Improving a P300-based Brain-Computer Interface
Effects of luminosity contrast, stimulus duration, interface type & screen size: From Yeuqing Li’s thesis (2009)
Brain-computer interface (BCI) is a new technology with the capability of allowing people with motor disabilities to live a better life. Though research has been conducted with non-disabled people, little to no research has been conducted regarding BCI usability with people with motor disabilities. This research focused on the usability analysis of P300-based BCI applications with four independent variables: luminosity contrast, stimulus duration, interface type, and screen size. Ten participants with motor disabilities and ten non-disabled participants were employed in the research. The difference in task performance between participants with and without motor disabilities was also analyzed. P300 Speller was used as the testbed in the research.
The results showed a significant effect or interaction effect on task performance for most independent variables, indicating that these variables are important to consider when developing and testing P300-based BCI applications.
In study 1, participants reached an average accuracy of 51.7%. Results indicated that luminosity contrast and stimulus duration had significant effects on task performance. Participants had better task performance in high luminosity contrast than low luminosity contrast. Non-disabled participants showed significantly better task performance than participants with motor disabilities. However, stimulus duration showed a different effect on a different group of participants.
In study 2, participants reached an average accuracy of 54.2%. Results indicated that the interface type and screen size had significant effects on task performance. Nondisabled participants showed significantly better task performance than participants with motor disabilities. However, interface type and screen size showed a different effect on a different group of participants.
Research showed satisfactory spelling accuracy for non-disabled participants: 76.5% in study 1 and 79.7% in study 2. However, participants with motor disabilities showed unsatisfactory spelling accuracy: 26.8% in study 1 and 28.7% in study 2. Therefore, two groups of participants demonstrated significantly different task performance, indicating that non-disabled people may not be considered as a substitute for people with motor disabilities in BCI research.
This research found that participants with motor disabilities had better performance in frequency-based interface with higher luminosity contrast, middle stimulus duration, and bigger screen size. Because participants in this study showed different use preferences in stimulus duration and screen size, optimal use condition should be provided to the users for better task performance. This research intends to provide invaluable empirical data and meaningful insights to the future research of P300-based BCI applications.