Ever wondered why your dog only seems to recognize his own name half the time, but will come tearing from a different floor of the house any time you rustle a resealable bag?
Both the dog’s name and the rustle of the treat bag are cues. One you may have taught on purpose—though the dog may not have learned what you thought you were teaching. The other you probably taught by accident—but you did a better job than you know.
Some definitions:
- A cue is a signal in the environment that tells the animal reinforcement is available for a certain behavior or sequence of behaviors now. “Commands” from you can be cues, but all cues are not commands. In fact, the great majority of your dog's cues come from the environment, not you. A dry mouth might cue the dog to walk to the toilet. A closed lid on the toilet might cue the dog to walk to the water bowl. An empty water bowl might cue the dog to bang the bowl with a paw.
- A reinforcer is anything that, as a consequence of behavior, increases the future probability of that behavior.
What most people want their dog to do when they call his name is for the dog to look at or come to them. So ideally, for the dog, the name should predict: “If you look/come here, you’ll get something or get to do something you really like.”
But how often is that true? What percentage of the time does your dog’s name predict a treat, play, access to the outdoors, petting of a sort your dog genuinely loves, or other great reinforcers?
What percentage of the time does your dog’s name predict the end of play, the end of digging, losing sight of a squirrel, being grabbed abruptly by the collar, kennel time, departure time, a nail trim, a squirt of ear cleaner, petting of a sort your dog doesn't like, or just nothing in particular?
Estimating these ratios will give you a pretty good picture of the current reliability of your dog’s response to his name. And you can apply this same math to figure out how reliable any other cue is. The "math" is a crude version of the Matching Law, which essentially says animals exhibit behaviors in proportion to how much reinforcement has been available for those behaviors in the past.
Now: About what percentage of the time would you say the aforementioned rustle of a bag predicts the dog will get food?
How often does it predict anything other than food? (If your answer is anything less than 99%, BTW, you're a sadist.)
So what can you do to increase the chances that your dog will respond to a his name the same way he does to the rustle?
Make sure your cues predict an outcome that your dog will really look forward to. And make sure they don't start to mostly predict outcomes your dog wants to avoid. This is a common problem with recall training: owners teach it briefly with great stuff, but then start to practice only when they need it. This means that every time the dog responds, he will be giving up something he wants. People also start to get stingy with the reinforcers because they think "he knows what it means," but what he is learning is that this cue means the cost is not worth it.
More broadly the answer is to ensure your cues predict enough reinforcement to make the behavior "worthwhile" under a variety of conditions—be that reinforcement cheese or avoidance of an electric shock. But ethical trainers try very hard to teach dogs using stuff they will work to get, and using stuff that evokes joy has the additional benefit of creating joyful associations with both behaviors and the cues that signal them. Does your dog just mechanically walk over when you rustle the bag? Or does he bound over with soft eyes and a broad tail wag and his mouth lolling open, maybe a drop of drool on his tongue? When you give your recall cue, you want that look to come over his body as soon as he hears it.
When you establish a strong predictive relationship between the cue and the outcome, behavior often evolves naturally as a by-product of the association. If you teach a dog what reliably signals that you are going to produce something he loves, and then you give that signal when you're not standing right next to him, he will figure out what behavior to do to get where the good stuff is. He already knows how to run. (See also: The Dog Already Knows How.)
Already have a problem? It's often best to simply jettison cues that have poor predictive value, that have been weakened by overuse or poisoned. Replace them with new ones that are more reliable predictors of reinforcement.
You probably don’t want to change your dog’s name—I don't either, though frankly I'm human, and so it's my least effective way to get her attention. But do be aware of, and try to bolster, its relationship to reinforcement, and then maybe teach another cue, a fond nickname or a "secret password" that you’ll use when you need a more reliable response.
Even then, though, there are still obstacles to reliability. Let’s take one of our own strongest behaviors: stepping on the gas at the sight of a green light. For experienced drivers, it feels like the most "automatic" of responses. But it's still a choice, and there are still times when we will choose to do otherwise.
First, you have to be able to perceive the cue—which you might not if you’re texting (stop texting!), squinting into the sun, or trying to retrieve the Cheez-It you dropped into your crotch.
What’s more, there may be competing cues in the environment—say, an adorable dog on the corner, or a witless pedestrian in the middle of the crosswalk.
When you call your dog’s name, and he doesn’t respond, it could be that he isn’t sure what it predicts, or there are competing cues in the environment, or both. Hey, Dad’s calling my name over there, which could mean a biscuit, a bath, or nothing at all, but the strong scent of rabbit poop right under my nose is heralding the taste of rabbit poop, so . . . no contest.
Such hurdles can be largely overcome by anticipating and building tolerance for various competing cues into your training. If your dog can respond enthusiastically to his name at home, master it in the yard. If you’ve mastered it in the back yard, work the front porch. Take it on the road, and move it on down the street. Work it far away from rabbit poop, and then work incrementally closer.
Remember too that there are often multiple cues working together to evoke a given behavior—not just your word. Example: Ella's dad started teaching recall in their local meadow from an early age. He paired the cue, "come," with frozen liverwurst chunks, crisped salmon skin, and other delights, and frequently after she came, she was turned back out to play with the other dogs who met up in the field every morning. By the time she was in adolescence, she hd a stellar response to the cue in the field. But at home in the back yard, not so much. Why? When she came to the same cue there, she was shut inside and given a biscuit or some kibble. She likes both of those things, but not as much as she likes being outside.
To remedy this, we taught her a new cue, "inside." With Ella just outside the back door to start, we yelled the word, then scattered freeze-dried turkey hearts around the mudroom. When she was done eating, most of the time we sent her back outside. Then we gradually increased the distance she needed to travel. Here's the result:
Finally, the point of training a recall cue is to be able to call your dog off a potentially life-threatening distraction. But during training, train—don't test (a mantra I came across while reading an article about gundog trainer Mike Stewart a couple years ago and immediately stole). Set the dog up to earn reinforcement over and over again, because (remember our definitions?) it's what strengthens future behavior. Don’t bother giving the cue when you won’t be able to make it pay off for the dog; don't turn your cue into another signal that means "maybe she will, maybe she won't." And reinforce most generously when looking at you costs the dog access to something else really great.
This post was originally written for One Tail at a Time. It has since been revised and updated multiple times.
