Affiliation: School of Mechanical Engineeing, Shanghai Jiao Tong University, 200240,China. Name: Liu Guangquan, Huang Gan, and Zhu Xiangyang. email: quanguangliu@sjtu.edu.cn 1. The EEG data were band-pass filtered by a 5-order IIR filter. Frequency bands were optimized for each subject according to the result of a ten-fold cross validation in the calibrating data set. All the frequency bands used did not exceed 8-30Hz.By this processing, EOG artifacts were also removed. 2. For each two of the three classes (treating idle state as one class), CSP was applied to find the spatial filters W (resulting in 3 different W). Log-variances of the eight best components were used as features, resulting in a 3*8=24 dimensional feature vector at each time point. 3. Fisher's LDA was used to reduce the dimension from 24 to 2. 4. The Gaussian likelihood was caculated to determine the output result at each time point. 5. We think data sets 'c' and 'd' are artificially generated.