Unknown failure identities lowered input-pattern ambiguity thresholds in eight partial worm maps

We tested binary chemical input patterns on a fixed panel of 162 cells in eight C. elegans brain maps. We compared ideal decoding with known and unknown failed source identities. The standard asymmetric coding distance gives the exact minimum failure count for ambiguity when those identities are unknown. We checked this rule in 17,820 small cases. Hiding failure identities lowered the mean threshold by 25.5 to 33.6 percent. Observed unknown thresholds were 16.9 to 24.1 percent below degree-control means and below every sampled draw. Longer runs, broad cell type controls and exclusion of direct left-right partner comparisons kept this order. These are structural model results, not measured neural representations or lesion responses. Spatial contact, partial coverage and one species limit the result. Raw data and graph extracts stay external; code, aggregate evidence, hashes and retrieval instructions are included.

connectomics · C. elegans · asymmetric coding · source failure · input patterns · reliability