Eyes-Closed Increases the Usability of Brain-Computer Interfaces Based on Auditory Event-Related Potentials

This data-set contains the raw electroencephalography (EEG) data for the study:

Hübner, D., Schall, A., Prange, N. and Tangermann, M., 2018. Eyes-Closed Increases the Usability of Brain-Computer Interfaces Based on Auditory Event-Related Potentials. Frontiers in Human Neuroscience, 12, p.391.

The zip file contains data (a header file (*.hdr) containing general information, an EEG file (*.eeg) containing the time series information and a marker file (*.vmkr) containing time-stamps of events) from 12 subjects. Each subject participated in a single EEG session. A session consisted of a total number of 18 runs and each run contained six trials. Runs, where subjects had their eyes closed, were followed by runs where subjects had their eyes open and vice versa. The current condition was indicated to the user on the screen. To prevent systematic errors, the condition used in the first run alternated between participants. In each trial, one out of six bisyllabic words (length=300ms) were cued by a sentence as target stimuli (indicated by the markers 200-205 in the marker file) before presenting a sequence of word stimuli with stimulus onset asynchrony (SOA) of 250 ms.

During the sequence, each speaker played a different distinct word 15 times, resulting in a class-wise ratio of 1:5, with 15 target and 75 non-target stimuli per trial. Target words are indicated by the markers 111-116 while non-targets words are indicated by the markers 101-106 in the marker file. A total of 810 targets and 4050 non-targets were recorded per subject and condition (eyes-open / eyes-closed). In a run, each of the six stimuli was chosen exactly once as a target, while the other stimuli served as non-targets. We pseudo-randomized the ordering in which the stimuli were presented and in which the targets were selected. The mapping from stimulus to loudspeaker was also performed pseudo-randomized.

EEG activity was recorded and amplified by a multichannel EEG amplifier (BrainAmp DC, Brain Products) and with 63 passive Ag/AgCl electrodes (EasyCap). The channels were placed according to the 10-20-system, referenced against the nose and grounded at channel AFz. Electrode impedances were kept below 15kOhm. Eye signals were recorded by Electrooculography (EOG) with an electrode below the right eye of a subject. The signal was sampled at a rate of 1kHz.

In addition, 1-minute resting state EEG recordings from before and after the session are also available for each subject.

Additional scripts are available from the authors upon request.