A cognitive brain-computer interface monitoring sustained attentional variations during a continuous task

Area of Science:

• Cognitive Neuroscience
• Brain-Computer Interfaces
• Human-Computer Interaction

Background:

• Sustained attention is crucial for cognitive tasks, but its variations are challenging to monitor.
• Existing methods for assessing attention may lack real-time, objective measures.
• Cognitive brain-computer interfaces offer a potential solution for objective attention monitoring.

Purpose of the Study:

• To develop and validate a cognitive brain-computer interface (cBCI) for monitoring visual sustained attention.
• To introduce a multi-level difficulty continuous performance task (CPT) to probe attention variations.
• To identify electroencephalographic (EEG) features predictive of cognitive load and task difficulty.

Main Methods:

• A visual sustained attention CPT with three difficulty levels was designed.
• EEG data were recorded during task performance.
• Orthogonal forward regression was used for feature selection.
• Leave-one-subject-out cross-validation was employed for classification accuracy assessment.

Main Results:

• Specific EEG features, including prefrontal theta, gamma, and fronto-central beta and alpha power, best predicted cognitive load.
• Pairwise classification of task difficulties achieved 75% accuracy with 5-second epochs.
• Accuracy improved to 85% with 30-second epochs, demonstrating robust generalization.

Conclusions:

• The developed cBCI effectively monitors variations in visual sustained attention.
• EEG-based feature analysis provides objective measures of cognitive load and task difficulty.
• This approach holds promise for applications requiring real-time attention assessment.