By Sheeva Azma
When neuroscientist Dr. Laura Cuaya moved from Mexico to Hungary, she wondered if her dogs would notice the fact that everyone was suddenly speaking a different language.
She designed an experiment to answer the question: can dogs distinguish between different languages? One of her test subjects was her own dog, Kun-Kun.
Magnetic resonance imaging or MRI is a type of brain scanning technology that allows you to study the structure and function of one’s brain. Functional MRI studies (fMRI) is a type of MRI that helps scientists understand what parts of the brain may be involved when doing specific tasks.
In her experiment, the dogs listened to the children’s story, The Little Prince. They listened to the story in both Spanish and Hungarian.
Her research showed that dog brains can detect speech, and that dogs’ brains show different activity patterns to familiar and unfamiliar languages. Her study has since been all over the international media, at NBC News, Gizmodo, SciShow, and more news outlets worldwide. It has also graced the cover of NeuroImage, the academic journal where it was published.
.@Lauveri’s research showed that dog brains can detect speech and that dogs’ brains have different activity patterns to familiar and unfamiliar languages.Tweet
An Interview with Dr. Laura Cuaya, Dog MRI Expert
I recently chatted with Dr. Cuaya, who is a postdoctoral researcher at the Neuroethology of Communication Laboratory at Eötvös Loránd University in Budapest, Hungary, and the lead author on this study. “Fortunately, Kun-kun is still humble,” she says regarding the media attention.
From being an MRI researcher myself, I know that studying your dog with MRI is no easy feat. For one thing, most software is developed with humans, not dogs, in mind. Another challenge is getting your participant – whether a person, dog, or other animal – to stay still in the scanner. Too much movement can hinder image quality, rendering the data unusable. That’s not even to mention the everyday challenges of designing an MRI experiment, which I am all too familiar with. So, I asked Dr. Cuaya about it all. Keep reading for our conversation!
Sheeva: What was it like putting your dog, Kun-kun, in the MRI scanner for this study? Does he like being in the MRI scanner?
Dr. Cuaya: Kun-kun and his brother, Odín, are my first dogs. When I started my Ph.D., studying awake and cooperative dogs using functional MRI, they were the first participants.
Kun-kun finished his training in 2015. He always participates in our experiments and many behavioral projects at the Department of Ethology. So, for this study, he didn’t require any extra training. Indeed, the researchers in Budapest were impressed because, on our first weekend in Budapest, Odín and Kun-kun participated in an experiment successfully.
Here, I would like to add that for us, a “successful” experiment is when dogs move a maximum of 3 mm during the experiment (around 5 minutes). They are amazing dogs!
There are always one or two familiar people inside the room during the scanning. In this way, we are sure that the dog is relaxed. We consider that their psychological state inside the scanner is more comparable to humans than other animals with physical restraint or anesthetized.
From my perspective, dogs enjoy participating in our studies. Besides seeming happy, they lay voluntarily and don’t leave the session (they can leave the scanning at any point).
Sheeva: Where did you get the idea for this study?
Dr. Cuaya: First, it was a personal motivation. When I moved to Budapest, Hungarian was very salient for me. In Budapest, people are extra friendly with dogs. My dogs were delighted to be greeted anywhere. I wondered if they also noticed that people in Budapest speak a different language.
Happily, this question fit in with the Neuroethology of Communication Lab goals. In several studies, the lab has shown how dogs perceive speech. Now, we have added another little piece to the big picture.
Sheeva: Do you think pets really learn languages or pick up on our prosody – the way we say things governed by the rise and fall of our voices? How did you design your study to test that dogs notice the language itself?
Dr. Cuaya: We chose fragments of The Little Prince for our study because the content wasn’t directly relevant for the dogs. We know that dogs’ brains process the meaning of relevant words (Andics et al., 2016), so we wanted to avoid words that dogs could identify (like praise words). A great and cleverly-designed experiment found that dogs present “the cocktail party effect,” in which, even in noisy environments, dogs reacted to their name, even if it was spoken by an unknown speaker (Mallikarjun et al., 2019).
In our study, dogs listened to The Little Prince as read by two speakers, one who spoke Spanish and one Hungarian. We made sure that both speakers were unknown to the dogs, since dogs’ brains process the speaker’s identity (Boros et al., 2020). Finally, we controlled for acoustic properties between languages. In this way, we could rule out that acoustic properties, such as the rise and fall of one’s voice, drove the dogs’ distinction between languages we observed in our data.
While we found different activity patterns in the secondary auditory cortex for the languages, it is important to note that we don’t find different activity patterns for both scrambled stimuli. Our results could suggest that dogs’ brains distinguished prosody in the languages. However, as we did not have a systematic variation in prosodic cues, we can not conclude what prosodic cues were relevant. We conclude that the dog’s secondary auditory cortex activity patterns distinguished between language – regularities and not only two complex audios (i.e., scrambled stimuli).
Sheeva: Most MRI analysis software is for human brains, and most work by referencing a brain that is developed from averaging many human brains together and fitted to Talairach coordinates. Was it challenging analyzing a dog’s brain data when most research studies have been done on humans?
Dr. Cuaya: This is an excellent question. Fortunately, fMRI software for humans can process dogs’ data. However, some of the pre-processing steps are more manual than in humans. Also, in the last few years, different papers have implemented improvements. For example, a stricter motion correction (Prichard et al., 2018) or proposing that a tailored dog hemodynamic response function (HRF) can increase detection power in the visual modality (Boch et al., 2021). In 2019, Czeibert and colleagues published a detailed – and great – dog atlas. I am happy to say that I met the dog who is the basis of this atlas, Barney.
Sheeva: What were some of the other challenges of this study?
Dr. Cuaya: We had problems with the linguistic terms. Our control stimuli were called “scrambled” because we cannot be sure what linguistic properties keep or lose compared to the speech. And then, we conclude that dogs detected “language regularities” because we weren’t sure if “prosody” adequately described the difference between stimuli. In future studies, we would like to systematically explore the language’s acoustic properties to understand better what dogs detected in our study.
Sheeva: Do you have plans for more studies for Kun-kun and the other dogs?
Dr. Cuaya: In contrast to findings in humans, we did not find a higher cerebral response to speech than scrambled speech in dogs. In future experiments, we would like to explore if the dog brain has a speech sensitivity by presenting a broader selection of stimuli to dogs.
Sheeva: What are your overall goals for this research?
Dr. Cuaya: I would like to contribute to how people see dogs. Our study joins the evidence that dogs learn about their social environment. Although we never teach our dogs their familiar language sounds, their brains know it. As someone with dogs in my family, it is great to know that dogs are picking up subtle cues of their social environment all the time. I hope people give dogs opportunities to continue learning by involving them in family activities.