Mechanisms underlying reinforcement learning of motor skills
(2023)
Files
PhD_VASSILIADIS_FINAL.pdf
Restricted Access - Adobe PDF
- 7.16 MB
Details
- Authors
- Supervisors
- Abstract
- Motor learning allows animals, including human beings, to acquire skills that are essential for efficient interactions with the environment. This ability to learn new motor skills is of great practical relevance for daily-life activities (such as when learning to drive), but also for motor rehabilitation after a lesion of the nervous system (such as a stroke). For a long time, motor learning has been mainly conceptualized as a process allowing to iteratively correct movements based on sensory information (e.g., visual, somatosensory). Importantly though, in the last years, there has been an increased appreciation that motor learning also results from other mechanisms including reinforcement learning, a process through which appropriate actions are selected through outcome-based feedback (e.g., success or failure). As such, recent evidence shows that reinforcement feedback and mo-tivation can be beneficial for motor learning both in healthy individuals and neurological populations. Despite the potential importance of these findings to improve current rehabil-itation protocols, the mechanisms underlying reinforcement-related improvements in mo-tor learning remain largely unexplored. This PhD aimed at providing deeper mechanistic understanding of reinforcement learning of motor skills through behavioral analyses, neuroimaging and non-invasive brain stimulation. In Study 1, I found that enhancing motivation (by providing monetary reward for good performance) during a motor training can lead to persistent improvements in performance that are not obtained with reinforce-ment feedback only, and are related to an increased regulation of motor variability based on previous outcomes. In Study 2, I investigated the effect of reward timing (i.e., the delay between the end of movement execution and reward receipt) on motor learning and found that delaying reward by only a few seconds could strongly influence motor learning dynamics and consolidation. Finally, in Study 3, I investigated the causal role of the stria-tum in reinforcement motor learning. Here, I show, by combining an innovative non-invasive deep brain stimulation approach called transcranial electric temporal interfer-ence stimulation and neuroimaging, that a specific mechanism relying on striatal high gamma oscillations is causally involved in reinforcement learning of motor skills. Overall, this work characterizes key mechanisms underlying the effect of reinforcement on motor learning, paving the way towards the incorporation of optimized reinforcements in motor rehabilitation protocols.
- Affiliations
Citations
Vassiliadis, P. (2023). Mechanisms underlying reinforcement learning of motor skills. https://hdl.handle.net/2078.5/101456