# Logic Principles of the Tripartite Synapse Model These are the principles that govern the model's logic — not the syntax in which it is expressed, but the reasoning that shapes every variable, every behavior, every transition. A note on language. This document does not say "the system." There is no system — only local components, each reading arrived signals and acting on its own state. "System," "whole," "network," "the organism's memory" are names applied from outside, by us, describing what coupled local components do together. The first principle states this; the rest honor it by never speaking in the voice of a whole that does not exist. Where a sentence seems to want "the system does X," it is rewritten as "local components, contextualized thus, do X locally" — because the form of the document should enact its central claim, that locality goes all the way down and all the way up, with no privileged vantage anywhere inside. The ten categories run foundation-first. The two closing panels — the Logic of DAY and the Logic of NIGHT — put the whole vocabulary to work, dissecting the two modes the local components actually run in. They come last because they use everything established before them. --- ## I. There Is No System; Holism Is Real but Only Described **There is no system — only local components.** Nowhere in the model is there a controller, a global plan, a representation of the whole, or a vantage from which the whole is seen. There are only components, each reading the signals that physically reached it and acting on its own state. Every behavior is local; every evaluation uses only local state and arrived signals; every trace is a local record; every commitment a local draw on a shared pool. "System" is our word for the aggregate, spoken from outside. No component is the system, occupies its standpoint, or can read it — not the smallest bouton and not the highest integrating actor. **And yet what we describe as holism is real.** Memory, rhythm, selection, consolidation, sparsification — these are real behaviors, not illusions. The puzzle is that they are real without any whole existing to bear them. The resolution: they are *enacted* by coupled locals, never *encoded* in any one of them or in any representation. The holism is in the doing — the ongoing competitive, signal-mediated, scope-alternating process of many local components — not in any part and not in any model the parts contain. **What couples the locals is what we then describe as a whole.** Three couplings do all the work. *Shared pools*: the only thing every component touches is the finite resource it competes for; when one draws, the others have less, when one returns, the others have more — local actions become mutually consequential through a common, capped resource. *Cross-scale coincidence*: a lasting change requires confirmations from larger scales that no component can produce for itself, so every lasting change records an agreement across scales that no single scale authored. *Signals*: components cannot read each other, but they emit and receive, so they are coupled without any one gaining access to another's interior. Through these three, purely local action acquires what we describe — from outside — as global organization. **The properties we call holistic belong to no component.** Rhythm, equilibrium, memory, the joint selection for significance-and-sustainability — none exists in any single component. They are descriptions of the coupled population over the DAY/NIGHT alternation. The gap between "what a component knows" (only its own state) and "what we describe the population as doing" (choosing what to keep) is not bridged by any component knowing more. It is bridged by the coupling itself. Understanding, here, is enacted, not encoded — and "the system understanding" is only ever our shorthand for coupled locals enacting, faithfully local at every step, holistic only in our description. --- ## II. Two Contextualizations, and the Loop Between Them **DAY and NIGHT are not two phases of a system; they are two contextualizations of the same local components.** A component does one kind of thing always: it reads arrived signals and acts on local state. What changes between DAY and NIGHT is the *context* that fixes what those signals mean, who the component's counterparties are, and what is scarce. By "DAY" we name the contextualization in which the relevant environment is the external world and the component's information is about exogenous events. By "NIGHT" we name the contextualization in which the relevant environment is the internal economy and the information is about endogenous resource-state. The component cannot tell which it is in — it has no access to "the scope" any more than to "the system." It reads local signals, and context makes them mean what they mean. **Each contextualization has its own environment and its own information.** NIGHT is not the absence of an environment — it is interaction redirected from the world to the self. By day a component interfaces outward, perceiving sensory-driven events and reward; by night it interfaces inward, perceiving how much resource reached it, what its peers are claiming, what coheres. Both are full perceive-act loops against a real environment. Resource levels and demand-signals are the night's perceptual field exactly as glutamate and dopamine are the day's. "Internal" and "external" are relative to the subject one fixes; from a component's own standpoint there is only "my environment" — the signals reaching me — which happens to be the world by day and the economy by night. **A component is thus two contextualizations sharing one structure.** By day it is an environmental interface; by night an economic agent. The two share only the structure, and the structure is exactly what carries between them: the night-agent builds the ceiling the day- interface will operate within, and the day-interface generates the evidence (the tag) that authorizes the night-agent to build. The tag is information becoming a resource-claim at the boundary — the single unit that crosses from the day's information-context to the night's resource-context. **The move between the two is an emergent local transition, not an imposed clock.** There is no global day/night switch. Each component enters its NIGHT when its own activity is low and an arrived sleep-pressure signal is high, and returns to its DAY when that signal falls. Both conditions are read locally — own activity, arrived signal — so the transition is itself a local decision. Components therefore cross over at different times: a wave, not a switch (local sleep). The signal that carries them is itself the product of locals: activity generates fatigue; fatigue, integrated by one component (the hypothalamic actor) that does nothing but integrate fatigue and emit pressure, raises sleep-pressure; high pressure plus a component's own quiet opens its restructuring window; restructuring discharges fatigue; discharge lowers pressure; the component re-enters DAY. DAY and NIGHT are the two phases of one homeostatic loop the local components run on themselves — neither imposed from outside nor scheduled from above. The component never knows it is "in NIGHT"; it reads a signal level and its own activity, and what results we call night. **The mechanistic root of the alternation: behavior and restructuring exclude each other.** A component cannot rebuild its structure while it is busy behaving — the two compete for the same substrate. Only when its activity is low does it have the access to its own architecture that restructuring requires. The brief low-activity gaps within a day permit small adjustments; the sustained, widespread quiet that the sleep-pressure signal creates permits the large ones. NIGHT is not an arbitrary time for consolidation — it is the condition under which consolidation is *possible*, because quiet is what grants a component access to itself. --- ## III. Locality and Signals **Only local evaluation.** Every decision a component makes — to act, to deposit a trace, to register an interrupted success — uses only information physically present in it. It cannot read another component's interior. The presynapse does not know the postsynapse's calcium; the dendrite does not know which distal spines are active; the astrosynapse does not know whether the postsynapse is waiting. Each judges from its own state alone. **A component cannot read the whole, either.** The completion of locality: not only can no component read another's interior, none can read "the system," "the scope," or the global state. There is no aggregate vantage available anywhere inside — not even to the integrating actors, who read only the summed emissions that reached them and emit signals in turn, with no more access to the whole than a bouton has. Locality holds up the hierarchy as strictly as across it. **Cross-component influence travels only as signals that arrive and become local.** Information crosses a boundary only by being emitted — feedforward transmission, retrograde messengers, neuromodulatory and sleep-pressure broadcast, the demand and recycled resource of the night economy. A signal in transit is invisible; a signal that has arrived is local and can be read. Downstream reaches upstream by emitting; upstream never reaches into downstream. Every coupling in the model is of this form — an emission that becomes, on arrival, another component's local state. **Everything emits; nothing is a pure sink.** No component only consumes. Each, whatever it receives, emits something a neighbor will read — its transmitter, its retrograde feedback, its fatigue, its demand, its recycled material, or simply its activity for an integrating actor to sum. The direction of emission reverses with context (see the two Logic panels): outward and downstream by day, inward and upstream by night. But the invariant holds in both — there are no leaves and no sinks, only a reversal of which way "out" points. --- ## IV. Resource and Conservation **Nothing is free.** Every behavior consumes a resource. There is no operation that does not draw something down. This is not a constraint added on top of the logic — it is its foundation. Selectivity, competition, and forgetting all follow from the single fact that resources are finite. **Resources are redistributed, not created.** The pools are bounded by external ceilings. Within them, resource is only moved — from one site to another, from a dismantled structure back to the pool. No internal process manufactures capacity; it only reallocates. A gain anywhere is paid for by a loss elsewhere — coupling that is not designed but is the automatic consequence of drawing from a common pool. **Two resources, two conservation laws.** Energy is a flow — produced and consumed, gone after use. Material is a stock — incorporated into structure and recovered when structure is dismantled. Different sources, different recovery. A behavior can be energetically affordable yet materially limited, or the reverse. Keeping them distinct is what makes the accounting honest. **Material circulates; energy ratchets — energy is the one irreversible flow.** This is the sharp form of the distinction, and it is the arrow of time. Material cycles indefinitely: spent into a ceiling, recovered when that ceiling decays, returned to be spent again. Energy does not. It is produced fresh at the roots, burned irreversibly on the work of building and behaving, never recovered. Everything else cycles or relaxes — traces decay and reform, occupancy fills and drains, material recycles — but energy only ever goes down per unit produced, capped externally by glucose. The total learning the local components can ever do is bounded by their lifetime energy throughput, and no internal cleverness lifts that bound. The irreversibility of energy is what makes them age. **Every economy has a single capped root.** Each resource traces to one producer with a hard ceiling — the astrocyte cell body for synaptic energy, the soma for neuronal material. Everything downstream competes for shares of that capped production. The ceiling is set from outside and is the ultimate arbiter of how much can be done. **Scarcity is what forces choice, and choosing is learning.** What is consolidated is the outcome of bounded demand (the standing tags and endurance needs) meeting bounded supply (the energy and material produced each night), and the match need not clear — a night may run out of energy before it runs out of demand. Unmet demand is not discarded but carried forward and retried. The selective pressure is, at bottom, this repeated failure of supply to fully meet demand: it is *because* not everything can be afforded that there must be choosing, and the choosing is the learning. --- ## V. The Timescale Ladder The spine. Every quantity sits on one of four nested tiers, and timescale is not incidental to it — timescale *is* its meaning. **Four tiers.** FAST traces (ms–s): residual calcium, synaptic current — the immediate response. MEDIUM occupancy and evidence (s–min): the filled receptor surface and channel coupling, the accumulating possible-tag, the endurance need. The SLOW tag (hours): the validated bridge to consolidation. PERSISTENT capacity (written only at night, drifting over days): the structure and budget ceilings. A quantity's decay constant is what it means — a fast-decaying quantity is a momentary signal, a slow one a commitment, a non-decaying one a capacity. Putting two timescales in one variable destroys both meanings, which is why a quantity carrying both a momentary and a lasting role is split into two. **Capacity and occupancy are two rungs, not a separate principle.** What was once stated as "night builds containers, day fills them" is simply this: the PERSISTENT tier is written at night and bounds the MEDIUM tier, which fills by day within it. The same physical quantity — receptor count, vesicle coupling, fuel level — has a fast/medium component (how full, occupancy) and a persistent component (how big, capacity), governed by different processes at different tiers. Short-term change is occupancy; long-term change is capacity; they never do each other's job. **Two capacities, two drives, one pool.** Structure is the capacity for strength — how powerfully a behavior can act. Budget capacity is the capacity for endurance — how long it can be sustained. Both are persistent ceilings, both filled competitively at the medium tier, both drawn from the same finite material and energy, so strength and endurance compete. A ceiling of either kind is never free even by day: filling it costs a competitive share of a shared pool, and a high ceiling makes a large standing claim satisfiable only by out-competing neighbors. Capacity that cannot be filled is wasted. **Structure shapes form, not just maximum.** A ceiling does not merely cap — it conditions the transfer function at every moment. Tighter calcium-channel coupling makes each spike more reliably coupled to release; more anchoring slots convert each pulse more faithfully to current; tonic co-agonist keeps the gate primed. The persistent tier shapes the quality of behavior continuously, not only its peak. **The tiers are a ladder: each rung's output is the next rung's input, in both directions.** *Evidence ascends* — fast traces accumulate into medium evidence, which bridges (on validated coincidence) into the slow tag, which commits at night into persistent capacity; nothing reaches a slower tier without accumulating through the faster ones. *Capacity descends* — persistent structure bounds medium occupancy, which bounds fast behavior; the ceiling at each level was set by the level above, on a slower timescale. Both the strength pathway (trace → possible-tag → tag → structure) and the endurance pathway (trace → endurance-need → budget ceiling) are the same upward climb, differing only in what validates it: associative dopamine for strength, homeostatic fuel-shortfall for endurance. **A pool's recovery timescale is what its exhaustion means.** The ladder governs pools as well as traces. A *fast* pool (the readily-releasable vesicle pool) depletes and recovers fast, so its shortfall is transient — short-term depression, self-correcting once activity slows. A *medium* pool (operational budget) recovers at the medium scale, so its shortfall is a standing constraint worth recording as endurance evidence. *Persistent* capacity changes only at night, so its "shortfall" is a structural limit unfixable in the day. This is why a behavior's failure mode can be read off which pool ran dry: fast-pool exhaustion is depression, medium-pool exhaustion is endurance evidence, persistent limit is structural. Depletion-and-recovery is the pool-side mirror of creation-and-decay — drawn down by behaving, refilled toward a ceiling — and in both, the timescale is the meaning. --- ## VI. Causation Circulates — Emergence Up, Constraint Down, Command Nowhere The model has causation in two directions, and the whole point is that they coexist without either becoming control. **Emergence flows up.** What we describe as global organization — sparsification, normalization, winner-take-more, the allocation of fuel — is nowhere computed centrally. It emerges from many local components drawing on shared pools. No allocator decides which synapses to fuel; the synapses' own demands, each purely local, competing for capped production, produce the allocation. No allocator exists; the allocation is real. The local emissions sum, through the pool, into an aggregate no component authored. **Constraint flows down.** An integrating actor holds an aggregate its constituents cannot see — total weight, territory demand, accumulated fatigue — and broadcasts it back as a constraint each constituent then interprets locally. The neuron, summing emitted activity it never reads interiors to obtain, broadcasts a renormalization that each component applies to itself. This is top-down, but it is constraint, not command: the higher actor sets a bound; the lower still decides locally within it. **Command exists nowhere.** This is the load-bearing claim. The downward direction never becomes a higher component deciding for a lower one, nor a lower one deciding for itself in isolation. No actor authorizes its own restructuring — each is put in the position to restructure by the actor above it, which holds the aggregate it cannot see and opens the quiet window it cannot open, then *broadcasts*, never reaching in. The soma cannot decide within the soma; it is put in position by the neuron — which is itself put in position by the integrated fatigue, which is itself the sum of what the components emitted. Causation circulates — up as emergence, down as constraint — and the circle closes with no controller anywhere on it. Every integrating actor is itself only another local component reading summed arrived signals; none is "the system" in disguise. **The recursive grant repeats at every scale.** The relationship is the same all the way up: the astrosynapse is put in position by the astrocyte; the soma, bouton, spine, branch, axon by the neuron; the neuron and astrocyte by the hypothalamic fatigue signal; the synaptic strengthening by the organism's dopamine and the assembly's replay. Each scale grants its constituents the conditions they cannot grant themselves — an aggregate they cannot see, a window they cannot open — and grants it by signal, never by reaching in. The hierarchy is real, the circulation is closed, and at no point does it produce a seer of the whole or a commander of the parts. --- ## VII. Selection and Asymmetry **Building is the active drive; weakening is its shadow.** All the machinery is oriented toward strengthening what is significant and sustaining what is fuel-limited. There is no symmetric machinery for weakening. Weakening happens to whatever the building machinery did not select, as a consequence of the resources building consumed. The orientation is toward learning; forgetting is its cost. **Depression is never explicit — it is what happens when building does not.** No signal says "weaken this." Ceilings of both kinds decay continuously and are held up only by maintenance; when building consumes the shared resource, unmaintained ceilings drift down. Depression is the absence of maintenance, not the presence of a depression signal — and the same is true of lost endurance, idle capacity removed for lack of use, and of occupancy, which drifts back the moment its driving trace decays. **Validation enters from beyond the part; the part cannot validate itself.** A component cannot know whether its own activity was significant — that information exists only at a larger scale and arrives as a signal (the neuromodulatory broadcast for the organism's verdict, replay for the assembly's). Cheap reversible change is autonomous; expensive lasting change requires authorization that enters from outside the component being changed. This is why lasting change always records a coincidence across scales: each scale confirms what the one below cannot know about itself. **Strength is associative; endurance is homeostatic.** Strength requires significance — the dopamine coincidence saying "this was worth saving." Endurance requires only that fuel, not structure or significance, was the binding constraint on a forming success — it needs no external validation, because metabolic sustainability is not the organism's to judge but the component's own to register. Two independent criteria, and selection requires winning on both: activity without significance is not saved; significance without sustainable fuel cannot be maintained. The conjunction filters for connections both valuable and viable. **Equilibrium is the residual of imperfection.** Where alignment or balance is reached, the success removes the very signal that drove it, allowing slow drift back, which regenerates the signal. The soma that aligns to its input rhythm stops generating the mismatch that aligned it, drifts, and re-aligns. The component that builds enough endurance stops depleting, loses the signal, and lets capacity decay until depletion returns. Each component hovers near its own optimum, never resting there, corrected continuously by the small errors its own imperfect state produces. What we describe as a stable population is the sum of these local never-quite-settlings. --- ## VIII. Coupling, Openness, and Boundedness **Couplings create trajectories, not just states.** Some variables, once moved, make further movement the same way easier — the astrosynapse wrapping tighter after potentiation, easing future potentiation. These self-reinforcing couplings give momentum: components do not merely occupy states, they follow trajectories, deepening whatever direction they have begun. The astrosynapse is the strongest such coupling — the gain control that reshapes the input itself, amplifying whatever trajectory a synapse is on. **The same signal serves opposite functions through different receivers.** Glutamate spillover brakes the presynapse while exciting the astrosynapse — one ligand, two receptor types, opposite cascades, simultaneous opposite effects. Function is set by the receiver, not the signal. One event coordinates several responses with no coordinating mechanism. **Metabolic availability is a selective pressure parallel to validation.** Beyond the explicit activity-and-reward gating, the bare availability of fuel continuously selects which components can participate: one that cannot be fueled cannot generate the activity that would let it be tagged. Metabolism silently shapes what can be learned, independent of and parallel to the plasticity machinery. **Finite and open, not infinite and closed.** The components are bounded and their state space is bounded, and they receive inputs they cannot generate from within — sensory drive, neuromodulatory and replay validation, metabolic supply. Because they are finite, their self-modification generates no infinite regress. Because they are open, their highest validation comes from outside any component being changed. **The fixed points are explicit, not hidden.** The quantities the components cannot modify from within — thresholds, the vascular ceiling, the neuromodulatory and replay signals — are declared as given. They are the boundary with what the components did not set and cannot inspect. Making them explicit is the honest acknowledgment that every self-modifying process operates within constraints it did not choose. Correctness is never certified internally: whether a change was good is answered by the organism's later experience in the world, fed back as signal. The fixed point lies outside — the components act, the world responds, and the response, not any internal check, determines what was worth keeping. --- ## IX. The Logic of DAY — Outward Perceive-Act in Three Relationships By day, a local component interfaces outward, against the external environment. Its information is about exogenous events; its currency is information, which is cheap and gathered passively from whatever the world does; its direction is evidence ascending. The day-logic has three parts, given by the three relationships a component stands in — the same three the night-logic has, but turned outward. **Vertical (what is handed down).** The component behaves within conditions set from above: structure from the last night fixes its ceiling, neuromodulators set its context. It does not choose its sensitivity, its active-zone size, its threshold — it operates within whatever architecture was left and whatever modulatory context has arrived. The persistent tier, built above on a slower timescale, descends as the bound the day's behavior runs inside. **Lateral (coordination with peers, in the moment).** The synapse's behavior is a real-time negotiation among pre, post, and astro through the cleft: the presynapse emits transmitter, the postsynapse integrates and emits retrograde feedback, the astrosynapse gates with co-agonist and brakes with spillover. Coincidence detection is lateral — the postsynapse's large calcium requires the simultaneous arrival of presynaptic glutamate and astrocytic co-agonist. What the day gathers as significant is what several lateral peers confirmed at once. **Local (acting on own state, gathering evidence).** The component deposits its own fast trace, fills its own occupancy toward its ceiling, judges its own interrupted success by its own proxy. This is the purely local perceive-act: sense the environment through arrived signals, act, leave a trace. Evidence climbs from here — fast trace to medium possible-tag and endurance-need, toward the slow tag — but only if the lateral and vertical confirmations arrive in their windows. The day's work is to gather, through these three relationships, the evidence that the night will or will not be able to afford to keep. --- ## X. The Logic of NIGHT — Inward Negotiation in Three Phases By night, the same local component interfaces inward, against the internal economy. The external world is gone; no sensory drive, no reward, no new event-information arrives. What arrives instead is resource and the signals of the economy — how much material reached me, what the constraint from above is, whether my peers cohere. The currency is resource, genuinely scarce and fought over; the information is resource-state; the direction is capacity descending and demand ascending. A NIGHT is not one act but a cycling of three phases, given — like the day's three parts — by the three relationships, here turned inward. The phases are retried in rounds because coherence is reached by different components at different times; one monolithic pass could not express coherence-when-ready. **Phase 1 — make room (vertical: higher actor ↔ its components).** The decision lives above, the execution below. The integrating actor holds the aggregate no component can see — total committed weight, territory load — and, against the night's fixed external allowance, broadcasts how much room exists. Each component receives the verdict and applies it to itself: occupancy is reset multiplicatively toward baseline, total weight renormalized, what was not earned pruned. This is where the fixed external constraint is translated into per-component room, and where the rule is enforced that *what persists must have earned persistence* — occupancy that earned no tag returns to baseline; only ceilings carry forward. **Phase 2 — check coherence (lateral: component ↔ peers).** No resource is drawn. Each component, by signaling with its peers and reading the replay broadcast, determines whether its standing change finds coherent partners *this round* — whether pre, post, and astro agree, whether the synapses that fired together are re-presented together. This costs only a round of time, the night's cheap currency. Its outcome is a verdict: in sync, or not yet. **Phase 3 — draw and commit (local: component ↔ itself and its arrived resource).** Only a component whose Phase 2 found coherence draws energy and material and burns them into structure, spending its tag as it commits. The draw is local — *I* commit, within what reached me — but gated by the lateral verdict. A component that found no coherence draws nothing: it holds its tag, leaves the resource in the pool for peers that did cohere, and waits for a later round when its own peers may have come into sync. This deferral is the reason the cycle exists and is non-destructive — a change is never spent on until it is worth the spend, and energy, the irreversible currency, is burned only at the moment of coherent commitment. The three phases spend three progressively dearer currencies — position, then time, then energy — and the dearest is spent last and only when justified. **The cycle converges; waking ends it.** Round by round, more components find coherence and commit, each commit reshaping the aggregate so the next Phase 1 recomputes room from the new state. The night ends not by a counter but emergently: as commitment discharges fatigue, the sleep-pressure signal falls, and components re-enter their day. A well-rested set of components discharged its demand; an overloaded one wakes with tags unspent, which carry forward and compete again the next night. A change that never finds coherence before its tag decays is forgotten — not pruned after spending, but simply never worth the spend. And here too nothing is a pure sink: by night a component's emissions run inward and upward — its demand, its fatigue, its recycled material climbing back toward the integrating actors — the same invariant as the day, with the direction of "out" reversed.