Neuromemoriology: A Closed-Loop Hypothesis of Understanding with Retention
Revision 2, 30 October 2026. Rebuilt around the closed-loop model. The original (neuromemoriology_hypothesis.md) is unchanged. No literature citations have been added yet; every "established" item below still needs a primary source attached before this is shared.
1. The model (the operator's hypothesis)
Memory and understanding are one closed loop with two halves and a linkage:
- Half A, neural: the central nervous system's cells (neurons, plus glia and their lipid handling).
- Half B, spinal and fluid: the spinal cord and the cerebrospinal fluid (CSF) system, including the glymphatic route (the brain's fluid-based waste-clearance pathway).
- Linkage: the brain, which joins A and B.
The loop is redundant and cross-referencing: each half can check or support the other. It operates at every level, from nightly cleanup and low-order motor support, through high-order motor function, up to the highest abstraction. The operator's collective name for what the loop produces is understanding with retention.
Status: hypothesis. The sections below separate what published biology supports from what the model adds.
Gaps the model must close before it is a mechanism
- Define "understanding with retention" as one measurable thing. At present it covers motor support and abstraction alike. The model must say what differs between levels, otherwise the claim has no content.
- Name what crosses between the halves. Candidates: CSF-borne signals, lactate and other soluble substances, lipids, electrical activity. One carrier needs to be chosen, or each assigned to a level.
- State the dissociation prediction. If the halves are redundant, impairing one should leave function largely intact, and impairing both should fail sharply. If the loop fails as a single unit whichever half is hit, it is a description, not a mechanism.
2. Established biology (supports the neural half)
| Claim |
Status |
| Astrocytes wrap synapses and respond to neuronal activity (the "tripartite synapse") |
Established |
| Neurons export peroxidised (oxidatively damaged) lipids; astrocytes take them up and store them in lipid droplets, in an ApoE-dependent process |
Established in animal and cell work |
| The ApoE4 variant (the strongest common genetic risk for late-onset Alzheimer's disease) impairs astrocyte lipid handling |
Established association; mechanism still under study |
| Myelin is lipid-rich insulation and sets conduction speed along an axon |
Established |
| Lipid-derived messengers (endocannabinoids) act retrogradely across synapses |
Established |
| Nerve conduction speed depends on temperature (falls measurably with cooling) |
Established |
| Astrocyte-derived lactate is needed for long-term memory formation in rodents |
Established in rodents |
3. Plausible but contested (needs care)
| Claim |
Why it is contested |
| The astrocyte-neuron lactate shuttle is the main way neurons are fuelled during activity |
Neurons also raise their own glucose use when active; the size of the shuttle's role is debated |
| Lactate signals plasticity genes |
Evidence is stronger for Arc and related genes via NMDA receptor signalling than for CREB |
| Glymphatic clearance is faster in sleep |
Original findings said yes; some 2024 mouse work found clearance slower under sleep and anaesthesia. The nightly-cleanup link should not be load-bearing until this is settled |
| Sensory endings in tissue act locally as well as reporting inward |
Plausible: peripheral sensory terminals release peptides such as CGRP locally. This is "acts locally and reports", not "terminates" |
| Sympathetic nerves form close junctions on fat cells |
Sympathetic innervation of fat is established; the term "half-synapse" is this project's own label |
4. Speculative (original Sessions 5 to 7 and part of 8)
These were the original document's most confident claims and its least supported. They are kept as an appendix to the model, not its foundation.
- Fat as a distributed heater for the whole nervous system. Against it: adult humans have little brown fat; core temperature and blood flow already hold brain temperature; limb nerves vary with environment. A test would be whether animals or people with impaired brown-fat heat production show degraded motor learning. No supporting evidence is known.
- Large glial lipid droplets as a protective forcefield. Against it: in healthy tissue astrocyte droplets are small and sparse; large accumulations appear in ageing, ApoE4 and neurodegeneration. Droplets as waste containment is supported; "enormous droplets as protection" is contradicted by the pathology.
- Sensory signalling "terminating" at fat as a design feature. A sensory ending is the start of the afferent path (cell body in the dorsal root ganglion, signal on to the spinal cord). See the local-release point in section 3.
- Lactate as a "complex synthesis from signal data". Lactate comes from glycolysis, one of the simplest pathways. The real, supported part is the coupling: glutamate uptake brings sodium into the astrocyte, which switches glycolysis on. That is regulated metabolism, not synthesis of a new carbohydrate from signal data.
5. Corrections to the original
- Session 5 said myelin insulation leads to the nerve-fat junction. Sympathetic fibres to fat are unmyelinated. Removed.
- Session 5 conflated "neural adipocytes" with peripheral white, beige and brown fat. The adult brain has no true adipocytes; the original's own Session 4 says so. The brain-side equivalent is the astrocyte lipid droplet system.
- The "ten-step glycolysis bottleneck" argument (Session 8) is wrong: the astrocyte runs the same steps and passes lactate on so the neuron can skip them.
- Subtitle claimed neuroimmunity; the content has one passing mention. Dropped.
- Typos fixed: broken arrow in Session 1, "Astrophic" in the Session 3 title.
6. How this sits with the lipid-droplet model
The operator's separate model holds that the lipid droplet is the only memory store, and that memory is reconstructed across the brain. This document supports glia and lipid droplets as maintenance for memory machinery, not as a store. That tension is open and should be stated, not smoothed over:
- If droplets are the store, the evidence in section 2 is relevant only indirectly, and the model needs its own evidence.
- If droplets are maintenance, the closed loop explains retention without a store, and the two models are compatible.
7. Suggested tests (what would count against the model)
- Impair CSF and glymphatic clearance while leaving neural signalling intact; look for a retention deficit that appears before any neural loss.
- Impair astrocyte lipid-droplet handling (as in ApoE4 models) while leaving clearance intact; compare the pattern with test 1. If the two deficits are indistinguishable, the halves are not separable and the redundancy claim fails.
- Compare effects across levels: low-order motor learning, high-order motor learning, abstract learning. The model predicts a different balance between the halves at each level; no difference would undermine the "every level" claim.
Source note: biology in sections 2 and 3 is stated from general knowledge as of June 2026 and still needs primary citations.