Yi-Chieh Lee, Wen-Chieh Lin, Fu-Yin Cherng, Li-Wei Ko.
IEEE Transactions on Neural Systems and Rehabilitation Engineering, Vol.24, No. 3, 2016, pp.399-408.
Attention detection is important for many applications. Automatic determination of users’ visual attention state is challenging because attention involves numerous complex and internal human cognitive functions. Behavioral observations, such as eye gaze or response to external stimuli, can provide clues for users’ visual attention state; however, users’ cognitive state cannot be easily known. Conventional EEG-based methods detect attention by observing the dynamic changes in the frontal lobe of the brain, especially in the anterior cingulate cortex (ACC). However, that area in the brain is associated with many functions, some of which correlate with conscious experience but are not directly related to attention. In this study, we design an attention monitoring system to detect whether the brain experiences a visual stimulus consciously.Our experiments verified the feasibility of our design, and the average classification rate ranged from 72% to 82%.
In this work, we present a new approach to monitor human’s visual covert attention. In particular, we exploit the neuroscience finding that attention modulates the SSVEP power to detect the user’s attentional state. Our results demonstrated that SSVEP power is a reliable feature for attention monitoring and the average classification rate is about 72% to 82%.
Compared with previous SSVEP research, the contributions of this study are:
1) achieving plausible classification of attention state based on single-trial EEG data,
2) determining that the O1 and O2 channels are effective features for SSVEP-based visual attention detection, and
3) verifying that covert attention could be detected in high-frequency SSVEP.
The following two plots verify that average SSVEP power attenuates as inattention occurs.
Left : Average SSVEP power with 13 Hz flash stimuli. Right: Average SSVEP power with 30 Hz flash stimuli.