Profile of Mood States 2nd Edition-based Emotion Intensity Estimation by Electroencephalogram and Heart Rate...

Area of Science:

• Neuroscience
• Psychophysiology
• Affective Computing

Background:

• Accurate emotional state assessment is vital for mental health care, stress management, and affective computing.
• Current methods often rely on subjective self-reports, limiting objectivity.
• Developing objective, quantitative methods for emotion recognition is a key research area.

Purpose of the Study:

• To classify and estimate the intensity of multiple emotional states using physiological signals.
• To investigate the efficacy of electroencephalogram (EEG) and heart rate variability (HRV) for emotion recognition.
• To identify key physiological indicators for accurate emotion classification.

Main Methods:

• Elicited emotions using a jigsaw puzzle task.
• Recorded electroencephalogram (EEG) and heart rate variability (HRV) signals.
• Employed Support Vector Machines (SVMs) with Principal Component Analysis (PCA) or Autoencoders (AE) for classification and Recursive Feature Elimination (RFE) for feature selection.

Main Results:

• Classification models achieved a kappa coefficient over 0.9 with PCA or AE preprocessing.
• Identified key features: mean RR interval (MRRI), low-frequency (LF) and high-frequency (HF) power, LF/HF ratio, and prefrontal alpha asymmetry.
• Determined less important features including standard deviation of RR intervals and F7α-F8α asymmetry.

Conclusions:

• EEG and HRV signals can simultaneously classify and estimate multiple emotional states.
• The findings support the development of objective emotion recognition systems.
• This research offers potential for non-invasive mental well-being assessment and advanced human-computer interaction.