The .mat file has a five-way tensor electroencephalogram (EEG) data.
Please see the reference paper for the detail.


--
data_joint_5s_offline.mat

size(data) =  [NUm. of channels, Num. of sampling points, Num of stimuli, Num. of blocks, Num of subjects]
Num. of channels 	: 9;
Num. of sampling points : 1251; (5s gazing + 0.14s transition time)
Num. of stimuli 	: 40;
Num. of blocks 		: 6;
Num. of subjects 	: 12;


--
data_joint_05s_offline.mat

size(data) =  [NUm. of channels, Num. of sampling points, Num of stimuli, Num. of blocks, Num of subjects]
Num. of channels 	: 9;
Num. of sampling points : 160; (0.5s gazing + 0.14s transition time)
Num. of stimuli 	: 40;
Num. of blocks 		: 12;
Num. of subjects 	: 12;

*The order of the stimulus frequencies in the data:
[8:1:15 8.2:1:15.2 8.4:1:15.4 8.6:1:15.6 8.8:1:15.8] Hz

Reference:
X. Chen, Y. Wang, M. Nakanishi, X. Gao, T. P. Jung, S. Gao,
"High-speed spelling with a noninvasive brain-computer interface",
Proc. Natl. Acad. Sci. U. S. A., 112(44): E6058-E6067, 2015.

Masaki Nakanishi, Swartz Center for Computational Neuroscience, UC San Diego