Related Experiment Video
Updated: Nov 28, 2025
04:27
Using Facial Electromyography to Assess Facial Muscle Reactions to Experienced and Observed Affective Touch in Humans
Published on: March 15, 2019
11.5K
The Role of Features Types and Personalized Assessment in Detecting Affective State Using Dry Electrode EEG
Paruthi Pradhapan1, Emmanuel Rios Velazquez1, Jolanda A Witteveen1
1imec The Netherlands/Holst Centre, 5656 AE Eindhoven, The Netherlands.
Sensors (Basel, Switzerland)
|December 2, 2020
Summary
This study used electroencephalography (EEG) to assess affective states, finding that combining linear and nonlinear features improved accuracy. This approach outperformed subjective ratings in a naturalistic setting.
Area of Science:
- Neuroscience
- Affective Computing
- Machine Learning
Background:
- Electroencephalography (EEG) shows promise for assessing human affective states but struggles with real-world reliability.
- Challenges include setups impacting affect processing and reliance on generalized affect models.
- Subjective affect assessment as ground truth is often debated.
Purpose of the Study:
- To explore the use of a convenient EEG system in a naturalistic setting to capture affective reactions.
- To evaluate the performance of machine learning models in classifying affective states using EEG data.
- To compare EEG-based affect assessment with subjective self-assessment.
Main Methods:
- Utilized a convenient electroencephalography (EEG) system with 20 participants viewing affective movie clips.
- Employed a state-of-the-art machine learning approach combining linear (symmetry, single-channel) and nonlinear (multiscale entropy) features.
- Performed binary classification for valence and arousal.
Main Results:
- The highest performance was achieved by combining linear and nonlinear EEG features.
- The best F1-scores were 0.71 for valence and 0.62 for arousal.
- Performance was 10-20% higher than using independent raters' subjective assessments.
Conclusions:
- Combining linear and nonlinear EEG features enhances affective state assessment accuracy.
- Affective self-assessment may be underrated, and individual differences in perception and physiological response are crucial.
- This approach offers improved reliability for real-life affective computing applications.
