Then they tested the dogs by placing a treat-holding container on either side of the subject—one box on the right and one on the left. When they gave the “left” cue, the dog got the food reward if she ran to the box on the left. If she ran to the wrong box, she got no reward. Once dogs consistently responded correctly to verbal and visual cues alone, the cues were given together, with a twist. The researchers gave a verbal signal for one direction and a visual signal for the other to see which one the dogs would follow. For anyone whose dog competes seriously in agility, the results were a no-brainer: The dogs consistently followed the visual pointing cue and ignored the verbal cue. This dynamic plays out on every agility course—a dog will usually go where her handler’s body is pointing rather than where the handler might be verbally trying to send her.
This bias toward the visual as opposed to the verbal can pose problems for dogs even in everyday life, says Mills, “This simple example emphasizes that when training dogs, we have to realize that dogs may be reading signals we’re not aware of.” So when your voice tells the dog to do one thing but your body tells her to another, she’s not being stubborn—she may just be reading a different message than the one you think you’re sending.
Even when we’re purposefully sending visual commands to our dogs, such as in the obedience trial ring or field trials or other long-distance work, there’s more to the signal than we might think. Says Mills, “In a similar study, we looked at the dog’s response to different visual right-and-left cues. We compared eye movement and head movement to the right or left with pointing right or left, but keeping the eyes and head looking forward.” Using six dogs, they found that dogs found the hidden food source faster when the two signals were presented together which, Mills says, suggests that “Dogs are taking in the whole picture of what’s going on.” That is, they don’t look at our hands or our head, they look at our entire body. As a result, if all signals are not consistent, dog can become confused.
Do these studies mean we should scrap verbal commands altogether and focus on the visual signals? Obviously, dogs can learn verbal commands, because we use them all the time and some dogs respond correctly on a regular basis. But perhaps even those who respond don’t know the cues as well as we think. Mills and his colleagues performed a series of studies to test this, too. First, they tested slight variations in the commands to see if dogs recognized them as the same words. They taught dogs to stand and stay, and then, from five feet away, the trainer gave either a “come” command or a “sit” command.
Once the dogs were reliable about responding correctly, the researchers changed the command words slightly. In place of “sit,” they used “chit,” “sat” and “sik,” and in place of “come,” they used “tum,” “keem” and “kufe.” The results? In general, dogs did not respond as well to the similar-sounding words; or, taken from another viewpoint, they were able to recognize that the similar-sounding words were not the same as the commands they had learned. This sounds like no big deal, but, says Mills, “From a practical point of view, due to slight differences in how handlers pronounce words, obedient response to one handler’s commands won’t necessarily transfer to another unless the phonemic characteristics are mimicked.”
You might think you could get around this by tape-recording the command and just playing it back, but Mills found that dogs don’t respond to tape-recordings as though they were a real-time human voice. In yet another experiment, a “come” or “sit” command was given in one of four conditions: from a person sitting in a chair; from the same person wearing sunglasses to prevent visual cues; and both conditions, but the command issuing from a tape recorder behind the person. Says Mills, “Dogs made many more errors when the tape recorder was used.”
Such errors could be attributed to the dogs distinguishing a difference between the tape-recorded and live voice command, but another hypothesis is that dogs also rely on lip movement or some other indication that the human is speaking to them. In fact, in a fifth variation, the handler uttered the “come” or “sit” cue while looking away from the dogs, and they again made many errors, indicating that orientation of the handler is important.