PhD Defense: Implicit Brain-Computer Interfaces for Adaptive Systems: Improving Performance through Physiological Sensing
When humans interact with systems, we are limited to the direct commands to the system and results from these commands. While a machine is discrete and deterministic, the human is rich and unpredictable and may give off visible or physiological signs as to his or her current state. However, the system cannot detect these cues, thus leaving this fertile source of input untapped. Physiological sensing can remove this bottleneck by allowing a system to know more about a user's current state and react accordingly. In this thesis, I use functional near-infrared spectroscopy (fNIRS), a non-invasive brain-sensing technology, to determine user state with no additional effort from the user and adapt intelligent systems in real time. I present work showing that we can improve user performance by modifying the goals for a user or by modifying the user interface. I discuss design principles and present a taxonomy of adaptive strategies to best utilize this input. This research aims to make brain-sensing more practical and efficient for adaptive systems.