Dog Brains Can Tell When You’re Angry, Scared, or Sad

Dogs have long been celebrated for their ability to sense human moods. Owners often describe how their pets seem to know when they are happy, stressed, or upset, offering comfort or adjusting their own behaviour accordingly. Until recently, much of the evidence for this skill came from behavioural observations. A new study using functional magnetic resonance imaging (fMRI) has now provided the first clear neural evidence that dogs do not simply divide human facial expressions into “good” and “bad.” Their brains generate distinct activity patterns for specific negative emotions, including the ability to differentiate fear from both anger and sadness.
Researchers at the University of Vienna, led by Raúl Hernández-Pérez and Laura V. Cuaya, trained pet dogs to remain completely still inside an MRI scanner while viewing photographs of human faces. The animals stayed awake throughout the sessions, wore protective headphones to block scanner noise, and were free to leave at any time. The team published their findings in the journal iScience in August 2026 under the title “Dog brain representations of human facial expressions: Encoding happiness and differentiating specific negative expressions.”
In the first experiment, eight dogs looked at images of happy and neutral faces belonging to unfamiliar people. Happy expressions produced stronger responses in a right temporal cluster that extended into the caudate nucleus, a brain region associated with reward processing. The same network becomes active when dogs smell a familiar person or anticipate a treat. This finding suggested that a human smile carries intrinsic positive value for dogs, even when the face belongs to a stranger.
The researchers then expanded the study to twelve dogs, mostly border collies and golden retrievers, with one mixed-breed participant. These dogs viewed faces displaying four basic emotions: happiness, anger, fear, and sadness. Using the temporal-caudate region identified in the first experiment as a region of interest, the team applied machine-learning classifiers to the brain data. The classifiers successfully distinguished happy faces from each of the three negative expressions. However, within that specific network the algorithm could not separate the negative emotions from one another. In other words, the reward-related circuitry treated smiles as special while treating the negative faces more similarly to each other.
A different picture emerged when the scientists examined activity across the entire brain. Representational similarity analysis revealed distinct neural patterns that differentiated fearful faces from angry ones and fearful faces from sad ones. These differences appeared in areas including the right mid ectosylvian gyrus, the left splenial gyrus, and the right rostral suprasylvian gyrus. Notably, the analysis did not find reliable differences between angry and sad faces. The results indicate that dogs’ brains encode at least some negative human expressions as separate categories rather than collapsing them into a single “unpleasant” signal.
“This is the first demonstration that the dog brain can distinguish between emotions from the same valence,” Cuaya has explained. Hernández-Pérez added that humans are highly skilled at detecting subtle facial details, and dogs appear to share that sensitivity. The study builds on earlier behavioural research showing that dogs prefer objects associated with happy human expressions and respond differently to emotional cues in voice, posture, and scent. By restricting the stimuli to static photographs of strangers, the researchers isolated the contribution of facial features alone.
The findings carry implications for understanding the human-dog relationship. Dogs and humans have co-existed for tens of thousands of years. During domestication, selection pressures likely favoured individuals that could accurately interpret human social signals. Reading faces would have helped dogs anticipate human intentions, avoid conflict, and cooperate more effectively. The neural evidence that dogs process specific negative expressions differently supports the idea that their social cognition is more nuanced than a simple positive-versus-negative distinction.
At the same time, the researchers are careful not to overstate what the data show. Distinct patterns of brain activity demonstrate discrimination, not necessarily the same subjective experience of emotion that humans have. Dogs may register a fearful face as signalling a different kind of potential threat or need compared with an angry face, without “feeling” fear or anger in a human-like way. In everyday life, dogs also rely on multiple channels of information. Body language, vocal tone, and scent cues typically accompany facial expressions and may help resolve ambiguities such as the difference between anger and sadness that the brain scans could not clearly separate.
The experimental constraints themselves are worth noting. The dogs saw only still images, not dynamic faces or full social interactions. The faces belonged to unfamiliar people rather than the dogs’ own owners. Familiarity can influence emotional processing, and future work may examine whether responses differ when the faces belong to people the dogs know well. The sample sizes, while respectable for awake canine fMRI, remain relatively small. Larger studies will be needed to confirm the robustness of the distinctions and to explore individual differences across breeds and experience levels.
Despite these limitations, the research adds meaningful depth to the scientific understanding of canine social intelligence. Previous imaging work had shown that dog brains respond more strongly to human faces than to dog faces in certain temporal regions and that emotional faces activate areas linked to reward and memory. The new study advances the field by demonstrating that the representation of human facial expressions extends beyond valence alone. Happiness receives preferential processing in a frontotemporal-caudate network, while certain pairs of negative expressions produce separable whole-brain patterns.
For dog owners, the results offer a scientific basis for everyday observations. Smiling at a dog may genuinely activate reward circuitry in its brain. A fearful expression may register differently from an angry or sad one, potentially shaping the animal’s subsequent behaviour. Cuaya has suggested that people can reward dogs not only with food but also with happy facial expressions, noting that even strangers’ smiles appear to carry positive significance.
The work also raises broader questions about how domestication has shaped the canine brain. Unlike primates, which share a more recent common ancestor with humans, dogs followed a separate evolutionary path yet developed sophisticated abilities to read human social cues. Studying these abilities provides a comparative window into the neural foundations of interspecies communication and cooperation.
In practical terms, the findings may eventually inform training methods, animal-assisted interventions, and welfare assessments. Understanding how dogs process human emotional signals could help handlers communicate more effectively and reduce misunderstandings that arise when people assume their pets interpret expressions exactly as humans do.
The study leaves several avenues open for future investigation. Researchers are interested in combining visual and auditory cues to see how multimodal information affects neural representations. Examining responses to dynamic facial expressions or to the faces of familiar caregivers could reveal additional layers of processing. Comparative work with other domesticated species remains challenging—fMRI studies in cats, for instance, have proven far more difficult—but the canine model continues to offer unique insights.
What remains clear is that dogs are not merely reacting to the presence or absence of a smile. Their brains register finer distinctions among the expressions humans display. When a person looks frightened, the dog’s neural response differs from the pattern elicited by anger or sadness. That capacity for discrimination, revealed through careful brain imaging, deepens the scientific appreciation of the bond between humans and their oldest animal companions. Dogs have spent millennia learning to read us. The latest scans show just how carefully they are watching.