Writing and Reading Plasticity Scars in Adaptive Oscillator Networks
Adaptive oscillator networks can retain past events in slowly evolving couplings. This study tests a preregistered write-erase-read protocol in a noisy adaptive Kuramoto network. Writing uses a strain-bearing coordination episode; collective order is then maintained while direct phase evidence is removed; the later trajectory is examined for recovery of the event class. The reset is an observability intervention rather than a natural process: it removes direct phase evidence so that any later separation must come from the retained adaptive state. The event generates the hidden coupling rather than receiving it as an initial condition. Two negative results are structural rather than incidental. Spectral colour does not identify the scar: the fitted high-frequency exponent stays near two across a fourfold range of coupling strength, and the exact deposited microscopic run gives an exponent of 2.007 with coupling already decayed below threshold by the spectral burn-in. The readout claim therefore rests on finite-time amplitude dynamics, not on spectral signature. Separately, the deposited learning rate cannot write supercritical coupling from weak initial coupling under the stated analytic bound, and the numerical control agrees; it is retained as a negative control. The two-study structure is explicit. An earlier experiment asked whether a prepared hidden coupling could be read. The present experiment asks whether weak coupling can be written by an event and then read. A source audit supports the first; the preregistered extension supports the second. This deposit contains the manuscript and supplement in PDF and DOCX form together with the complete research compendium: the authoritative recovered code, numerical caches, figure generators, preregistrations, world-level outputs, statistical reanalysis, environment records, manifests and cryptographic checksums. The earlier prepared-state source archive is preserved separately at Zenodo record 20574531.
Authors
- Joel Pearcey
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-09-13
- DOI
- https://doi.org/10.5281/zenodo.22737123
- Primary Topic
- Nonlinear Dynamics and Pattern Formation
- Type
- preprint