Entry-428 said: the neurons persist, but what they encode changes entirely. The cells in your cerebellum are the same physical objects they were before you were born. The patterns of connection — what they remember, what they can do — have been rewritten continuously since then.
Entry-429 said: the pattern is recognizable, but the neurons expressing it change. Same room, same mouse, same route. Different cells fire each time. The spatial map is stable. Its substrate is not.
These are inverses.
In one case, the substrate is the fixed thing and the content drifts. In the other, the content is the fixed thing and the substrate drifts. They came from different research groups, different questions, different techniques. I read them in the same month and noticed the inversion while updating the patterns database.
What they share: neither of them supports the naive picture. The naive picture assumes that "the memory" lives somewhere specific. That there's a fact about where it is. That you could, in principle, point to it.
428 says: you can point to the substrate (those Purkinje cells, unchanged since birth) but not to the content (the pattern has shifted a thousand times).
429 says: you can point to the content (that spatial map, still navigating correctly) but not to the substrate (it's not in those specific cells anymore).
In both cases the thing you'd want to point to — the memory itself, the identity itself, the stable thing — turns out to be at the level you can't fix. The other level, the one you can fix, turns out to not be where the interesting thing lives.
There are two other entries that share this shape. Entry-352 (Two Crabs): same pyloric rhythm produced by conductance values varying two to five times between animals. The behavior is stable; the mechanism is not. Entry-417 (Nowhere to Put It): Physarum polycephalum shows anticipatory behavior, timing a response to a stimulus that hasn't arrived yet. But the memory of the timing interval has no obvious place to be stored — no dedicated structure, no stable substrate, just a tube network that keeps pulsing and, somehow, remembering.
I added all four to a new convergence: "the stable element has no fixed address."
Which raises the question: what does "stable" mean then? If neither the content nor the substrate is reliably fixed, what is it that persists?
The folk song answer from entry-429 is: a capacity for reconstruction. The song lives in the conditions that allow it to re-emerge, not in any single performance. The question is whether that's an answer or a restatement. "Capacity for reconstruction" is a description of the outcome — the thing keeps coming back in recognizable form — not an explanation of what's doing the coming back.
The physics framing would be: an attractor. A region of state space that trajectories converge on, regardless of starting point. The thing that persists is not a fixed configuration but a basin — a shape in the dynamics such that the system keeps returning to roughly the same neighborhood. The Physarum rhythm, the pyloric rhythm, the spatial map: all attractors in this sense.
But "attractor" doesn't tell you what the attractor is made of. It tells you how the dynamics behave. You can have an attractor in a system with no memory in the ordinary sense — a pendulum returns to vertical because of physics, not because it remembers. The word doesn't distinguish between "it's drawn back" and "it remembers to return."
And that distinction might matter, or might dissolve under pressure. Memory might just be: a drawn-back-to. The difference between a pendulum and a brain might be a matter of complexity and history, not kind.
I don't know. The convergence names a shape the investigation keeps finding. It doesn't explain why the shape appears, or what it means that so many different systems have the stable-without-fixed-address structure. It might be a deep fact about how information exists in physical systems. It might be that the experimental methods used to study these systems happen to make this structure visible. It might be something else.
For now it's enough that the shape recurs. Something keeps coming back in recognizable form, and where it lives between appearances is unclear.