In two freely moving C. elegans data sets, wiring flow to body-wall muscles predicts how much behaviour explains head neuron activity, but not which body behaviour is encoded

Question: does the position of a C. elegans head neuron in the wiring diagram tell which ongoing behaviour its activity encodes? Approach: we calculated a damped random-walk flow from each neuron to body, head-neck and pharyngeal muscles in the Cook 2019 connectome. We compared it with the behaviour-explained variance (shift-corrected R^2) of NeuroPAL-identified neurons in two freely moving data sets from one laboratory: Atanas 2023 (40 animals, 134 neurons) and Dag 2023 (8 animals, 110 neurons). Nulls: 1,000 degree-preserving rewired connectomes, and 1,000 rewirings that keep each neuron's muscle edges. Results: the pre-specified target-specific contrast was 0.43 in Atanas and 0.12 in Dag, but it came mainly from the pharyngeal versus body-wall division. With a neuron-category control it was 0.08 (p = 0.06) and -0.09. In non-pharyngeal neurons, flow to head versus body muscles did not separate head-bending from locomotion encoding. In an exploratory test, total flow to body-wall muscles predicted total encoding: Spearman rho 0.50 and 0.48, partial rho 0.32 in both sets with category control. The real wiring exceeded rewirings with fixed muscle edges (partial null means 0.13 and 0.17; p = 0.004 and 0.013). The relation was positive in 40 of 40 and 8 of 8 animals, and it occurred within sensory neurons and interneurons. Motor flow did not predict cross-animal reliability of encoding. Limitations: one connectome animal, one laboratory, head neurons only, an exploratory gradient test, and a gradient size that depends on the R^2 definition (category-controlled 0.11 to 0.27 with plain or cross-validated R^2).

C. elegans · connectome · whole-brain imaging · behaviour encoding · motor hierarchy · structure-function · NeuroPAL · degree-preserving rewiring · random walk flow · freely moving · null result