EEG-Based Decoding of Selective Visual Attention in Superimposed Videos

Yuanyuan Yao , Wout De Swaef , Simon Geirnaert

IEEE Journal of Biomedical and Health Informatics +

|

June 16, 2025

View abstract on PubMed

Summary

Researchers decoded selective visual attention from electroencephalography (EEG) signals using natural videos. This advancement in neuroscience could enhance brain-computer interfaces by interpreting complex visual dynamics.

Area Of Science

Neuroscience
Cognitive Science
Signal Processing

Background

Selective attention is crucial for efficient visual processing, allowing enhancement of relevant stimuli and filtering of irrelevant information.
Understanding visual attention is vital in neuroscience, with implications for developing advanced brain-computer interfaces.
Existing methods often rely on artificial stimuli, limiting the study of attention in naturalistic, dynamic visual environments.

Purpose Of The Study

To investigate the feasibility of decoding selective visual attention from electroencephalography (EEG) signals using naturalistic video stimuli.
To develop and validate a novel free-viewing paradigm for studying attention to dynamic visual content.
To determine if EEG can capture neural responses modulated by attention to irregular, real-life motion patterns.

Main Methods

A free-viewing paradigm was employed where participants attended to one of two superimposed videos.
A stimulus-informed decoder was trained on electroencephalography (EEG) data to identify components correlated with attended motion.
Analysis included correlating EEG decoding with eye movements and exploring complementary information from EEG and gaze data.

Main Results

EEG signals successfully decoded selective visual attention to naturalistic motion with above-chance accuracy.
Eye movements correlated with attended motion but did not solely drive the EEG-based decoding, indicating complementary neural information.
EEG signals appear to capture neural responses to both attended and unattended stimuli, even with spatial overlap.

Conclusions

Electroencephalography (EEG) responses to naturalistic motion are modulated by selective visual attention.
This study demonstrates the first successful EEG-based decoding of selective visual attention using natural videos.
The findings open new avenues for experimental design in attention research and brain-computer interface development.