Decoding covert visual attention based on phase transfer entropy.

Area of Science:

• Neuroscience and Brain-Computer Interfaces (BCI)
• Signal Processing and Machine Learning

Background:

• Covert attention, a new BCI control signal, requires decoding from human brain activity (EEG) to direct processing resources.
• Existing methods for decoding covert attention have limitations, necessitating novel approaches.

Purpose of the Study:

• To design and evaluate a novel expert system for decoding covert visual attention using EEG signals.
• To investigate the efficacy of Phase Transfer Entropy (PTE) as a decoding feature for the first time in this context.
• To compare binary and multi-class classification systems for attention decoding.

Main Methods:

• Developed an expert system using EEG signals from 15 subjects performing a visual attention task with color changes.
• Employed Phase Transfer Entropy (PTE) for feature extraction and decoding.
• Evaluated performance across binary and multi-class systems, focusing on Alpha (8-13 Hz) and Beta1 (13-20 Hz) frequency bands.

Main Results:

• Achieved high two-class classification accuracies of 91.87% (Alpha band) and 89.53% (Beta1 band).
• Multi-class classification accuracies reached 65.11% (Alpha band) and 63.38% (Beta1 band).
• PTE demonstrated superior performance compared to previous phase synchronization methods, with the Alpha band from the posterior region showing optimal results.

Conclusions:

• The novel expert system effectively decodes covert visual attention using EEG and PTE.
• PTE is a promising feature for attention decoding, outperforming traditional methods.
• The Alpha frequency band, particularly from the posterior brain region, is crucial for accurate covert attention decoding.