Dogs’ brains, like humans’ brains, separate words in a continuous stream of speech, according to a recent study by staff at Eötvös Loránd University’s Department of Behavioral Sciences, published in the journal Current Biology.
Using vigilant EEG and fMRI (functional magnetic resonance imaging), researchers have now been able to demonstrate for the first time that non-human species of mammals can perform complex statistical calculations to recognize word boundaries, the university reports to MTI.
Children learn to recognize words in continuous speech before associating meaning with them. To determine word boundaries, children make complex calculations on the patterns of occurrence of syllables. Syllables that appear together regularly are words, and other syllable relationships are not. The latest discovery by ELTE researchers makes it likely that dogs may also be able to recognize these complex speech patterns.
Speech processing requires very complex mathematical operations. To extract words from the flow of speech, it is not enough to note how many times each syllable occurs together. We also need to recognize the probability of two clips appearing together. The brains of children who are learning language take these possibilities into account when breaking down continuous speech into words
Mariana Borros, one of the first authors of the article pointed out.
We first measured the brain activity of dogs in the family by means of an electroencephalogram. Interestingly, the dogs’ brainwaves differed when they heard the common words, i.e. the connections of frequent syllables and rare words. We were even more surprised that the brain waves were different even when we compared the always occurring clips with those that only appear occasionally, but at a high frequency. So it seems that dogs take into account not only how often a syllable’s connection appears, but also how likely the syllables in it are to follow each other. Infants also use this last, more complex arithmetic to break down continuous text into words, and this is the first time it has been shown that other mammalian species are capable of doing this.
– says Lilla Magyari, the other first author of the article and the pioneer of the methodological foundations for EEG scans in stimulated and non-invasive dogs.
We’re just beginning to understand that some of the computational and neural processes required to learn a language are not unique to humans, said Attila Andiks, head of the research group, adding that we still don’t know why they evolved similar brain mechanisms.
