Wiring terms cut common calcium response error by 1.9% but gave no gain for left–right differences in a test across 102 C. elegans worms

We tested whether wiring predicts response differences between left and right partner neurons measured during the same stimulus. We matched 12,551 observations from 1,224 events in 102 C. elegans worms. Each test animal stayed outside model fitting and penalty selection. The main difference model selected no wiring terms in all five folds. On the same rows, wiring reduced error for the partners' mean transformed response by 1.90%. The central 95% range from 199 animal resamples with model refits was 0.64–3.18%. Folds and selected penalties stayed fixed. A later test forced finite wiring penalties for the difference outcome. Error rose by 0.15%, with a refit range from a 0.50% rise to a 0.09% fall. Common-response gains remained with stricter labels, a longer response window, and no ASE or pharyngeal endpoints. The models used cell identities, measured source response, and fluorescence controls. These are conditional estimates, not forecasts before stimulation. The reference anatomy came from other animals. Unequal cell gains, calcium dynamics, measurement error and missing physiology limit interpretation. The results do not establish biological symmetry. The package preserves matched data, source hashes, code, predictions, model states, figures and repeat instructions.

connectome · Caenorhabditis elegans · signal propagation · bilateral neurons · paired measurements · calcium imaging · animal generalization · structure and function · open data