Alpha-modulation induced by covert attention shifts as a new input modality for EEG-based BCIs
* Introduction: By shifting covert attention to certain locations in the visual field, the processing of visual stimuli appearing at the attended position is facilitated, while the processing of competing stimuli is actively suppressed . In electro- and magnetoencephalography (EEG,MEG), these shifts of attention (before appearance of the target stimulus) are reflected by changes in the oscillatory α-band (8-14 Hz) activity over posterior sites in accordance with the direction of attention . Following a recent MEG study , in which α-band modulations induced by covert spatial attention shifts to four different locations were used as an input modality for a brain-computer interface (BCI), we set the next step by investigating its feasibility using EEG. * Methods & Results: Eight healthy participants had to shift covert visual attention to one of six different target locations, while strictly fixating the center of the screen. A visual cue indicated the target location, where one of two symbols briefly appeared after a variable duration (500-2000ms) and participants had to press a left or right button to indicate which symbol they perceived. Only trials with a target latency of 2000ms have been used for further analysis. The topography of α-modulations varied systematically with the locus of attention (Figure 1). In the earlier interval (600-900ms after cue onset), there is an α-decrease (desynchronization) over the visual cortex with peaks contralateral to the attended locations, whereas in the later interval (1500-1700ms) there is an α-increase (synchronization) with peaks ipsilateral to the attended locations. Separability of locations was estimated as r²-values and showed maximums reaching from 0.005 to 0.015, which promises good classification ability. * Conclusion: In this study, it was demonstrated that modulations in the α-band related to shifts of covert spatial attention constitute a promising new input modality for EEG-based BCIs. Using the sgn r²-value as an indicator of class separability, attended locations can be discriminated in certain time intervals and electrode channels. Offline classification accuracy is currently under investigation.