New evidence suggests dogs may ‘picture’ objects in their minds, similarly to people

When a dog follows a command or fetches a ball, it’s hard to know what’s really going on inside its canine cranium. Do dogs understand and respond to tone of voice, the syllables of words, accompanying hand motions and body language, or just the situational context? Behavioral studies have offered some clues, but new research brings additional evidence that our favorite furry friends really do grasp the meaning behind words. 

Dogs show a pattern of neural activity that seems to indicate they can differentiate between words for different objects, and are even surprised when presented with words and objects that don’t match up, according to a study published March 22 in the journal Current Biology. A team of neuroscientists and animal behavior researchers used non-invasive electroencephalogram (EEG) testing to measure the electrical pulses inside 27 pet dogs’ brains during an experiment involving the dogs’ owners and some well-loved toys. They found an electrical impulse pattern similar to a known signal in humans. The findings shed light on canine noggins and also add to our knowledge of the origins of complex language. 

“It’s wonderful to have studies like this,” says Ellen Lau, a neuroscientist studying linguistics at the University of Maryland who was not involved in the new research. Applying EEG to dogs, instead of the more invasive techniques that are often used to study animal brains, allows for more direct comparisons between humans and non-humans, she explains. “If we want to understand what’s common across humans and animals, we need to have more of this kind of data.”

A vocabulary test for babies, adapted to dogs

Among animals, family pups are unique for how much exposure they get to human language. “You can probe a lot of interesting questions about language experience with dogs, because they’re some of the only animals that live in our houses and pay attention to us,” says Amritha Mallikarjun, a neuroscientist researching canine cognition at the University of Pennsylvania who was not involved in the new research. 

The study scientists set out to test if dogs grasp the relationship between words and their corresponding objects. “We were interested in whether dogs understand words the way humans do,” says Lilla Magyari, co-lead author and a cognitive neuroscientist and psychologist at the University of Stavanger in Norway. Some standout dogs are able to demonstrate their vocabulary through behavioral tests, but not all dogs are as abiding, capable, or well-behaved. The scientists wanted to know if even dogs that don’t display exceptional abilities still have some language sense. 

People have internal references for what words mean, or the ability to “picture” an object inside their minds’ eye from memory. However, it’s unclear if any other animals share this capacity to imagine something that’s not there from an associated sound. To explore this question, Magyari and her colleagues adapted a cognitive test previously used in studies of infants. The assessment compares EEG readings from a subject told a word or phrase, and then either shown a corresponding object or an object that doesn’t match the description. 

In humans, even those too young to speak, an observable effect called the N400 appears on EEG read-outs when people encounter language and other stimuli. It’s a characteristic signal that peaks around 400 milliseconds after a stimulus is presented, and gets larger when objects or images and words don’t match up. The bigger the surprise, the bigger the signal. Many scientists interpret the effect as evidence of understanding and proof of an internal reference for a word’s definition, even in non-verbal subjects.  

In order to make the test canine friendly, Magyari and her co-researchers made some careful adjustments, controlling for the dogs’ comfort, potential variability in voice, and other movements or communication signals between dogs and owners that might influence results. 

“I think this study is beautiful,” Lau says–noting the thorough and well-considered design. “I think they really did everything you need to be doing in animal cognition work.” 

The dogs, all healthy companion animals, were recruited via social media and were selected based on an owner’s assessment that their pet understood at least three object words. After a period of acclimating to the lab, owners and dogs were separated by an electronic window that could quickly toggle between transparent and opaque. The scientists attached electrodes to the dogs’ heads at key points. Over multiple trials, the pets were played recordings of their owners’ voices calling their attention to one of five familiar objects (e.g. ‘Fido, look, the ball’), while being shown their owners’ faces through the window. Then, after a brief period of opaque blankness, the window would reveal the owner holding up one of the objects–either a match to the previously played phrase, or a mismatch. Meanwhile, the EEG recorded the electronic pulses going on inside their brains. 

Out of the 27 dogs that started the experiment, 18 were included in the final analyses. Nine were excluded, mostly because they wouldn’t sit still enough to yield clean EEG data. But even accounting for the challenges of wiggley animal subjects, the scientists still found clear patterns in their results. 

When there was a mismatch between the auditory stimulus and the object presented, the dogs’ EEG readouts routinely showed a significant signal peak between 200 and 600 milliseconds later–indicating that even average domestic canines can distinguish between the meanings of some words. The dogs had the largest brain response when the most well-known words were paired with mismatched objects, based on owner reports and further bolstering the findings. 

The timing of the pulse suggests it could be analogous to the human N400 signal, though follow-up research would be needed to verify this hypothesis, says Marianna Boros, co-lead study author and a cognitive neuroscientist and psychologist at ELTE University in Budapest, Hungary. Mallikarjun notes that “EEG is very finicky,” and it’s possible that the documented brain wave in the study is something unique, because human and dog brains are so different, she explains. Nonetheless, Boros is eager to continue probing the potential connection. “Our study is just the first in non-human animals, testing this mismatch effect. We have to conduct many more,” she says. “It’s pretty exciting to see that there might be some evolutionary continuity.”

Piecing together the building blocks of language

Because the experiment used objects familiar to the dogs involved, the study doesn’t show that canines can generalize a word to mean an entire category of object–another key aspect of human language, says Colin Phillips, a linguistics professor at the University of Oxford and the University of Maryland. He was also not entirely convinced that the time delay within the research proved dogs were referencing a mental image or memory. “They’ve associated sounds with specific objects,” Phillips says. “It’s impressive… it’s a carefully controlled study,” he adds–but ultimately not a very surprising one. “We kind of already know dogs can do this.”

Language is more complex than just noun recognition, and the study does not suggest that dogs are anywhere near as capable language learners as humans, say Boros and Magyari. Rather it hints at what abilities among mammals might have preceded humans’ extraordinarily complex linguistic system, they note. 

Mallikarjun agrees that studying our pets can provide insight into ourselves. Through research like this, we can better learn what is and isn’t unique to human cognition, and come to understand language development, she says. 

And at the same time, it’s also a good reminder that dogs and other animals are special in their own ways. “Communication has many different components,” says Lau. “Just because humans have a particular kind of unique communication system that’s not fully shared with any other animals, that doesn’t mean other animals don’t have very complex communicative abilities too.”