Brain Machine Interfaces (BMIs)

A brain-computer interface (BCI) enables direct control of computer applications and robotic devices through brain activity, potentially enhancing the quality of life and independence of individuals with motor disabilities. Motor imagery, the mental representation of a movement, is proven to show similar neurological features as the actual execution of the movement. The beneficial effects of MI implementation are demonstrated in sports training and rehabilitation after trauma or stroke.  Thus, its use in BCIs is widely studied; machine learning algorithms are developed, to learn the user’s mental features and identify them online as he imagines the execution of the movement. However, the implementation of MI BCIs is still far from occurring in real life, because of a lack of consistency in performance. A great effort has been made in the improvement of brain signal handling and processing, but the system’s accuracy in predicting the imagined movement still depends significantly on the user. There are “good” and “poor” users; 15-30% of the users are thought to be unable to generate the “proper” brain features even after training.  This phenomenon is called “BCI illiteracy” or “BCI inefficiency”.  BCI research aimed at enhancing accuracy can target either technological factors, such as advancements in algorithms for feature extraction and classification, or human factors, which influence how effectively a person generates quality EEG patterns.

Our current research focuses on the capacity of the user to maintain mental imagery and its relation to BCI accuracy. Our purpose is to answer the following questions:

How can we maintain mental imagery or improve MI methodology?

Do meditators perform better?

How is sustained attention related to motor imagery success?

Feedback implementation: Can intention and motivation support or enhance mental imagery?