EEG
EEG data from channels O1, Oz and O2 were referenced to the average of the mastoids and bandpass filtered using a 1st order Butterworth filter with a passband from 0.1Hz to 50Hz. Data were then epoched using the photo-stimulus marking the time of cue onset, retaining data 500s preceding and 1s after the cue. Epochs with a standard deviation greater than 20mV in any channel were removed from analysis (0.5% of epochs). Data were then z-transformed within each channel & epoch, and any epochs with an observation in any channel whose absolute value was >4z (signifying an extreme value relative to the rest of that channel's data in that epoch) were removed from analysis (7.8% of epochs). After these data exclusions and excluding any trials excluded from the analysis of the behavioural data, data from the O1 and O2 channels were re-labelled as either ipsilateral or contralateral for each epoch as determined by that epoch's cue side. This data was first treated to a traditional event-related analysis whereby the data from each site and trial were baseline corrected (using the mean of the 500ms window preceding the cue), then within participant, channel, and time point, a mean voltage was computed across trials. The resulting data were then submitted to a generalized additive model (GAM) predicting voltage for each channel as a smooth function of time, yielding the posterior predicted curves shown in Figure 3.