Philosophy → Folding

Folding: Reality as a Self-Referential Address

From homeostasis in metabolism and reward to the fold, the address, and the up-spiral.

Marin Kokona · Working paper · September 2026 · Nothing here is claimed as proved

Abstract

Natural Process argued that balancing is what nature does, and left open what living systems balance toward. I answer with homeostasis: balancing against a stored setpoint. Through reward and insulin I show that the setpoint is physical, and that it can drift. I then follow the same form upward. Folding, the compression of structure into a recoverable seed, appears in fiction, algorithms and biology. I argue that physical limits on folding are not a flaw in the idea but the reason it is needed, that every definition (the number e is the example) is a fold of an unfinished, self-generating structure, and that what any definition leaves out is more than what it holds. A fold is then best understood as an address into a full space of possibilities, read by nothing but itself, which is homeostasis at the level of reality by analogy rather than by proof. Two consequences follow: the "reset" is a projection artefact of an up-spiral, and paradoxes generally dissolve when the missing dimension is added back. The argument explains continuation but not felt experience, and stops honestly at the hard problem. The result is offered as a modelling stance, not a proof.

Keywords: homeostasis · setpoint · folding · self-reference · holographic bound · hard problem

1Introduction: the question Natural Process left open

The Natural Process page [1] argued that balancing is what nature does. Heat flows from hot to cold, ions settle into salt, and the pancreas releases insulin in proportion to blood glucose. It ended with three questions it could name but not fully answer:

The monitor. If every step in the insulin cascade is plain chemistry, what makes the whole look purposeful? The page's answer was self-reference: the system models its own state and acts on the discrepancy. But a model of one's own state compared against what?
Equilibrium. Prigogine's warning [5] was that perfect equilibrium is death. Living systems balance, yet they stay far from equilibrium. So what exactly are they balancing toward?
The mind. At what point does balancing become awareness?

This paper answers the first two with one word, homeostasis, and then follows that word much further than I expected it to go. It turns out to run through metabolism, through habit, and finally through a picture of reality itself as a folded, self-reading structure. The third question is not answered. Section 8 marks exactly where the argument stops.

The path, in one line: balancing becomes homeostasis once there is a setpoint; a setpoint is stored structure; stored structure is a fold; and a fold that reads itself in whatever way keeps it unfolding has the same form as homeostasis, one level up.

Nothing here is claimed as proved. It is a structure I find coherent, laid down one base at a time and walked to the edge where the questions run out. Where the argument makes empirical contact (neuroscience, metabolism, algorithms, physics) I have tried to state that contact accurately and say where it is contested.

2Homeostasis: balancing toward a setpoint

2.1Balancing is not yet homeostasis

Sodium and chloride balance toward equilibrium. Nothing in the reaction stores where it should end up. The end state is simply wherever the forces cancel. The pancreas is different. It returns blood glucose to a reference value and defends that value against disturbance in both directions. Cannon named this homeostasis [2], and Jason Fung made the word central to his account of metabolism [3].

That difference answers the questions Natural Process left open:

Answer to the monitor question

The self-reference Natural Process pointed to is concrete: it is a setpoint, a stored piece of structure that the system compares itself against. The "global monitor" is not a separate supervisor. It is the reference value, written into receptor densities, channel thresholds and hormone sensitivities, and read continuously by the chemistry it governs.

Answer to the equilibrium question

A living system does not balance toward equilibrium. It balances toward its setpoint, which is a steady state held far from equilibrium and paid for with energy. This is exactly Prigogine's point [5]: balancing at one level sustains disequilibrium at the level above.

The information-content distinction on the Natural Process page falls out naturally. NaCl formation processes zero bits because there is no reference to compare against. The insulin response processes a continuous variable because it carries one.

2.2When the setpoint moves

Once balancing has a reference, a new failure becomes possible. The system can balance perfectly toward the wrong target. Two ordinary cases show the same architecture.

Reward. Once a reward has been learned, dopamine neurons fire at the cue that predicts it, not at the reward itself. A fully expected reward produces almost no response, and a missed one produces a dip [6]. Dopamine tracks anticipation, not arrival. A reward loop designed never to close, with always another level, rank or unlock, keeps anticipation running indefinitely. An unclosed loop is felt as a debt, and a debt as a duty.

The real question is why anyone keeps refilling. The answer is that a repeated flood raises the baseline, so that ordinary life registers as a deficit. You refill to escape lack, not to gain pleasure. In addiction research this shift of the reference point is called allostasis [7], and its physical form includes reduced dopamine-receptor availability, which partly recovers with abstinence [8].

Metabolism. The pancreas from Section 2.1 shows the same drift. Chronically high insulin blunts the cells' response to it, and the body answers by secreting still more: insulin resistance [4]. The system is still balancing faithfully, only now toward a reference that has moved.

On the evidence

Receptor down-regulation is well established for drugs of abuse [8]. How far behavioural rewards such as games produce the same change is still debated. The argument here needs only the architecture, a defended reference that can drift, and not a claim that the two cases are equal in magnitude. The popular "dopamine fast" is best read the same way: it removes the cue, not dopamine itself.

The practical side of both setpoints, how insulin and reward sensitivity are actually restored, is covered on my health site: Health for metabolism, and Dopamine and Gaming for reward.

2.3What actually moves a setpoint: articulation

Facts that are already known often do nothing. What makes them land is precise articulation: seeing the mechanism stated so clearly that it can be held in one piece. The lever is clarity, not emotional charge. It is the difference between unindented code and well-organised code. The logic is identical, but one is maintainable and the other is unusable.

For most people, clarity and motivation are separate systems. For me, clarity is sufficient motivation. That is an uncommon signal, not a general rule. It is also, I will argue in Section 9, a small instance of the paper's main idea, because a clear model is a good fold.

Metabolism and reward therefore reveal a law about defended reference points. The rest of this paper asks how far that law can be followed.

3Three observations of folding

The word reached me first through ten-fold cross-validation, where a dataset is split into ten folds and each is held out in turn [9]. There, a fold is simply a part. That is partition, not compression, and it is not what this paper means. The idea itself was sparked by something else, and I then found the same move in two unrelated places.

The Sophon. In The Three-Body Problem [10] a single proton is unfolded from higher dimensions into a surface the size of a planet, etched with circuitry, and folded back to subatomic scale. It is a planet-sized computer folded into nothing, with its function intact. This is the image that started the paper.
Union-find with path compression. On naive trees, a sequence of $n$ unions and finds can cost on the order of $n^2$ steps. Path compression, together with union by rank, folds each tree flat as it is traversed. Tarjan showed that this brings the amortised cost per operation down to $O(\alpha(n))$ , where $\alpha$ is the inverse Ackermann function [11]. That function stays below 5 for any $n$ that fits in the physical universe, so a deep structure is folded to almost nothing.
A seed. An entire plant's life is folded into something tiny. Add water and time and it unfolds into the whole organism.

Fiction, algorithms and biology all show the same move. That pervasiveness is what made me suspect that folding is not a metaphor borrowed between fields, but an operation reality itself performs.

Definition (working). To fold is to compress a structure into a smaller form from which the original can be recovered by unfolding. A fold is good to the degree that it preserves the structure that matters.

4The claim, and its limits

Claim 1

Folding is a fundamental operation of reality. The Sophon is less unrealistic than it sounds: we judge it impossible because we reason in everyday terms of space and time, but at the quantum scale and near a black hole those are not fixed limits.

Gravity already folds matter far beyond intuition. Compressed within its Schwarzschild radius, the Sun would be about six kilometres across, and the Earth would be smaller than a marble, about 18 mm. Reality folds prodigiously. Two objections, both accepted, sharpen the claim.

Objection 1: crushing is not preserving

A black hole folds by destroying accessible structure. The Sophon folds while preserving functional structure. These are opposite problems, and only the second is interesting here.

Objection 2: there are floors

"Anything can be folded infinitely small" overreaches. The Planck length, about

\;1.6 \times 10^{-35}

m, and the Bekenstein bound [12], which caps the information a region can hold in proportion to its size and energy, set real floors.

4.1Why the floors matter

These floors are real, and they are not a flaw in the idea of folding. They are the reason for it. The information a region can hold is finite, and the holographic bound says it scales with the area of the region's boundary, not its volume. If memory were infinite, nothing would need folding.

Claim 2: the limit is why folding is needed

Folding is folding with structure preserved, inside whatever limits exist. The limits do not argue against folding. They are what make it necessary.

My thinking about reality beneath its appearance owes something to Donald Hoffman's interface theory of perception [13, 14]. I credit it for that; no argument here rests on it.

So the frontier is folding with retrievable structure preserved. The seed and the path-compressed tree do this. The black hole does not. The next section asks what the thing being folded is, if it is not physical.

5The substrate question

The whole line of thought comes back to one question: does structure need a place to stand? Does information require a physical substrate, or can it exist free of any location?

Premise (not proved)

Suppose structure is primary. The world is then a program: pure structure, which has no physical size and so needs no place and no time in which to be stored. The matter we see is a rendering of that structure. I do not prove this here. Wheeler and Tegmark argue versions of it properly [15, 16]. What follows is what the premise buys.

5.1A symbol is a fold: the number e

The clearest example I know is a number we treat as finished:

e \;=\; \lim_{n \to \infty} \left(1 + \frac{1}{n}\right)^{n} \;=\; \sum_{k=0}^{\infty} \frac{1}{k!} \;=\; 2.71828\ldots

We write a single letter, and it looks like a thing that is done. It is not. Its digits never end and never repeat. The number is transcendental, so no finite algebraic recipe produces it either [17]. What the symbol actually names is a process, a procedure that can always be run one step further and is never complete. The letter is the fold, and the never-ending expansion is what it unfolds into [18].

And the structure is self-generating. The function

e^{x}

is the only function, up to a constant factor, that is its own derivative:

\frac{d}{dx}\, e^{x} = e^{x}

This is why e appears wherever something grows in proportion to what it already is: compound interest, populations, radioactive decay. It is the signature of a structure that produces more of itself from itself.

This is the general point. What we define in our reality, whether numbers, objects or laws, is a short name for a complicated, self-generating structure that never finishes. We hold the fold. The unfolding is never done.

5.2Nothing is fully defined

If every definition is a fold of something unfinished, then every definition leaves something out. Every theory, however good, leaves something out as well. What is left out is not one stray item. It is more than everything inside the definition: the rest of the infinite thing.

Mathematics gives this a precise form. A definition is a finite string of symbols, so there are only countably many definitions. Cantor showed that there are uncountably many real numbers [19]. So almost every real number can never be named by any definition at all. Whatever our definitions capture, what they miss is strictly larger. Gödel found the same shape inside a single formal system: any consistent system rich enough for arithmetic leaves true statements it cannot prove.

5.3Fullness, not incompleteness

I do not contradict Gödel. I go up a level. Reality is not one consistent system but the full space that holds all of them: every branch and its negation, the one and the not-one, side by side. Each definition, theory or system is incomplete. The space that contains them all is not.

Call it the space of possibilities. All branches already exist, in the way that e has all of its digits whether or not anyone computes them. The question is then not what builds a branch, but which branch you are in.

6The address, and who reads it

6.1The fold as an address

Claim 3

The fold, meaning the seed or the compressed thing, is an address into the space of possibilities. It is not physical and not a number. Thinking it must be a coordinate in space and time is the same mistake as thinking the Sophon needs room to exist. The address is infinitely small and at the same time an infinite description, in the way that the single letter e names an expansion that never ends.

"Infinitely small" does not collide with the floors of Section 4. Those floors limit what physical matter can hold, and the address is not physical.

A life, everything that is experienced, is that address unravelling. Living is the fold opening: the seed becoming the plant, the proton unfolding into the planet-sized computer.

6.2The fork: determinism or steering?

If a life is an address unravelling, the obvious objection is determinism. The plant is already in the seed, so nothing is chosen and you only watch it read out. The alternative, that intention jumps you to a different address, owes an answer to the question of what does the jumping.

I dissolve the fork rather than choose a side. The address is neither read out mechanically nor steered from outside. It is read in a way favourable to its own unfolding: the reading sustains itself, keeps opening and keeps going.

This also gives a precise meaning to an old phrase. Eternal life is not a duration on a timeline. That would be the bounded-circle picture of a loop again. It is the unfolding that never finishes. We picture a "loop" as a circle with limits because we think in space and time. A real loop is infinite: an unfolding that keeps generating more of itself without closing, as e^x does.

6.3Favourable to what, and chosen by what?

Claim 4: the reader is the address

There is no external reader and no substrate coming back in through the door. The reader of the address is the address, a self-referential structure that reads itself in whatever way keeps itself reading. "Favourable" simply means continuity.

A biological precedent

This is less exotic than it sounds. Von Neumann showed that a self-reproducing machine needs a description together with a constructor that reads it [20]. Life solved this by writing the constructor into the description itself: the genome encodes the ribosomes and polymerases that read the genome. The reader is part of the address.

Cost, acknowledged

This buys continuity without intention. It is a selection principle with no chooser, close to anthropic reasoning [21]: only self-sustaining patterns persist, so the persisting patterns are self-sustaining. That borders on tautology, and it gives up any steering wheel. Section 7 looks for where steering can come back in.

6.4Is this homeostasis?

A structure that continues because continuing is what it does needs no wanter and no chooser. It is tempting to call this homeostasis at the level of reality, but the claim has to be stated carefully. Homeostasis in Section 2 has two parts: a setpoint, and correction of errors against it.

At the scale of reality, the only candidate setpoint is continuation itself. Unfoldings that keep unfolding persist, and the rest stop. That has the same form as homeostasis, a reference defended by the structure that carries it. What is missing is a demonstrated mechanism of error correction. So I hold it as a strong analogy, not as an identity.

Homeostasis runs through metabolism and reward as a mechanism, and through reality itself as an analogy I find hard to resist. It is also what Natural Process was reaching for when it named self-reference as the hidden variable.

7Against the reset: the up-spiral

In Memento [22], Leonard solves his crux and then deliberately plants the next false lead, so the search starts again from zero. A common reading is that this is simply what happens once the crux is solved: you press the reset button.

Position

Reject the reset. The error is the assumption that continuity has to return to zero.

The up-spiral. A helix, not a circle. Nothing is wiped, everything accumulates, and each pass sits higher than the last. Memento is the flat circle.
The reset is a projection artefact. It is what a higher-dimensional spiral looks like when it is viewed along its axis (Figure 1).
No privileged zero. To see only zero and infinity is to look at half the spectrum. The full range runs from negative infinity to positive infinity, and without a privileged zero there is nothing to reset to.

Figure 1. One helix, two projections. Viewed along its axis, the path closes on itself and every turn looks like a return to zero. Viewed from the side, nothing is ever revisited. The reset belongs to the viewpoint, not to the path.

7.1Where choice enters

Section 6 gave up the steering wheel. It comes back, in a modest form, at the breaks. Continuity has dead zones: interruptions, forgetting, gaps between one reading and the next. Choice lives there, in picking the path along which the important parts make it through. Reading back what was said so that it is encoded better is a small, concrete example.

7.2A caution: continuity includes others

Re-encoding an idea makes it one's own in a true and modest sense. It does not license the stronger claim that the world is one's own creation. "Favourable to continuity" has to count other people as real continuers too.

8The honest open edge

This is the third question from Section 1, and the model does not answer it. Homeostasis explains continuation, meaning why self-sustaining unfoldings persist. It does not explain why there is anything it is like to be one. A thermostat regulates temperature and, as far as anyone can tell, feels nothing. Why is the self-maintaining structure experienced rather than running dark? This is the hard problem [23].

One attempt to dissolve it was that "inside" and "outside" are spatial words, and space has already been set aside, so the question is malformed. That was a legitimate strike, but the question survives being restated without spatial language: why is there felt experience at all, rather than none?

Hofstadter argues that a self-referential loop is what an "I" is [24], and the self-reading address is exactly such a loop. That accounts for the self. It does not, by itself, account for the felt-ness.

The honest claim is that we walked to the edge of the hard problem, not that we solved it. It is not a summit with a flag. It is a mirror, the point where the thinking turns and sees itself thinking. I am the executed program asking whether programs need executing. The question folds back into the asker.

9Conclusion: a modelling stance

Put simply, the master move is to ask why we are looking at it as a two-dimensional circle. Every paradox met along the way (the reset, the fork, the zero) was an artefact of viewing a higher-dimensional thing in too few dimensions. Folding is the movement between dimensional views.

We never explain the whole. We explain a model. A model is a deliberate projection, so explaining through a model is itself an act of folding. That is why precise articulation moves setpoints (Section 2.3): a clear model is a fold that preserves the structure that matters.

So the result is not a finished cosmology. It is a stance:

Don't flatten to the circle.
Don't smuggle in the zero.
When a paradox appears, add the dimension back.
Hold every model as a fold, never as the whole, because what it leaves out is more than what it holds (Section 5.2).

Understanding stays forever unfinished, and each model is a release.

I arrived at these ideas from the particulars above, but several of them have serious precedents. They are listed here so the reader can check the argument against them.

The fold in philosophy. Leibniz's monads each contain the whole universe from one point of view [25], which comes close to an address that is infinitely small and yet a complete description. Deleuze made the fold the central figure of his reading of Leibniz [26].

Enfolded order in physics. Bohm distinguished an implicate (enfolded) order from the explicate (unfolded) order we observe [27]. That is close to "matter is a rendering of structure", in the vocabulary of folding itself.

Structure without substrate. Wheeler's "it from bit" makes information prior to matter [15]. Tegmark's Mathematical Universe Hypothesis holds that physical reality is a mathematical structure, and that all such structures exist [16]. That is the most formal version of the "fullness" in Section 5.3.

Spacetime as interface. Hoffman's interface theory of perception holds that spacetime and physical objects are a species-specific user interface rather than the structure of reality [13, 14]. I credit it for shaping how I think about reality beneath its appearance; no argument in this paper rests on it.

The undefinable. Cantor's diagonal argument [19] and Gödel's incompleteness theorems are the precise ancestors of Section 5.2: whatever a system of definitions captures, it leaves out more.

Homeostasis and allostasis. The term is Cannon's [2]. The moved setpoint of Section 2.2 corresponds to allostasis in the addiction literature [7]. The far-from-equilibrium framing is Prigogine's [5], as discussed in Natural Process.

Self-reference and selection. Von Neumann's self-reproducing automata [20], Hofstadter's strange loops [24] and the anthropic principle [21] are each one face of "the reader is the address".

References

M. Kokona. "Natural Process." Philosophy, this site, 2026. link
W. B. Cannon. The Wisdom of the Body. W. W. Norton, 1932.
J. Fung. The Obesity Code: Unlocking the Secrets of Weight Loss. Greystone Books, 2016.
A. M. Freeman, L. A. Acevedo and N. Pennings. "Insulin Resistance." StatPearls, NCBI Bookshelf, 2023. link
I. Prigogine and I. Stengers. Order Out of Chaos: Man’s New Dialogue with Nature. Bantam, 1984.
W. Schultz, P. Dayan and P. R. Montague. "A neural substrate of prediction and reward." Science 275 (5306): 1593–1599, 1997.
G. F. Koob and M. Le Moal. "Drug addiction, dysregulation of reward, and allostasis." Neuropsychopharmacology 24 (2): 97–129, 2001.
N. D. Volkow, G. F. Koob and A. T. McLellan. "Neurobiologic advances from the brain disease model of addiction." New England Journal of Medicine 374: 363–371, 2016.
M. Stone. "Cross-validatory choice and assessment of statistical predictions." Journal of the Royal Statistical Society B 36 (2): 111–147, 1974.
Liu Cixin. The Three-Body Problem. Trans. Ken Liu. Tor Books, 2014 (Chinese original 2008).
R. E. Tarjan. "Efficiency of a good but not linear set union algorithm." Journal of the ACM 22 (2): 215–225, 1975.
J. D. Bekenstein. "Universal upper bound on the entropy-to-energy ratio for bounded systems." Physical Review D 23 (2): 287–298, 1981.
D. D. Hoffman, M. Singh and C. Prakash. "The interface theory of perception." Psychonomic Bulletin & Review 22 (6): 1480–1506, 2015.
D. D. Hoffman. The Case Against Reality: Why Evolution Hid the Truth from Our Eyes. W. W. Norton, 2019.
J. A. Wheeler. "Information, physics, quantum: the search for links." Proceedings of the 3rd International Symposium on Foundations of Quantum Mechanics, Tokyo, 1989.
M. Tegmark. "The mathematical universe." Foundations of Physics 38 (2): 101–150, 2008.
C. Hermite. "Sur la fonction exponentielle." Comptes rendus de l’Académie des Sciences 77, 1873.
L. Euler. Introductio in analysin infinitorum. 1748.
G. Cantor. "Über eine elementare Frage der Mannigfaltigkeitslehre." Jahresbericht der Deutschen Mathematiker-Vereinigung 1: 75–78, 1891.
J. von Neumann. Theory of Self-Reproducing Automata. Ed. A. W. Burks. University of Illinois Press, 1966.
B. Carter. "Large number coincidences and the anthropic principle in cosmology." In IAU Symposium 63: Confrontation of Cosmological Theories with Observational Data, 1974.
C. Nolan (dir.). Memento. 2000.
D. J. Chalmers. "Facing up to the problem of consciousness." Journal of Consciousness Studies 2 (3): 200–219, 1995.
D. R. Hofstadter. I Am a Strange Loop. Basic Books, 2007.
G. W. Leibniz. The Monadology. 1714.
G. Deleuze. The Fold: Leibniz and the Baroque. Trans. T. Conley. University of Minnesota Press, 1993 (French original 1988).
D. Bohm. Wholeness and the Implicate Order. Routledge, 1980.

← Natural Process

Philosophy index →