High-density ear-EEG for understanding ear-centered EEG

Abstract Background: Mobile ear-EEG provides the opportunity to record EEG unobtrusively in everyday-life. However, in real-life, the EEG data quickly becomes difficult to interpret, as the neural signal is contaminated by other, non-neural signal contributions. 
Due to the small number of electrodes in ear-EEG devices, the interpretation of the EEG becomes even more difficult. 
For meaningful and reliable ear-EEG, it is crucial that the brain signals we wish to record in real life are well-understood and that we make optimal use of the available electrodes. Their placement should be guided by prior knowledge about the characteristics of the signal of interest.
Objective: We want to understand the signal we record with ear-EEG and make recommendations on how to optimally place a limited number of electrodes.
Approach: We built a high-density ear-EEG with 32 electrodes spaced densely around one ear. We used it to record four auditory event-related potentials (ERPs): the mismatch negativity, the P300, the N100 and the N400. With this data, we gain an understanding of how different stages of auditory processing are reflected in ear-EEG. We investigate the electrode configurations that carry the most information and use a mass univariate ERP analysis to identify the optimal channel configuration. We additionally use a multivariate approach to investigate the added value of multi-channel recordings.
Main results: We find significant condition differences for all ERPs. The different ERPs vary considerably in their spatial extent and different electrode positions are necessary to optimally capture each component. In the multivariate analysis, we find that the investigation of the ERPs benefits strongly from multi-channel ear-EEG.
Significance: Our work emphasizes the importance of a strong theoretical and practical background when building and using ear-EEG. We provide recommendations on finding the optimal electrode positions. These results will guide future research employing ear-EEG in real-life scenarios.