For the active trials, the classifier is fit with a k-fold cross-validation – training on active trials and testing on active trials. For the passive trials, the classifier is trained on all of the active trials and tested on the 320 passive trials. Note that the actively presented letter/digit stimuli (4, 6 vs H, E) were different from the passively presented ones (9, 8 vs. A, C), thus any significant generalization of neural responses demonstrates a true ‘letter/digit’ percept, which is not specific to particular pixel arrangements. Thicker lines in the decoding timecourses show the timing of significant above-chance decoding for passive trials and active trials separately. Shaded grey areas show when the two timecourses significantly diverge. Both statistical tests used a temporal permutation cluster test and clusters are displayed that are significant at p<0.05.