Small bounded errors in slow observations can hide gap-removal differences in stable linear connectome models
We tested whether known chemical filters ensure reliable prediction after electrical connections are removed. We derived a bound for a stable linear model with reciprocal gaps and filtered chemical currents. Cell parameters and the first-order observation filter were known. The bound covers every current history with a fixed amplitude limit. A four-cell example used 1 ms chemical and 0.5 s observation times. A response error of 0.01 permitted either intact circuit. Their steady gap-removal responses differed by exactly 10/21 in normalized voltage units. Any rule can therefore have error of at least 5/21 in one compatible circuit under the stated bounded-error model. The same change on the full 302-cell Cook neural support gave a sufficient shared-data error radius of 0.01720. Its gap-removal difference was 0.47448. Contact counts define support only. All functional weights are synthetic. The result extends prior fast-filter examples to complete continuous responses with fixed, known chemical times. It gives no trial-count limit for independent random noise and does not predict worm drug effects. The package contains the proof, a certificate tool, source facts, model matrices, figures, checks, and offline repeat instructions.
connectome · electrical synapse · chemical synapse · practical identifiability · observation filter · bounded error · intervention prediction · C. elegans · linear model
References
7- usesWith known leaks and reciprocal gaps, full voltage responses leave 26 coupling changes unresolved in a Cook C. elegans model with instantaneous chemical transmission sci:sha256:0c1c93618d6713502f3eae4aefb7443ba21acd429acb804f7548932c222ef542
We use its verified factual contact tables, balanced known-leak change and synthetic cycle recipe. We add chemical and observation dynamics and complete-response error bounds.
- supportsIn a calibrated 302-neuron model with one-pole chemical filters and source-shared times, one exact sine response matrix identified electrical and chemical weights sci:sha256:4c6684e491b7114c3026aa10e0fffbaf46f72a4cf8ed123866a25a37cf6a6112
It already shows weak fast-filter recovery and different gap-removal predictions from three response summaries. We extend this question to every bounded input history with fixed, known chemical times and a slow observation filter.
- usesdoi:10.1038/s41586-019-1352-7
The corrected July 2020 Cook and Jarrell workbook supplies the original contact support. We independently recheck all 12 section counts of the selected motif.
- doi:10.1002/dneu.22410
Prior work explains parallel electrical and chemical paths and different mechanisms with similar circuit function.
- doi:10.1093/bioinformatics/btp358
Prior work distinguishes structural and practical identifiability.
- doi:10.1016/0024-3795(75)90112-3
The diagonal-dominance inverse bound is standard. We use it in the frequency argument and give a time-domain proof separately.
- doi:10.1038/s41586-023-06683-4
The stimulation and calcium response atlas motivates the observation question. We use none of its response values.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.