All EEG signal preprocessing procedures were done in the EEGLab Toolbox under a common pipeline. Channels that contained large long-term artifacts were excluded from the subsequent analysis after raw data of all the channels were demeaned. We applied high-pass filtering at 0.5 Hz to remove any slow drift. Independent component analysis (ICA) was applied and manually assessed to remove components that contained eye movement artifacts, channel noise, line noise, ECG artifacts, and major muscle artifacts. Artifacts from transitory muscle activities that contaminated the EEG signals were visually detected and marked in the raw data. For EEG source analysis, the data was re-referenced from the original reference (CPz) to the average of all channels. To eliminate the high-frequency interference, we also applied a 300-Hz low-pass filter [EEGLab default finite impulse response (FIR) filter]. Subsequently, all datasets were segmented into epochs of 3-s duration (−1 s to 2 s relative to the peripheral movement onset). To ensure equal treatment of the resting dataset, we randomly generated and inserted 150 fake markers into the original raw datasets, and the epoching information was saved
