Shared input noise changed from a loss to a gain with larger readouts in two passive adult worm models

We tested uniform shared input noise in passive electrical models of two partial adult worm maps with 180 common cells. Half of each independent input noise power was replaced by one shared current, with input variances fixed. At unit coupling and no readout noise, median one-cell ratios of mean contrast information were 0.625 and 0.617. Two-cell ratios were 1.270 and 1.274. Full observation gave an exact twofold gain for balanced inputs. A wiring-free control also gave a two-cell gain. With readout variance 0.1 and coupling 10, two-cell median ratios fell below one. We supply the covariance and information identities, random-panel scores, direct numerical checks and a regeneration method. This is a model result. It does not measure coding in worms. Uniform leak, uniform shared current, exact decoders and artificial contrast inputs limit biological use. External source tables and record-level transforms are excluded.

connectome · electrical coupling · shared noise · Fisher information · partial observation · C. elegans