A study has shown that dogs are able to pick out individual words in sentences. This is thanks to the same computations and brain regions used by human babies.
We learn to spot new words first as infants. Then, we begin to learn the meanings of each word.
To tell where each word ends and another begins, babies use complex calculations that keep track of which syllables appear together — and thus likely form words.
By using a combination of brain imaging techniques, experts led from Hungary’s Eötvös Loránd University have shown that dogs are capable of similar feats.
This is the first time that statistical learning, or the ability to apply it to non-human mammals has been demonstrated.
These findings are in line with a study that revealed that dogs tilt their heads while listening to music because it helps them hear and process information better.
Scroll down for video

A study has shown that dogs can pick up individual words in sentences when they are spoken to. This is due to the same brain regions and computations as human babies.
‘Keeping track of patterns is not unique to humans — many animals learn from such regularities in the surrounding world, this is called statistical learning,’ explains paper author and ethologist Marianna Boros of the Eötvös Loránd University.
Complex computations are what make speech unique. It is not enough to count the number of times certain syllables are used together in continuous speech to learn new words.
“It is easier to calculate how likely these syllables are to occur together.
‘This is exactly how humans, even 8-month-old infants, solve the seemingly difficult task of word segmentation — they calculate complex statistics about the probability of one syllable following the other.
“Until now, we didn’t know if other mammals could use such complex computations in order to extract words from speech. We decided to test the brain capabilities of family dogs for statistical learning using speech.
‘Dogs were the first domesticated animal species. They are also the one we most often refer to. However, we don’t know enough about their word-learning abilities and the neural processes behind them.
The study involved the measurement of electric brain activity in dogs using an electroencephalogram, or EEG.
The scans revealed key differences among dogs’ brain waves when it comes to rare and frequent words.
Lilla Magyari (an author of the study) explained: “We saw differences between dogs’ brain waves when they were able to process frequent words and rare words.
“But even more surprising, we also observed brain wave differences for synlables that always occurred together, compared to syllables which only occasionally did, even though total frequencies were the exact same.
“So it turns that dogs track not only simple statistics (how often a word occurs), but also complex stats (how likely a word’s parts will occur together).
“This has never been seen before in any other non-human mammals.” It is precisely the kind of complex statistics human infants use for extracting words from continuous speech.
The researchers then used functional MRI scanning for comparison to determine how similar brain regions in dogs that are responsible for complex computations to human brains.

The researchers measured the electric brain activity in dogs using an electroencephalogram, or EEG.
As with the EEG scans, the tests were performed on awake, cooperating, unrestrained animals, although the dogs involved in the fMRI experiments were previously trained to lie motionless for the duration of the scans.
“We know that both language-related and general learning brain regions participate in this process in humans,” Dr Boros explained. Dr Boros said that we also found the same duality in dogs.
“Both a generalist brain region and a specialist one” [the basal ganglia and auditory cortex, respectively]However, the activation patterns of the two were quite different.

Functional MRI scanning was used by the researchers to examine how similar brain regions in dogs that are responsible for complex computations to those in humans.

The fMRI scans were conducted on the same animals as the EEG scans. However, the dogs in the fMRI experiments had been trained to remain motionless for the duration of the scans.
The generalist brain region responded better to a random speech stream (where there were no words that could be detected using syllable stats) than to a structured stream (where words could be easily identified by computing syllable stats).
“The specialist brain area showed a different pattern: we saw brain activity rise over time for structured speech streams but not random speech streams.
“We believe that this activity is the trace word learning leaves on to the auditory cortex.

We learn to recognize new words from speech streams as infants. However, we don’t actually learn the meaning of each word until we become adults. To tell where each word ends and another begins, babies use complex calculations that keep track of which syllables appear together — and thus likely form words
According to the researchers, the findings suggest that neural processes that are key to human language acquisition may not be just for humans.
Attila Andics (an author of the study) added that “But we still don’t know how these human-analogue brain mechanisms to word learning emerged in dogs.
Are they a result of skills acquired in a language rich environment or over thousands of years of domestication or are they an ancient mammalian ability?
“By studying speech perception in dogs, including better breeds of dogs with different communication abilities, and other close-to-human species, we can trace back the roots of speech perception specializations.”
The complete study was published in Current Biology.