A POMDP Approach to P300 Brain-Computer Interfaces*

Most of the previous work on brain-computer interfaces (BCIs) using P300 has been focused on feature extraction and classification algorithms to achieve high performance for the communication between the brain and the computer. While significant progress has been made in such lower layer of the BCI system, the issues in the higher layer have not been addressed sufficiently. Existing P300-based BCI systems use a random order of stimulus sequence for eliciting P300 signal for identifying users‟ intentions. This paper is about computing an optimal sequence of stimuli in order to minimize the number of stimuli, hence improving the performance. To accomplish this objective, we model the problem as a partially observable Markov decision process (POMDP) with observation delays. Through simulation and human subject experiments, we show that our approach achieves a significant performance improvement in terms of the success rate and the bit rate.