Multimodal consumer choice prediction using EEG signals and eye tracking

Area of Science:

• Neuroscience
• Marketing Science
• Computer Science

Background:

• Neuromarketing enhances traditional marketing by analyzing consumer brain activity and emotions.
• Electroencephalogram (EEG) is commonly used, but Eye Tracking (ET) remains underexplored in this field.
• Predicting consumer choices requires advanced analytical methods integrating multiple data sources.

Purpose of the Study:

• To develop and validate a novel multimodal approach for predicting consumer choices.
• To integrate electroencephalogram (EEG) and eye tracking (ET) data for enhanced predictive accuracy.
• To address the gap in utilizing eye tracking data within neuromarketing research.

Main Methods:

• EEG and ET data were preprocessed, including noise reduction (bandpass filter, ASR, FORCE) and artifact handling (SMOTE).
• Feature extraction involved both handcrafted (statistical, wavelet, fixations, saccades) and automated methods (CNN-LSTM, LeNet-5).
• A meta-learner ensemble classifier (Random Forest, XGBoost, Gradient Boosting) was employed for buy/not buy classification.

Main Results:

• The multimodal approach achieved 84.01% accuracy in predicting consumer choices.
• The model demonstrated 83% precision in identifying positive consumer preferences.
• Performance was evaluated using accuracy, precision, recall, and F1 score, showing superior results.

Conclusions:

• Integrating EEG and ET data offers a powerful method for predicting consumer choices in neuromarketing.
• The proposed multimodal approach significantly improves the accuracy and precision of consumer behavior prediction.
• This research highlights the potential of combining neurophysiological and behavioral data for more effective marketing strategies.