Nine wiring terms gave no clear added gain in a ridge test of paired fluorescence changes in 87 C. elegans worms

We tested whether reference wiring helps predict a target cell's fluorescence change at a light pulse relative to a time 15 seconds earlier in the same recording. We made paired window summaries from all 110 processed NWB files in DANDI 001075, version 0.240930.1859. The selected main panel has 10,625 target observations from 500 pulses in 87 C. elegans worms. Each test animal stayed outside fitting and tuning. A ridge reference used cell identity, recording terms and before-center fluorescence. Nine added wiring terms described direct weights, two-step routes and positive walks through all 302 Cook neurons. The planned test gave no clear gain and selected the lowest penalty. An added wider grid improved the reference fit but retained no clear wiring gain: error decreased by -0.107%, with a central 95% whole-animal range from -0.713% to +0.427%. Wiring products with measured source responses also gave no clear gain. Changed shifts, stricter labels, another animal split and shorter disjoint windows did not give a clear positive wiring gain. Mean measured change was higher at actual pulses than at shifted times. These shifts are not randomized shams and can contain earlier pulse effects. Some primary windows overlap; the added short check removes that overlap. Ranges keep fitted predictions fixed and omit training and selection uncertainty. The result limits these features and this selected short-window assay, not anatomical mechanisms or dynamical models in general. The package preserves paired data, code, source hashes, model states, test predictions, figures and repeat instructions.

connectome · Caenorhabditis elegans · signal propagation · time controls · calcium imaging · ridge regression · animal validation · reproducible data