Cluster Firing

When a city center becomes a neural network.

Three mechanisms, one chain

In the brain, three mechanisms work together as a single chain. They are often explained separately, which makes them seem like independent concepts. They are not. They are three stages of the same event: a signal traveling from trigger to response.

Here is the chain:

Remove any one of these and the whole chain breaks. Without the action potential, nothing fires. Without myelin, the signal crawls instead of jumping. Without saltatory conduction, the signal visits every node instead of skipping to the important ones.

Together, they answer three questions:

Should this fire? (action potential — does the evidence cross the threshold?)
How fast should it travel? (myelin — how thick is the insulation on this path?)
Where should it go? (saltatory conduction — skip to the next thick node, ignore the thin ones)

How this works in the brain

You smell coffee. That smell is a signal. It reaches your olfactory neurons, and if it's strong enough, they fire — that's the action potential. The signal now needs to reach the parts of your brain that know what coffee means: the kitchen, the morning routine, the café on the corner.

The path from your nose to those memories is wrapped in myelin — because you've smelled coffee thousands of times. Thick myelin. The signal doesn't crawl along the nerve fiber. It leaps from gap to gap, skipping the insulated sections entirely, regenerating only at the nodes. That's saltatory conduction. In milliseconds, the smell of coffee has activated your entire morning routine.

Now imagine smelling something you've only encountered once — a rare spice in a market in Marrakech. The signal fires (action potential), but the path has almost no myelin. The signal travels slowly, continuously, without jumping. It takes longer to find the memory. You might not find it at all.

The difference is not in the signal. It's in the path.

How this works in ThetaOS

The Life Lens System contains a database of an entire life. Transactions, locations, people, photos, text, check-ins. Every connection between two entities is a synapse. Every synapse has a thickness — determined by how many evidence layers confirm it and how often it has been activated. That thickness is the myelin.

When a trigger arrives, the system must decide: what fires, how fast, and where does the signal go? Those are the same three questions.

A city center as neural network

On 13 April 2026, I stepped off a bus in Haarlem and walked into the city center. What follows is real data.

The system knows 29 shops and restaurants in the Haarlem city center. Each one has a myelin thickness — measured by the number of bank transactions over the years. Here are the top ten:

Location	Transactions	Myelin
Apple Store	950
Etos	935
Bakker van Vessem	728
HEMA	~500
Jamin	342
Darras	216
IJssalon Garrone	187
Athenaeum Books	150
Gall & Gall	146
Solo's Hairstudio	112

And the bottom five:

Location	Transactions	Myelin
Van Vuure (cigars)	20
Restaurant Parck	15
Holland & Barrett	7
So Low	2
Barista Café	~10

Look at the difference. Apple Store has 950 transactions. So Low has 2. That's not a small difference — it's a 475x difference in myelin thickness. In neural terms: one path is a superhighway. The other is a dirt track.

The firing sequence

That day, I visited six shops. The bank recorded the transactions. Here's what happened in the system, described as a neural firing sequence:

Step 1 — Action potential. The first transaction arrives: €13.96 at Simon Levelt, Haarlem. The system recognizes the location as part of the "Haarlem city center" cluster. The trigger crosses the threshold. The cluster fires.

Step 2 — Saltatory conduction. The signal doesn't activate all 29 locations equally. It jumps to the thickest paths first: Apple Store (950), Etos (935), Van Vessem (728). These are the nodes of Ranvier — the points where the signal regenerates. The system checks: is there a transaction at any of these today? Apple Store: yes, €299.95. The signal strengthens.

Step 3 — Inhibition. The 23 locations without a transaction today are dampened. Not forgotten — dampened. They stay in the cluster but don't fire. Just like in the brain: inhibition prevents one signal from activating everything. Without it, every query would return all 29 locations every time. That's epilepsy, not intelligence.

Step 4 — Payload. Each fired location delivers its payload based on type:

Six transactions. One trigger. The system knew which shops to expect, which ones to check first, and what to do with each answer. It didn't search 29 locations one by one. It jumped along thick paths and ignored thin ones. That's saltatory conduction on a shopping street.

Three types of clusters

The city center is one type of cluster. We've identified three so far:

Location cluster — same place, different shops. Haarlem city center: 29 locations, ~4,700 transactions. Triggered by "went into town" or by a transaction at any member location.

Experience cluster — same place, different people and memories. Vlieland: 385 days on the island, dozens of people (Peter Ros, family, friends), locations (De Zafant, Stortemelk, the ferry), activities (cycling, dinners, writing). Triggered by "Vlieland" or by a ferry booking.

Transit cluster — a hub that connects other clusters. Harlingen: the ferry terminal, the parking lot, the supermarket for last-minute supplies. Thin myelin on its own, but it's the gateway to the thick Vlieland cluster. The signal passes through on its way somewhere else.

Each type fires differently. A location cluster fires wide — many locations, shallow depth. An experience cluster fires deep — fewer locations, but rich with people, emotions, stories. A transit cluster fires fast — it's a relay station, not a destination.

Why this matters

Every personal knowledge system stores connections. Links in Obsidian. Relations in Notion. Contacts in a CRM. But they all treat every connection the same. A link is a link. A contact is a contact.

The brain doesn't work that way. It has thick paths and thin paths. Fast signals and slow signals. Clusters that fire together and connections that stay silent unless specifically summoned. The brain is not a database — it's a weighted, myelinated, selectively firing network.

That's what the Life Lens System builds. Not just the connections, but the speed of the connections. Not just the data, but the readiness of the data. When I walk into a city center, the system doesn't wait for me to tell it what happened. It sees the first transaction and fires the whole cluster — thick paths first, thin paths inhibited, payloads ready.

The three mechanisms are not three features. They are one chain:

Structure enables behavior. Behavior waits for a trigger. The trigger activates the chain. One mechanism, three stages, real data.

The 29 shops in Haarlem are not a list. They are a neural network, weighted by eight years of bank transactions, ready to fire the moment I step off the bus.

Written 13 April 2026. Data: 29 locations, ~4,700 transactions (2018–2026), SQLite database on a single VPS. Total storage cost: €6.92/month.