Biologists have decoded and classified animal emotions.

Many contemporary studies in the field of biology focus on the emotional states of animals. The authors of one study proved that our smaller brothers are, in fact, capable of experiencing emotions, while another explored what a dog feels when interacting with humans.

Experts from the University of Copenhagen in Denmark and Harvard University in the USA developed an artificial intelligence model that can determine emotions from vocalizations with 89 percent accuracy—these are the characteristic sounds made by animals. The results of their work were published in the scientific journal IScience.

The authors of the publication analyzed thousands of sound signals from cows, goats, horses, Przewalski's horses (the only wild horse species), sheep, pigs, and wild boars. Among the key features identified for distinguishing positive and negative emotions were changes in the duration of vocalization, fundamental frequency, and amplitude characteristics. Notably, the indicators were similar across different species: the researchers suggested that the primary ways to express emotions vocally may be evolutionarily retained.

The researchers trained an artificial intelligence model on the obtained data and tasked it with classifying vocalizations based on the emotions they express. The accuracy of the resulting "animal emotion decoder" was 89.49%.

"Understanding how animals express emotions can help us improve their well-being. If we detect stress or discomfort at an early stage, we can intervene before it escalates. Equally important is that we could also encourage positive emotions. This would change the game for animal protection," noted the authors of the scientific article.

The researchers hope they have taken a step closer to a future where technology allows people to understand and respond to animal emotions. This would open up new possibilities not only for wildlife conservation but also for science.