Comparing Neural Correlates of Human Emotions across Multiple Stimulus Presentation Paradigms

Area of Science:

• Neuroscience and Cognitive Science
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Existing electroencephalography (EEG)-based emotion recognition systems predominantly use single stimuli, leaving the impact of varied stimulus presentation paradigms on neural correlates unanswered.
• Publicly available datasets are widely used, but novel experiments are needed to address emotion classification across diverse stimulus-presentation paradigms.

Purpose of the Study:

• To investigate subject and stimulus-independent neural correlates for emotions evoked by different stimulus presentation paradigms.
• To evaluate the effectiveness of classifying emotional states (fear, neutral, joy) using EEG data across four distinct paradigms.

Main Methods:

• Recorded EEG data from 25 participants experiencing fear, neutral, and joy emotions evoked by emotional imagery, pictures, sounds, and audio-video movie clips.
• Extracted features using Common Spatial Pattern (CSP) and classified emotions via Linear Discriminant Analysis (LDA).
• Analyzed classification performance across different spectral bands and evaluated topographical maps of neural patterns.

Main Results:

• Higher frequency spectral ranges generally yielded better emotion recognition across all paradigms.
• Joy emotions were classified more effectively than fear emotions.
• Beta and alpha oscillations, along with frontal and temporal brain regions, showed the most significant results irrespective of paradigm and spectral band.

Conclusions:

• This study is novel in its consideration of four different stimuli paradigms for EEG-based emotion recognition.
• The findings contribute to designing robust EEG systems for real-time emotion recognition applicable across various scenarios.
• Frontal lobe activity and higher spectral frequencies are crucial for distinguishing emotional states.