Original title: Emotion Recognition from Physiological Signals Using Machine Learning Algorithms Under Controlled Emotional Stimuli
Aditi Site, Annariina Lohiranta, Tarmo Lipping · September 2026
Computers often struggle to detect how a person feels. To solve this, researchers tested whether body signals can reveal emotions. They tracked heart activity, called an electrocardiogram, and sweat levels on the skin, called galvanic skin response. First, the researchers pulled out specific timing and wave patterns from these body signals. Next, they fed that information into machine learning models, which are computer programs trained to spot patterns. They found that tree-based models worked well when sorting simpler groups of emotions and tracking arousalHow intense or calm an emotion is, from low to high energy., which is your body's level of excitement. Specifically, combining heart and sweat signals gave the best performance. The models identified discrete emotions with 52.8 % accuracy and arousal with 53.6% accuracy, showing body signals can track emotional states.
Imagine trying to guess if a friend is bored or excited just by watching their breathing and how much they fidget. You might get it right sometimes, but it is harder if you only look at one thing, like breathing alone. In this paper, the researchers use both heart and skin signals together, like watching breathing and fidgeting at once, to make better guesses about emotions.
Emotion recognition from body signals could let computers respond to your mood without you saying a word—useful for mental health monitoring, adaptive learning, or safer driving. This study shows that simple tree-based models can pick up on basic emotional states from heart and skin data, bringing cheap, wearable emotion sensing closer to real use.
The abstract does not report performance on valence classification, nor does it compare against deep learning methods. Accuracy is modest (around 53%), and the controlled stimuli may not reflect real-world emotional responses. Generalizability across individuals or contexts is not addressed.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Emotion Recognition from Physiological Signals Using Machine Learning Algorithms Under Controlled Emotional Stimuli — Aditi Site, Annariina Lohiranta, Tarmo Lipping
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.