BME Seminar Wickenden 321
Sponsor: Dawn Taylor
Presenter: Amy L. Orsborn, Ph.D. Scientific advisor, Meta Reality Labs
Title: Predicting and shaping user-device interactions in co-adaptive neural interfaces
Abstract: Neural interfaces can restore or augment sensorimotor capabilities by converting biological signals into control signals for an external device via a decoder algorithm. Users adapt their behavior in these interfaces via sensorimotor learning and decoder algorithms can adapt via machine learning. Leveraging user and decoder adaptation presents opportunities to improve usability and personalize devices. But we have limited understanding of how user and decoder learning interact. We also lack principled methods to model and optimize these complex two-learner dynamics. In this talk, I'll first briefly present work suggesting that adaptive decoder algorithms influence brain learning in a motor brain-computer interface. I'll then present new computational methods based on control theory and game theory that allow us to analyze and generate predictions for user-decoder interactions, and experimental validation of these predictions in human myoelectric interface experiments.