Study: Dogs’ Brains Distinguish Human Anger, Fear, and Sadness

Summary: New research shows that dogs have specialized brain networks that process human facial expressions. Using functional magnetic resonance imaging (fMRI) and machine-learning analyses, scientists monitored the brains of awake pet dogs as they viewed photos of human faces showing happiness, anger, fear, and sadness.

The study found that happy human faces preferentially activate a network that includes the temporal cortex — a region linked to higher-order sensory and social processing — and the caudate nucleus, a central structure in the brain’s reward system. Whole-brain analyses further showed that dogs do more than classify expressions as simply positive or negative: their brains form distinct neural activity patterns that separate fear from anger and sadness.

These results offer direct neural evidence that domestication and long-term coexistence with humans have equipped dogs with the neural architecture to decode important human emotional signals.

Key Facts

  • Reward center response to smiles: When dogs viewed smiling faces, the caudate nucleus became strongly active, suggesting that human smiles carry emotional and motivational value for dogs.
  • Distinct neural signatures for negative emotions: Machine-learning pattern analysis revealed that dogs display unique brain activity for fear, anger, and sadness rather than treating all negative expressions as the same.
  • Convergent adaptation through domestication: Dogs appear to have evolved social-cognitive skills that let them interpret human expressions independently of the primate lineage, likely driven by thousands of years of coevolution with people.
  • Temporal cortex specialization: The temporal cortex in dogs is central to processing faces and higher-level social cues and forms a functional circuit with the caudate nucleus tuned to human facial signals.
  • Multimodal, real-world processing: Although static photos produced clear neural responses, dogs in everyday life combine facial information with body posture, vocal tone, and scent to interpret human states.

Source: Cell Press

For years, dog owners and researchers have observed that dogs respond to our smiles, frowns, and fearful looks, but the neural basis of this sensitivity has been hard to document—until now.

Published in the Cell Press journal iScience on August 7, the study used fMRI to scan awake pet dogs while they viewed human faces expressing different emotions. The scans revealed a clear neural pathway for processing happy faces and, for the first time, evidence that dog brains distinguish between specific negative human expressions such as anger, fear, and sadness.

This shows a dog.
Canine brain networks process happy human faces in reward centers and distinguish between negative expressions. Credit: Neuroscience News

“Humans are very skilled at reading subtle facial cues, and dogs show a similar sensitivity,” says Raúl Hernández-Pérez, first author from the University of Vienna. “Studying dog brains helps us identify the neural adaptations that support their social understanding of people.”

The researchers began by examining happiness, an emotion prior work suggested is particularly meaningful for dogs. In the first experiment, eight dogs were shown photos of strangers with happy or neutral expressions. Happy faces elicited strong activation in the right temporal cortex and the caudate nucleus — areas associated with visual- social processing and reward, respectively.

Laura Cuaya of the University of Vienna, who continued the project through collaborations at other universities, notes, “Activation in the caudate hints that a smiling human face may be a rewarding or emotionally relevant cue for a dog.”

Next, the team tested whether the same brain network responded to other emotions. In a second experiment, 12 dogs viewed faces showing happiness, anger, fear, or sadness while researchers applied machine-learning classifiers to the fMRI data. Using the temporal-caudate cluster identified in Experiment 1 as a region of interest, classifiers reliably distinguished happy faces from each negative emotion. That finding shows the network is tuned specifically to happiness.

Importantly, whole-brain representational analyses revealed distinct patterns that separated fear from anger and sadness in different cortical areas. In other words, the dog brain does not simply sort expressions into “good” and “bad” — it forms separable neural representations for specific negative expressions.

The authors emphasize that these neural distinctions do not mean dogs experience emotions exactly as humans do. Static images are only part of how dogs assess human emotional states: in everyday interactions dogs combine facial cues with posture, vocal signals, and scent information.

“Comparisons with primates often reflect shared ancestry,” says senior author Laura Cuaya. “Dogs took a separate evolutionary route, but through domestication they developed the social skills needed to read and cooperate with humans.”

All participating animals in this study were family pets trained to remain calm in the scanner. The authors point out that free-ranging dogs living without regular human contact might process human cues differently because their social environments vary widely.

Future work will take a more dog-centered approach to understand how these neural signals guide dogs’ real-world behavior and will directly compare dog and human brain responses to the same emotional cues.

“I want people to appreciate dogs as emotional partners,” says Cuaya. “Recognizing that they are tuned to our expressions can change how we communicate with and care for them.”

Funding:

This research received support from the Austrian Science Fund, the European Research Council, the Hungarian Academy of Sciences, and the National Brain Programme 3.0.

Key Questions Answered:

Q: How can researchers scan a dog’s brain with fMRI without sedating the animal?

A: The dogs in these experiments are family pets trained with positive reinforcement to lie still in the scanner. They wear ear protection and voluntarily rest their heads on custom chin supports while watching screens, which allows scientists to record brain activity without anesthesia or undue stress.

Q: Why is caudate activation important when a dog sees a smile?

A: The caudate nucleus is a core component of the brain’s reward circuitry, associated with positive expectation and pleasure. Its activation indicates that a human smile may be intrinsically rewarding or emotionally meaningful for dogs.

Q: Does this mean dogs feel emotions exactly like humans?

A: No. The study demonstrates that dogs’ brains can differentiate human facial expressions, but it does not prove that dogs experience or conceptualize those emotions in the same way humans do. It shows that domestication has provided dogs with neural tools to interpret human emotional cues.

Editorial Notes:

  • This article was edited by a Neuroscience News editor.
  • The full journal paper was reviewed.
  • Additional context was provided by editorial staff.

About this animal psychology and emotion research news

Author: Julia Grimmett
Source: Cell Press
Contact: Julia Grimmett – Cell Press
Image: Image credit: Neuroscience News

Original Research: Open access.
“Dog Brain Representations of Human Facial Expressions: Encoding Happiness and Differentiating Specific Negative Expressions” by Karl Kristjan Kaup, Javier Hidalgo Jiménez, Jaan Aru.
DOI: 10.1016/j.isci.2026.116900


Abstract

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

Dogs can distinguish human facial expressions, particularly happiness, but the underlying brain mechanisms were previously unclear. Using fMRI, researchers ran two experiments in awake pet dogs to probe how canine brains respond to human emotional faces.

In Experiment 1 (n = 8), happy faces produced stronger responses than neutral faces in a right temporal cluster extending into the caudate nucleus, including parts of the rostral Sylvian gyrus. In Experiment 2 (n = 12), dogs viewed faces expressing happiness, anger, fear, or sadness.

Using the cluster from Experiment 1 as a region of interest, a machine-learning classifier distinguished happiness from each negative facial expression, indicating differential BOLD responsiveness to happy faces. Whole-brain representational similarity analyses revealed activity patterns that separated angry vs. fearful faces and sad vs. fearful faces in distinct cortical regions, providing evidence that dog brains can distinguish between specific negative facial expressions.