Michael Tangermann (BrainLinks-BrainTools, University of Freiburg) | Auditory Brain-Computer Interfaces
| When |
Oct 22, 2013
from 05:15 PM to 06:45 PM |
|---|---|
| Where | Lecture Hall, Hansastr. 9a |
| Contact Name | Ad Aersten |
| Add event to calendar |
 vCal iCal
|
Abstract
Our auditory system is capable to process information rapidly, and our ability to use spatial auditory attention is well-trained by every-day tasks.
Motivated by vision problems in some of the patients in need of a Brain-Computer Interface (BCI), research on the auditory modality for BCI started in 2005. Exploiting the ability of a BCI user to focus her/his attention onto external tone stimuli, the evoked potentials of the EEG reveal differences between attended stimuli (targets) and un-attended stimuli (non-targets). Compared to the well-known auditory oddball paradigm, the experimental setups for BCIs comprise rather rapid stimulation sequences and typically more than just two types of tones. Furthermore, BCI systems make intense use of supervised and unsupervised machine learning models in order to process brain signals on a single-trial basis.
In my talk I will show examples of how the real-time classification of auditory ERP responses can be used to create a communication interface via BCI technology, which is independent of the visual modality. Analyzing such systems, the role of a spatial encoding of the presented tones will be touched as well as improvements of auditory BCI paradigms by recruiting language processing areas, by optimizing SOA (stimulus onset asynchrony) and by breaking with the oddball-dogma of randomized stimulus presentation.
If of interest, the scientific potential of this neurotechnology can be discussed with the audience (e.g. for the purpose of monitoring dynamic attention processes in real-time).
Keywords
Brain Signal Analysis, Machine Learning, ERP (Event-Related Potential), AEP (Auditory Evoked Potential), EEG (Electroencephalogram), BCI (Brain-Computer Interface)