HRV-Based Recognition of Complex Emotions: Feature Identification and Emotion-Specific Indicator Selection.

Area of Science:

• Psychophysiology
• Affective Computing
• Biomedical Engineering

Background:

• Current emotion recognition systems often use discrete states and contact-based sensors.
• Complex emotions are mixtures of basic emotions, influenced by valence direction.
• Heart rate variability (HRV) is a potential physiological marker for emotional states.

Purpose of the Study:

• To differentiate four compound emotions (Positive Surprise, Negative Surprise, Positive Sadness, Negative Sadness) using HRV.
• To evaluate a contactless camera-based system (Deep Health Vision System, DHVS) against a reference device (Polar H10).
• To explore the relationship between HRV, subjective emotion reports (Valence, Arousal, Dominance), and emotion composition.

Main Methods:

• Ten healthy adults viewed emotion-inducing video clips.
• HRV was recorded using both a Polar H10 chest strap and the DHVS webcam system.
• Standard HRV indices were computed, and statistical analyses examined changes, correlations, and device concordance.

Main Results:

• Subjective reports confirmed distinct affective profiles for the four compound emotions.
• HRV changes correlated with Valence, Arousal, and Dominance, with specific indices sensitive to emotion category (LF/HF) and valence (Total Power).
• The DHVS system showed partial concordance with the Polar H10, particularly under positive valence conditions.

Conclusions:

• HRV captures distinct autonomic signatures of complex emotions, linked to subjective evaluations.
• LF/HF and Total Power provide complementary insights into emotion category and valence.
• The DHVS demonstrates preliminary feasibility as a contactless platform for complex emotion recognition, requiring further validation.