Facial Expressions Reveal Emotion In Mice

Do non-human animals experience emotions in the same way humans do? If so, can we detect them? This recent study published in Science finds evidence that mice experience distinct emotional states similar to those of humans. It also finds that these emotional states are reflected in distinct facial expressions that can be detected by computer algorithms.

Understanding the emotive capacity of animals is important to animal advocates for at least two reasons. First, it improves our understanding of the similarities between humans and non-human animals, a key philosophical facet of modern human-animal studies. Second, understanding animals’ emotional state can help us improve their well-being more effectively.

In the study, researchers exposed mice to sensory stimuli that were assumed to produce an emotional reaction. For example, a sugar solution produces a sensation of pleasure, a bitter solution produces a sensation of disgust, and small tail shocks produce pain. As the mice experienced these stimuli, researchers recorded their facial expressions with small cameras, and their brain activity using an imaging technique called two-photon microscopy.

The results suggested that:

1. Mice experience distinct emotional states that can be determined from facial expressions. Researchers trained a machine learning classifier on videos of mice experiencing the different stimuli (assumed to cause different emotional reactions). They found that the algorithm was able to accurately predict the stimulus that each video corresponded to. The fact that the algorithm could differentiate between the videos suggests that each stimulus produced a distinct response, and that these responses could be detected automatically. These responses might indicate the presence of different emotional states.
2. Mices’ facial expressions reflect “fundamental features of emotions.” While the classifier results found that different stimuli produced different responses, it was not clear whether these responses reflected emotions or simple reflexes. (Unlike reflexes, emotions have characteristics such as “intensity, valuency, generalization, flexibility, and persistence.”) To test this, researchers varied the stimuli strength (for example, increasing the amount of sugar in the sucrose solution). They found that as each stimulus became stronger, the mices’ reaction also became stronger, suggesting that the reactions had the “intensity” property of emotions. Researchers next gave the mice a salt water solution, which is appealing to mice at low concentrations but unpleasant at higher concentrations. Researchers found that at the appetizing low concentrations, mice reacted with the same “pleasure” facial expression as the sucrose solution, but when the concentration increased, mice reacted with the same “disgust” facial expression as the bitter solution. This suggests that the mices’ facial expressions generalize between different types of experiences, another key aspect of emotion.
3. Mices’ facial expressions reflect past experience. Another characteristic of emotion is that it reflects past experiences — for example, we feel melancholy when we see something that recalls a sad memory. To test whether this characteristic held in mice, researchers first gave mice an appealing sucrose solution, then injected them with a chemical that induced malaise. This procedure induces “conditioned taste aversion,” leading mice to avoid the sucrose solution in the future the same way we might avoid a restaurant that gave us food poisoning. While before this procedure sucrose induced the “pleasure” facial expression, researchers found that after the malaise-causing injection caused mice to associate sucrose with unpleasant feelings, they instead reacted with the “disgust” facial expression. This suggests that mices’ facial expressions, and thus emotions, reflect past experience.
4. Mices’ facial expressions correlate with emotion-relevant parts of the brain. Finally, researchers conducted several experiments that investigated how facial expressions related to activity in the brain. Using two-photon brain imaging, they identified separate neurons that correlated with facial expressions and sensory stimuli. They also found that activating portions of the insular cortex, a portion of the brain known to be relevant to emotion in humans and other animals, could produce the same facial expressions associated with “pleasure” and “disgust.” This suggests that the neural basis for emotion in mice may be similar to that in humans.

There are, of course, ethical questions surrounding the use of captive animals in scientific studies. It is important to recognize the suffering animals in these studies endure, even as they provide new scientific insights that can help us improve the well-being of both humans and other animals. Faunalytics does not support experiments using captive animals, and shares the results of such work only to highlight ways it might be used to help reduce or eliminate their use.

Animal advocates will take particular interest in the evidence provided by these findings, that animals may exhibit behaviors and emotions often thought to be exclusive to humans — a finding that provides a stronger basis for giving animals moral standing more equivalent to that of humans. These results also have the potential to more directly help us improve animal welfare by assessing animals’ well-being using inexpensive cameras. This may be particularly helpful for improving the welfare of farmed animals, whose well-being is often difficult to scientifically measure.