Independent contact loss gave higher mean pair reachability than shared site loss in eight partial worm map models

We tested directed route survival in eight public C. elegans brain maps on a fixed panel of 162 neurons. Each chemical synaptic site had one source and one or more targets. We compared loss of whole sites with independent loss of each site-target contact. Both models had the same survival probability for every directed edge. At an assumed failure probability of 0.5, independent contact loss raised mean pair reachability by 0.22 to 4.70 percentage points. All eight nominal 95 percent Monte Carlo intervals excluded zero. There were 4,000 draws per model, map and failure probability. Five probabilities gave 320,000 graph draws. A conditional replacement proof shows that independent contact loss cannot lower single-target reachability under the stated assumptions. A fork example shows that shared loss can instead raise joint reachability of two targets. Prior percolation work already shows that equal edge probabilities need not imply equal propagation. Our contribution is a checked site-resolved application and finite-map effect sizes. Failure rates and complete sharing are assumptions. No release, activity or behavior was measured. Partial maps and eight different animals prevent a developmental or cross-species claim.

connectomics · C. elegans · polyadic synapses · reachability · correlated failures · reliability