Study shows dogs can distinguish between human emotions including fear, anger and sadness

Brain scans reveal dogs distinguish human emotions

Researchers at the University of Vienna, working with colleagues from the National Autonomous University of Mexico and Eötvös Loránd University in Hungary, used functional magnetic resonance imaging (fMRI) to scan the brains of awake, unrestrained dogs as they viewed still images of human facial expressions. The study, published on 10 August 2026 in the Cell Press journal iScience, comprised two experiments involving pet dogs volunteered by their owners, who received no compensation. The research built on prior findings that dogs can differentiate between very different emotional states such as happiness and anger. The novelty of the new work lay in testing whether the canine brain also produces distinct patterns for emotions that humans themselves categorise as all negative, such as fear, anger, and sadness.

First experiment: happiness activates reward circuits

In the first experiment, eight pet dogs were shown images of unfamiliar human faces displaying either happy or neutral expressions. Happy faces elicited a stronger response in a region of the right hemisphere spanning the temporal cortex and extending to the caudate nucleus, areas associated with higher cognitive processing and reward circuits, respectively. Laura Cuaya, a postdoctoral researcher at the SCAN-Unit of the University of Vienna and co-author of the study, noted that this activation pattern resembles what is seen when a dog expects a snack or hears praise.

The activity in the caudate nucleus is particularly interesting. It suggests that happy human faces may have emotional significance for dogs.

Second experiment: telling fear from anger and sadness

The second experiment involved 12 pet dogs shown images expressing happiness, anger, fear, and sadness. While the temporal cortex and caudate nucleus responded specifically to happy faces, the dogs did not react in the same brain regions to negative emotions. A whole-brain analysis using machine learning revealed distinct neural patterns separating fear from anger and fear from sadness, providing the first evidence that dogs can distinguish between specific negative facial expressions rather than simply dividing the world into positive and negative.

As far as we know, it's the first time there is proof that dogs can distinguish between faces expressing two [similar] emotions.

Raúl Hernández-Pérez, a postdoctoral researcher at the SCAN Unit, University of Vienna, and first author of the study, described the broader significance of the work.

Humans are really good at picking up small details on the face, and so are dogs. By studying their brains, we can understand what adaptations have developed to support their remarkable social and emotional understanding of humans.

Methodology and limitations

The dogs, which included border collies, a Labrador retriever, and a golden retriever, were trained to lie still in the scanner and wear headphones to protect their hearing. Each dog had to spend a minimum of six minutes in the machine for a proper scan. The animals were awake and unrestrained, and could leave the sessions at any time. Some dogs had previously participated in similar experiments and were already trained for scanner procedures. The research was described as fairly small, with over a dozen young dogs volunteered by their owners.

Interpreting the results

The researchers cautioned that the findings do not mean dogs experience emotions exactly as humans do. In real life, dogs also process human body language, voice nuances, and scent information. Cuaya noted that comparing dogs with primates involves examining abilities that may stem from a common ancestor, whereas studying dogs offers insight into adaptations developed specifically for interacting with humans. The study was started at the National Autonomous University of Mexico and later expanded to Eötvös Loránd University in Hungary before completion at the University of Vienna.