Dogs' brains can distinguish between human fear, anger and sadness, fMRI study shows

A first look inside the canine brain

A study published in iScience on 10 August 2026 by a team led by researchers at the University of Vienna provides the first evidence that dogs can distinguish between different negative human facial expressions, not merely positive from negative. Using fMRI scans, the team showed that dogs' brains produce distinct activity patterns when viewing happiness, anger, fear and sadness. The history of dog domestication began roughly 33,000 years ago, and a conversation made of human gestures has accompanied these animals ever since. A face can anticipate a caress, a reward or a threat, which is why dogs have evolved to read facial expressions closely.

First experiment: happy versus neutral faces

The first experiment involved 8 dogs, including 6 border collies, 1 Labrador retriever and 1 golden retriever, all pets volunteered by their owners without compensation. The dogs were shown photographs of unfamiliar human faces with either happy or neutral expressions, each displayed for 7 seconds while the dog lay in an MRI scanner. The faces were of strangers to rule out familiarity effects and were of the same gender as each dog's primary caregiver. Each set of five happy and five neutral faces was repeated five times per dog. Happy faces activated the temporal cortex, associated with higher cognitive functions, and the caudate nucleus, associated with reward processing, in the right hemisphere, along with the piriform cortex.

Laura Cuaya, a postdoctoral researcher at the SCAN Unit of the University of Vienna, explained that the location of this activity matters: if it had appeared only in early visual areas, it could have been explained by differences in shapes, colours or contrasts. The involvement of higher-level regions suggests dogs process emotional meaning, not just visual features.

There is a beautiful analogy with the resonance. It is as if the brain lights up and fills with colors.

Second experiment: separating negative emotions

The second experiment involved 12 dogs shown new sets of unfamiliar male and female faces bearing four expressions: happiness, sadness, fear and anger. The researchers used machine learning to analyse the brain activity. The temporal cortex and caudate nucleus network responded exclusively to happiness, while distinct activity patterns across the whole brain allowed the team to distinguish between all four emotions. The finding shows the canine brain does not simply classify the world into good and bad.

This is the first demonstration that the dog brain can distinguish between emotions from the same valence.

Cuaya added that by merely seeing people's faces, "dogs can discriminate emotions and form expectations."

Training and methodology

All 14 dogs across both experiments were trained to lie still, awake and unrestrained, in a sphinx position within the scanner. Seven of the eight dogs in the first experiment had prior fMRI experience; the new participant, Molly, received training using the same protocol. A chin rest supported their heads and ear muffs provided noise protection, and the dogs could leave the sessions at any time. A minimum of six minutes in the machine was required for a proper scan. The reward circuitry activated by happy faces is the same part of the brain that responds to tasty food or praise.

Raúl Hernández-Pérez, first author of the study from the University of Vienna, 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.

He also noted an open question: "We don't know if these regions are related just to happy faces or to emotional faces in general."