125° visual

angle; starting position at 3.8° eccentricity; velocity of 5°/s), but on audiovisual trials a click-sound (duration, 20 ms; volume, 60 dB SPL) was played at the moment of bar overlap via a central loudspeaker. Subjects reported their percept of the ambiguous stimulation via button-press (left and right thumb) after fixation-cross offset. The percept-response mapping was counterbalanced across subjects. The study was conducted in accordance with the Declaration of Helsinki and informed see more consent was obtained from all participants prior to the recordings. We recorded the continuous EEG from 126 scalp sites referenced against the nose tip. Electrode impedances were kept below 20 kΩ. For artifact cleaning, we split the data set into two frequency bands (low frequencies, 4–34 Hz;

high frequencies, 16-250 Hz). While eye movements and heartbeats cause low frequency artifact, muscle activity induces high-frequency artifact of the EEG signal. Separating these two artifact regimes allowed for more efficient artifact detection and removal. After filtering, the data were cut into trials of 2.5 s duration (−1.25 to 1.25 s). Trials with eye movements, eye blinks, or strong muscle activity were identified by visual inspection and rejected for both frequency bands. Navitoclax nmr To reduce remaining artifacts (e.g., small eye movements, muscle twitches, and cardiac artifacts), we applied independent component analysis (Hyvarinen, 1999 and Jung et al., 2000), separately for high and low frequencies, and rejected components that reflected signal artifacts. The selection of artifact components was based on careful inspection of their topography, power spectrum, and relation to the temporal structure of the experiment (mean ± SD number of rejected components: high frequency, 38 ± 10.5; low frequency, 14.5 ± 8.2). Preprocessing resulted in 179 ± 38.3 (mean ± SD) bounce trials and 167 ± 39.6 (mean ± SD) pass trials per subject. For all analyses, we recombined the data of the low- and high-frequency bands after the transformation to

the frequency domain. To control for potential microsaccade artifacts (Yuval-Greenberg et al., 2008), we repeated all tests for coherence modulations within Histone demethylase the identified cortical networks (see below) after removing data that were confounded by microsaccades (EOG based detection; Keren et al., 2010). All spectral estimates were performed using the multitaper method based on discrete prolate spheroidal (slepian) sequences (Mitra and Pesaran, 1999 and Thomson, 1982). The mean frequencies and bandwidth of experimentally observed brain oscillations typically follow a linear progression on a logarithmic scale (Buzsaki and Draguhn, 2004). Accordingly, we computed spectral estimates across 23 logarithmically scaled frequencies from 4 to 181 Hz (0.25 octave steps) and across 23 points in time from −1.1 to 1.1 s (0.1 s steps).