Brain scan study reveals dogs attend to word meaning, not just intonation

Imagine if we could capture the words of an angry dog owner holding a chewed-up shoe – "How could you? You terrible dog!" – and digitally alter the tone to sound praising. Would the dog be oblivious to the reprimanding content of the message? I should admit that, until quite recently, I thought that the answer was yes ­– that no matter how chastising the words you used, you could convince a dog that it is being showered in praise, simply by adopting an affectionate tone. But a recent study published in Science indicates that many of us might be vastly underestimating canine listening skills. The findings reveal that dogs do not rely exclusively on intonation when judging the reward value of human speech, but that they also recognise the meanings that we assign to words.

Attila Andics at Eötvös Loránd University in Budapest and his colleagues used functional magnetic resonance imaging (fMRI) to examine patterns of brain activity in 13 specially trained dogs as they listened to their trainer speaking either words of praise or neutral words. The two types of phrase were selected based on their containing words of similar frequency in Hungarian, which allowed the investigators to assume that any distinct effects that praising phrases might have on the brain could be more reasonably attributed to their particular meaning rather than the animals' greater familiarity with them. Importantly, the dogs got to hear each type of phrase being delivered in both a praising and neutral tone on different occasions. This was absolutely critical for disentangling any distinct effects that word content and intonation might have on the canine brain. Indeed, the setup led to several interesting discoveries, including fascinating similarities in the ways speech is processed in the dog and human brain.

The fMRI scans revealed that words of praise, as opposed to neutral words, provoked particularly strong activations in the dogs' left hemispheres, irrespective of whether they were actually delivered in a typical praising fashion or in an entirely neutral tone. This indicates that dog brains are capable of extracting the arbitrary symbolic content that we assign to words and learning the special importance of words with rewarding meanings. On the other hand, intonation was found to be the province of the opposite hemisphere, as brain scans revealed that right-side regions, which normally process auditory information, responded to neutral and praising tones of delivery with different levels of activation regardless of actual word content.

This anatomical dichotomy of word meaning and intonation in the canine brain has some striking parallels with what we know about the brains of humans. In the late 1800's, Karl Wernicke discovered that patients with damage to part of the left hemisphere suffered from a bizarre disorder of language in which their speech was fully fluent but entirely devoid of meaning, and that they also had difficulty understanding the meaning of other people's words. More recently, it's been discovered that patients with certain types of right hemisphere damage understand the meaning of words, but struggle to interpret people's emotional states or attempts at humour, which are usually carried by variations in intonation. It's interesting to discover that the two hemispheres of dog brains appear to be similarly specialised for interpreting communications using word meaning and tone.

However, when it comes to dogs deducing what we are trying to say, how important is intonation of voice as opposed to word content? After all, the evidence that dog brains can differentiate words' meanings and intonations in separate brain regions does not necessarily give us reasonable ground to believe that speaking to dogs in a deceptive tone would not cause the true meaning of words to elude them.

It turned out that dogs use tone to prepare their brains to the possibility of hearing some rewarding content, but integrate both sources of information to decide whether they should assume that they are indeed being praised. This interpretative process was indicated by the researchers' use of functional connectivity analysis, which measures how strongly different brain regions are communicating with each other at any given time. When dogs heard a praising tone, this substantially increased the strength with which auditory regions of their right hemisphere communicated with the caudate nucleus – a deep-brain cluster of neurons, found in most vertebrate species, that acts as a central component of the brain's reward system.

In fact, this effect doesn't seem to be too different from what happens in the human brain when we assess whether something we are perceiving is pleasant. A relatively recent human study found that auditory regions connect more strongly with parts of the reward system when individuals evaluate how much they enjoy a particular piece of music, and that the strength of this connection during listening predicts how much money people are subsequently willing to spend on buying a song.

Thus, from a dog brain's perspective, a praising intonation appears to compel the reward centres and auditory regions to talk to each other, as if alerting the animal to the possibility that it might really like what it's about to hear (eg. "Well done!"). However, only when both intonation and word meaning were consistent with praise did the dopamine-releasing neurons of the dog brain's reward system produce a strong level of activity that is associated with providing a sense of satisfaction.

The similarities between how canine and human brains interpret speech by integrating distinct sources of information about word meaning and intonation might be surprising. But perhaps they point to the fundamental influence that we have had on animal brains since the dawn of canine domestication some 35,000 years ago. It is possible that, in the process of selecting dogs with characteristics that made them better companions to the human race, we have inevitably also selected for brains that process communications in a similar way to ours. Ultimately, similarly tuned brains might be better suited for connecting with each other. Perhaps brain scans of wolves could offer us some further insights?

About the author

Post written by Sofia Deleniv for the BPS Research Digest. Sofia holds a degree in Experimental Psychology and is currently working towards a PhD in Neuroscience at the University of Oxford, where she investigates multisensory processing in the mouse brain. In 2015, she decided to try her hand at science writing by starting her blog 'The Neurosphere'. You can visit her Facebook page or Twitter feed for updates on her written work and other exciting bits of science.

Further reading

—Neural mechanism for lexical processing in dogs

Top image credit Borbála Ferenczy