Related Experiment Video
Updated: Jun 30, 2026
05:38
Interaction between Phonological and Semantic Processes in Visual Word Recognition using Electrophysiology
Published on: June 29, 2021
EEG-based classification models reveal differential neural processing of words and images
Neda R Morakabati1, Alison S Thiha1, Eitan Schechtman1
1Department of Neurobiology and Behavior and Center for the Neurobiology of Learning and Memory, University of California, Irvine, Irvine, California, USA, 92697.
Biorxiv : the Preprint Server for Biology
|June 29, 2026
Summary
Support vector machines trained on electroencephalography (EEG) data successfully decode object categories from images and words. Image trials showed higher accuracy than word trials, demonstrating EEG
Area of Science:
- Cognitive Neuroscience
- Machine Learning
- Neuroimaging
Background:
- Machine learning methods analyze neuroimaging data to monitor neural representation activation.
- These methods discern brain networks processing specific item categories.
- Functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) are common neuroimaging modalities.
Purpose of the Study:
- Introduce a novel task and analytical pipeline for investigating category representations using EEG.
- Assess the efficacy of EEG data in decoding object categories presented as images and words.
Main Methods:
- Utilized a dataset of 30 participants viewing images and words across five object categories.
- Trained support vector machines on EEG data to classify item categories.
- Implemented a consecutive category presentation task with participant responses.
Main Results:
- Achieved significant category classification accuracy for both image and word trials using EEG data.
- Image trials demonstrated higher classification accuracy than word trials.
- Parietal and Left Temporal electrodes were crucial for image classification; category patterns generalized across participants for images.
Conclusions:
- The developed pipeline and EEG data yield high classification accuracies, particularly for image trials.
- EEG data is valuable for neural decoding of category representations.
- These methods can explore neural representation activation during wakefulness and offline states.
