The influence of movement speed on reward-based motor learning

Abstract

Human movement is inevitably variable. This variability can be seen as a constraint to overcome, but it may also be a feature: being variable may result in the discovery of better movement solutions. Especially when feedback is limited to binary information on movement success or failure, variability is key for discovering which movements lead to success. Since moving faster increases variability, we aimed to answer the question whether movement speed can be harnessed to improve such reward-based motor learning. Subjects performed a stepping task in a slow and a fast session. They had to learn the gain between their step lengths and visual target distances on screen based on binary reward feedback. We successfully manipulated movement speed between sessions and participants could learn the gain in both sessions. We found no difference in learning between speed sessions, despite the fact that variability in gain increased in the fast relative to the slow session. To distinguish between different sources of variability, we estimated inevitable motor noise from the variability following successful trials. We estimated exploration as the additional variability following non-successful trials relative to following successful trials. We found no relation between variability sources and learning. In conclusion, reward-based motor learning is possible in a gain-learning task. In this task, moving faster did not lead to higher learning. Since the role of variability may differ between experimental tasks, whether movement speed can be harnessed to improve motor learning needs to be tested in other experimental tasks.