AI & ML
Hebbian self-organization of monosynaptic reflexes in a biologically realistic multiarticular limb
arXiv cs.RO0 views
arXiv:2611.23441v1 Announce Type: cross Abstract: Spinal monosynaptic reflexes depend on patterns of connectivity between muscle spindle afferents and motoneurons that should reflect musculoskeletal dynamics. One possibility is that the details of these circuits are specified genetically, but this requires an unlikely coevolution of musculoskeletal and neural elements. Here we show how biomechanical structure can determine the connectivity of self-organized spinal circuits. Using a neuromechanical model combining a multi-joint limb, proprioceptive feedback, and Hebbian plasticity driven by motor babbling, we find that biologically realistic connectivity emerges spontaneously when babbling is neither too sparse nor too frequent. This results in strong homonymous projections, weaker heteronymous projections among functional synergists, and very weak connections among antagonists. This outcome also depends on musculoskeletal morphometry. Heteronymous synergies form when proximal muscles are strongest---the characteristic gradient of mammalian limbs--whereas excessively strong biarticular muscles lead to unphysiological antagonist coupling. The mechanisms differ: homonymous connections arise from direct sensorimotor correlation during fusimotor coactivation by beta motoneurons, while heteronymous connections depend on multi-joint dynamics that emerge only after homonymous reflexes stabilize and increasing muscle strength generates multisegmental motion. Overall, this study provides a mechanistic basis for fetal sensorimotor development and generates testable predictions about the consequences of naturally occurring and experimentally induced variations.
Source:https://arxiv.org/abs/2611.23441PreviousNeural Circuit Architectural Priors for Rat LocomotionNextDOCK: Dual-Stream-Inspired Self-Supervised Object-Centric Representations Grounded in Robot Kinematics
Related posts
TrajWAM: Trajectory-Conditioned Future Learning for Vision-Language Driving
0arXiv cs.RO
Adaptive Action Horizon Selection via Flow-Field Uncertainty
0arXiv cs.RO
Object Pose: The Link between Weak Perspective, Paraperspective, and Full Perspective
0arXiv cs.RO
Attention Is Not Relevance: Challenging the Causality of Attention in Autonomous Driving
0arXiv cs.RO
DOCK: Dual-Stream-Inspired Self-Supervised Object-Centric Representations Grounded in Robot Kinematics
0arXiv cs.RO
Neural Circuit Architectural Priors for Rat Locomotion
0arXiv cs.RO