Dogs can match some words with objects, study suggests
(CNN) - Dogs can understand that certain words refer to specific objects, according to a recent study, suggesting that they may understand words in a similar way to humans.
It offers the first evidence of brain activity for this comprehension in a non-human animal, researchers said, though the study's conclusion has faced scrutiny from other experts in the field.
It has long been known that dogs can learn commands like "sit," "stay," or "fetch" and can respond to these words with learned behaviors, often with the help of a treat or two, but untangling their understanding of nouns has proven more difficult.
To understand dogs' language skills, Lilla Magyari, an associate professor at Stavanger University in Norway and researcher at Eötvös Loránd University in Hungary, and Marianna Boros, a postdoctoral researcher at Eötvös Loránd University, were inspired by studies investigating the comprehension of infants before they can speak. They decided to mimic these experiments with dogs, they said.
As the study's lead authors, they devised an experiment in which 18 dog owners said words for objects their dogs already knew. Then, the owners held up either the matching object or a different one while small metal discs harmlessly attached to the dogs' heads measured brain activity in a process known as electroencephalography (EEG).
In this way, scientists observed that brain activity in 14 of the 18 dogs was different when they were shown an object that matched with the word, compared to one that was mismatched. They said that the resulting brain activity was the same as those produced by humans in similar experiments.
"Our claim is to say that a dog understands a word, it means in the absence of the object, the dog activates a so-called mental representation," Boros said. "We can imagine it as the memory for that object.
"When the owner shows the object which is not matching that mental representation, then there is a very typical brain response we observed in the dog's brain that in humans is widely accepted as an index of... semantic understanding."
There was a two-second gap between owners saying the word of an object and showing it, a condition favoring the interpretation that dogs understood the words rather than simply associated them with the object, the researchers argued in the study.
Words that dogs knew better - as determined by their owners - also produced a bigger mismatch effect when the wrong object was shown, which researchers said strengthened their hypothesis.
Previous experiments testing dogs' understanding of nouns had involved them fetching specific objects when asked, according to a statement released by the Eötvös Loránd University.
This method suggested that dogs only fetched the correct object at a rate expected by chance though, as Magyari noted, dogs can perhaps be unmotivated or distracted during studies.
By using EEG, there was no need for this behavioral response and researchers were able to test the dogs' "passive understanding because maybe they can reveal more than they are able to exhibit or show," she added.
But the true extent of the dogs' comprehension is still unknown, even by the study's authors, since the dogs were responding to their own toys and objects that the owners brought to the lab.
"In this study, we only know that when they heard the words they were expecting their (own) objects," Magyari said.
"So we don't know how much (understanding)... they have about the relationship between the word and the object, whether it also reflects categorical knowledge, which means whether they think the ball refers to many ball-like things not only to their own ball. This is something further studies need to look into."
Clive Wynne, a professor at Arizona State University and director of the university's Canine Science Collaboratory lab, told CNN the experiment was a "clever" concept but it showed dogs understood a "stimulus" followed by an "important consequence" rather than the meaning intrinsic to a word.
He said the time delay in the experiment "was neither here nor there, if it's conditioning there can be a gap of some seconds" and that only familiar words would elicit a response explaining the greater mismatch effect.
He said that dogs lacked the two areas of the brain crucial for human understanding of language therefore the EEG pattern highlighted by the researchers was not shared by humans.
"If we're making a claim that the pattern of the brainwave shows you that it must be an understanding of words, you need it to be the same pattern," he said.
The study was published in the journal Current Biology on March 22.