Selection errors committed by the posthoc evaluation method after having processing the data of one entire block.

<p>The data displayed stems from the same blocks as in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0102504#pone-0102504-g006" target="_blank">Fig.6</a>, which had been recorded while feedback was given by the unsupervised method. With the exception of three difficult blocks (first blocks of users <i>nbe</i> and <i>jh</i>, and third block of user <i>nbb</i>) the posthoc re-analysis obviously outperforms the original online performance gained by the unsupervised method (see <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0102504#pone-0102504-g006" target="_blank">Fig.6</a>). It effectively corrected most initial mistakes at the beginning of each block, thus recovering communication from the very first trial on. Both unsupervised methods (online and posthoc) had trouble selecting a good performing classifier for the three difficult blocks.