ocrampal/organism
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update 2026-06-29-tripartite-synapse_v17.md

Browse Source
This commit is contained in:
ocrampal
2026-07-13 11:55:05 +02:00
parent b173478e1b
commit 2210be7086
1 changed files with 70 additions and 52 deletions
Download Patch File Download Diff File Expand all files Collapse all files
elements/neuron/appunti/2026-06-29-tripartite-synapse_v17.md
+70 -52
View File
@@ -1,7 +1,3 @@
---
include_toc: true
---
# Tripartite Synapse — Pseudocode v17
> Companion: `tripartite_synapse_v17_biology.md` · principle: `logic_principles_v5`.
@@ -134,7 +130,7 @@ its structure rebuilt) OR energy is spent (overloaded — unspent tags carry to
## Conventions
```
SCOPE = {DAY, NIGHT} CONTEXT = {AP, NOT_AP, NOT_SPIKE_TRAIN, bAP, NOT_bAP, CONTINUOUS}
SCOPE = {DAY, NIGHT} DAY CONTEXT = {AP, REFRACTORY⊂NOT_AP, NOT_AP, NOT_SPIKE_TRAIN, CONTINUOUS} NIGHT CONTEXT = {NON_REM_1, NON_REM_2, REM}
THE THREE CATEGORIES (see logic_principles "The Ring"): (ACTION ⇄ RECOVERY) × many, then PREPARATION
ACTION the component's defining deed; deposits the fast trace. ALWAYS LOCAL to the acting
@@ -157,7 +153,23 @@ CONTEXT LICENSES PHASE (context = the imposed condition; phase = the work it per
NOT_SPIKE_TRAIN no spike AND no train ⊂ NOT_AP → adds PREPARATION (runs on top of NOT_AP)
A process runs in the SHORTEST quiet its timescale fits: fast-trace decay and partial pool refill
in NOT_AP (they ride the train and set short-term depression); tag formation, full refill, and
occupancy read-out in NOT_SPIKE_TRAIN. (Components without trains use bAP / NOT_bAP, or continuous.)
occupancy read-out in NOT_SPIKE_TRAIN. (Components without trains use continuous flow.)
WHO EMITS THE CONTEXT (a context is a signal from ABOVE, not self-generated):
AP, REFRACTORY emitted by SOMA — its own firing state (AP = it fired; REFRACTORY = the recovery
window after, when it cannot fire again yet). What was "NOT_AP" is REFRACTORY.
NOT_SPIKE_TRAIN emitted by the NEURON — sustained absence of soma spikes can only be tracked from
ABOVE the soma (a component cannot easily know "I have not fired for a while");
the neuron keeps count of the gap and broadcasts it down.
NON_REM_1/2, REM emitted top-down within the NIGHT context (see DAY/NIGHT switch) — the night's
sub-phases: NON_REM_1 → ACTION (build⇄release), NON_REM_2 → RECOVERY
(import+free), REM → PREPARATION (replay+measure+prime).
POST INVERTS THE DAY MAP. For the postsynapse the deed is RECEIVING, which happens whenever no
descending spike is arriving, so: NOT_AP → ACTION (open AMPA/NMDA, respond), AP → RECOVERY (the
brief descending spike, driven from above, restore + boost traces), NOT_SPIKE_TRAIN → PREPARATION.
POST's ACTION continues through the soma's REFRACTORY (refractory is the soma's state, not POST's;
from POST it is NOT_AP), which is how the spine charges up to ready the next spike.
VARIABLE TIERS (timescale = meaning; see logic_principles "The Timescale Ladder")
FAST (ms–s) immediate response fast_trace
@@ -459,63 +471,67 @@ The postsynaptic spine is the synapse's primary memory locus: it detects coincid
runs the calcium dynamics that decide potentiation versus depression, and requires the most
validation (three coincidences) before committing.
**POST's ACTION is the synaptic event (context NOT_bAP).** Integration is graded and ongoing
rather than spike-punctate, so POST's action-context is "glutamate present, no bAP": three calcium
sources feed the fast trace — AMPA current (small Ca, begins ejecting the NMDA Mg block) and NMDA
(large Ca, only on the local coincidence of depolarization + astrocyte D-serine + glutamate). The
action deposits the calcium trace and emits the two retrograde messages. Because POST's receptors
are physical coincidence detectors, two of its three tag-coincidences (astro D-serine, and the bAP
below) are detected *in the action*; only the organism's dopamine coincidence is left to evaluation.
**POST's ACTION is the synaptic response (context NOT_AP).** Integration is graded and ongoing
rather than spike-punctate, so POST's action-context is "no descending AP arriving": it opens its
channels and responds to whatever glutamate has arrived. Three calcium sources feed the fast trace —
AMPA current (small Ca, begins ejecting the NMDA Mg block) and NMDA (large Ca, scaled by the local
coincidence of depolarization + astrocyte D-serine + glutamate). The action deposits the calcium
trace and emits the two retrograde messages. Crucially POST's ACTION **continues through the soma's
refractory period**: the refractory belongs to the soma, not to POST — from POST it is simply NOT_AP,
so the spine keeps responding and integrating, charging up so the neuron is ready to fire again once
its refractory ends.
**bAP is a second, vertical action-context.** The soma's back-propagating spike arrives, adds
depolarization and calcium, and supralinearly amplifies an existing candidate — the soma's
confirmation that it fired, detected in the action-moment (instantaneous coincidence).
**POST's RECOVERY is the descending AP (context AP) — a brief event.** When the soma's spike arrives
it adds depolarization and calcium, supralinearly boosts an existing candidate (the soma's
confirmation that it fired), and resets the spine: calcium extrudes, the NMDA Mg-block re-establishes
as Vm falls, and the trace relaxes — restoring the ability to respond again. This is the *alter-ego*
of the response, and it is short (the spike event), not the whole refractory window.
**EVALUATION FOLDS INTO PREPARATION.** POST runs the same three categories as every component.
ACTION is the synaptic response (integrate glutamate, detect the instantaneous coincidences, emit
the retrogrades). RECOVERY restores the ability to respond again — calcium extrusion and NMDA
Mg-block re-establishment, the fast alter-ego of the response. PREPARATION shapes what comes next:
fills AMPA surface toward the slot ceiling from accumulated calcium (short-term potentiation, no
**EVALUATION FOLDS INTO PREPARATION (context NOT_SPIKE_TRAIN).** PREPARATION shapes what comes next:
it fills AMPA surface toward the slot ceiling from accumulated calcium (short-term potentiation, no
dopamine — for the next response), climbs the tag on the dopamine coincidence (for the night),
refills budget, and settles. A fuel shortfall while calcium was climbing toward a tag is endurance
evidence (a preparation deposit made during action); a surface already at its ceiling is a
structural limit, not endurance.
**During NIGHT — the same three categories, turned inward.** RECOVERY imports material/energy and
primes AMPA responsiveness from the standing tag. PREPARATION replays: POST responds to the
re-evoked glutamate exactly as it responds by day (same AMPA/NMDA machinery, same endurance
deposit into the same trace), reading the response as participation rather than transmitting — no
dopamine. ACTION is the structural change: general homeostatic lowering, then rebuild where the tag
stands and participation was confirmed, consuming the tag. Both ceilings draw the same finite pool
and compete; unmaintained ceilings drift down.
**During NIGHT — the same three categories, turned inward, in sleep-stage contexts.** NON_REM_2
(RECOVERY) imports material/energy and primes AMPA responsiveness from the standing tag. REM
(PREPARATION) replays: POST responds to the re-evoked glutamate exactly as it responds by day (same
AMPA/NMDA machinery, same endurance deposit into the same trace), reading the response as
participation rather than transmitting — no dopamine. NON_REM_1 (ACTION) is the structural change:
rebuild where the tag stands and participation was confirmed (consuming the tag) ⇄ release where
participation was low. Both ceilings draw the same finite pool and compete; unmaintained ceilings
drift down.
```
// PARAMETERS K_AMPA · AMPA_Ca · AMPA_cost · NMDA_cost · bAP_cost · pka_cost · traffic_cost
// PARAMETERS K_AMPA · AMPA_Ca · AMPA_cost · NMDA_cost · AP_cost · pka_cost · traffic_cost
// req_cost · Mg_eject · K_Ds · Ca_STP · Ca_TAG · eCB_thr · drift · baseline
// NO_synth_cost · eCB_synth_cost
// INTERFACE
// EMIT retro_NO (+), retro_eCB (−) → PRE
// RECEIVE (signals) glutamate ← PRE ; astro_Dserine ← ASTRO ; bAP ← DEND/SOMA ; dopamine
// READ glutamate ; astro_Dserine (GAIN, not gate) ; bAP (dend_structure.bAP_fidelity) ; dopamine
// RECEIVE (signals) glutamate ← PRE ; astro_Dserine ← ASTRO ; AP (descending) ← DEND/SOMA ; dopamine
// READ glutamate ; astro_Dserine (GAIN, not gate) ; AP via dend_structure.bAP_fidelity ; dopamine
// OWN post_structure{slot_ceiling, spine_volume, reserve_ceiling} ; post_budget_ceiling
// SUPPLY astro_lactate[syn] ← ASTRO ; dend_ship_post ← DEND ; post_material ← DEND(NIGHT) ; post_energy ← SOMA(NIGHT)
// EMERGENCY shockwave_lockdown ← ASTRO
// NOTE POST endurance is own-state only (own Ca climbing); no arrived feedback term.
// coincidences sort by timescale: D-serine/bAP detected IN ACTION (instantaneous); dopamine in PREPARATION.
// THE THREE CATEGORIES (same at DAY and NIGHT; here DAY, subject = arrived glutamate / bAP):
// ACTION respond: integrate glutamate, detect instantaneous coincidences, emit retro (NOT_bAP);
// bAP is a second, vertical action-context (soma's spike confirms)
// RECOVERY restore the ability to respond: Ca extrusion + NMDA Mg-block re-establish
// PREPARATION shape what comes next: AMPA fill (next response) + tag climb (NIGHT) + refill + settle
// Coincidences sort by timescale: D-serine/bAP detected IN ACTION (instantaneous); dopamine in PREPARATION.
// THE THREE CATEGORIES (same at DAY and NIGHT; here DAY, subject = arrived glutamate / AP):
// ACTION (NOT_AP) respond: open AMPA/NMDA channels, integrate glutamate, detect the
// instantaneous coincidence, emit retro. This is POST's defining deed and it happens
// ALWAYS EXCEPT during an arriving AP — INCLUDING through the soma's refractory period:
// refractory is the SOMA's state, not POST's; from POST it is simply NOT_AP, so POST
// keeps responding and integrating, charging the spine so the neuron is ready to fire
// again once refractory ends.
// RECOVERY (AP) the BRIEF descending-spike arrival (ms): POST is driven from above, restores its
// ability to respond (Ca extrusion, Mg-block re-establishes as Vm falls), and deposits/
// boosts traces. Short event, not a long window.
// PREPARATION (NOT_SPIKE_TRAIN) shape what comes next: AMPA fill (next response) + tag climb (NIGHT)
// + refill + settle.
// Coincidences sort by timescale: D-serine detected IN ACTION (instantaneous); dopamine in PREPARATION.
// ===== ACTION =====
DAY | bAP: // second action-context (vertical): soma confirms
Vm += bAP_depol × dend_structure.bAP_fidelity; post_budget -= bAP_cost
if post_possible_tag > Ca_TAG: post_fast_trace += bAP_Ca_boost() // amplify only if candidate present
DAY | NOT_bAP: // respond to arrived input (the defining deed)
// ===== ACTION (NOT_AP: open channels, respond — continues through the soma's refractory) =====
DAY | NOT_AP: // respond to arrived input (the defining deed)
a = sat(glutamate, K_AMPA)
AMPA_current = a × AMPA_surface; Vm += AMPA_current; post_budget -= AMPA_cost // SOURCE 1 AMPA
post_fast_trace += AMPA_Ca·AMPA_current // @cut affords: channel-opening frequency, response magnitude (quantity of response) [ms]
@@ -526,13 +542,15 @@ DAY | NOT_bAP: // respond to arrived
if Vm > eCB_thr:
retro_eCB += eCB_emit(Vm); post_budget -= eCB_synth_cost // EMIT − brake
// ===== RECOVERY (restore the ability to respond; alter-ego of the response) =====
DAY | NOT_bAP · recovered: // Ca extrusion + Mg-block re-establish
post_fast_trace *= decay(ms) // FAST — Ca extruded, trace relaxes
// ===== RECOVERY (AP: brief descending-spike arrival — restore + boost traces) =====
DAY | AP: // soma's spike arrives (ms event), driven from above
Vm += AP_depol × dend_structure.bAP_fidelity; post_budget -= AP_cost
if post_possible_tag > Ca_TAG: post_fast_trace += AP_Ca_boost() // boost only if a candidate is present
post_fast_trace *= decay(ms) // Ca extruded, trace relaxes — ready to respond again
// (NMDA Mg-block re-establishes as Vm falls — implicit in the Vm>Mg_eject gate next response)
// ===== PREPARATION (shape the next response AND the NIGHT) =====
DAY | quiet: // sustained quiet
// ===== PREPARATION (NOT_SPIKE_TRAIN: shape the next response AND the NIGHT) =====
DAY | NOT_SPIKE_TRAIN: // sustained quiet (emitted by NEURON — above SOMA)
// for NIGHT: climb the tag; dopamine is the integrable coincidence (#3)
if post_fast_trace > Ca_TAG: post_possible_tag += post_fast_trace; post_budget -= pka_cost // @cut affords: participation / consistency of co-activity (running average) [min]
if dopamine > dop_thr and post_possible_tag > tag_thr:
@@ -557,7 +575,7 @@ DAY | quiet: // sustained quiet
// ACTION (same AMPA/NMDA machinery, same endurance trace), read for participation not significance.
// ===== ACTION (build ⇄ release; participation gates direction; tag funds build, persists across cycles) =====
NIGHT | build or release:
NIGHT | NON_REM_1:
if rest_permission and post_tag > tag_expiry and post_participation ≥ MEDIUM: // BUILD (slice)
Δ = min(slot_batch, post_material, post_energy·f_cap, post_tag) × post_participation
post_structure += Δ; post_material -= Δ; post_energy -= Δ·assembly_cost
@@ -573,7 +591,7 @@ NIGHT | build or release:
// else: HOLD — tag waits for its pattern
// ===== RECOVERY (acquire material/energy in CONTENTION with other components; receive freed material) =====
NIGHT | import + free:
NIGHT | NON_REM_2:
post_material += transit(post_material_ship, τ_transport_spine) // import (contested pool)
post_energy += transit(post_energy_ship, τ_transport_spine)
post_material += post_freed_material; post_freed_material = 0 // reclaim what release freed
@@ -584,7 +602,7 @@ NIGHT | import + free:
post_structure += min(post_maint, maint_cost) // hold up what is used
// ===== PREPARATION (REPLAY the response — SAME machinery as day ACTION; multi-timescale) =====
NIGHT | replay + measure + prime:
NIGHT | REM:
// FAST: replay the response to the re-evoked input (probe)
if arrived_glutamate_replay or arrived_replay_AP:
a = sat(glutamate, K_AMPA)