Scientists Put Awake Dogs in MRI Machines. What They Found Changes Everything

BLOG OVERVIEW & KEY TAKEAWAYS
Key Takeaway 1: Gregory Berns at Emory University trained dogs to lie still in an fMRI scanner without sedation, becoming the first researcher to measure dog neural activity in real time while the dogs were conscious. His findings: the caudate nucleus, the brain region associated with positive expectations and reward anticipation, activates specifically and reliably in response to the owner’s scent and the hand signal for reward. Dogs have a dedicated neural pathway for positive experiences involving humans.

Key Takeaway 2: When dogs were shown the scent of their owner versus strangers, the owner’s scent produced the strongest caudate activation of all stimuli tested. The same brain region that activates in humans when they encounter something they love activated in dogs for their owner’s scent. This is the closest thing to an objective neurological measurement of the love a dog has for their person.

Key Takeaway 3: The MRI research also revealed that dogs process human faces in a dedicated facial processing region of the brain similar to the fusiform face area in humans. They are not just seeing a face and recognizing it as familiar. They are processing it with specialized neural architecture. Your face has its own dedicated neural real estate in your dog’s brain.

The Day Dog Science Changed Forever

In 2012, a neuroscientist at Emory University named Gregory Berns did something that had never been done before.

He trained dogs to lie perfectly still inside a functional MRI machine, a noisy, confined cylinder that most humans find profoundly uncomfortable, while fully awake and unsedated. He taught them using positive reinforcement, building up their tolerance over weeks of careful training. He designed a chin rest that held their head position stable enough for clean neuroimaging data.

And then he looked inside their brains while they were conscious.

What he saw there answered questions that dog owners have been asking forever. Questions that had previously been dismissed as sentimental anthropomorphism. Questions about whether the connection you feel with your dog is real, reciprocal, and neurologically meaningful.

The answers, published across multiple peer-reviewed papers and later in Berns’ book How Dogs Love Us, are remarkable.

The Training Process: A Story of Partnership

Before discussing the findings, it is worth understanding what it took to produce them. Because the process itself says something important.

Berns and his team could not sedate the dogs. Sedation alters brain activity. They needed natural, awake brain states. Which meant the dogs had to voluntarily, calmly lie still in a machine that produces approximately 95 decibels of mechanical noise, in a confined space, with their head immobilized.

The only way to accomplish this was through the same positive reinforcement training that forms the backbone of modern dog training. The dogs were not forced or coerced. They were trained patiently, rewarded at every step, and given the ability to leave whenever they chose.

Several dogs who began the training ultimately decided not to participate. They were not brought back.

The dogs who did participate completed hundreds of scan sessions voluntarily. They were walked to the scanner, assumed the position, and held still through multiple 10-minute scan protocols.

What this means: the neurological findings were produced by dogs who chose to be there. Every scan result represents a dog who voluntarily submitted to the procedure for the rewards and the relationship with their training partner.

The science about dog love was produced through an act of dog trust.

Finding 1: The Caudate Nucleus and What It Means

The caudate nucleus is a brain structure associated with anticipatory reward processing. In human brain imaging studies, it activates in response to things people value highly: food, money, pleasant social interaction, and the anticipated presence of loved ones.

When Berns and his colleagues presented dogs with five different scents inside the scanner (their own scent, a familiar human who was not their owner, an unfamiliar human, a familiar dog, and an unfamiliar dog), the owner’s scent produced stronger caudate activation than any other stimulus.

The owners were not present. The dogs had not seen them recently. There was no food present. The only stimulus was the molecular signature of the person the dog lived with.

And the reward-anticipation center of the brain lit up.

Your scent activates the same region of your dog’s brain that activates in humans when they encounter something they love. That is not metaphor. That is a measurement.

Berns’ interpretation: the caudate activation in response to owner scent suggests that dogs have positively valenced associations with their primary person that are neurologically similar to the positive associations humans have with things they love.

He chose his words carefully because he is a scientist. But the implication is clear: dogs do not just tolerate their owners or seek food from them. Their brains register their owners as something deeply positive in a way that shows up on the same brain imaging patterns as human love.

Finding 2: Dogs Process Human Faces With Dedicated Architecture

A subsequent study by Peter Cook and colleagues at Emory, also published using the MRI methodology, examined how dogs process faces.

In humans, there is a region of the temporal cortex called the fusiform face area (FFA) that is specifically activated by faces. It processes faces differently from other objects. It activates more strongly for faces than for comparable non-face stimuli. It is specialized neural architecture dedicated to face processing.

When dogs in the scanner were shown short video clips of human faces versus the backs of human heads, and dog faces versus the backs of dog heads, a region in the dog’s temporal lobe showed preferential activation to faces over non-face stimuli.

Dogs have a face-preferential brain region. Your face is processed in the dog’s brain through specialized neural machinery that treats it as categorically different from other visual stimuli.

Furthermore, the region showed activation for both dog and human faces, suggesting it is a general face-processing region rather than species-specific. Dogs recognize the category “face” and process it through dedicated neural systems.

Your face has its own neighborhood in your dog’s brain.

Finding 3: The Voice-Processing Discovery

Researcher Attila Andics at the Eötvös Loránd University in Budapest used the same fMRI methodology to study how dogs process human voices.

The finding: the dog brain has a region in the auditory cortex that responds specifically to voices, similar to the voice-sensitive regions in human temporal cortex. And critically, this region shows differential response to emotionally positive versus emotionally negative voices, and responds more strongly to familiar voices than unfamiliar ones.

Your dog’s brain has neural architecture that specifically processes your voice as distinct from other sounds, recognizes its emotional content, and responds differently to your voice depending on your emotional tone.

When you speak to your dog with affection, their brain activates differently than when you speak with frustration. Not just behaviorally. Neurologically.

Finding 4: Reward Signals For Praise Versus Food

One of Berns’ most telling experiments involved comparing caudate activation in response to food rewards versus verbal praise from the owner.

The dogs were trained to associate two different objects (a toy car and a toy horse) with different rewards: one predicted food, the other predicted verbal praise from the owner.

The majority of dogs showed similar or higher caudate activation in response to the praise-predicting object compared to the food-predicting object.

About 20% of the dogs activated more strongly for the food predictor.

The remaining majority valued their owner’s attention neurologically at least as much as food. For many dogs, praise from their person activated the reward system more strongly than eating.

This is a direct neural answer to the question of whether a dog’s attachment to their owner is truly social or primarily food-motivated. For most dogs in this study, it is genuinely social.

What the Research Cannot Tell Us

Berns is careful about this. What the MRI findings confirm is that dogs have neural activation patterns in response to their owners that resemble the patterns humans show toward things they love. What the research cannot directly establish is the subjective experience behind that activation.

We cannot know, from brain imaging data alone, whether a dog’s subjective experience of their attachment to their owner is phenomenologically similar to human love. That question may never be definitively answerable because we cannot directly access another being’s subjective experience.

What Berns argues, and what the research supports, is that the neural machinery is there. The activation patterns are there. The functional equivalence of the response to what we call love in humans is there.

Whether that constitutes love in the philosophical sense, he leaves to the reader.

Most readers, having read the research, have already decided.

Q&A

Q: How did they teach dogs to lie still in an MRI?

Through systematic desensitization and positive reinforcement training conducted over weeks. Dogs first practiced the chin rest position. Then they were habituated to the scanner sounds using audio recordings at increasing volume. Then to the scanner aperture using a mock scanner. Then to brief sessions in the actual scanner, building duration slowly. The dogs who completed the full protocol did so entirely voluntarily and left training sessions eagerly returning.

Q: Were these dogs special? Would any dog do this?

The initial dogs in Berns’ lab were his own dogs who happened to be highly trainable. Subsequent studies expanded to dozens of dogs of various breeds showing that the methodology is replicable across a broad population. The limiting factor is training time and the dog’s individual temperament, not any rare special ability.

Q: Has this research changed how scientists view dogs?

Significantly. The MRI data provided the first direct neurological evidence for the emotional richness of dog cognition and the specific neurological reality of their bond with humans. It shifted the scientific conversation from “dogs appear to prefer humans” to “dogs have dedicated neural architecture for processing human faces, voices, and scents, and their reward systems activate specifically for their owners.” That is a meaningful scientific shift.

Final Thought

A dog lying still in a loud, uncomfortable machine, voluntarily, for the sake of the relationship with their training partner.

A brain scan showing the reward center lighting up for the smell of the person they live with.

Dedicated neural architecture for processing your face. Your voice. Your presence.

The science has been looking for evidence of the bond you already knew existed. It found it. It found it in the architecture of your dog’s brain itself.

They were built to love you. And they do.

Scientists Put Awake Dogs in MRI Machines. What They Found Changes Everything.