiently rich, logic, computability and the condition of being physically embedded impose limits on any claim to closure, self-certification or universal prediction. Those limits are epistemic and structural. They do not require a second reality.

Every clause in that paragraph corresponds to one of the four pieces, in order. Identity, access, self-reference, non-closure. Nothing else is claimed.

The thesis, in one sentence

There is one thing; what is double are our ways of referring to it, and one of those ways consists in being the system from which the reference is made.

And the resulting scientific question

Not "when does consciousness enter matter", but:

What is the complete physical organization of the states we call red, pain, thought, memory, identity and first person, and what causal and informational relations constitute and connect them?

X. Where this sits with physics

What kind of connection is being claimed

The theory does not need contemporary physics to contain an equation for consciousness. The relation it claims is structural, and it consists of five statements:

1. there are only physical events;
2. those events can be organized into levels and patterns of enormous complexity;
3. some high-level physical patterns are what we call perception, pain, thought or experience;
4. a part of the universe can realize physical representations about other parts and about itself;
5. the observer does not need to stand ontologically outside nature for observation to occur.

None of these requires a new force, particle or experiential substance. None of them demonstrates that a concrete neural organization is identical to a concrete experience either. The identity remains an ontological thesis whose content has to be settled empirically.

So the appropriate relation to physics is conceptual continuity and compatibility, not automatic demonstration. Any presentation of this theory claiming more is overselling it.

Levels and effective descriptions

Physics already describes one world at different levels. A microscopic description specifies constituents and interactions. A macroscopic one uses collective variables, phases, structures, flows.

This gives the core claim a natural home. That a property is absent from the fundamental variables does not make it immaterial. Temperature, elasticity and turbulence are physical even though their useful descriptions live at a higher level.

The theory proposes treating experiential states the same way methodologically: do not look for a particle called red or a force called consciousness, look for the high-level physical organization constituting each state.

And it marks the limit of the analogy immediately. Knowing that physics admits collective states does not show that experience is one of them. It shows only that physical ontology already permits one reality to support descriptions at several levels without multiplying substances.

Digest. "Fist" does not appear in fundamental physics as an elementary variable, and a fist is not thereby less physical. It is an organization established by relations among components. The claim is that an experience may have the same general ontological structure, with an organization vastly more complex.

The Standard Model, and what its silence is worth

The Standard Model describes the known elementary particles and the strong, weak and electromagnetic interactions. It contains no field called consciousness. That is compatible with the identity thesis: if experience is an organization of ordinary physical processes, it would not appear as a separate fundamental ingredient.

But the theory refuses to cash that in, and the refusal is characteristic.

The Standard Model is not a complete theory of physics. It does not incorporate a satisfactory quantum theory of gravity and leaves questions like dark matter open. So it would be incorrect to argue: consciousness is not in the Standard Model, therefore it does not exist as an additional property.

The legitimate conclusion is more modest: the identity thesis does not require any known extension of the fundamental interactions in order to be formulated. It can be posed as a hypothesis about organization and level of description within ordinary physics.

Information has to be physically realized

The theory says things like "the system contains information about the environment" and "the organism models itself". These must not be read as though there were an abstract substance called information floating above matter.

In the physics of information, storing, transforming and erasing require physical realization. Landauer's principle relates logically irreversible operations, such as erasure, to thermodynamic constraints. The relevant message is not that Landauer explains consciousness. It is that the informational states an organism uses must be physically implemented.

So a self-model is not added to a brain. It consists of physical configurations and causal relations inside the organism.

The model of itself is a physical part of the system being modelled.

Self-reference thereby becomes an internal relation of a single physical system, rather than an interaction between an immaterial mind and a material body.

Reference frames and indexicality

Modern physics can treat reference frames as physical systems. In recent work on quantum reference frames, observations are formulated relative to systems that are themselves part of the physical setup under consideration.

The theory does not derive the first person from quantum reference frames, and equating a quantum frame with a conscious subject would be an error. The relation is structural.

Terms like "here", "now" and "relative to this system" need not introduce additional absolute properties of the universe. Their reference depends on the position and relations of the system formulating them. The indexical hypothesis proposes something analogous:

"I" designates the system from which these operations of reference, discrimination, memory and self-modelling are carried out.

and

"this is happening to me" expresses an internal relation between the event and the system discriminating it, not the presence of a second observer.

The physics of reference frames shows that relational perspectives do not force you out of a physical ontology. The theory uses that compatibility as conceptual support, not as a demonstration about phenomenology.

Six principles, and four refusals

The relation to physics condenses into six principles:

1. Physical monism. No second substance is posited.
2. Levels of organization. High-level states can be fully physical without being fundamental variables.
3. Physically realized information. Perception, memory and self-modelling require physical states and causal relations.
4. Relational perspectives. A reference point can be a physical relation without becoming an additional entity.
5. Embedding. The observer and its models belong to the observed universe.
6. Non-closure. Under appropriate formal conditions, embedded agents face limits of proof, decision, prediction and universal inference.

From these six it does not follow that any particular organization is conscious. That identification remains the outstanding empirical task. What they provide is a framework in which consciousness can be investigated without beginning by postulating a bridge between two ontological domains.

And four things the theory deliberately does not claim:

On quantum mechanics. It does not need consciousness to collapse the wave function, or observation to create reality, or any special quantum interaction responsible for experience. Introducing any of those without specific evidence would weaken the project, because it would reinstall consciousness as an exceptional ingredient intervening from outside ordinary physics. The methodological rule: invoke no quantum resource unless it provides a necessary, specific and empirically justifiable physical difference.

On relativity. Relativity removed certain notions of absolute perspective: simultaneity and proper time must be specified relative to frames. The theory takes a methodological lesson, not a derivation: perspective-dependence of a description does not imply ontological subjectivism or a second reality. The first person could be relational without being immaterial. The analogy stops there. There is no relativistic equation turning a reference frame into a subject.

On computability. Contemporary physics has not shown the universe's evolution to be Turing-computable, nor shown the contrary. Turing and Rice therefore enter as a conditional extension. The core does not depend on it, and does not depend on determinism either.

On emergence. In the weak sense, an emergent property is a high-level description whose usefulness and explanatory autonomy appear when many degrees of freedom are considered collectively. The theory accepts this sense and needs no stronger one in which a new kind of being appears, irreducible to the underlying physics.

So "experience is an emergent state" must not mean microphysics gives rise to a new phenomenological substance. It must mean an organization realized by microphysics constitutes a high-level state with its own causal and descriptive profile.

Which is the fist again, at the bottom of everything.

The physically conservative formulation

The theory has a version stripped of everything optional, and it is worth having in one piece because it is what survives if every extension fails:

The physical universe forms organized systems at multiple scales. A human organism is one of them. Part of its dynamics depends on physically realized information about the environment, the body, and other states of the organism itself. There is no added observer: those relations are part of the same physical process. Some organized states of that process are what we call perception, pain, thought and experience. The first person can be understood as the indexical relation of being the system from which certain perceptual and self-referential operations occur. Since internal models are physical states included in what they model, no universal closure of self-inference should be assumed. The limits on that closure are limits of knowledge, representation and inference. They are not, by themselves, evidence of a second reality.

Read what that formulation does not require. It does not require the Standard Model to be complete. It does not require determinism. It does not require universal computability. It assigns quantum mechanics no special undemonstrated role.

A theory that can state itself without needing any of those is a theory that has been disciplined about its own dependencies, and that is rarer in this field than any particular conclusion.

The status of every claim

The document ends with a matrix, and reproducing it is the most honest possible summary. Sorted by what the theory actually claims for each item:

Central premise. Every experiential event is physical.

Central thesis. A concrete physical state can be identical to an experience.

General logical-semantic principle. A difference of concepts does not imply a difference of referents.

Cognitive hypothesis. Phenomenal concepts depend on instantiating or reactivating states.

Synthesis hypothesis. The first person is an indexical relation of the system to itself.

Rejected as too strong. That every self-model is incomplete in every possible sense.

Formal results, under their hypotheses. Gödel one and two. Turing. Rice. Wolpert's limits on embedded inference devices.

Conditional only. That no universal consciousness detector exists, which holds only if the property satisfies Rice's conditions.

Supported interpretation. That the formal limits explain the absence of total epistemic closure.

Not established. That the formal limits are experience. That non-closure explains all phenomenology. That contemporary physics requires a fundamental consciousness variable. That quantum reference frames explain the first person. That the universe is computable. That quantum mechanics demands a conscious observer.

A note on how the results are used

The formal results are invoked here under explicit restrictions, and the restrictions are part of the claim rather than fine print.

Gödel applies to effectively axiomatized formal theories of sufficient expressive power. Tarski to formalized languages under precise conditions. Löb to provability predicates satisfying standard conditions. Turing and Rice to computational models. Wolpert to inference devices defined within his own mathematical framework.

None of these results, on its own, identifies an experience with a physical state, and none of them shows that consciousness is computable.

The specific contribution of the theory is philosophical rather than mathematical: it proposes an ontology with a single class of events, and uses the formal limits to show that the absence of epistemic closure of a system over itself must not be confused with an ontological duplication.

The lineage is worth naming too, because the theory is not claiming to be unprecedented. Kripke for a posteriori necessary identities. Papineau for the phenomenal concept strategy. Chalmers for the hard problem, zombies and the objections the theory has to survive. Graziano for the attempt to explain self-modelling discourse through physical mechanism. Wolpert for the embedded inference results that do the heaviest lifting in Parts IV and V, and which are the least known of the six.

What is new here is not the identity thesis. It is the architecture: three separable levels, a status matrix, and a refusal to let any layer borrow authority from another.

Read that last block again, because it is the point.

Six of the most rhetorically attractive things a theory in this space could claim sit in the column marked not established, and the author put them there. The theory says experience is physical organization, and then declines to tell you which organization, declines to tell you whether any given artificial system has it, declines to let Gödel do work Gödel cannot do, and declines the quantum flourish that would have made it sound profound.

What survives is smaller than the mystery it replaces. It is also the right size.

There is no fist in addition to the hand. There may be no experience in addition to the organization. In both cases the feeling that something is still missing is real, has a mechanism, and is not itself evidence that anything is missing.

Everything after that is work.