Transcutaneous Spinal Cord Stimulation Disrupts Conscious Ankle Proprioception and Produces a More Constrained Locomotor Pattern in Unimpaired Adults

Transcutaneous spinal cord stimulation (tSCS) modulates spinal sensorimotor circuits primarily through activation of afferent networks. While prior work has emphasized locomotor performance and spinal excitability, how tSCS affects conscious proprioceptive perception and the extent to which such effects parallel changes in locomotor control remain unclear. We investigated the acute and training-related effects of tSCS on ankle proprioception and gait in unimpaired adults (n = 14), with an independent control group (n = 14) completing identical proprioceptive training without stimulation. Proprioception was quantified using a bilateral robotic assessment of dynamic ankle localization ability (Crisscross), gross motor output using maximum dorsiflexion strength, and gait during normal and tandem treadmill walking using spatiotemporal, trunk-sway, and mediolateral center-of-mass (CoM) excursion measures. Acute tSCS increased ankle proprioceptive error (p < 0.001) while dorsiflexion strength was unchanged (p = 0.30). Gait shifted toward a modestly more constrained locomotor pattern, characterized by reduced step width and ML CoM excursion (p < 0.05). With continued training under stimulation, proprioceptive error decreased and, unlike the control group, the tSCS group showed progressive improvement that persisted after stimulation ended. Sagittal-plane gait measures recovered toward or beyond baseline, whereas mediolateral measures remained constrained, revealing a direction-dependent reorganization of locomotor control. Together, these findings show that tSCS influences multiple aspects of the sensorimotor control loop, disrupting conscious proprioception while reshaping locomotor behavior, and that the nervous system can adapt to altered afferent input through training.

Improvement 1: Adaptive Afferent Perturbation Modeling for Motor-Learning Agents

• What to build: An AI training framework that injects controlled, temporary noise into the agent’s proprioceptive feedback (simulating tSCS-induced error) during early training, then gradually removes it.

• What the improved AI can do: Learn robust motor policies (e.g., for exoskeletons or prosthetics) that generalize to degraded sensory conditions, and show faster post-perturbation recovery than agents trained without such noise—mirroring the human tSCS group’s persistent improvement after stimulation ceased.

Improvement 2: Direction-Dependent Gait Constraint Encoder

• What to build: A neural network controller for bipedal robots that explicitly separates sagittal-plane (forward/backward) and mediolateral (side-to-side) control loops, with a learnable “constraint weight” for each plane.

• What the improved AI can do: Automatically tune lateral stability (step width, CoM excursion) to be more conservative during uncertain sensory conditions, while allowing sagittal speed/stride to recover to baseline—replicating the human result where ML measures stayed constrained but sagittal recovered.

Improvement 3: Proprioceptive-Error-Aware Reinforcement Learning Reward Shaping

• What to build: A reward function that penalizes the difference between predicted and actual joint position (proprioceptive error) during training, not just task success.

• What the improved AI can do: Develop internal models that explicitly account for altered afferent input, leading to more stable learning under sensor noise. The AI would show a “training-after-effect” where performance continues to improve even after the noise is removed, analogous to the tSCS group’s post-stimulation gains.

Improvement 4: Dual-Objective Training Scheduler for Sensorimotor Adaptation

• What to build: A curriculum that alternates between (a) high-error proprioceptive tasks and (b) gross motor output tasks (e.g., strength or speed), with a shared latent representation.

• What the improved AI can do: Learn to dissociate conscious perception from motor output—so that a robotic limb can maintain force/strength while its state-estimation module is being recalibrated. This prevents the common failure mode where improving one degrades the other.

Improvement 5: Predictive CoM Excursion Regulator for Balance Under Sensory Perturbation

• What to build: A model-predictive controller that uses mediolateral CoM excursion as a hard constraint (like the human’s persistent narrowing) while leaving sagittal dynamics unconstrained.

• What the improved AI can do: For humanoid robots or balance-assist devices, this AI will automatically tighten lateral margins when proprioceptive confidence drops, but not over-constrain forward motion—leading to safer, more natural walking during sensor failures or external perturbations.

Transcutaneous spinal cord stimulation (tSCS) modulates spinal sensorimotor circuits primarily through activation of afferent networks. While prior work has emphasized locomotor performance and spinal excitability, how tSCS affects conscious proprioceptive perception and the extent to which such effects parallel changes in locomotor control remain unclear. We investigated the acute and training-related effects of tSCS on ankle proprioception and gait in unimpaired adults (n = 14), with an independent control group (n = 14) completing identical proprioceptive training without stimulation. Proprioception was quantified using a bilateral robotic assessment of dynamic ankle localization ability (Crisscross), gross motor output using maximum dorsiflexion strength, and gait during normal and tandem treadmill walking using spatiotemporal, trunk-sway, and mediolateral center-of-mass (CoM) excursion measures. Acute tSCS increased ankle proprioceptive error (p < 0.001) while dorsiflexion strength was unchanged (p = 0.30). Gait shifted toward a modestly more constrained locomotor pattern, characterized by reduced step width and ML CoM excursion (p < 0.05). With continued training under stimulation, proprioceptive error decreased and, unlike the control group, the tSCS group showed progressive improvement that persisted after stimulation ended. Sagittal-plane gait measures recovered toward or beyond baseline, whereas mediolateral measures remained constrained, revealing a direction-dependent reorganization of locomotor control. Together, these findings show that tSCS influences multiple aspects of the sensorimotor control loop, disrupting conscious proprioception while reshaping locomotor behavior, and that the nervous system can adapt to altered afferent input through training.