A brain-computer interface (BCI) is an augmentative communication mechanism
that does not rely on peripheral nerves or muscles. Current BCIs are
error prone and slow with error rates of 10 to 30% and transmission rates of
10-25 bits/min, however, error recovery and correction in BCI has largely been
neglected. The focus of this thesis is the development of a method to automatically
recover errors in BCI using the P3 brain signal for response verification.
The existence of the P3 signal in responses to controlled goal items is shown
in an evoked potential BCI used to control items in a virtual apartment. A
reduced response exists when items are accidentally controlled. Offline experiments
were run, and with a theoretical mean improvement in accuracy from 78%
to 85%, there was a statistically significant improvement (P < 0.008, Wilcoxon
signed rank test) in accuracy of 3% using a correlation algorithm for P3 signal
detection on responses. The presence of the P3 signal in responses to goal
items indicates it can be used for automatic error recovery without requiring
additional time, which will improve the speed and accuracy of brain-computer
interfaces.