Hands Off My Bone!


This post was chosen as an Editor's Selection for ResearchBlogging.orgResearchBlogging.orgDifferent dog growls mean different things, right? Probably. But can you tell the difference?

Here's a dog growling when someone is trying to steal its food: link to mp3.
Here's a dog growling at a stranger: link to mp3.

You probably can't tell the difference. Neither could I. Dogs can. Surprised? Probably not. But, this is the first experimental evidence that dogs use different communicative vocalizations during social encounters with other dogs and with humans, depending on the situation.

The researchers recorded dog growls in three contexts: food guarding from another unfamiliar dog (Figure 2, panel A), playing tug of war with an unfamiliar human (Figure 2, panel B), and being approached in a threatening manner by an unfamiliar human (Figure 2, panel C). At least 10 growls were recorded in each recording session. A total of 36 dogs were recorded growling, and were controlled for age, weight, height, breed, and sex.

i-b1a5dafdb2b1974dd0067fa061dfe182-recording dogs.jpgFigure 2: Recording sessions with the dogs.

Then, they tested what other dogs would do when played the recordings of dogs growling in different situations. It went like this:
(1) Owner enters room with dog on leash. There is nobody else in the room; the experimenters are watching on closed-circuit television monitors.
(2) The dog is led around the room by the owner, to get acquainted and to discover a hidden bone in the room.
(3) The dog was allowed to sniff at the bone from a distance of 20cm, but not allowed to touch it.
(4) After allowing the dog to sniff the bone, the owner would lead his/her dog 3 meters away from the bone and stay there.
(5) Upon hearing a signal from the experimenter, the owner released the dog from the leash. The owner was instructed not to talk, touch, or look at the dog.
(6) If the dog approached the bone, then when it was within 5cm (just before it was in reach), the experimenter played back one of the pre-recorded growls. The trial was concluded if the dog left the bone unattended for more than 90 seconds, or started to chew on the bone, or moved the bone to a different location.

When the food guarding growl was played to the dog, 11 of 12 dogs withdrew from the bone within 15 seconds. Compare this to only 2 of 12 withdrawing upon hearing the threatening stranger growl, and 4 of 12 withdrawing upon hearing the play growl. Seven of the dogs who heard a food guarding growl stayed permanently away from the bone, compared with only one each in the threatening stranger and play growl conditions. Even though there wasn't actually another dog present in the room!

bar doggrowl.jpgFigure 3: Why don't these people include graphs in their papers? Lucky for you, I graphed the data.

Dogs avoided the bone when they heard the food guarding growls, but the other two types of growls were not effective deterrents. If the dogs had reacted differently only in the case of the play growl, then the different emotional state in the growl-er (play versus defensive aggression) could explain the differences in behavior in the dogs' behavior, but this was not the case. The dogs clearly differentiated two types of aggressive growls, one guarding food, the other when approached by a threatening stranger. The dogs' responses to the growls was only significant in the appropriate context.

It has been shown that humans are able to categorize dog growls and barks into emotional categories: fearful, aggressive, and playful. But humans haven't been able to successfully categorize more subtle variations in growling, which this research indicates dogs can perceive.

But I find one important problem with this study. Another difference between the food guarding growl recordings, and the other two conditions, is that the food guarding was the only growl recorded that was produced and directed toward another dog. The threatening stranger and social play growls were produced while interacting with humans. Maybe it doesn't make a difference, but it seems like something that's pretty important to investigate.

What I'd like to see is a combination of this study, and the one I wrote about previously, on play behaviors. Do dogs use different growls with each other in various social contexts? Let's record dog growls while playing with other dogs, while demonstrating aggression against other dogs, and during sex. Then let's play those growls to a new set of dogs and see what happens. I think it would be really interesting to know if dogs intentionally use different growls to communicate their inner states (fear, anger, hunger, etc), and if they intentionally deceive other dogs (or humans) by producing a certain type of growl or bark to manipulate the behavior of the other individual, even when the growl doesn't match up with the emotion state of the communicating dog.

Also, do graduate students growl at each other to guard their equipment from invaders? Discuss.

Farago, T., Pongracz, P., Range, F., Viranyi, Z., & Miklosi, A (2010). 'The bone is mine': affective and referential aspects of dog growls. Animal Behaviour, 79, 917-925. DOI: 10.1016/j.anbehav.2010.01.005

Try this dog fans: copy your dog's growl (or any dog's) indicating that you want it to leave you and your food alone. Add a lip snarl; don't look poochie in the eye. After a few tries, your mimicry will be good enough to get your dog to flop to the ground, look away from you, and quit that annoying begging. Imitating your dog's cues is so much easier than getting it to learn English.

Interesting. I could tell the difference between the two growls, but I don't know if I could do the same without the comparison. But I'm definitely going to give the mimicry thing a go.


I've certainly found that feline-aggressive body language and violent hissing deals with aggression from my mother's cat quite nicely.

By stripey_cat (not verified) on 09 Aug 2010 #permalink

Fascinating growl comparison!

I've noticed the difference between a dog barking at strangers, and a dog barking to get in the house. It's actually pretty easy.

The first bark is prolonged, loud, and aggressive-sounding, but the second is usually on or two quick barks followed by a long pause of about 20-30 seconds before the dog repeats herself again with one or two barks, and so on until you let her in the house.

By Charles Sullivan (not verified) on 09 Aug 2010 #permalink

Imitating your dog's cues is so much easier than getting it to learn English.


Some years ago, while I was exploring my yard at night (you can find neat stuff in the dark), I was shocked to find myself the subject of aggression by a neighborhood dog. Barking angrily, he charged at me. I was on my knees and in no position to flee, but on a momentary whim, I barked back.

The dog turned and fled.

I don't know what I said, but it did the trick.

By Michael Suttkus, II (not verified) on 09 Aug 2010 #permalink

I would like to see some real science added to this type of study - record the sounds to be analysed acoustically in a spectrogram. It should be possible to clearly distinguish differences, if any.

By Robert Wright (not verified) on 18 Aug 2010 #permalink

Actually, Robert (@7) this has been done - including in the paper I discussed in this post.

I was more interested not in the question of acoustic analysis, but in the question of whether and how other dogs were able to distinguish different vocalizations of their conspecifics.

I didn't include the acoustic analyses in my post, but the paper is fairly accessible, and I'm happy to send you a PDF if you'd like.

That aside, even the part of the paper that I did review in this post is absolutely "real" science; science is simply the application of the scientific method to a question about the world.

@7 - If the research question was about the sonic properties of dog growls, that would have been an appropriate method to focus on. In fact, the research was about the social function of dog growls elicited in different circumstances, and no sonic analysis of the sounds would be necessary to study this. Even so, the authors provided one, and, as the spectrographic differences between the growls did not actually answer their research question (which was about behavior, not acoustics) they still rightly focused on behavioral experiments. The method of "real" science that you mention is a minor detail in this article, and isn't necessary to the author's case.

If you can look at spectrograms and tell what behavior those sounds will elicit in different species without running behavioral tests, please do it and publish on it. The world of bioacoustics would love to know how it's done. Until then (and even after,) please try to avoid groundlessly belittling behavioral science.