Scott Gavett

Date of Award

January 2012

Document Type


Degree Name

Master of Science (MS)


Electrical Engineering

First Advisor

Reza Fazel-Rezai


This thesis was written to compare a few different paradigms for the brain computer interface (BCI) virtual speller using the P300 signal. The paradigms consist of electrodes to record electroencephalogram signal (EEG), software to analyze the data, and a computer where the subject's EEG is the input for a virtual keyboard. There were three experiments that were constructed to test the accuracy, region error, and adjacency error among the paradigms. The first experiment was the comparison of four paradigms: the single character (SC), row/column (RC), region based 1 (RB1), and region based 2 (RB2) paradigms. Six subjects were considered for that experiment and the accuracy of each paradigm and region errors were considered. The second experiment was designed to determine the errors per region for the region based paradigm. Eight subjects were considered for this experiment and the results concluded that region 4 (middle of the paradigm) had the most errors. The last experiment performed was the comparison of the SC, RC, and RB2 paradigms. This experiment took into consideration the accuracies of each paradigm, region errors, along with errors due to the adjacency problem. Overall, the three experiments shared the same results with the RB paradigms being slightly better than the RC paradigm in accuracy and both the RC and RB paradigms being statistically better than the SC paradigm.