Cortical alpha changes during visuospatial attention: a deep learning-enriched EEG analysis

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Brain Imaging

Background:

• Covert visuospatial attention modulates alpha-band brain activity.
• Specific cortical regions involved in this modulation are not fully understood.

Purpose of the Study:

• To investigate the specific cortical regions involved in alpha-band modulation during cued visuospatial attention.
• To present a novel combined approach using conventional analysis and deep learning for brain oscillation research.

Main Methods:

• Whole-cortex analysis of alpha-band changes using source-level electroencephalographic (EEG) signals.
• Integration of conventional alpha power analysis with a deep learning technique (interpretable convolutional neural network - CNN).
• Discrimination of attention direction from EEG signals using the CNN to identify discriminative brain regions.

Main Results:

• Conventional analysis indicated selective left parietal lobe involvement and broader right hemisphere involvement.
• The CNN approach confirmed dominant left parietal lobe engagement and limited right parietal lobe involvement (supramarginal gyrus).
• Findings suggest a tonic engagement of the right parietal lobe, limiting its dynamic alpha modulation range.

Conclusions:

• The left parietal lobe plays a dominant role in visuospatial attention.
• A combined EEG and deep learning approach effectively characterizes alpha-band attention-related changes.
• This study enhances the understanding of neural mechanisms underlying attention and brain oscillations.