(NaturalNews) Researchers have developed a computer software that can identify why a dog is barking, according to a study published in the journal
Animal Cognition.
A research team from Eötvös Loránd University in Budapest, Hungary, recorded more than 6,000 barks from 14 different Hungarian sheepdogs in six different situations: alone, ball, fight, play, stranger and walk. Those barks were then fed into a software developed by researchers from the Ecole Polytechnique Fédérale de Lausanne in Switzerland.
On average, the software correctly identified the dog's situation from its bark 43 percent of the time, better than the average human performance of 40 percent. It was able to identify which dog was barking 52 percent of the time, also a higher success rate than the human one.
The software's accuracy varied widely, however, depending on the situation the dog was in. The software did best at identifying the "stranger" bark, with a success rate of 63 percent. It was only able to correctly identify the "play" bark 6 percent of the time. The software's ability to identify individuals also depended on the type of bark, ranging from a high of 60 percent for a "play" bark to a low of 30 percent for a "stranger" bark.
According to lead researcher Csaba Molnár, the findings suggest that there is less variation between the way dogs bark at strangers than the way they bark while playing. He suggested that this could be due to the fact that historically, dogs were bred for their guard ability, which necessitated a guard bark that is easy to identify by humans. There was no need to breed dogs for consistent play behavior, however.
Juliane Kaminski of the Max Planck Institute suggested an alternate interpretation - the play barks might have evolved over time in response to the different reactions that they elicit from humans.
"In fight situations, the dog barks aggressively and the
human runs away, end of story," she said, which would explain the lack of variation in those barks.
Have comments on this article? Post them here:
people have commented on this article.
Email this article to a friend
Now Available from 
