46 KiB
Table of Contents
- The Logic of the Tripartite Synapse Model — v5
- PART I — Why This Is Not a Model but a Way of Making Models
- PART II — The Unifying Principle
- PART III — What the Physics Is, and How It Works
- 1. Locality — The Only Thing That Exists Is a Local Component
- 2. The Ring — One Act in Three Phases
- 3. The Two Turnings — Day and Night
- 4. The Timescale Ladder
- 5. Scarcity Decides — Collaboration by Day, Competition by Night
- 6. Causation Circulates — Emergence Up, Constraint Down, Command Nowhere
- 7. The Four Operations — How the Local Is Multiplied Into a Whole
- Coda — The Seven as One, and the Why Beneath Them
include_toc
| include_toc |
|---|
| true |
The Logic of the Tripartite Synapse Model — v5
A synthesis of the principles the pseudocode enacts. The document is ordered why → what → how: it opens with why this is a different kind of object than an ordinary model (Part I), states the single principle its content obeys (Part II), then descends through seven categories that specialize that principle (Part III). The why comes first because it is the reason everything else matters — without it, a reader could take the categories for a description of a synapse and miss that they describe a physics that writes itself.
What changed in v5. The old "evaluation" phase is retired — it was always preparation aimed at the other scope. The ring is recut into three categories: ACTION, RECOVERY, PREPARATION. The obsolete subject-mapping (lateral/local/vertical) is dropped. New findings are folded in: the rhythm is (ACTION ⇄ RECOVERY) × many, then PREPARATION; every category spans all timescales; night PREPARATION replays the day ACTION with the same machinery; build and release compete within a component while material competes between components; there are two independent forgettings; collaboration by day versus competition by night follows from the rivalry of each scope's currency; behavior is legible and meaning is assigned by the reader not the signal; and the three categories are the three modulable dimensions of behavior. Nine categories are consolidated to six, a seventh is added (the four operations), and — new in this revision — the "why" (formerly a closing note) is corrected and promoted to the front as Part I.
PART I — Why This Is Not a Model but a Way of Making Models
Before the principles, one question: what kind of object is this? The answer is unusual, and it governs everything that follows. This is not a model you can write down and run. It is a generator of models — a rule that turns each history into a different fixed model, and only once that history has been lived. History is not a variable inside the model; history is the model. This part earns that claim, because stated cold it sounds like mysticism, and it is not — it is a checkable fact about what the coupled components do.
The pseudocode is a physics written in the grammar of an algorithm. The companion pseudocode
reads like a program — assignments, conditionals, loops — but every line leans on something code
cannot supply. Its primitives — the calcium influxes, the fluctuations, the clearances — name
physical processes, not computations; mini_Ca() is a placeholder for "whatever the matter does
here." Every ·Δt is a differential equation in disguise: the discrete step is our notation, the
thing itself is continuous. And every coincidence — the three-way gate, the tag, the build — assumes
its inputs are present at the same instant at the same place, which the physical cleft supplies for
free by diffusion but which an if can only presuppose. The imperative grammar is a transcription;
the content is a dynamical system. The pseudocode is faithful to the model exactly where it is
unfaithful to computation.
The natural objection: surely it can still be simulated. Nothing here is non-computable in principle. The dynamics are differential equations with thresholds, which computers integrate routinely. If "implement" means "numerically approximate a trajectory," computation suffices. This objection is correct as far as it goes — so the question is what happens when you try to act on it.
A first answer that is true but philosophical: the simulator occupies the vantage the model denies. The model's content is that there is no global state — no component reads another's interior, no place holds the whole, holism is enacted and never encoded. But to compute the system you must hold every component's state in one memory and step them in one loop: the simulator is the forbidden global observer. To order the updates it needs a scheduler (a central order-giver) or a synchronous clock ticking all components together — the "command from above" that "causation circulates, command nowhere" denies. And it must count time as a variable, where the model insists time is suffered — read off the decay of stores, kept by forgetting. So a computed simulation gets the trajectory right and the ontology backwards. This is real, but on its own it can be waved away as metaphysics. The decisive answer is concrete.
The decisive answer: there is no one model to simulate — only a way of making models. Compare two cases.
Where simulation works — pricing a financial option. You have one fixed model: a stochastic equation with fixed parameters, the same rule on day 1 and day 200. You run 100,000 random price paths through that same equation. Each path differs, but all are samples of one stationary object — the fixed distribution the equation defines. Average the payoff over them and it converges: 100,000 paths give a good estimate, 200,000 barely move it. It works because the paths are variations on a single system — noise around a stable structure. History matters within a path but never changes the model; every path runs the same equation. The model is one object; the paths are its samples.
Where the same recipe breaks — this model. Take the four steps in turn. (1) There is no one fixed model. The equation is not the same on day 1 and day 200: night 1 rewrites it into a new equation, night 2 rewrites that. Each path runs a different, self-modified equation by day 10 — there is no fixed rule to sample from. (2) The paths are not variations on one system; they are different systems. In option pricing, two paths are the same stock behaving differently. Here, the path where synapse X won an early material competition and grew, and the path where its neighbour Y won instead, have physically different structures — different synapses exist. They are not two runs of one model but two different models a shared early history produced. (3) There is no center to converge to. The average final price is a real thing; the "average" of X exists, Y pruned and Y exists, X pruned is not a valid configuration — it is a blend of two incompatible circuits, corresponding to no possible state. (4) More samples stabilize nothing. More option paths tighten the estimate; more runs here yield more distinct circuits, never a better estimate of one, because there is nothing for them to estimate.
In one line: in Monte Carlo, history varies within a fixed model, so samples estimate the model; here history is the model — each history builds a different system — so there is nothing the samples jointly estimate. That is the precise content of "there is no one model, only a way of making models." The pseudocode is not a model you sample; it is a generator of models, one per history, knowable only once the history is complete.
And Monte Carlo is not the only rescue that fails — every acceleration method fails, for the same reason. Each general way to compute a system faster than living it out relies on some stable invariant to exploit, and this model, by construction, holds none.
- Closed-form solution needs the future to be a computable function of time; here it is a function of the whole history — no formula takes a path as input and skips it.
- Coarse-graining / renormalization (physics' strongest tool, and tempting given the fast-day / slow-night split) needs the fast variables to settle, at fixed slow parameters, to a stationary average the slow dynamics can see. But the day's dynamics never settle history-independently — which patterns can fire depends on structure built by every prior night — and the coupling is bidirectional and same-order: the slow change is made of specific fast events (which pattern replayed), not their average. Coarse-graining discards exactly the individuating detail the model consolidates. The micro-detail here is the signal, not the noise.
- Dynamic programming / memoization needs state recurrence to cache and reuse; irreversible ratcheting (energy spent, structure pruned) means no configuration is ever revisited — nothing repeats, so nothing can be cached.
- Surrogate / learned models need cross-history regularity to generalize; the histories are incommensurable individuals with no shared structure, so there is nothing to learn that is cheaper than running the history.
Every method needs one of: time-parametrizability, scale separation with stationary fast statistics, state recurrence, or cross-history regularity. This model has none — it is history-parametrized, its fast and slow are same-order coupled, it never recurs (irreversible ratchet), and its histories are incommensurable. The methods do not fail by bad luck; each needs the stable, reusable structure that "the specification is continuously rewritten by its own running" abolishes.
And here the exponential appears — not as the obstruction, but as its price. Suppose you refuse all of the above and insist on simulating anyway. To simulate is to fix a structure: a simulation is a set of variables updated by fixed rules, and you cannot write the loop without committing to what the variables are. But the real structure changes every night. So you face a forced choice. Freeze one structure and you have committed to a single branch — one accidental history, a measure-zero sample of a thing that is not a distribution. Stay faithful while keeping a fixed substrate and you must instead carry every structure the system might occupy as an enumerated set — and that set multiplies each night, growing exponentially in the number of nights, of changing dimension, non-factorable. This exponential is not a property of the model; the model never enumerates, it simply becomes one structure. The exponential is the shadow the fluid, self-rewriting model casts on a fixed substrate — it arises if and only if you demand the stable structure that simulation requires. The need for stable structure is what converts self-rewriting into exponential enumeration; drop the demand and the exponential vanishes, leaving only a physics living one history.
Three concrete faces of the obstruction. The foreclosed synapse: a synapse pruned on night 3 is gone; a pattern that would have used it on night 50 breaks at that link and cannot replay, so its downstream components lose participation and drift toward pruning too — one cheap early pruning deterministically forecloses a family of patterns fifty nights later, and you cannot know night 50's structure without having run nights 3–49 in order. The two histories that never reconcile: run from the same start twice; because material is conserved and structure capped, X-growing starves Y, and by night 20 the runs have disjoint sets of synapses — not noisy versions of one answer but two incompatible circuits with no meaningful average. No shortcut: because each night's structural change feeds the next day's dynamics feeds the next night's change, with no scale separation to exploit and no recurrence to cache, the one honest trajectory must be computed night by night, in order, in full — it is its own shortest description. The only way to know the state at night N is to run all N nights.
Why this is one insight, not several. The deep cause is that the model abolishes the separation between program and data. Structure (the equations) is built from the accumulated traces of behavior; behavior runs on structure. Night turns data into program; day turns program into data. There is no stable specification anywhere, because the specification is continuously rewritten by its own running — which is just "holism enacted, not encoded" and "no global state," seen over time. A computation requires the split: the program is, by definition, the stable part. A thing with no stable program cannot be captured by one.
What the physics does instead — and why the synapse is its own faithful implementation. The physical synapse escapes all of this not by being non-computable but by never needing an invariant. It does not compute which structure obtains tomorrow; it becomes it, by undergoing its night. It realizes exactly one history in real time — the real one, not a sampled one — needing no global memory (each component holds only its own state), no scheduler (time sequences everything at once, everywhere, for free), no counted clock (its stores keep time by decaying). So the faithful implementation of this model is not a program but a material: something that, by its own constitution, undergoes these dynamics with locality, simultaneity, continuity, and suffered time, without a controller. You can compute a life — one honest history, in full, in order, incompressibly — but never the model, because there is no "the model": there is a rule that makes one model from each history, and the synapse is the matter that runs that rule by being it.
Two honest limits. This says faithful acceleration is impossible, not that useful approximation is — a coarse model can teach you things, it just would not be this model. And it holds for the model as specified (irreversible, non-recurring, individuating); whether real neural tissue is secretly more regular, with statistics one could exploit, is an open empirical question, not something these principles can foreclose.
Everything below is what this self-writing physics is (Part II) and how it works, category by category (Part III).
PART II — The Unifying Principle
Watch one presynaptic bouton for a day and a night. By day it releases neurotransmitter, restocks its vesicles so it can release again, and — in the quiet after a burst — stocks a trace that records how much this release mattered. By night it does the same three things at a slower tempo: it changes its structure, restocks the material to change again, and replays the release as a probe to measure whether the change is still warranted. Nothing supervises it. It reads only its own state and the signals that reach it. What we call the synapse, the neuron, the memory, the organism is nowhere inside the bouton — it is only the name we give to many such boutons, coupled.
That is the whole model in one instance. Stated generally:
There is only the local component and its one repeating act. Everything we call a system — the synapse, the neuron, the assembly, the organism — is that act, multiplied and coupled, and described from outside. The act has one shape (act, recover, prepare) run in two directions (outward by day, inward by night), and the relations between components are set by what is scarce. Holism is real, but it is enacted by the coupling, never encoded in any part.
This is why the model is a generator rather than a fixed object (Part I): because the specification is never encoded in any part but enacted by the running, it is rewritten by that running, so no fixed model exists — only the rule and the history.
Every category in Part III is this principle, turned to face one question: What is a component? (locality), What is its act? (the ring), What are its two directions? (the two turnings), At what speeds does it act? (the ladder), How do components relate? (scarcity), Who is in charge? (causation — no one), and By what operations is the local multiplied and coupled into a describable whole? (the four operations). None adds a new assumption; each specializes the one above.
A note on language. This document does not say "the system." There is no system — only local components, contextualized by their neighbors. Where the phrase appears, it is inside quotation marks, naming the thing we are denying: an actor that stands above the parts, holds the whole, and acts on it. No such actor exists here.
PART III — What the Physics Is, and How It Works
1. Locality — The Only Thing That Exists Is a Local Component
Everything the model contains is a local component: the bouton, the spine, the astrocytic process, the dendrite, the soma, the axon. The actors we call higher — neuron, astrocyte, organism — are not additional things. They are descriptions of many components' coupled activity, spoken from outside. This is the direct reading of the unifying principle, and the rest of the category is its mechanics.
A component reads only its own state and the signals that arrive. It cannot read another component's interior, and it cannot read "the whole." When the bouton needs to know whether its release reached a responsive target, it does not inspect the spine; it waits for a retrograde signal the spine emitted. Coordination is never achieved by a component consulting a global state, because there is no global state to consult. It is achieved by signals crossing between components, each read locally and made to mean something by the local context that receives it.
Everything emits; nothing is a pure sink. A component that only consumed would be invisible to the rest and could not participate in coordination. Even the leaves of the daytime chain — the bouton, the spine — emit: by day they emit fatigue upward and retrograde messages laterally; by night they emit freed material into the shared pool and demand upward. To exist in the model is to be readable, and to be readable is to emit.
Behavior is legible: acting leaves a readable mark, sent or not. What a component emits is not always an intended message. Some emissions are signals — sent to be read (glutamate, the retrograde messages, D-serine). Others are traces — the physical residue of acting, read by others though never "sent": spillover glutamate is the consequence of a bouton releasing more than the cleft can clear, and that overrun is itself information about the bouton's power. There are no silent acts. Acting and informing are inseparable, because behavior displaces the shared medium and the displacement is readable. This is why coordination needs no broadcast of intent: a component that simply behaves is already legible to whoever shares its medium.
Meaning is assigned by the reader, not carried by the signal. A signal is a physical fact — a molecule, a voltage, an overrun. It has no intrinsic meaning; its meaning is fixed by the local context that reads it. The same endocannabinoid is a brake to the bouton (reduce release) and a report of postsynaptic excess to the astrocytic process (a pressure cue for its own structural control). The same nitric oxide is confirmation to strengthen for the bouton and this coincidence was real, keep the capacity for the astrocyte. The same spillover is lost transmitter to no one and my presynapse has outgrown my volume to the astrocyte. One emission, many readers, many meanings — and none of the readers consults the others to agree on the meaning. This is the locality principle at the level of semantics: because no component can read another's interior, all it ever has is the shared physical facts, which it must interpret unilaterally. Coordination is achieved without shared meaning — each component reads the common medium and assigns its own.
Coupling is openness, and openness is bounded. A component is open — it takes in signals and supply, gives out signals and product — but its openness is bounded by what it can physically reach: its own cleft, its own supply lines, the neighbors it is wired to. It is neither sealed (that would make coordination impossible) nor unbounded (that would make it the whole). The bounded openness is what lets many local components compose into something we can describe as a whole without any of them being that whole.
Holism is real but only described. The re-evoked pattern at night, the neuron's total activity, the memory a synapse carries — these are real. But they are not stored anywhere. The pattern is not in any component; it is what happens when many primed components ignite each other. The neuron's "excitability" is not computed by anyone; it is the coincidence of many components' own lowered thresholds. Holism is enacted by the coupling and read off by us as observers — it is never encoded in a part, because if it were, that part would be the system, and there is no system.
2. The Ring — One Act in Three Phases
The local component's act has one shape, and it is the same shape everywhere: ACTION, RECOVERY, PREPARATION. This is the specialization of the principle to the question what is the act?
The three phases.
- ACTION is the component's defining deed — the thing that makes it the component it is. The bouton releases; the soma fires; the spine responds; the axon and dendrite propagate. Action is the only phase that spends irreversibly and reaches outside the component.
- RECOVERY is the fast alter-ego of action: it restores the capacity to act again. Vesicles refill, sodium channels de-inactivate, calcium clears. Recovery undoes the local depletion the action caused, so a next action is possible. It looks backward — it repairs what was just spent.
- PREPARATION shapes what comes next. It faces two futures at once: the next action in this same scope, and the action of the other scope. Setting the release machinery for the next spike is preparation for this scope; stocking the tag that the night will spend is preparation for the other. Preparation is provisioning, not judging — which is why the old "evaluation" phase was a misnomer and has been retired. Depositing a trace does not render a verdict; it lays down a provision that a later phase may or may not draw on. What we once called evaluation was always preparation aimed at the other scope.
The rhythm is (ACTION ⇄ RECOVERY) × many, then PREPARATION — then again. The act is not one pass through three phases. Action and recovery alternate rapidly — a tight inner loop, release-and-restock many times over — and only when that alternation subsides does preparation run, punctuating the bout and setting up the next. A spike train is exactly this: release ⇄ refill, release ⇄ refill, then, in the sustained quiet, the preparation that stocks the tag and tunes the next train. The inner loop is fast; preparation is the slower punctuation.
Every category spans all three timescales. The three phases are not three speeds. Each phase is a kind of work — deed, restore-capacity, provision — and each kind happens fast, medium, and slow. Preparation especially is multi-timescale: it contains a fast loop (probe and restock), a medium adjustment (tuning the release machinery from the tag), and a slow settling. A category names what kind of work, never how fast.
Action is always local; recovery and preparation may be contextual. A component necessarily has its own action — the deed just is the local event occurring in it, and it cannot be performed on another's behalf (that would be signalling, not acting). But the recovery and preparation of an action can live in other components. A calcium channel's action is letting calcium in; its recovery-and-preparation live in the presynapse and above. So the ring is a property of coupled components, not of the individual: the ring must close — every action recovered from and prepared for — but no single component need run all three phases itself. What is necessary is the closing of the ring, not its co-location.
The three categories are the three modulable dimensions of behavior — which is why the synapse has three parts. Ask what about a behavior can be changed, and there are exactly three answers: how hard (intensity), how soon again (timing), and where (spatial extent — which connections exist, how isolated they are). These are not an arbitrary list; they are the three categories seen from outside. Intensity is the magnitude of the ACTION — a bigger release is a bigger deed. Timing is set by RECOVERY — how fast the capacity to act is restored is the temporal window and the readiness for the next deed. Space is set by PREPARATION — which structure is built or pruned is the configuration future action will run on. To modulate a dimension is to modulate the corresponding phase; there is nothing to change about a behavior except its three phases, so there are exactly three dimensions, in one-to-one correspondence.
This is why the synapse is tripartite and not bipartite. Three separable dimensions want three independent controllers, and the parties divide them: the presynapse owns the clean intensity knob (how much it releases), the postsynapse owns sensitivity (how strongly it responds), and the astrocytic process owns timing and space (its clearance sets how fast transmitter is cleared — shorter dwell, sharper temporal window — and its coverage sets spillover and isolation). A two-party synapse could set intensity but could not independently sharpen timing or bound space; the third party exists precisely to control the dimensions the two coinciding parties cannot. In the category language, the astrocytic process is the recovery-and-preparation specialist of the synapse — it owns how-soon and where — while the pre and post are action specialists — they own how-hard. The tripartite structure and the three-phase act are therefore two expressions of one three-way partition: three phases of the deed, three dimensions of what can be changed, three parties to change them.
The correspondence is not perfectly symmetric, and the asymmetry is instructive. Intensity and timing each have a live mode — they are modulated moment to moment by the action and the recovery — and also a provisioned mode, the persistent ceiling on them, set slowly. Space has no live mode: a connection cannot be added mid-behavior; spatial structure is inherently slow. So preparation owns space outright and also sets the ceilings for intensity and timing, while action and recovery hold the live knobs. This is why "evaluation" was never a fourth category — there is no fourth dimension for it to modulate. Behavior has three modulable dimensions; the act has three phases; a would-be fourth phase would have nothing to change, which is exactly why it collapsed into preparation.
3. The Two Turnings — Day and Night
The one ring is turned in two directions. This specializes the principle to what are the component's two scopes? — and it is where the model's deepest duality lives.
Two contextualizations, two currencies. By day the component faces outward, against the world (the cleft); its currency is information — cheap, gathered passively, and non-rival (see category 5). By night it faces inward, against the economy; its currency is material and energy — scarce, conserved, and rival. The component does not know it is in "day" or "night" as a global state; each turning simply runs against whatever environment is present, and the environment differs.
The rotation: the same physical event is a different phase in each scope. This is the sharpest form of the duality. Neurotransmitter release is the day's ACTION — the outward deed. The same release, run at night, is PREPARATION: the component releases not to transmit but as a probe, to replay a behavior and measure how much it participates in the re-evoked pattern. And the structural change, which the day can only mark (the tag is an inert claim pointing at a restructuring that never happens by day), is the night's ACTION — its irreversible defining deed. So the defining act of one scope is the measuring-instrument of the other: release is day-action / night-preparation; restructuring is night-action / day-inert-mark. The scopes do not merely run the ring in two directions — they swap which event is the deed and which is the provisioning. Because it is a ring, each scope simply enters at a different phase.
Night PREPARATION replays the day ACTION — the same machinery. Because preparation-at-night is a replay of the behavior, it runs the very code the day action runs: the same release, the same capacity and vesicle checks, the same endurance deposit into the same trace. Endurance discovered in replay is as real as endurance discovered in behaving. Only two things differ: there is no dopamine (significance is already settled), and the released transmitter is a probe — it carries the pattern onward to the next component and its own trace is read as participation. The action machinery is written once and serves as the deed by day and the measurement by night.
The tag is the payload that crosses between the turnings; the fatigue loop is the switch. Each scope's PREPARATION mints what the other scope will consume. Day-preparation mints the tag — a token of confirmed significance — which the night spends on structure. Night-preparation measures participation, which gates that spending. The tag is one token with three roles: by day it is the significance bridge; at night it lowers the component's own threshold so its pattern can re-ignite, and it funds the build, a slice at a time. Distinct from this payload handoff is the switch — the fatigue loop that decides when a component crosses between scopes. Activity accrues fatigue; a single continuous integrator (the one actor that never sleeps) reads the aggregate fatigue and emits a pressure; when a component's own activity falls and pressure is high, it crosses into night; when pressure discharges, it crosses back. No scheduler; no clock. The switch says when to turn; the tag says what crosses when it turns. One ring, two turnings, stitched by the tag and switched by fatigue.
Two independent forgettings. Because night ACTION is build ⇄ release (category 5), two distinct things can be lost, by two distinct mechanisms. Structural pruning sheds built structure a component no longer uses — driven by low participation, regardless of any tag it holds. Intention decay is the tag itself decaying unspent — a planned strengthening that never found the participation to license it. The tag is patient: it is sliced by building and never touched by releasing, so it survives across non-participating cycles and cashes in when its pattern finally re-evokes. Disuse prunes structure; unspent intention fades on its own slow clock. The two are independent, and both are forgetting.
4. The Timescale Ladder
Orthogonal to the ring is the ladder of timescales. This specializes the principle to at what speeds does the component act? The ring says what kind of work; the ladder says how fast; they compose — every phase of the ring occurs at every rung of the ladder.
The rungs. FAST (milliseconds to seconds): the immediate trace a single action leaves. MEDIUM (seconds to minutes): occupancy and evidence — the running average of fast traces, and the eligibility climbing toward a tag. SLOW (hours): the tag, the consolidation bridge. PERSISTENT (written only at night): the structural ceilings, and the two conserved stocks — energy, which does not return, and material, which does.
A tier's timescale is set by both its creation and its decay. A fast trace is deposited as a point event and relaxes in milliseconds. A medium trace ramps while a condition holds and settles over minutes. The timescale is not a label attached to a variable; it is the joint consequence of how the variable is written and how it fades. This is why the same climb appears in every component: each action leaves a fast trace; the average of fast traces over seconds fills occupancy (short-term strength); the average of that average over minutes, gated by dopamine, raises the tag. Occupancy is the fast-and-medium memory of participation; the tag is its slow, validated distillate.
Evidence ascends the ladder; capacity descends it. By day, information climbs from fast trace to tag — evidence accumulating upward. By night, capacity is written downward from the tag into persistent structure. Each rung also has its own failure meaning, set by its timescale: a fast pool running dry is transient depression; a medium pool constrained is a standing endurance need; a persistent ceiling reached is a structural limit. Depletion and recovery at each rung mirror the creation and decay of its trace — the same timescale governs both the evidence and the capacity at that level.
5. Scarcity Decides — Collaboration by Day, Competition by Night
How components relate to one another is not an independent fact; it follows from what is scarce. This specializes the principle to how do components relate? — and it unifies conservation, selection, and the collaboration/competition character of the two scopes into one causal chain.
Two conserved currencies, two rules of flow. Energy ratchets: it is spent irreversibly, the arrow of time in the model — a component that burns energy into structure cannot get it back. Material circulates: it is freed by one component and reclaimed by another, conserved as it moves. Scarcity of both forces choice — two ceilings (structure and endurance) compete for one finite night pool, and what is not maintained drifts back down.
Rivalry of the currency sets the relation. By day the currency is information, which is non-rival: a bouton releasing glutamate does not use up the spine's ability to receive it; a trace here does not deplete a trace there. When producing for others costs nothing, the natural relation is collaboration — and the day is exactly that: each component acts so the next can act, releasing, integrating, clearing, passing activity along the chain, co-producing the pattern and the tags. By night the currency is material and energy, which are rival and conserved: every unit one component builds into its structure is a unit another cannot have, and the total is capped. When what one takes another loses, the natural relation is competition — and the night is exactly that: components contend for the shared pool, build what they win, and free what they shed back into contention.
But night's competition is adjudicated by collaboration. The relation is subtler than "day collaborate, night compete." The replay that arbitrates the night's competition is itself a collaborative act: a pattern re-evokes only if every component along its loop is primed and ignites the next — mechanical coherence, a collaboration all the way around, one un-primed link breaking it. Participation — the measure that gates who gets to build — is a measure of successful collaboration in that re-enactment. So a component earns its share of the scarce material in proportion to how well it collaborated in replaying the pattern. Collaboration is primary in both scopes: by day it produces the shared, non-rival good; by night it adjudicates the competition for the rival one. The register is economic, not martial — components do not fight; they contend for a conserved resource, and the contention is settled fairly by a collaborative criterion.
Two competitions at two loci. Within the night, competition appears twice, cleanly separated. Within a component, build and release contend over its own structure, arbitrated by participation: high participation builds (funded by the tag, a slice per cycle), low participation releases (freeing material, the tag untouched), and in between the component holds. Between components, this one and its peers contend for the shared material and energy during recovery. The internal tension (grow or shrink?) is settled by the replayed pattern; the external tension (can I get material?) is settled by contention with neighbors. Selection under scarcity is the sum of these: what survives a night has both earned its tag by day and won its material by night, and what neither participates nor is maintained returns to the pool. Selection is not a judge's verdict; it is what scarcity leaves standing.
6. Causation Circulates — Emergence Up, Constraint Down, Command Nowhere
The final category specializes the principle to who is in charge? — and the answer is no one. Causation moves in two directions across the coupling, and neither is command.
Emergence ascends; constraint descends. By day, evidence and activity emerge upward: components act locally, and their emitted activity is what a higher description (the neuron, the assembly) is made of. Nothing reaches down to make them act. By night, constraint descends: a higher actor broadcasts a bound — a renormalization target, a downscale factor — but it does not reach in. It emits a signal; each component reads that signal and scales itself. The neuron never edits a synapse; it announces a total, and the synapses each renormalize their own structure against it.
No actor authorizes its own restructuring. A component cannot open its own night. It is put in position by the actor above — which holds an aggregate the component cannot see and opens a window the component cannot open — and then, within that window, the component acts locally on its own state. The soma cannot decide within the soma which of its synapses matter; the synapses decide that locally, by their own thresholds. And the synapses cannot ignite their pattern alone; the soma's firing does that. Each is put in position by the other; neither reads the other's interior. This is the recursive grant: act locally, be enabled hierarchically.
Command is nowhere. There is no actor that both holds the whole and acts on it — that would be the system, and there is no system. What looks like top-down control is always a broadcast constraint scaled locally; what looks like bottom-up assembly is always local emission summed from outside. The neuron that "renormalizes" only announces a number. The assembly that "replays" is only coincident local thresholds propagating through coupling. Causation circulates — up as emergence, down as constraint — but it never concentrates into command. This is the unifying principle in its final form: because there is only the local component and its one act, there is no one to be in charge, and the whole is enacted by the parts, never encoded above them.
7. The Four Operations — How the Local Is Multiplied Into a Whole
The six categories describe what a component is and does. This one is a different cut: it asks by what operations the local is coupled into something we can describe as a whole. There are four — integrate, coincide, broadcast, inject — and together they are the entire vocabulary by which scale is crossed. The previous category named the two directions of causation; this one names the mechanisms, and adds the two the directional view misses.
Integrate — make the distributed present. A quantity spread over time or space has no instantaneous local existence. Flow is nowhere emitted; it is the accumulation of many releases across a duration. Frequency is not present at any instant — a single spike has no rate; rate lives only in the relation between events separated in time. Total activity is not held by any component; it is the sum over many. The system reads none of these directly — it cannot, because they are not anywhere. It reads them by transducing the distributed into a store whose instantaneous level is the quantity's present shadow. The fast trace is the device: each event deposits a quantum, the store leaks, and its standing level encodes recent frequency — high when deposits outran decay, low when they did not. Spatial integration does the same across space: the dendrite summing its spines, the soma summing its dendrites, the astrocyte summing its processes each make a spatially-distributed quantity locally present at one site. This is how a distributed system verifies what is expressed nowhere and by no one: it never reads the quantity, it reads the store the quantity filled. And it is how the whole "knows" what no part knows — not by computing, but by being the place where the parts accumulate.
Time itself is read this way. The system has no clock; it does not count duration. Time enters only as the decay of stores — "how long ago" is how far a trace has fallen, "how fast" is how high it stands against its leak. Time is not represented; it is suffered, and the store's level is the readout. Every leaky store is a little clock that keeps time by forgetting rather than by counting.
Coincide — read several present-made stores at one site, and get an event. Integration produces quantities; coincidence produces events — the meaningful happenings the components act on. And nothing significant is caused by a single signal: significance is always the co-occurrence of several. The postsynaptic calcium event requires glutamate and depolarization and the astrocytic gain together; the tag requires accumulated eligibility and validation together; the night's build requires a standing tag and confirmed participation together; the astrocytic spike requires many processes' calcium together. A coincidence is two or more stores being high at the same instant — which can only be read where all of them are present. So every coincidence needs a meeting-site that owns none of the signals it compares: the site where the transduced-present stores overlap. This is why the synapse is tripartite (the coincidence detector needs a third input neither coinciding party owns), and the pattern recurs at every level — each has its coincidence and its meeting-site. Integration makes the distributed present; coincidence reads several presences together. They are one mechanism in two steps: transduce, then compare.
Broadcast — distribute one state to many, without addresses. The third operation sends a single state outward to a whole population at once: the back-propagating spike to all a soma's spines, the action potential to all its boutons, the renormalization to all a neuron's synapses, the priming field and the calcium spike to all an astrocyte's processes. Broadcast is the descending partner of integration, and like integration it is addressless — integration destroys location by summing (the sum does not say which input), broadcast destroys it by spraying (the signal does not select which target). Neither is a message from one component to another specific component; there is no addressed communication across scale, only summation up and spraying down. Crucially, almost every broadcast is endogenous and reflective: it carries a quantity integrated from the components' own locals and sends back down. The back-propagating spike carries "the soma fired," which is the integral of dendritic input, reflected to the spines. The renormalization carries the integrated total weight. These are top-down in delivery but bottom-up in origin — the components talking to themselves across scales, closing the loop that integration opened.
Inject — import the one thing that cannot be built from within. Reward is different, and the difference is not its direction. It, too, descends as a broadcast, like the spike and the renormalization — so top-down delivery is not what sets it apart. What sets it apart is its origin: every other broadcast reflects a quantity assembled from the components' own activity, but no amount of integrating the components' own activity can produce whether the behavior was good for the organism in its world. That fact is exogenous — it comes from outside the components' own self-talk, from the organism's encounter with its environment. Reward is the single channel by which information that could not have been integrated from below enters the coupling at all. This is the precise sense in which it is the opposite of integration: not top-down versus bottom-up, but exogenous versus endogenous — a global that is irreducible to the locals, against a global that is made of them. And it is necessary, because significance is defined at the organism's scale: locality can compute what happened (activity, load, coincidence) but never whether it mattered, so that verdict must be injected. The tag is exactly the meeting-site where endogenous evidence (eligibility, built from local activity) coincides with this exogenous value — consolidation is the marriage of the components' self-knowledge to the world's verdict, and it is the one place the model reaches outside itself.
So four operations, and they divide cleanly: integration makes quantities (by transducing the distributed into present stores, time included, as decay); coincidence makes events (by reading several such stores at a site that owns none of them); broadcast distributes (mostly the components' own integrated state, reflected back down, addresslessly); injection imports the one global — organism-in-world value — that no integration could produce. The first two build meaning from the inside; the third circulates it; the fourth admits the one thing meaning cannot be built from within.
Coda — The Seven as One, and the Why Beneath Them
Read downward, the seven categories are one principle refracted seven ways. A component is local (1); its act has one shape, the ring (2); the ring turns in two directions, day and night (3), at every rung of the timescale ladder (4); the relations between components are set by what is scarce, collaborative where the currency is free and competitive where it is conserved (5); causation circulates between components without ever concentrating into command (6); and the local is multiplied into a describable whole by four operations — integrate, coincide, broadcast, inject — none of which is a component reading another's interior (7). Remove any one and the principle loses a facet; none stands apart from it.
And all seven serve the why of Part I. Each is a way the specification refuses to sit still in any part: locality forbids a global copy; the ring builds structure from behavior and behavior from structure; the two turnings make the night rewrite what the day runs; scarcity makes the rewriting irreversible and history-locked; causation-without-command leaves no controller to hold a fixed program; the four operations cross scale only by summing and spraying, never by encoding the whole anywhere. Together they are why there is no fixed model to run — only the rule and the history. There is only the local component and its one repeating act; everything else is that act, multiplied, coupled, and described from outside — and because it is only ever enacted, never encoded, it must be lived to be known.