How dogs think

The companionship of dogs has added meaning to human lives for thousands of years. Amid declining birth rates and a weakening of traditional support systems, that bond may now be more important than ever.

“Dogs are increasingly stepping into this emotional and relational gap, not merely as pets, but as surrogate children or primary companions,” said Enikő Kubinyi, PhD, a professor and head of ethology (the study of animal behavior) at Eötvös Loránd University (ELTE) in Budapest, Hungary (Current Directions in Psychological Science, online first publication, 2025opens in new window).

Research led by Kubinyi shows that dog-human relationships combine the upsides of best friend relationships and parent-child bonds, making them more supportive and positive than most relationships between humans (Scientific Reports, Vol. 15, 2025opens in new window).

“Clearly, there’s something remarkable and unique about how dogs live with people. What is the underlying process that makes this relationship so special? How do people and dogs fall in love with each other?” said Clive Wynne, PhD, a professor of psychology at Arizona State University who studies the behavior of dogs.

Answering those questions, and understanding how to optimize the relationship between dogs and humans, is at the heart of the growing field of dog cognition. On top of studying how dogs understand human gestures and language, psychologists and others are exploring dogs’ powerful sense of smell, how they choose which humans to cooperate with, and even what their memories are made of. Researchers are also using magnetic resonance imaging (MRI) and electroencephalography (EEG) to peer inside the minds of dogs, as well as collecting brain tissue and genetic data to understand aging and disease.

Studying the unique relationship between dogs and humans may even offer insights about how social and cognitive abilities evolved in humans.

“Dogs are special because they were the first domesticated animal. Looking at their brains is almost like looking back in time,” said Gregory S. Berns, MD, PhD, a professor of psychology and distinguished professor of neuroeconomics at Emory University. “We hope this can offer clues about how they came to not only coexist but to thrive with humans.”

Lessons from the canine mind

The following prompts (organized by topic) can support high school and undergraduate educators in exploring dog cognition and the human–canine bond in psychology courses.

1. Motivation and emotion 
Dogs often form strong emotional bonds with humans and show signs of empathy and stress detection. How might classical and operant conditioning explain dogs’ emotional responses to human behavior?
Reinforces: Conditioning, emotional learning, and animal behavior

Decades ago, dogs caught the interest of comparative psychologists after studies showed they could outperform apes on certain tasks. While apes could better use logic to find hidden food, dogs gained more from social and communicative cues, such as an experimenter pointing to the location of food (Bräuer, J., et al., Journal of Comparative Psychology, Vol. 120, No. 1, 2006; Hare, B., & Tomasello, M., Journal of Comparative Psychology, Vol. 113, No. 2, 1999). Early research on dog domestication led by Brian Hare, PhD, a professor of evolutionary anthropology and psychology and neuroscience at Duke University, also sparked interest in the humanlike social cognitive skills of dogs (Science, Vol. 298, No. 5598, 2002opens in new window; Trends in Cognitive Sciences, Vol. 9, No. 9, 2005opens in new window). Since then, peer-reviewed publications in the field have surged (Aria, M., et al., Animal Cognition, Vol. 24, No. 3, 2021opens in new window).

Initial efforts were met with some resistance. Many believed dogs operated solely on Pavlovian conditioning and were not worth studying. But pioneers such as Ádám Miklósi, PhD, who leads research on dog cognition at ELTE, began to uncover dogs’ unique social abilities—helping the subject gain scientific credibility.

Now, the highly collaborative field uses a clever suite of methods to explore the inner lives of dogs. In the lab, short problem-solving games allow researchers to quickly collect data from their subjects, who are typically volunteer pet dogs or service dogs in training.

“I often describe this type of research as a theatrical production for dogs,” said Zachary Silver, PhD, an assistant professor of psychology at Occidental College in Los Angeles. With a dog on one side of the room, humans on the other side interact, hide food, or perform other actions, while researchers observe how the dog responds.

Because pet dogs are so common, researchers increasingly rely on citizen science. Thousands of pet owners conduct simple experiments at home, adding further richness to the data on dog behavior.

“It’s an interesting research culture where studies don’t just happen quietly in a lab, but people can watch and participate directly,” said Evan MacLean, PhD, an associate professor of veterinary medicine and founder of the Arizona Canine Cognition Center at the University of Arizona.

Large-scale international research collaborations such as the ManyDogs projectopens in new window—an international consortium of researchers working in canine science—are also pooling data from labs around the world to answer the field’s most pressing questions, including why dogs follow the human pointing gesture.

The research boom has also offered the chance to reexamine some assumptions baked into the field. For example, virtually every early study of dog cognition used visual stimuli, with less than 2% focused on olfaction (Animal Cognition, Vol. 23, No. 1, 2020opens in new window). That comes down to a human bias toward examining the world with our strongest sense, said cognitive scientist Alexandra Horowitz, PhD, who runs the Horowitz Dog Cognition Lab at Barnard College, but it has long limited our understanding of how dogs think.

Horowitz has adapted visual techniques used with infants and other animals to study the sense of smell in dogs. For example, the mirror test checks for self-recognition in animals: While anesthetized, an animal is marked—typically on the forehead—then given access to a mirror when awake. If the animal investigates the mark, that is seen as a form of self-recognition. Horowitz designed an olfactory version of the test, which found that dogs can distinguish their own odor from that of other dogs and notice when their odor has been modified (Behavioural Processes, Vol. 143, 2017opens in new window).

She also studied the full body “shake” behavior—the motion dogs use to shake off water. Horowitz found that it typically preceded a change in behavior, such as when two dogs walk away and stop interacting (Animals, Vol. 14, No. 22, 2024opens in new window). Other research has delved into the head-tilt, finding that it happens when dogs process meaningful external cues, such as the name of a toy they know (Sommese, A., et al., Animal Cognition, Vol. 25, No. 3, 2022opens in new window).

“The fact that straightforward behaviors like ‘shake’ and ‘head-tilt’ aren’t well understood means we may have gotten ahead of ourselves with dog cognition work and forgotten to ask what some basic things really mean,” Horowitz said.

What adaptations in dogs allow them to fit so well into human society? Researchers disagree. Wynne and others argue that dogs succeed around people because of their capacity to form strong emotional bonds with members of other species—and that their intelligence stems from a strong ability to make mental associations (e.g., between words and objects). From this perspective, many dog behaviors can be seen as conditioned responses. Others say dogs have evolved a more sophisticated form of social intelligence—one that parallels that of humans.

The simple question of why dogs follow the human pointing gesture aims to tease out the underlying process. Do dogs understand that pointing conveys meaning—or are they simply conditioned to follow a finger? The first major study from the ManyDogs consortium found that dogs did not respond differently to pointing alone versus pointing plus social cues, including eye contact and calling of the dog’s name. That finding appears to support a conditioning-based view of dog behavior over a social cognitive one (Espinosa, J., et al., Animal Behavior and Cognition, Vol. 10, No. 3, 2023opens in new window).

Another key question at the heart of the dog cognition debate: How do dogs process human language? All dogs can learn some words and some highly gifted dogs can learn hundreds of words (Kaminski, J., et al., Science, Vol. 304, No. 5677, 2004opens in new window). But are dogs really forming mental representations of an object or action—or are they just very skilled at making associations?

An EEG study from the ELTE group found that dogs may indeed form mental representations of what words mean. Using a “semantic violation paradigm,” researchers spoke a word that dogs were trained to know (such as “ball”) while showing them an object. Sometimes, the object matched the word; other times, they were mismatched. In humans, the mismatch (and resulting realization that the word and object don’t belong together) triggers a distinct electrical response in the frontal lobe known as the N400 effect. In the 2024 study, dogs showed a comparable EEG response (Boros, M., et al., Current Biology, Vol. 34, No. 8, 2024opens in new window).

“These findings suggest that upon hearing these words, dogs activate corresponding mental representations, indicating a capacity for referential understanding,” Kubinyi said.

Other studies have found parallels between dog and human brains, offering some of the first evidence of such similarities beyond primates. Berns found an area linked to facial processing in the canine temporal lobe (PeerJ Life & Environment, Vol. 3, 2015opens in new window). Claudia Fugazza, PhD, an ethology researcher at ELTE, found evidence of episodic-like memory in dogs. She and her colleagues first trained dogs to repeat actions on command (using the cue “Repeat!”), then tested whether dogs could remember and repeat spontaneous actions they had performed earlier. When given the “Repeat!” command, even up to an hour after performing a spontaneous action, dogs recalled and reproduced those behaviors. This suggests that their memory involves recalling past experiences rather than simply linking commands to specific actions (Scientific Reports, Vol. 10, 2020opens in new window).

Researchers are also further investigating why some “gifted dogs” have cognitive abilities that far surpass their peers. Researchers at ELTE have proposed a canine “g factor,” a score that represents a dog’s cognitive abilities and varies between individuals and over the life span. A series of seven tasks, measured with 129 pet dogs and longitudinally with a smaller subset, revealed individual differences in problem-solving and learning abilities (Bognár, Z., et al., GeroScience, Vol. 46, No. 6, 2024opens in new window).

Those findings point to important parallels between dog and human intelligence, Kubinyi said, and suggest that dogs may be a valuable model for research on cognitive aging in humans.

While some research aims to pin down the fundamentals of dog behavior and cognition, other studies are exploring the nuances of their interactions with humans. We know that dogs can reduce emotional distress, increase life satisfaction, and even help treat post-traumatic stress disorder (Matijczak, A., et al., Emotion, Vol. 24, No. 2, 2024; Gmeiner, M. W., & Gschwandtner, A., Social Indicators Research, 2025opens in new window; Leighton, S. C., et al., JAMA Network Open, Vol. 7, No. 6, 2024opens in new window). But how does the special relationship between dogs and humans unfold—and what can humans do to improve it?

Silver is investigating how dogs read human social cues to decide who to cooperate with. In one study, dogs watched an experimenter interact with two other humans: a helper and a non-helper. When the experimenter reached for a clipboard just out of reach, the helper always handed them the clipboard; the non-helper always moved the clipboard further away. Dogs quickly developed a preference for the helper (Animal Cognition, Vol. 24, No. 1, 2021opens in new window).

“It only takes a few occurrences before dogs develop a pretty substantial preference for the prosocial person,” Silver said.

Dogs also have a knack for adapting to human behavior and emotions. Research has shown that dogs synchronize their behavior with both children and adults and that they produce significantly more facial movements when a human is paying attention to them (Wanser, S. H., et al., Animal Cognition, Vol. 24, Vol. 4, 2021opens in new window; Kaminski, J., et al., Scientific Reports, Vol. 7, 2017opens in new window). When dogs and children interact, oxytocin levels rise in both parties (Gnanadesikan, G. E., et al., Psychoneuroendocrinology, Vol. 169, 2024opens in new window).

Other research seeks to understand how dogs respond to human stress. Jeff Stevens, PhD, the Susan J. Rosowski Professor of Psychology and head of the Canine Cognition and Interaction Lab at the University of Nebraska–Lincoln, is observing whether dogs act differently around odors collected from humans before and after a stress task.

“In the past, dogs probably did better if they avoided stressed-out humans,” Stevens said. “Have they now evolved to respond to human signals that are relevant to them?”

For example, do dogs spend time near stress odors or show stress-related behaviors (such as licking their lips or soliciting attention from their owners)? Stevens and his team are now analyzing video they collected from 73 dogs.

Humans, on the other hand, may be worse at reading dogs’ emotions than we think. Wynne and his doctoral student, Holly Molinaro, recorded videos of Molinaro’s father interacting with their family dog. In some videos, he triggered a “happy” reaction by offering a treat; in other videos, he triggered an “unhappy” reaction by offering a cat. When participants viewed the original videos, they correctly predicted the dog’s emotional state. But when Molinaro recut the videos so the dog’s reaction no longer matched the original trigger, participants still rated the dog’s emotion based on the trigger they saw. That suggested their answers relied more on context than on the dog’s actual behavior (Anthrozoös, Vol. 38, No. 2, 2025opens in new window).

“There’s a cautionary tale here,” Wynne said. “We have these powerful intuitions that we know how our dogs are feeling, and yet the evidence suggests that people are actually really bad at this.”

Researchers are also studying dog training methods, both as a way to strengthen the human-dog bond and to gain further insights about how dogs learn.

A large-scale developmental study with service dog puppies, led by Hare, helps home in on what makes training effective. Hare and his colleagues tracked the development of more than 100 puppies over a 7-year period, including their cognitive skills between 8 and 20 weeks of age. They compared dogs reared in a “super-enriched environment” (family homes) with those reared in an “extreme-enriched environment” (Duke’s campus, where dogs interacted with thousands of students).

Hare found that these two rearing environments did not impact dogs’ memory, self-control, communication, or cognitive abilities differently. The research also showed that basic abilities, such as paying attention to human gestures, develop around week 8 of a dog’s life. Self-control, including skills needed for house training, develops between weeks 12 and 14. That has an important implication for dog welfare, Hare said: If a dog barks at night before 14 weeks of age, let it out to do its business.

The research, published in Puppy Kindergarten: The New Science of Raising a Great Dogopens in new window, also found that playing a simple game with puppies helps them develop a skill critical to effective training and a healthy dog-human relationship: eye contact. First, show a puppy it can get food from a box and allow it to succeed a few times. Then, lock the box. After unsuccessful attempts to open the box, most dogs eventually stop, look up, and make eye contact to ask for help.

“Dogs who repeatedly played this game, which takes about 5 minutes, made twice as much eye contact at 20 weeks of age than dogs who had not played the game,” Hare said.

In her research, Horowitz also found that playing “nose work” games (scavenger hunts that rely on smell) with pet dogs increased their positive judgment bias, a proxy for optimism in dogs (Applied Animal Behaviour Science, Vol. 211, 2019opens in new window).

“Sniffing scavenger hunts are an easy thing you can add to a dog’s life, which is enriching for them—and maybe also for you,” Horowitz said.

New approaches are allowing researchers to start pinpointing what genetic differences underlie cognitive and behavioral differences between dogs. MacLean partnered with Canine Companions, a nonprofit service dog provider that keeps detailed family trees, allowing researchers to know the exact genetic relationships between dogs. Their first study indicated that about 40% of the variation in some puppy social skills, including eye contact with people and point-following tendencies, could be explained by genetics (Current Biology, Vol. 31, No. 14, 2021opens in new window).

MacLean is also part of the Dog Aging Projectopens in new window, which is studying cognitive aging in 50,000 dogs across the United States with the hope that it can improve life for senior dogs and also offer clues on treating Alzheimer’s disease in humans. At ELTE, the Canine Brain and Tissue Bank is also supporting research on neurodegeneration, as well as on how the gut microbiome influences cognition and behavior.

“New brain imaging and genetic tools are letting researchers explore how dogs think and feel in ways that weren’t possible before,” Kubinyi said. “We have only begun to understand the complexity of the canine mind and the human-canine bond.”