1. EEG data: 16 subjects tarred together. They were initially acquired at 5000 Hz (in the MRI) but were downsampled to 500 Hz for processing and to fit within Dryad's 10 GB limit. They were initially read in from BrainProducts format to Matlab using Fieldtrip (ft_preprocessing). Continous data was extracted into trials around when stimulation was presented. This period the MRI was 'silent' (meaning no RF pulses or gradient switching, thus is free of these artifacts). The data does however contain BCG artifact. The raw.trialinfo refers to: % trialinfo: column one: 81 means attend left, 82 means attend right % column two: last digit is condition (0=0back, 1=1back, 2=, 3= , 4= catch trials) % column three: '1' means response, '91' is just fixation return Triggers were messed up for participant f14: cfg.trialdef.preeventvalue='S 26'; cfg.trialdef.posteventvalue='S 28'; cfg.trialdef.offsetvalue={'S 1','S 2'}; But for all other participants, cfg.trialdef.preeventvalue='S 90'; cfg.trialdef.posteventvalue='S 92'; cfg.trialdef.offsetvalue={'S 81','S 82'}; Data were also demeaned relative to baseline window of -0.9 to -0.3 s prior to the zero time point, and re-referenced over channels to the average reference (excluding ECG).