Every cell remembers, and knows where it sits

Generated from this site's git history. with Claude means the commit carries a co-author line for an Anthropic model: an LLM took part in that change, not necessarily in every word of it, and git cannot say which. The same history on GitHub →

1. v10 2026-10-16 06:39 Refine speculations and homepage feature list Joshua Anderton +4 −2 words

   The second is what happens when that scales up, to neurons and to specialised cells like photoreceptors. My picture is that every cell in the body stores memory, and the body shares it like a decentralised network, pulling up what it needs when it needs it. The work of attending to a stimulus, recalling, or expressing genes leaves a byproduct, and I think that byproduct is long-term memory across the senses.

   I also suspect this is coordinated by a morphogenetic rule: some mathematics tied to how many cells there are and what it takes to hold them together. That rule would pick out the cells at the edge, and give every cell a sense of where it sits in the whole. If so, memory and spatial awareness aren'tmight not be extras. They could come built in once cells are coordinated.

   Alone, a cell probably meets the world much as it does in a crowd. But pack in thousands or millions, and network effects appear. Every cell polarises, trades ions and carries a small electric field, so it can't help affecting its neighbours. Packed densely enough, it has to pass on what it holds, and the network keeps passing it on until it looks like replay.

   Commit 6c53aaa on GitHub →

2. v9 2026-09-29 05:24 Remove legacy redirects and footer wordmark Joshua Anderton +0 −10 words

   ---

   redirect_from: /rants/every-cell-remembers-and-knows-where-it-sits/

   title: "Every cell remembers, and knows where it sits"

   Commit ee0d9c6 on GitHub →

3. v8 2026-09-28 16:20 Rename Rants to Speculations (/speculations/), with redirects from /rants/ Joshua Anderton with Claude Opus 5.5 +10 −0 words

   ---

   redirect_from: /rants/every-cell-remembers-and-knows-where-it-sits/

   title: "Every cell remembers, and knows where it sits"

   Commit bb62419 on GitHub →

4. v7 2026-09-25 08:37 Rants: publish both Joshua Anderton with Claude Opus 5.5 +0 −2 words

   tier: SPEC

   published: false

   ---

   Commit 05f64cd on GitHub →

5. v6 2026-09-25 08:32 Rant: plainer sentences in "Every cell remembers" Joshua Anderton with Claude Opus 5.5 +133 −258 words

   ---

   There are two things I need to understand. The first is habituation,: and whetheris it is the wayhow every single cell stores information.? My guess is that onea single cell can hold long-term informationmemory, and that habituation is the formatform it holds it intakes.

   The second is what happens when that isscales extrapolated outward: to different kinds of cellsup, to neurons, and to specialised cells like photoreceptors that specialise in light. My working picture is that every cell in the body stores memory, and the wholebody distributesshares that memoryit like a decentralised network, coordinatingpulling theup informationwhat it needs when it needs it. The processingwork itselfof (attending to a live stimulus, to memoryrecalling, toor geneexpressing expression)genes hasleaves a byproduct, and I think that byproduct is long-term, multisensory memory across the senses.

   I also suspect that the way a cell saves habituation, and salience, which is roughly meaning,this is coordinated automatically by somea morphogenetic rule: a kind ofsome mathematics tied to thehow number ofmany cells inthere a systemare and the coordinationwhat it takes to hold them together. Part of thatThat rule would bepick identifyingout the cells at the periphery.edge, Andand give every cell, not only specialised ones, would carry an embeddeda sense of where it sits in the whole. IIf think that may be a requirement rather than a feature: once cells are coordinatedso, memory and spatial awareness aren't extras. They come built in once cells are coordinated.

   AAlone, singlea cell probably meets itsthe surroundingsworld much as it does in mucha the same way whether it is alone or notcrowd. But groupedpack within thousands or millions, of others,and network effects fallappear. out of the way aEvery cell works. A cell that polarises, exchangestrades ions and carries a small electric field, affectsso it can't help affecting its surroundings by definitionneighbours. Packed densely enough, it is forcedhas to pass on what it holds on to its network, and the network iskeeps forcedpassing toit propagate that memoryon until it becomes somethinglooks like replay.

   TheHoneybees are the comparison I keep coming back to is honeybees and other colonies. A systemcolony gets big enough that it has to communicate, and thereat seems to be asome threshold, morphological and mathematical, where it suddenly takes a shape. The shape I have in mindimagine is a pair oftwo feedback loops running across the whole network, one for salience and one for habituated memory, that the whole network runs together. The coordinatedshared signal might be what you get when everyeach cell is requiredhas to exchangeswap its habituated state for the unhabituated state of a signalfresher one nearby: to tradetrading its version of the story for the waggle dancer's.

   The last piece is the peripheryedge. I think the outer cells becomeare the first line of defence. They pick outsort external reward, external stimulus, and externalstimulus informationfrom things that can be habituated, which is to sayor silenced. Silencing may be one of their most important jobs, becausejob: once the outside is ruled out, the whole system can findspot deviantwhat signalsis unusual inside itself. It can coordinate its own timing and structure so that every cell, not just the peripheral ones, can isolate what differs from what it expects to experience: vibrationVibration, bioelectric signalling, coupling, and firing. Allall of that hashave to be damped, or the system feeds back on itself and loses coherence.

   ## Where I'd start

   The pieces to simulate, or falsify: that single cells save information; that they coordinate it; that each cell knows its positionplace in the network; and that past some threshold of size, structure, geometryshape and complexity, a network is mathematically requiredhas to start communicating like this. The places to look are the networks that did or did notdidn't cross itthat line: slime mould, honeybees, antants, and termite coloniestermites, robot swarms.

   Commit 6e110e7 on GitHub →

6. v5 2026-09-25 08:32 Rant: remove repetition from "Every cell remembers" Joshua Anderton with Claude Opus 5.5 +96 −169 words

   ---

   There are two things I need to understand. The first is habituation, and whether it is in fact the way every single cell stores information. IMy wantguess to showis that one cell can hold long-term information, inand whateverthat format a cellhabituation is capable of holding it. My guess is that the format isit habituationholds it in.

   The second is what happens when that is extrapolated outward: to different kinds of cells, to neurons, to photoreceptors that specialise in light. My working picture is that every cell in the body is a system in which every cell stores memory, and the whole distributes that memory like a decentralised network., The network coordinatescoordinating the information it needs, when it needs it. But theThe processing itself (attending to a live stimulus, to memory, to gene expression) has a byproduct, and I think that byproduct is long-term, multisensory memory.

   I also suspect that the way a single cell saves habituation, and salience, which is roughly meaning, is coordinated automatically by some morphogenetic rule: a kind of mathematics tied to the number of cells in a system, and to the coordination it takes to hold them together. TherePart of that rule would be an automatic process for identifying the peripheral cells ofat athe systemperiphery. And itevery wouldcell, not only be specialised cells and neurons that track where the periphery is relative to their own location. Every cellones, would carry an embedded sense of where it sits in the whole. I think that may be a requirement rather than a feature: once cells are coordinated, memory and spatial awareness come built in.

   I think that may be a mathematical requirement rather than a feature: once cells are coordinated, memory and spatial awareness come built in. A single cell probably meets its surroundings in much the same way whether it is alone or not. But grouped with thousands or millions of others, there are network effects that fall out of the way a cell works. A cell that polarises, exchanges ions and carries a small electric field affects its surroundings by definition. Packed densely enough, it is forced to pass what it holds on to its network, and the network is forced to propagate that memory until it becomes something like replay.

   So the pieces I want to simulate, or falsify, are these:

   - single cells save information;

   - they coordinate it;

   - alongside it, each cell has to know its own context and position in the network;

   - the network is forced to propagate, or perturb, all of that memory in a way that turns it into something like replay.

   The comparison I keep coming back to is honeybees and other colonies. A system gets big enough that it has to communicate, and there seems to be a threshold, morphological and mathematical, where it suddenly takes a shape. The shape I have in mind is a pair of feedback loops, one for salience and one for habituated memory, that the whole network runs together. The coordinated signal might be what you get when every cell is required to exchange its habituated state for the unhabituated state of a signal nearby: to trade its version of the story for the waggle dancer's.

   ## Where I'd start

   WhatThe I want is a waypieces to showsimulate, or rule out,falsify: that asingle systemcells issave mathematicallyinformation; requiredthat tothey enter this kind of communication protocol oncecoordinate it; crossesthat aeach cell knows its position in the network; and that past some threshold of size, structure, geometry and complexity, a network is mathematically required to start communicating like this. The places to look are the networks that either did or did not cross it: slime mould, honeybees, ant and termite colonies, robot swarms.

   Commit 7cc0717 on GitHub →

7. v4 2026-09-25 08:32 Rant: fix terms in "Every cell remembers" Joshua Anderton with Claude Opus 5.5 +7 −8 words

   There are two things I need to understand. The first is habituation, and whether it is in fact the way every single cell stores information. I want to show that one cell can hold long-term information, in whatever format a cell is capable of holding it. My guess is that the format is habituation.

   The second is what happens when that is extrapolated outward: to different kinds of cells, to neurons, to cellsphotoreceptors that specialise in keeping information about light. My working picture is that the body is a system in which every cell stores memory, and the whole distributes that memory like a decentralised network. The network coordinates the information it needs, when it needs it. But the processing itself (attending to a live stimulus, to memory, to gene expression) has a byproduct, and I think that byproduct is long-term, multisensory memory.

   I also suspect that the way a single cell saves habituation, and salience, which is roughly meaning, is coordinated automatically by some morphogenetic rule: a kind of morphogenetic mathematics. Something tied to the number of cells in a system, and to the coordination it takes to hold them together. There would be an automatic process for identifying the peripheral cells of a system. And it would not only be specialised cells and neurons that track where the periphery is relative to their own location. Every cell would carry an embedded sense of where it sits in the whole.

   I think that may be a mathematical requirement rather than a feature: once cells are coordinated, memory and spatial awareness come built in. A single cell probably meets its surroundings in much the same way whether it is alone or not. But grouped with thousands or millions of others, there are network effects that fall out of the way a cell works. A cell that polarises, exchanges ions and carries a small electromagneticelectric field affects its surroundings by definition. Packed densely enough, it is forced to pass what it holds on to its network.

   So the pieces I want to simulate, or falsify, are these:

   The comparison I keep coming back to is honeybees and other colonies. A system gets big enough that it has to communicate, and there seems to be a threshold, morphological and mathematical, where it suddenly takes a shape. The shape I have in mind is a pair of feedback loops, one for salience and one for habituated memory, that the whole network runs together. The coordinated signal might be what you get when every cell is required to exchange its habituated state for the unhabituated state of a signal nearby: to trade its version of the story for the waggle dancer's.

   The last piece is the periphery. I think the outer cells become the first line of defence. They pick out external reward, external stimulus, and external information that can be habituated, which is to say silenced. Silencing may be one of their most important jobs, because once the outside is ruled out, the system can find deviantsdeviant signals inside itself. It can coordinate its own timing and structure so that every cell, not just the peripheral ones, can isolate what differs from what it expects to experience: vibration, bioelectric signalling, coupling, firing. All of that has to be damped, or the system feeds back on itself and loses coherence.

   ## Where I'd start

   Commit fa8105a on GitHub →

8. v3 2026-09-25 06:43 Essays: unpublish until the operator has previewed the layout Joshua Anderton with Claude Opus 5.5 +2 −0 words

   tier: SPEC

   published: false

   ---

   Commit 7e2253f on GitHub →

9. v2 2026-09-25 06:41 Essay: move timestamp before build time so Pages renders it Joshua Anderton with Claude Opus 5.5 +1 −1 words

   cells, past some size, has no choice but to start talking.

   date: 2026-09-25 0906:00:00 -0600

   tier: SPEC

   Commit bf8bfcf on GitHub →

10. v1 2026-09-25 06:37 Publish: A spiking network built to lose Joshua Anderton with Claude Opus 5.5 first version

    v1: the original version, as first committed. Read it on GitHub →

    Commit 141a921 on GitHub →