Search

Electrodermal activity (EDA) is one of the most widely used techniques for automatic emotion recognition. Lately, great efforts have been made to increase EDA information extraction from signal processing and feature engineering and selection tasks, with the main purpose of maximizing the predictive power of such models. However, far too little attention has been paid to understanding the emotional models underlying such affective states recognition systems with EDA and the inferential value of such statistical learning models. A systematic review and meta-analysis of the literature on emotion recognition from EDA will be conducted. Journal articles, conference papers, and electronic preprints published between January 2010 and December 2020 will be selected. We will primarily investigate the characteristics of the affective model used (e.g. categorical and dimensional models of affective states). Secondarily, the characteristics of the EDA measurements and statistical models used for the automatic recognition task will be explored. It is expected to analyze common features in the state of the art, and based on this to make recommendations for future work in the field.