Traditionally, people learn new gait patterns via extensive practice walking in a new environment or in response to instructions from a clinician. In our new preprint, we asked: could people learn new walking patterns without actually walking? 1/n https://www.biorxiv.org/content/10.1101/2020.04.02.021949v1

Across three experiments, we asked whether learning to step with only one leg (similar to skateboarding) could facilitate learning of a wholly new bilateral gait pattern (in our study, split-belt walking). The basic paradigm was as follows: 2/n

After baseline walking, healthy young participants performed a pre-learning task that varied across groups. We included a catch trial after pre-learning to assess any aftereffects, and then participants walked on a split-belt treadmill (followed by a washout). 3/n

In Exp 1, a Control group walked with the belts tied during pre-learning. We also tested a group (Alternating Unilateral) where each leg was trained to step individually; each leg stepped on a moving belt while the other leg was stationary in alternating intervals. 4/n

Importantly, these speeds matched the subsequent split-belt speeds. Alternating unilateral stepping resulted in learning that transferred to novel gait patterns: participants who performed the alternating unilateral stepping learned to walk on a split-belt treadmill faster. 5/n

This led us to test two competing hypotheses in Exp 2: did the participants learn to move their legs to match a specific speed (speed matching hypothesis), or did they learn about the relative difference in the belt speeds/motions of their legs (speed difference hypothesis)? 6/n

First, both hypotheses predict that a Fast Unilateral group that stepped with the R leg at 1.4 m/s (L = 0 m/s) during pre-learning should learn the split-belt (R: 1.4 m/s, L: 0.7 m/s) pattern faster b/c the R speed is matched and R is faster than L in both conditions. 7/n

We observed this to be true in our data, essentially replicating the findings from Exp 1. The faster learning observed during split-belt treadmill walking co-occurred with faster changes in spatial features of walking. 8/n

We then dissociated predictions of each hypothesis. The speed matching hypothesis predicts faster learning in an Opposite Slow group (R: 0 m/s, L: 0.7 m/s); the speed difference hypothesis predicts faster learning in a Slow Unilateral group (R: 0.7 m/s, L: 0 m/s). 9/n

Our data aligned with the predictions of the speed difference hypothesis. We observed faster learning in the Slow Unilateral group but not in the Opposite Slow group. This showed that stepping with the R leg facilitated faster split-belt learning when R speed > L speed. 10/n

Importantly, we also found that the size of the pre-learning aftereffect (i.e., amount learned during unilateral stepping) predicted smaller asymmetry (i.e., faster learning) during split-belt walking across all groups in Exp 2. 11/n

Exps 1 and 2 showed that people could learn a new walking pattern by 1) stepping with each leg independently and 2) stepping with only one leg. In Exp 3, we then asked whether active movement was necessary for locomotor learning at all. 12/n

We tested an Observational group that watched a video of another person learning to walk on split-belt treadmill during pre-learning. This group showed modestly faster learning when compared to the Control group, but the difference was not statistically significant. 13/n

Across all groups, we showed that the transfer of learning that occurred between the pre-learning tasks and the split-belt walking task was predicted by faster changes in spatial (e.g., center of oscillation difference) and not temporal (e.g., phasing) gait parameters. 14/n

In summary, we found that people could learn new walking patterns by stepping with only one leg. The adaptive learning that occurred during unilateral stepping transferred to learning of a wholly new, bilateral gait pattern (i.e., split-belt treadmill walking). 15/n

This study was informed by several split-belt treadmill studies and prior work showing that unilateral stepping causes changes in subsequent bouts of overground walking (Kahn and Hornby, 2009 Physical Therapy; Huynh et al., 2014 MSSE). 16/n

One note about the first author: Christine Song is a senior undergrad @HopkinsNeuro and collected all of this data over the past 2+ years (she was the first trainee in my lab!) without taking a single full-time research semester. She is a machine. 17/n

Thanks also to our co-authors @janstenum, @KristanLeech, and Chloe Keller! We appreciate any feedback on the preprint. Thanks for following along! 18/18