News Release

Machine learning offers glimpses into the emotional lives of mice

Peer-Reviewed Publication

American Association for the Advancement of Science (AAAS)

Using a machine learning algorithm to analyze mouse facial expressions, Nejc Dolensek and colleagues have uncovered the neurological origins of emotional states. Their work "provides an objective analysis tool that is essential to be able to understand the neurobiological mechanisms of emotions, to identify species-specific emotions, and to identify their variability across individuals," say Benoit Girard and Camilla Bellone in a related Perspective. The neurobiological origins of emotions remain mysterious to researchers. Scientists still do not fully understand how emotions arise in the complex circuitry of the human brain, and attempts to understand emotions in animals like mice have been held back by a lack of precise tools for gauging emotional states. To better understand the emotional states of our furry cousins, Dolensek et al. used an advanced machine vision technique to precisely classify facial expressions in mice reacting to emotion-inducing events. The researchers recorded the mice as they exposed them to sensory stimuli such as sweet and bitter tastes and fearful events and identified several facial expressions that consistently correlated with emotional descriptors such as pleasure, disgust and malaise. These facial expressions showed properties such as valence (engendering positive or negative reactions) and scaled with the strength of the stimulus, suggesting that they corresponded to internal emotional states and were not merely reflexive reactions. Using two-photon calcium imaging, the authors also characterized "face" neurons in the brain whose activity correlated to specific facial expressions in the mice. The work, the authors say, may assist in moving towards a more universal and evolutionary based definition of emotions and their neural underpinnings across species.


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.