The neural mechanisms behind movement have been puzzling for decades. While much focus has been on trying to explain the generation of locomotion little attention has been given to the ability to pause ongoing movement at any point while keeping posture. Here, we argue that within this ability to pause movement lies the key to understanding the generation of movement itself. The prevailing view suggests the existence of specialized modules for distinct functions. Nevertheless, recent observations, including rotational neural activity during rhythmic limb movements, challenge this perspective (Lindén 2022). To investigate how spinal neural networks execute locomotion and postural pauses, we utilized high-density electrophysiology in rat spinal cords during volitional locomotion, coupled with optogenetic perturbation of the pedunculo-pontine nucleus to induce movement arrests (Goni-Erro 2023). We present evidence to support the existence of a continuous attractor network (CAN) within the spinal network. We demonstrate that during locomotion, the neuronal manifold exhibits robust rotational patterns across animals and invariant to speed. Importantly, upon pause in movement the trajectory converges to a stable point-attractor precisely when the pause occurs and persists until the movement is continued. Through computational modeling, we argue that the network is analogous to a CAN, as observed in other parts of the nervous system e.g. grid cells in the entorhinal cortex (Gardner 2022). We finally suggest specific structural mechanisms by which the network is physically implemented to control locomotion and pause in the spinal cord of rodents, while accounting for the neuronal diversity.