Humans Guided Evolution of Dog Barks

It’s a question that tends to arise when a neighborhood mutt sees a cat at 3 a.m., or if you live in an apartment above someone who leaves their small, yapping dog alone all day: Why do dogs bark so much? Perhaps because humans designed them that way. “The direct or indirect human artificial selection […]
Image may contain Animal Dog Mammal Pet Canine and Coyote
In wolves, from which modern dog breeds are descended, barking is done mostly by pups. Tambako the Jaguar/Flickr

It's a question that tends to arise when a neighborhood mutt sees a cat at 3 a.m., or if you live in an apartment above someone who leaves their small, yapping dog alone all day: Why do dogs bark so much?

Perhaps because humans designed them that way.

"The direct or indirect human artificial selection process made the dog bark as we know," said Csaba Molnar, formerly an ethologist at Hungary's Eotvos Lorand University.

Molnar's work was inspired by a simple but intriguing fact: Barking is common in domesticated dogs, but infrequent if not downright absent in their wild counterparts. Wild dogs yip and squeal and whine, but rarely produce the repetitive acoustic percussion that is barking. Many people had made that observation, but Molnar and his colleagues were the first to rigorously investigate it.

Because anatomical differences between wild and domestic dogs don't explain the barking gap, Molnar hypothesized a link to their one great difference: Domesticated dogs have spent the last 50,000 years in human company, being intensively bred to fit our requirements.

Evolution over such a relatively short time is difficult to pin down, but Molnar reasoned that if his hypothesis were correct, two facts would need to be true: Barks should contain information about dogs' internal states or external environment, and humans should be able to interpret them.

To people who know dogs well, this might seem self-evident. But not every intuition is true. As Molnar's research would show, sheepherders -- people understandably certain in their ability to recognize their own own dogs' voices -- actually couldn't distinguish their dogs' barks from others.

Molnar tested his propositions in a series of experiments described in various journal papers between 2005 and 2010. The most high-profile, published in 2008 in the journal Animal Cognition, described using a computer program to classify dog barks (.pdf).

At the time, many journalists -- including this one -- glibly interpreted the study as a halting step towards dog-to-human translation, but its significance was deeper. Molnar's statistical algorithm showed that dog barks displayed common patterns of acoustic structure. In terms of pitch and repetition and harmonics, one dog's alarm bark fundamentally resembled another dog's alarm bark, and so on.

Intriguingly, the algorithm showed the most between-individual variation in barks made by dogs at play. According to Molnar, this is a hint of human pressure at work. People traditionally needed to identify alarm sounds quickly, but sounds of play were relatively unimportant.

By recording barks in various situations -- confronting a stranger, at play, and so on -- and playing them back to humans, Molnar's group then showed that people could reliably identify the context in which barks were made. In short, we understand them.

The findings support Molnar's original hypothesis, though more work is needed. Molnar started to cross-reference a phylogenetic tree of dog breeds with their barking habits, looking for an evolutionary trajectory, but never finished. He had been a student, and his thesis was complete. Unable to get more funding, he's now a science journalist.

According to Eugene Morton, a zoologist and animal communication expert at the National Zoo, Molnar's ideas are quite plausible. Morton noted that barking is a very useful type of sound, simple and capable of carrying over long distances. However, it could have been a side effect of humans favoring other, domestication-friendly traits in the wolves from which modern dogs descended.

"Barks are used by juvenile wolves, by pups. It's neotenic -- something derived from a juvenile stage, and kept in adults. That's probably what we selected for," said Morton. "We don't want dogs who are dominant over us. The bark might go along with that breeding for juvenile behavior. Or it could have come with something else we selected, such as a lack of aggression."

Molnar's research is now a fascinating footnote waiting to be pushed forward by other researchers. In addition to that phylogenetic tree of barking, Molnar would like to see analyses of relationships between breeds' bark characteristics and their traditional roles. If, as with the deep frightening rumble of mastiff guards, breeds' barks tend to fit their jobs, it would further support the notion of human-guided bark evolution.

The ultimate evidence, said Molnar, would be if human knowledge of bark structure could be used to synthesize barks. "If these barks, played to dogs and humans, had the same effects, it would be awesome," he said.

Image: Mr. T in DC/Flickr

See Also: