Fig. 1.

Experimental design and feature extraction. (A) Examples of stimuli of each category. The patient has to provide ‘true’ or ‘false’ judgments on the presented stimulus, except for the ‘Rest’ condition (fixation cross, presented for 5 s). The number of trials per condition is 48 in each run (except for ‘Rest’, 33 trials). In the present work, ‘Math’ is the condition of interest. The four other conditions (namely self-episodic, self-semantic, self-judgment and rest) are pooled together to form the ‘Non-math’ category. (B) Definition of “Events”: The event onset (displayed by vertical arrows) represents the beginning of visual stimulus presentation. Each stimulus is displayed on the computer screen until the participant presses a button (1 for ‘true’ or 2 for ‘false’), in a self-paced manner with varying lengths of reaction time (RT). The next stimulus is then displayed, after an inter-stimulus interval (ISI) of 200 ms. ‘Rest’ condition had fixed intervals and the next stimulus appeared without the participant pressing any buttons. In this work, the continuous time series was divided into 1 s windows. Each window was associated to a label (‘Math’ or ‘Non-math’) if a stimulus was the only stimulus present during the selected 1 s window (bottom line, labels). To avoid including non-task related signal, the event onset was rounded toward the nearest integer and the event duration (here corresponding to the RT) was rounded toward negative infinity (i.e., Matlab ‘floor’).