Predictive models of the EEG activity based on eye movements during reading

Area of Science:

• Neuroscience
• Cognitive Science
• Biomedical Engineering

Background:

• Reading efficiency is crucial for learning and cognition.
• Understanding the interplay between neural activity and eye movements during reading is key to optimizing reading performance.
• Previous research suggests links between electroencephalography (EEG) and eye-tracking metrics, but their comparative predictive power for reading efficiency remains unclear.

Purpose of the Study:

• To investigate the relationship between eye movements and EEG signals during reading.
• To compare the predictive accuracy of eye movements and EEG for reading efficiency.
• To explore the potential for reducing measurement complexity in real-world reading studies.

Main Methods:

• Forty healthy adults read a novel fragment while their eye movements and EEG data were recorded.
• Machine learning models (CatBoost, KNN, linear regression) were employed to predict EEG from eye movements, eye movements from EEG, and reading efficiency from both.
• Analysis focused on specific EEG frequency bands (theta, alpha, beta) and eye movement features like fixation number.

Main Results:

• Machine learning models, particularly CatBoost, demonstrated strong predictive capabilities.
• EEG activity, especially alpha activity, was accurately predicted from eye movements.
• EEG data provided a significantly better prediction of reading efficiency compared to eye movement data alone.
• Specific EEG bands (alpha, beta) and lead locations (frontal, posterior) were associated with distinct cognitive processes during reading.

Conclusions:

• EEG activity and eye movements are interrelated during reading.
• EEG signals are more powerful predictors of reading efficiency than eye movements.
• These findings support the potential for simplified measurement protocols in field settings with limited equipment, by prioritizing EEG data.