All experiments reported here were performed in accordance with the guidelines of the Helsinki declaration.
Data preprocessing
Data were preprocessed separately for each sensor type (EEG, MEG gradiometers and magnetometers) and were analyzed using a combination of EEGLab routines and custom code. Data were first high-pass filtered (cut-off 1 Hz), then a customized adaptive filter was applied to suppress line-noise. This followed by Artifact Subspace Reconstruction23, re-referencing to the mean for EEG, and low-pass filtering (cutoff 40 Hz). Next Infomax ICA24 was carried out. The resulting ICs were evaluated automatically for artifacts, by combining spatial, spectral and temporal analysis of ICs in conjunction with analysis of EOG channels where they were recorded (see below). ICs identified as containing ocular, muscular or cardiac artifacts were removed from the data25.
