Book Description

Over recent decades, virtual reality (VR) and robotic technologies have demonstrated the potential to enhance physical therapy. Despite their advantages, clinical adoption of these technology-based systems have been slow due to limited evidence that they are more effective than traditional therapy. Currently, VR and robotic technologies are used to automate conventional therapy. Hence, the success of technology-driven rehabilitation relies on the efficacy of conventional therapy. This efficacy, in turn, is limited by our knowledge of motor learning and recovery. The goal of this thesis is to address several open questions in motor learning necessary for enhancing the efficacy of VR and robotic rehabilitation. Currently these technologies are used to train simple movements, but VR and robotic systems also have to potential to assist the learning of more complex skills. Moreover, these technologies afford the ability to assess complex skill learning in a controlled fashion without needing to sacrifice task complexity. However, a basic understanding of skill learning is required for these systems to be used effectively. In order to identify motor learning principles that apply to relearning the motor skills relevant for everyday life, this research aims to better understand complex motor skill learning, in contrast to the more prevalent research on simple, highly-controlled laboratory movements. Results from this research demonstrate that humans are sensitive to the redundancy of a task and learn strategies based on their variability and noise. Complementary experimental and computational results further reveal how the guidance methods used for shaping behavior should be adapted to the specific task. For tasks without redundancy, assistance should focus on reducing noise and variability. Contrastingly, assistance for tasks with redundancy should guide subjects towards stable solutions where variability has minimal effect of task performance. Verbal instructions were also shown to affect complex skill learning depending on the control strategy and the type of task. Together, these results provide initial principles for guiding complex skill learning that is more akin to the learning of everyday skills. Furthermore, these results demonstrate how VR and robotic technologies can be harnessed to better understand motor learning and simultaneously identify new approaches for motor rehabilitation.