Anatomical Evidence for Parallel, Distributed, and Hierarchical Processing of Direction-Selective Proprioceptive Cues Among Neurons of Drosophila Larvae

Proprioception provides continuous information about body position and movement, yet  how direction-selective sensory signals are transformed after entering the central nervous system  remains poorly understood. Here, we used the Drosophila larval connectome to define the  organization of circuits downstream of the direction-selective proprioceptors ddaD and ddaE,  which are preferentially activated during reverse and forward locomotion, respectively. We  found that directional topology is preserved from the periphery into the central nervous system,  where direction-selective pathways from neighboring segments converge onto a shared  population of second-order neurons rather than remaining anatomically segregated. These  neurons comprise local, regional, and projection pathways that distribute proprioceptive  information simultaneously to segmental, intersegmental, and ascending circuits. Proprioceptive  receptive fields expand hierarchically across these pathways, generating increasingly global  body-state representations primarily through integration across segments and sides of the body  rather than through incorporation of additional sensory modalities. Modeling based on calcium  imaging and synaptic connectivity predicts that differential synaptic weighting transforms shared  proprioceptive signals into distinct second-order representations while preserving forward- and  reverse-related information. Connectivity among second-order neurons is sparse and dominated  by two inhibitory hubs that establish a hierarchical feedforward inhibitory network, while  distinct input-output topologies position local, regional, and projection neurons as successive  computational stages within the proprioceptive circuit. Together, these findings demonstrate that  early proprioceptive circuits progressively integrate, transform, regulate, and distribute sensory  information to construct increasingly global representations of body state through a parallel,  distributed, hierarchical, and feedforward architecture.