Figure 2.

Spatiotemporal decoding.

Note. Epoched data after preprocessing (1) are averaged together in random partitions of the same conditions to create supertrials (2). Time courses for each of the 64 channels are pooled in time, averaging voltages in 10 time-bins (3). The resulting 640 values are used as features to train and test an LDA classifier (4). The whole procedure is repeated 100 times.