A Study of Visual Search based Calibration Protocol for EEG Attention Detection.

Area of Science:

• Cognitive Neuroscience
• Brain-Computer Interfaces

Background:

• Visual attention is crucial for daily tasks and can be measured using EEG Brain-Computer Interfaces (BCIs).
• Current attention calibration methods, like the Flanker task, may be insufficient for long experiments due to subject fatigue or boredom.

Purpose of the Study:

• To propose and evaluate novel attention calibration protocols using simultaneous visual search and audio cues.
• To compare the effectiveness of these new protocols against baseline methods for EEG-based attention classification.

Main Methods:

• Developed new calibration protocols using visual search and audio directional changes for 'attentive' states and static white noise for 'inattentive' states.
• Collected EEG data from sixteen healthy subjects under both proposed and baseline calibration conditions.
• Utilized six basic EEG band-power features and a standard binary classifier for performance comparison.

Main Results:

• The new calibration protocol achieved a mean subject accuracy of 74.37 ± 6.56%, a slight improvement over the baseline.
• Post-experiment surveys indicated that the new calibrations were more effective in inducing desired perceived attention levels.
• No statistically significant differences in classification performance were found between the new and baseline methods.

Conclusions:

• The proposed attention calibration protocols show potential for improving EEG-based attention measurement.
• Further refinement of calibration protocols and attention classifier modeling is needed for enhanced attention recognition.
• The findings suggest a promising direction for more robust BCIs in applications like gaming and clinical settings.