Experimental paradigm of one trial.

<p>(A) At t = 0 s, four symbols representing four tasks and a fixation cross were presented on the screen (baseline). (B) At t = 2 s, one symbol was randomly highlighted and in case of the word association or mental subtraction task, an initial letter or the specific subtraction, respectively, was presented. (C) Users were asked to perform the indicated task for 6 s (i.e. imagery period from t = 2-8 s) while a blue bar graph was provided from t = 3-8 s (i.e. feedback period). In the feedback sessions, the blue bar graph represented the real-time feedback and was only shown if the classifier detected the correct task (i.e. from the cue indicated task) at this very moment. In the screening or calibration run, the blue bar graph was shown as sham feedback independently from performance in about 70% of the trials (see methods Sections 5 and 6). The relevant time for classification was from t = 4-8 s (i.e. classification period). (D) At t = 8 s, discrete feedback in form of a smiley was shown for 1 s if the given task was detected correctly (see methods section 6). If this was not the case, no feedback was displayed. (E) At t = 9 s, the screen remained blank for 2.5-3.5 s, before the next trial started.