Contributions of asymmetry and instability to transfer and retention following locomotor adaptation

Gait asymmetry resulting from neurological injury is more costly and less stable than healthy gait. Split-belt treadmills, which drive limbs at different speeds, lead to spatial and temporal gait asymmetries, and perturb walking balance, have been used to study locomotor adaptability and learning related to asymmetry and stability. This knowledge may be leveraged to design more effective rehabilitation protocols. In experiment 1, we asked how constraining stride-rates away from preferred during split-belt walking influences learning revealed in a retention test. We found that constraints to stride rate during asymmetric walking uncovered the capacity to leverage redundant degrees-of-freedom for walking control, largely conserving asymmetry and stability. We also found learning tended to be greater for those who performed worse during initial training. This work highlights the flexibility of human gait and further suggests that locomotor training benefits those who need it most. In experiment 2, we asked if the contributions of asymmetry and instability could be adapted and learned independently. We found evidence of both domain-specific adaptation and learning, but for only temporal control and backwards instability. However, this finding of domain-specific adaptation and learning was found simultaneous to significant associations between spatial, temporal, and stability measures. Overall, we found evidence of domain-specific learning, but strong relationships remain between measures of spatial, temporal, and gait stability control, highlighting their codependent nature. In experiment 3, we asked how added variability without added asymmetry in split-belt walking could facilitate learning for novel walking conditions. We found that variable training led to similar gait asymmetry and less instability despite greater variability than fixed split-belt walking. 24-hours later, we found similar learning between groups, restricted to measures of forwards instability and temporal asymmetry. Here, we demonstrated similar locomotor learning was achieved for temporal and stability control using a variable training paradigm, with only 20% of the time in asymmetry. In sum, we demonstrated the highly adaptable nature of human walking, highlight the capacity to leverage redundant degrees-of-freedom to achieve and learn more symmetric and stable gait in a way that may prioritize performance within one domain (i.e., stability) without compromising adaptability in others (i.e., asymmetry)