EEG data was exported to EDF from the native format using NPX Lab 2012 (publicly available software at www.brainterface.com) and analyzed in the EEGLAB environment (http://sccn.ucsd.edu/eeglab/index.html; Delorme and Makeig, 2004), a collection of analytical tools running under Matlab 7.7.0 R2010a (Mathworks Inc., Natick, MA). EEG signal was digitally band-pass filtered between 1 and 100 Hz (with a FIR filter) and re-referenced to the average reference. After visual inspection and manual removal of segments characterized by gross artifacts, non-cerebral source activities (eye blinks and movements, cardiac and electromyographic activity) were identified and rejected using a semiautomatic procedure (Medaglia et al., 2009; Porcaro et al., 2009). The EEG signal was first decomposed into independent components (ICs) using FastICA version 2.5 (Hyvarinen and Oja, 2000; http://www.cis.hut.fi/projects/ica/fastica). ICs corresponding to artifactual sources and brain activity were separated with a manual procedure (Medaglia et al., 2009; Porcaro et al., 2009). The electrical power line noise was removed on ICs using the CleanLine plug-in of EEGLab. After removal of artifactual non-cerebral ICs, the “cleaned” signal was reconstructed by retro-projecting only the ICs containing cerebral signal. Cleaned data were segmented in 2-s epochs for following analysis steps.
