Optimizing Biofeedback and Learning in an EEG-Based Brain-Computer Interface

Brain-computer interface technology establishes communication between the brain and a computer, allowing users to control devices, machines, or virtual objects using their thoughts. This study investigates optimal conditions to facilitate learning to operate this interface. It compares two biofeedback methods, which dictate the relationship between brain activity and the movement of a virtual ball in a target-hitting task. Preliminary results indicate that a method in which the position of the virtual object directly relates to the amplitude of brain signals is most conducive to success. In addition, this research explores learning in the context of neural signals during training with a BCI task. Specifically, it investigates whether subjects can adapt to parameters of the interface without guidance. This experiment prompts subjects to modulate brain signals spectrally, spatially, and temporally, as well differentially to discriminate between two different targets. However, subjects are not given knowledge regarding these desired changes, nor are they given instruction on how to move the virtual ball. Preliminary analysis of signal trends suggests that some successful participants are able to adapt brain wave activity in certain pre-specified locations and frequency bands over time in order to achieve control. Future studies will further explore these phenomena, and future BCI projects will be advised by these methods, which will give insight into the creation of more intuitive and reliable BCI technology.