Lots of animals are well aware that bigger means scarier. In stressful or aggressive situations, for example, the hair or fur of chimpanzees, rats, cats, and even humans stands up on end (in humans, given our lack of fur, this results in goose bumps) in an effort to dissuade a potential attack. Elephant seals use a display called "rearing up" to make themselves look bigger - as if they need to look bigger in the first place!
Since some animals tend to be good at looking bigger than they truly are, visual cues may not actually be a reliable method of sizing up another individual. In addition, using vision alone to determine body size is error-prone due to distance and visibility. It would seem prudent, then, that some animals would need an alternative mechanism to use in determining body size. It seems as if domestic dogs have just such a method: they listen.
Sound is also prone to distortion, though. From the right spot, one wolf can sound like an entire pack, thanks to the reflection of echoes off canyon walls. However, previous studies have indicated that sound remains a reliable indicator of body size, and that acoustic signals are fairly resistant to distortion over long distances.
A previous experiment of rhesus monkeys showed that they were able to match an given sound - an aggressive call from another rhesus - to a images of other monkeys based on size. However, a critical flaw in this experiment was that the sounds and images both also varied with age. It is possible, therefore, that the rhesus monkeys were determining the approximate age of the monkey in the auditory playback, and the approximate age of the monkeys in the images, and simply matching them. Whatever the method, though, it seems likely non-human primates are able to match auditory and visual information.
This left Anna M. Taylor and colleagues from the School of Psychology at the University of Sussex with several questions: is this ability limited to primates, or is it more widespread among mammals? And after the use of age as a means of matching sounds and images was made impossible, would mammals still be successful on this sort of task? They knew that domestic dogs use other dogs' growls in their decision-making processes, so dogs would be a good non-primate species to test.
Each dog was brought into the testing room, where two stuffed dogs were at either side of a central loudspeaker. The stuffed dogs - which were actual deceased dogs stuffed by a professional taxidermist - were used to avoid any complications that could arise by using real dogs as models. The large dog was a 60cm tall German shepherd, and the small dog was a 30cm tall Jack Russell terrier. While the dogs were kept on a very loose leash held by the owners, neither the dogs' owners nor the experimenters were in the dogs' field of vision during the experiment. The owners were also naive to the purpose of the experiment, reducing the possibility of providing their dogs with any subtle hints.
Importantly, the recordings that were used were from ten different dogs, of different breeds, ages, and sexes. Each recording was manipulated with a computer so that as many variables as possible would be the same, including duration, frequency, and amplitude. In addition, two versions of each growl were created by manipulating something called the formant dispersion, which is the variable associated with vocal tract length. Vocal tract length is in turn associated with overall body size. The formant dispersion was either set to 1500Hz, which approximates an 11.7cm vocal tract appropriate for a small 6kg dog, or to 850Hz, approximating a 20.6cm vocal tract appropriate for a large 40kg dog.
After allowing the dogs to become used to the environment, the trial would begin. Each recording lasted for 25 seconds, and consisted of two identical growls each from the same condition (i.e. either two "large" growls or two "small" growls), with 60 seconds between trials. The experimenters recorded which model the dog looked at first, which model the dog looked at longest, and how long the dog looked at each model.
The dogs looked more often at the correct model than at the incorrect model, and for longer periods of time. On average, they looked at the correct model 2.1 times per trial, and 1.4 times at the incorrect model, which is a statistically significant difference. They also looked for an average 3.9 seconds longer at the proper model than the wrong model: 7.6 seconds compared with 3.7 seconds.
These results suggest that domestic dogs are able to match size-related information based on sounds with corresponding information in the visual domain, extending this ability beyond primates alone. It is of course unlikely that dogs are only able to distinguish German shepherds from Jack Russell terriers. Since the sound recordings were manipulated to be equivalent on every variable with the exception of formant dispersion, it can be reasonably concluded that the dogs' performance was indeed a judgment of size, and not some other variable like sex, breed, or age.
Taylor and her colleagues speculate that this ability is probably not the result of domestication, but probably emerged among wolves prior to domestication - or earlier. It is also unlikely that this skill is limited to inferring body size, but more widely applicable. The ability to combine information from auditory and visual inputs and compare them is not as simple as it seems. The idea would be that an animal understands that the visual and auditory information are categorically related. It requires taking two different perceptual inputs and comparing them in a more abstract conceptual way. More studies should be conducted to flesh out these possibilities - not only among different species and genera, but also investigating the extent to which animals are able to combine different sorts of sensory inputs in abstract, conceptual ways.
Anna M. Taylor, David Reby, & Karen McComb (2011). Cross Modal Perception of Body Size in Domestic Dogs (Canis familiaris) PLoS ONE, 6 (2) : doi:10.1371/ journal.pone.0017069
Images: Dogs image source
For more on dog-dog interaction:
For more on vocal communication in animals: