This study aimed to determine whether inter-individual differences in learning rate of a novel motor task could be predicted by movement variability exhibited in a related baseline task, and determine which variability measures best discriminate individual differences in learning rate. Thirty-two participants were asked to repeatedly complete an obstacle course until achieving success in a dual-task paradigm. Their baseline gait kinematics during self-paced level walking were used to calculate stride-to-stride variability in stride characteristics, joint angle trajectories, and inter-joint coordination. The gait variability measures were reduced to functional attributes through principal component analysis and used as predictors in multiple linear regression models. The models were used to predict the number of trials needed by each individual to complete the obstacle course successfully. Frontal plane coordination variability of the hip-knee and knee-ankle joint couples in both stance and swing phases of baseline gait were the strongest predictors, and explained 62% of the variance in learning rate. These results show that gait variability measures can be used to predict short-term differences in function between individuals. Future research examining statistical persistence in gait time series that can capture the temporal dimension of gait pattern variability, may further improve learning performance prediction.
Keywords: Inter-joint coordination; Kinematics; Motor learning; Movement variability; Vector coding.