Emotion recognition from auditory autonomous sensory meridian response (ASMR) using multi-modal physiological signals

Area of Science:

• Neuroscience and Psychology
• Affective Computing
• Biomedical Engineering

Background:

• Autonomous Sensory Meridian Response (ASMR) is a sensory phenomenon characterized by tingling sensations triggered by specific auditory stimuli.
• Previous research primarily focused on ASMR's association with positive emotions like relaxation and calmness.
• This study investigates a wider spectrum of emotions, including happiness, sadness, and disgust, elicited by ASMR.

Purpose of the Study:

• To explore the range of emotions induced by ASMR auditory stimuli beyond relaxation.
• To investigate the effectiveness of multi-modal physiological signals in identifying ASMR-induced emotions.
• To apply deep learning models for classifying these emotions based on physiological data.

Main Methods:

• Collected multi-modal physiological data (EEG, PPG, EDA) from 23 participants experiencing ASMR.
• Exposed participants to various ASMR auditory stimuli designed to elicit happiness, sadness, calm, and disgust.
• Utilized rmANOVA for statistical analysis and Artificial Neural Network (ANN) and Convolution Neural Network (CNN) for emotion classification using Valence-Arousal-Dominance (VAD) dimensions.

Main Results:

• Statistical analysis revealed significant differences in physiological responses across the four induced emotions and a neutral state.
• Deep learning models achieved high classification accuracy: ANN at 96.12% and CNN at 74.25% for the four emotions using VAD dimensions.
• Results demonstrate distinct physiological signatures for different ASMR-induced emotions.

Conclusions:

• Multi-modal physiological signals combined with deep learning effectively classify ASMR-induced emotions.
• This research contributes to the field of emotion recognition, particularly for mental health and therapeutic applications.
• The findings validate the use of physiological data for objective emotion assessment in response to auditory stimuli.