Single Trial EEG Patterns for the Prediction of Individual Differences in Fluid Intelligence

Area of Science:

• Neuroscience
• Cognitive Science
• Machine Learning

Background:

• Assessing fluid intelligence is crucial for career counseling and clinical applications.
• Fluid intelligence and electroencephalography (EEG) evoked potentials in oddball tasks are linked to cognitive processing and attention.
• Previous methods for fluid intelligence assessment are limited.

Purpose of the Study:

• To propose a novel system for predicting an individual's fluid intelligence level using single-trial EEG signals.
• To investigate the correlation between neural activity during a visual oddball task and fluid intelligence scores.
• To evaluate the effectiveness of wavelet decomposition and Support Vector Machine (SVM) for fluid intelligence prediction.

Main Methods:

• EEG signals were recorded from 34 participants in low-ability (LA) and high-ability (HA) groups during a visual oddball task.
• Participants were categorized using Raven's Advanced Progressive Matrices (RAPM) test.
• Features were extracted using Haar wavelet transform on EEG signals (0.3-30 Hz) and classified using SVM.

Main Results:

• The proposed system achieved 100% and 98% prediction accuracies for distinguishing LA and HA groups using specific wavelet features from delta frequency bands (0.0-1.875 Hz and 1.875-3.75 Hz).
• Analysis revealed significant differences in neural activity between LA and HA groups within these frequency bands.
• Statistical tests confirmed no significant effect of 2D vs. 3D content on fluid intelligence assessment.

Conclusions:

• Single-trial EEG analysis combined with wavelet decomposition and SVM offers a highly accurate method for predicting fluid intelligence.
• The delta frequency band of EEG signals contains discriminative information for fluid intelligence assessment.
• The developed system demonstrates superior performance compared to existing state-of-the-art techniques.