Locomotor adaptation on a split-belt treadmill: mechanisms, modulation, and clinical utility

Human locomotion exhibits remarkable adaptability, allowing individuals to dynamically adjust their gait patterns in response to changing environmental demands. Locomotor adaptation on a split-belt treadmill has been a widely studied motor learning technique where two independent treadmill belts move at different speeds, generating adaptation of stepping symmetry over time. This review synthesizes current knowledge on how distinct neural substrates modulate gait in response to the split-belt treadmill through reactive and adaptive processes, highlighting the cerebellum's role in forward model recalibration driven by sensory prediction errors. Particular emphasis is placed on integrating findings across all investigated modulators of locomotor adaptation, including error size, sensory environment, visual feedback, neuromodulation, and cognitive demands, examining both well-established effects on adaptation dynamics and areas where knowledge remains limited. Despite considerable research on the locomotor adaptation paradigm with robust effects on the treadmill, the limited transfer of locomotor adaptation to overground walking remains a major clinical barrier, likely due to the sensory differences between walking contexts. Recent evidence supporting a credit assignment framework is discussed, which suggests that the nervous system attributes motor errors to either shared or context-specific forward models, influencing generalization. Understanding and manipulating this mechanism, with a focus on the sensory environment during adaptation, may be essential to improving the clinical utility of locomotor adaptation and enhancing neurorehabilitation strategies aimed at restoring symmetrical walking in neurological populations.