With twofold leak bounds and gain 0.001–1000, the Cook gap model could not fit 218 C. elegans means within 1.50 sample SEM

We tested whether different neuron leaks remove a mismatch between electrical wiring and calcium responses. We used all 302 neuronal slots and 1,091 electrical links in the corrected Cook map. The fixed sample had 218 positive response means from 109 reciprocal C. elegans pairs. Each neuron had free positive input and readout gains. Each leak could vary independently between 1/sqrt(2) and sqrt(2). The global gap gain could take any value from 0.001 to 1000. No model in this family could fit all means within 1.50 sample standard errors. A free reciprocal transfer fit the same means within 1.25. Entrywise inverse bounds and four-response products excluded all models in 56 closed scale intervals. An independent checker used outward rounding and positive resistor elimination. Every checked cycle product was below one. Direct source checks reproduced all used graph weights and response summaries exactly. The SEM bands are sensitivity sets, not confidence regions across animals. The result applies to the stated static model, weight ratios, leak bounds, and selected response panel. It does not identify a missing biological process. The package preserves complete model inputs, code, numerical certificates, figures, source hashes, exploratory searches, and repeat instructions.

connectome · Caenorhabditis elegans · gap junction · passive electrical model · membrane leak · calcium response · matrix inverse bounds · reproducible research