“…Data was downsampled to 250 Hz, and linear support vector machine (SVM) classifiers were trained for each time sample, based on single trials, using all posterior electrodes (O1/2, P7/8, P3/4, Pz, TP7/8, PO7/8, POz, Oz, PO3/4, P5/6, P1/2, TP9/10). These sites were chosen because previous work has shown that these electrodes contain the most informative signal for visual tasks (see 38 , 51 ). We focused on two binary comparisons where participants saw stimuli from the same visual category (i.e., faces or cars) but face expectation was manipulated (i.e.…”