🎗️Reading Emotions: How Your Dog Knows When You Are Sad Or Happy

Does your dog know when you are happy, afraid, or sad? A new study shows that dogs' brains do more than just perceive human facial expressions: they can distinguish between different emotions, including differentiating between various negative emotions. The discovery reveals that these animals may be even more attuned to our emotional cues than we previously thought.

Imagine looking at someone's face and, without them saying a word, realizing whether they are happy, afraid, sad, or angry. For us humans, this seems to happen automatically. But can dogs do something similar?

We have known for some time that dogs pay attention to human facial expressions and react differently to a smile versus an angry expression. What remained unclear, however, was the extent of this processing in the brain: do dogs merely perceive that a person is displaying a "positive" or "negative" emotion, or can they distinguish between specific negative emotions?

A new study investigated this question directly by observing dogs' brain activity as they viewed human faces displaying various emotions.

To discover what was happening in the animals' brains, researchers used a scan called functional magnetic resonance imaging (fMRI). This method allows for the observation of which brain regions become most active while a person or animal performs a specific task.

In this study, the participants were pet dogs that remained awake during the scan. Instead of sedating them, the researchers trained them to stay still inside the scanner while viewing images. This step is particularly important because animal movement could compromise the images and make it difficult to identify changes in brain activity.

The researchers also took an important precaution: they used faces of people the dogs did not know. This way, any brain reaction to a face could be linked to emotional expression rather than simply to the dog recognizing its owner or another familiar person.

The study was divided into two experiments. In the first, eight dogs observed images of happy human faces and faces with a neutral expression. The researchers wanted to find out which areas of the brain responded most strongly when the animals saw an expression of happiness. To do this, they compared the brain activity produced by the two types of faces.

The analysis revealed an especially important region on the right side of the brain, located in an area related to the processing of information from the face and other social information. This region extended from areas of the temporal lobe to the caudate nucleus, a structure located deeper in the brain and involved, among other functions, in processing rewards, motivation and learning.

In other words, the dogs' brains didn't seem to simply “see” the happy face: some regions had an especially strong response to it.

In the second experiment, carried out with twelve dogs, the researchers made the challenge greater. The animals observed human faces expressing four different emotions: happiness, anger, fear and sadness. The question now was more specific: would the dog's brain be able to recognize each emotion or just separate the faces into two general categories, such as “positive” and “negative”?

To answer this, the researchers used a machine learning technique. Essentially, the computer was fed data on the brain activity patterns produced when dogs viewed specific expressions and sought to identify patterns capable of distinguishing one emotion from another. It is akin to teaching a system to recognize a brain "signature" associated with each type of expression.

The results showed that the region identified in the first experiment could distinguish happiness from each of the three negative expressions. However, when the researchers compared only the negative emotions, that specific region could not clearly separate anger, fear, and sadness.

This might suggest that the dogs' brains were merely identifying the difference between a positive emotion and negative emotions. But the analysis did not stop there. The researchers examined activity patterns across the entire brain, looking for much subtler differences. This analysis revealed that some pairs of negative emotions also produced distinct patterns.

The brain could distinguish, for example, faces expressing anger from those expressing fear. It could also differentiate sadness from fear. However, the difference between anger and sadness was not identified clearly enough.

These results are important because they show that dogs' ability to interpret human emotions may be more sophisticated than simply perceiving whether someone is happy or displaying a negative emotion. The canine brain appears to represent different human expressions in partially distinct ways, even when the emotions belong to the same general category.

Happiness showed a particularly strong signature in a network involving temporal regions and the caudate nucleus, while different combinations of fear, anger, and sadness produced distinct patterns in other brain areas. This does not necessarily mean that dogs understand human emotions in exactly the same way we do, nor that they have the same subjective experience of sadness, fear, or anger.

What the study demonstrates is something more specific and equally interesting: when a dog looks at a human face, its brain holds enough information to distinguish between some of the emotions expressed on that face.

The finding helps explain why dogs are so adept at adapting their behavior to people and reinforces the idea that, through living alongside humans, they have developed a particularly sophisticated ability to interpret our social cues.

READ MORE:

Dog Brain Representations of Human Facial Expressions: Encoding Happiness and Differentiating Specific Negative Expressions

Karl Kristjan Kaup, Javier Hidalgo Jiménez, and Jaan Aru

iScience. 116900, August 10, 2026

DOI:10.1016/j.isci.2026.116900

Abstract: 

Dogs can distinguish human facial expressions, particularly happiness, yet brain processes remain unclear. Using fMRI, we conducted two experiments in awake pet dogs. In Experiment 1 (n = 8), happy faces elicited a stronger response than neutral faces in a right temporal cluster extending to the caudate nucleus, including the rostral Sylvian gyrus. In Experiment 2 (n = 12), dogs viewed faces expressing happiness, anger, fear, or sadness. Using the Experiment 1 cluster as a region of interest, a machine-learning classifier distinguished happiness from each negative facial expression, but not between negative pairs, showing differential BOLD responsiveness to happy faces. Whole-brain representational similarity analyses revealed activity patterns differentiating angry vs. fearful faces (right mid ectosylvian and left splenial gyri) and sad vs. fearful faces (right rostral suprasylvian gyrus) but not angry vs. sad faces. This provides direct evidence that dog brains can distinguish between two negative facial expressions.