Estimation of Emotional Valence toward Politicians using Physiological Signals. A pilot study.

Abstract

Emotional valence refers to the positive or negative quality of affective experience and plays a central role in shaping political perceptions, judgments, and participation. However, its objective measurement in political contexts remains underexplored. This article presents a pilot study to evaluate the feasibility of estimating emotional valence from physiological signals elicited by visual stimuli with political content. Physiological data were collected from participants during exposure to images of Colombian political figures. Photoplethysmography (PPG) and galvanic skin response (GSR) signals were recorded using a non-invasive wearable device (EmotiBit). Several machine learning classifiers were evaluated to classify emotional valence (negative, neutral, and positive). Due to class imbalance, the F1 score was used as the primary evaluation metric. The best-performing model was the decision tree-based Extra Trees classifier, achieving 72.73% accuracy and an F1 score of 0.71 in a 3-class classification task. The results show that physiological signals can be used to measure emotional valence in political contexts, using portable, non-invasive wearable devices.