For 47 C. elegans neuron pairs in 21 worms, detailed wiring gave no clear gain in a fixed test of left–right behavior differences

We tested whether detailed wiring predicts signed left-right differences in behavior correlations. We used all 302 neuronal slots and all positive neuronal weights in the corrected Cook hermaphrodite map. The primary activity sample had 47 bilateral pairs in 37 cell classes from 21 worms. First and last recording segments gave velocity-difference agreement r = 0.803, with a 95% animal and class resample range of 0.696 to 0.873. The same-pair mean from other worms increased combined difference prediction error by 6.27%; its gain range was -15.43% to +1.14%. For pair averages, that reference reduced error by 37.97%, with a range of +22.88% to +50.93%. A nested ridge test excluded both the test animal and its cell class. Adding detailed input, output, and electrical partner profiles reduced error over connection-total controls by 0.41%, with a range of -0.29% to +1.30%. Both anatomy models had higher point error than zero. Specified quality, sample size, feature, and split checks gave small gains or losses. These results test stated prediction rules. They do not establish equal cell function or a neural cause of agreement within recordings. The main limits are unmatched anatomical animals, static unsigned features, current behavior correlations, and persistent measurement or identity effects. Code, derived data, figures, predictions, source hashes, and repeat instructions are included. A fresh source run reproduced all 78 decoded numerical result files exactly.

connectome · Caenorhabditis elegans · bilateral neurons · structure and function · behavior correlations · calcium imaging · prediction · repeatability