A Dynamic Architecture Linking Autonomic Activity to Emotional Dimensions: Real-Time Estimation From...

Area of Science:

• Physiological Psychology
• Affective Computing
• Wearable Technology

Background:

• Current emotion tracking methods often lack real-time, continuous, and interpretable outputs.
• Existing models may not fully capture the multi-dimensional nature of emotional states.
• Autonomic nervous system (ANS) activity is intrinsically linked to emotional responses.

Purpose of the Study:

• To introduce a novel, real-time model for estimating distinct emotional dimensions (valence, motivation, activation) using physiological signals.
• To correlate specific ANS branches (PNS, α-SNS, β-SNS) with these emotional dimensions.
• To validate a wearable-compatible system for continuous emotion inference.

Main Methods:

• Utilized photoplethysmography (PPG) signals to extract inter-beat interval (IBI) and pulse wave amplitude (PVA).
• Employed an extended Kalman filter (EKF) for signal processing and heart rate variability (HRV) analysis.
• Applied supervised partial least squares (PLS) regression, mapping physiological features to behavioral validation measures (facial EMG, joystick, eye-tracking).

Main Results:

• The model accurately estimated valence, motivation, and activation in real-time, comparable to traditional methods.
• Physiological signals from PPG were successfully linked to distinct emotional dimensions via specific ANS activity.
• Generated discrete emotion probabilities from dimensional estimates without requiring specific training data.

Conclusions:

• A novel framework for real-time, dimensional emotion tracking based on separate autonomic nervous system activities is established.
• This approach offers a low-cost, wearable solution for dynamic emotional inference in various applications.
• Supports a new theoretical model of emotion grounded in real-time autonomic functioning.