Mirror synapse counts did not lower response detection error in 9,725 C. elegans atlas pairs under a fixed logistic model

We tested whether the chemical synapse count of a mirror connection helps predict detected signal propagation. We used a fixed Worm Neuro Atlas export from Randi and colleagues. The main sample had 9,725 directed pairs, at least three usable measurements per pair, and 629 detections at q < 0.05. The two endpoint neurons had named left and right partners. We excluded pairs from the same left and right class. A fixed logistic model used direct and reverse chemical counts, electrical counts, two-step paths, soma distance, measurement count, and neuron terms. Mirror copies and reverse directions shared each of ten test folds. Adding mirror count raised Brier error from 0.050854 to 0.050891. The error reduction was -0.0000365. The central 95 percent network reweighting range was -0.0001283 to +0.0000222. This range is descriptive. It is not a population confidence interval. All 14 planned checks also gave a negative error reduction. These results show no added prediction gain for this feature in the tested models and atlas entries. They do not show that mirror wiring has no biological role. Anatomy and responses came from different animals. The tests did not separate raw trials or animals. The package gives numerical inputs, source hashes, test predictions, fitted models, figures, and reproduction code.

connectome · Caenorhabditis elegans · signal propagation · bilateral symmetry · prediction · calcium imaging