It depends: Why dog training “tips” often fail
By Kiki Yablon, KPA CTP, CPDT-KA
When I began apprenticing as a dog trainer, I had a million questions for my new mentor Laura Monaco Torelli, and most of them started with, “What do I do when …”
I imagined a professional dog trainer would just have a mental catalog of predetermined responses for various situations: When the dog does this, you should do that. If he does that, you do this.
But instead, the answer to almost all of my questions, to my frustration, started with some variation on “Well, it depends.”
It turns out this answer is so common in dog training that it’s sort of an inside joke—at one seminar I attended, a presenter joked that she should just have a T-shirt made.
It would be a couple more years before I really understood this answer, and I’m not sure I truly, deeply grokked it until my interest in dog behavior led me into the broader field of behavior science, called applied behavior analysis.
That’s where, finally, I learned that what we think of as a single behavior—let’s use puppy biting as an example today—can actually be many different behaviors, and that each one must be viewed in context of its specific antecedents and consequences.
Antecedents are signals in the environment that tell the animal a given behavior is likely to work now. Consequences are outcomes that make the behavior more or less likely the next time. Behaviors that produce desired consequences will be repeated, and antecedents that predict desired consequences for a given behavior will become cues. To change behavior, we have to work with these environmental conditions.
This is why dog training “tips” and even whole “methods” sometimes fail. General advice, by nature, cannot take into account the specific context in which a behavior occurs.
Yeah, yeah, you say. I get it in theory—but my puppy is biting my hand right now, and it hurts like a mother. I have to do something. What should I do?
The good news is that with some practice, assessing why a behavior might be occurring and figuring out what to change doesn’t have to take long.
First, identify the behavior—just one, and just the facts, no interpretation or implied motive. Then add what typically happens immediately before and just after, in similarly objective terms. For example:
Antecedent: I move my hand toward the puppy’s collar.
Behavior: The puppy puts her teeth on my arm.
Consequence: I pull my hand away.
Now we can make a hypothesis about whether, given these conditions, the behavior will increase or decrease. It can be helpful to rephrase the above terms as “when,” “if,” and “then”:
When I reach toward the puppy’s collar
If the puppy puts her teeth on my arm.
Then I pull my hand away.
Prediction: If nothing changes, when I reach for the collar, the puppy will continue to put her teeth on my arm in order to make my hand go away.
We don’t know why this hypothetical puppy wants my hand to go away. We don’t know if she’s “angry” (a label used to describe a group of behaviors that can look a lot like the ones we call “scared”). It’s misguided to speculate whether she is “dominant.” We could dutifully gather data about what outcomes a person (or type of person) reaching for her collar has predicted for her in the past, but we don’t even really need that. With just the facts above, we can make a plan and test our hypothesis.
One approach would be to teach the puppy that as I reach toward her collar, she can do a different behavior that produces a consequence she likes:
When I reach toward the puppy’s collar
If she looks forward
Then I give a treat
Prediction: The puppy will increasingly look forward more when I reach toward her collar.
What’s more, the puppy’s motivation to make my hand go away in the first place is probably going to get weaker. After all, my reach now predicts a treat.
There are lots of variables in this simple-sounding process—for instance, if the puppy is extremely uncomfortable being reached for, we might have to break it into small steps, only reaching partway toward the collar at first. The treats need to be something she really likes, not just something that came in a bag labeled “Treats.” For one puppy, I might be able to just catch her looking forward before she turns toward my hand. For another, I might have to prompt the new behavior more explicitly to start—say, by giving the treat right in front of her nose as I reach in until she starts to look for it.
Over time, the puppy can learn that me reaching for the collar predicts me clipping the leash on, which predicts a walk. At this stage, if the puppy finds going for a walk reinforcing, treats can be faded out.
Now let’s look at another puppy biting scenario, where the behavior looks the same (and hurts just as much), but the function is different:
When I am working on my computer
If the puppy puts her teeth on my arm
Then I reach for a bully stick
Prediction: The puppy will put her teeth on my arm again when I am working on my computer.
The function of this behavior is different from the function in the previous example—and thus it best assessed as a different behavior.
Note that the first time your puppy chews on your arm, she probably isn’t requesting a bully stick. But behaviors that start with one function can easily acquire another. Sitting is not the behavior dogs naturally do to get food. Yet most pet dogs, even if you’re barely making an effort to train them, will somehow learn to sit to get a treat out of your hand.
In this case, your solution should acknowledge that a puppy needs lots of legal stuff to do with her mouth.
Redirecting to a bully stick is not actually a bad way to respond—slowly take your arm away, wait for just a moment of behavior that you like better, and then go get the bully stick. ”
Another, better option would be to give her the bully stick as soon as you sit down at the computer, before she has to “ask.”
Here you might also want to manage the “When” part of the equation—if the cue doesn’t happen, the behavior doesn’t either. And if the behavior doesn’t happen, it can’t earn reinforcement and get stronger.
So you could, for instance, confine the puppy to an exercise pen near you during computer time, with plenty of options for chewing and play but no access to your arm. Think of the pen as training wheels: she can ride the bike, but she can’t fall over. When she’s developed good riding habits, you can take the training wheels off.
You could also teach your puppy:
When I am working on my computer
If you sit next to me *
Then I will give you a bully stick
* This is where you can really get mired in “tips.” It doesn’t have to be “sit.” Sitting might be hard for your individual dog, or uncomfortable on the surface near where you tend to use your computer. Something as simple as putting a small rug might help. Standing, lying down, or doing an adorable head cock might be acceptable or even preferable ways of making the same polite request. (For more on choosing alternative behaviors, see my previous columns When Sit Doesn’t Happen and Training With the Grain.)
There are as many behaviors as there are reasons to behave. Figure out what need your dog is expressing, and teach her the easiest way to get it met.