Multiscale EEG feature fusion for recognizing 3D object shapes through active touch

Area of Science:

• Neuroscience
• Cognitive Science
• Human-Machine Interaction

Background:

• Tactile perception is crucial for brain function research and developing tactile sensing technologies.
• Existing electroencephalography (EEG) studies often use simple stimuli or single neural features, limiting understanding of complex tactile shape processing.
• A systematic EEG framework for active 3D shape discrimination with multi-domain neural analysis is needed.

Purpose of the Study:

• To investigate the neural mechanisms of human tactile perception of 3D object shapes during active touch.
• To develop and validate a multi-domain EEG analysis and feature-fusion framework for decoding tactile shape information.
• To establish a baseline for tactile-dominant multimodal intelligent interaction research.

Main Methods:

• Participants actively grasped objects with varying 3D shapes while EEG signals were recorded.
• Multi-domain EEG analysis included time-frequency energy, dynamical complexity, and brain functional networks.
• An attention-enhanced multi-channel convolutional neural network was used for feature-fusion-based decoding.

Main Results:

• Tactile stimulation presence/absence was identified via alpha and beta band event-related synchronization/desynchronization.
• Increased object edges correlated with decreased brain complexity during active touch.
• Brain network efficiency and clustering in the high-alpha band correlated with 3D shape categories.
• The framework achieved 92.20% accuracy in recognizing seven 3D shapes.

Conclusions:

• The study presents a systematic EEG framework for active 3D tactile shape perception.
• Multi-domain neural features provide complementary information for reliable shape decoding.
• Findings offer insights into neural representations of 3D tactile shape perception.