In an effort to improve prosthesis design, researchers from Northwestern University and the Rehabilitation Institute of Chicago investigated how EMG control of a virtual computer interface affects an individual’s motor adaptation.
Current EMG-powered prostheses have often struggled to replicate natural mobility, as they rely solely on a user’s attention and conscious movement rather than being able to learn through practice and feedback. As a result, users have become frustrated with their devices, leading some to desert them altogether.
Led by Ms. Reva Johnson, the research team sought to gain an understanding of how adaptive EMG-controlled systems can become through feedback and movement anticipation. To quantify this adaptation, trials were evaluated and compared based on feedback uncertainty, control interface, and mean error of control. The Delsys Bagnoli™ EMG System was used for EMG data acquisition.
In the study, the team investigated trial-by-trial adaptation with two different levels of uncertainty and three different control interfaces: joint angle, torque, and EMG. Subjects used elbow extension movements to control a virtual cursor along a circular track, starting on the left side of the circle and reaching a target on the right side. The three phases of each experiment (familiarization, training, and testing) utilized various levels of visual feedback to increase the uncertainty of each phase as the experiment progressed.
As a result, every subject demonstrated trial-by-trial adaptation using all three control interfaces. While the control interface did not significantly affect adaptation, it did influence mean error. Specifically, the mean errors for EMG-controlled trials were significantly higher than both joint angle- and torque-controlled trials. Also, feedback uncertainty significantly affected adaptation rate throughout all trials.
Put simply, the results showed that adaptation rates depended primarily on feedforward and feedback information, rather than the method of control. This could prove important for prosthesis design moving forward, as a better understanding of adaptation will help to improve sensory information strategies used in new prostheses.
Reva received a B.S. degree in mechanical engineering from Valparaiso University in 2009, and a M.S. degree in biomedical engineering from Northwestern University in 2013. She is working toward a Ph.D. in biomedical engineering at Northwestern University and completing her research in the Center for Bionic Medicine at the Rehabilitation Institute of Chicago. Reva is currently investigating how powered prosthesis users plan, generate and correct movements. The knowledge gained from this work will help us understand how an amputee’s brain makes decisions and how we can best improve learning in prosthesis control and feedback systems.
Johnson RE, Kording KP, Hargrove LJ, Sensinger JW. Does EMG control lead to distinct motor adaptation?Frontiers in Neuroscience. 2014;8:302. doi:10.3389/fnins.2014.00302.