When good advice becomes rote advice, sometimes the reason it was good advice in the first place gets lost. Here’s an example:
When your dog is doing something you don’t want him to do, ask him to do something else instead.
In a way, it’s wonderful that this has become rote advice, because it’s at least based in good behavior science. (I’ve written here before about why you want to replace, rather than punish, an unwanted behavior.)
The problem is that when this nugget is dispensed without any additional information, it often doesn’t work. I can’t count the number of times I’ve seen a frustrated dog owner repeating “sit!” over and over as the dog continues to jump or bark or growl.
The name for this type of behavior intervention is differential reinforcement. There are many variations of differential reinforcement, but most of the time, when people ask a dog to sit instead of doing something else, the one we’re talking about is differential reinforcement of incompatible behavior (commonly abbreviated as DRI). “Differential reinforcement” means selectively reinforcing one behavior while removing reinforcement for (or “extinguishing”) another. “Incompatible” means the new behavior can’t be performed at the same time as the old one.
In the human world, there are established guidelines* for how to pick a replacement behavior to use in differential reinforcement. Dogs learn the same way we do, and responsible dog trainers are increasingly applying the same guidelines in their work:
The first step in using differential reinforcement, according to these guidelines, is to determine the circumstances the behavior occurs under and what the dog gets out of the old behavior in those circumstances. Both the antecedents and the consequences are important, because a dog may bark or jump or growl for different reasons in different circumstances. If possible, the new behavior you choose should give the learner a more appropriate way to get what he was after.
So let’s say your dog growls at other dogs who come within 10 feet of him while he’s on leash. The function of growling is typically to make something go away, and indeed, the usual result of your dog’s behavior is that you drag him away or the other dog’s owner crosses the street.
Just because your dog probably doesn’t enjoy being dragged away doesn’t mean the distance this achieves isn’t reinforcing. How do you know if it’s reinforcing or not? Well, reinforcement is anything that, as a consequence of a behavior, increases the future probability of that behavior. So if the behavior keeps happening, being dragged away is probably working for your dog.
Promptly stopping on the spot and asking your dog to sit in no way satisfies the dog’s desire for distance from an oncoming dog. On top of that, sitting is a very low probability behavior in those circumstances. For many dogs, no treat in the world is worth remaining stationary as a perceived threat advances. (To paraphrase Laura VanArendonk Baugh: Please enjoy this candy bar while you are being mugged.)
The measures you may feel compelled to resort to when sit doesn’t happen, such as physically forcing the sit or tightening up dog’s leash to immobilize him, will likely only intensify the emotional reaction that’s behind the growling and may result in dog redirecting his aggression toward the closest target: you.
Jumping up during greetings is another behavior that dog owners often try, and fail, to replace with sit. If you’ve spent years “ignoring” jumping that persists, it’ll make perfect sense to you when I say that reinforcement for this behavior in this circumstance is frequently physical contact with the owner. Sit can achieve this end, but it may first need to be trained to a level of fluency most of us, frankly, aren’t otherwise motivated to achieve. (Dog owners respond to the laws of behavior too.)
Which brings us to another of the guidelines:
The replacement behavior should be something the dog already knows how to do. This seems painfully obvious when you say it out loud—how could a dog do something he doesn’t know how to do? But it’s trickier than it sounds.
I think there are a couple of reasons dog owners so often default to “sit.” First, they think their dog does already know it, and that’s sort of true—dogs are born knowing how to plunk their butts on the ground. But a sit in your kitchen as you’re holding your dog’s dinner bowl is a different behavioral unit—with different antecedents and consequences—than a sit on the sidewalk with another dog approaching, which requires additional, incremental training before you can say the dog “knows” how to do it.
Second, when many people think of dog “behaviors,” they think only of the capital-B behaviors to which we commonly assign a formal cue. But anything a dog can physically do is a behavior.
One small-b behavior that might be a better replacement for growling is simply continuing to walk with the owner in a curve to a distance of 12 feet. Remember, though, that that differential reinforcement has two parts: reinforcing the new behavior and removing reinforcement for the old one. That means the curve-out is a behavior best begun after the other dog has been spotted, but before your dog is already growling.
If you still really want a sit for some reason, it might be easier to get at 12 feet.
With the behavior of jumping on owners or guests, you can also reinforce a simpler behavior that, thanks to gravity and anatomy, all dogs are already doing before they jump: keeping their front paws on or near the floor.
Finally, the new behavior you choose should provide at least as much reinforcement as, and ideally more reinforcement than, the old one. Otherwise the Matching Law will ensure that the dog continues to choose the old behavior.
So if you curve out around that other dog, and your dog walks with you, you’ve provided at least the same reinforcement as you would have by dragging your dog away. But don’t stop there—this would be a good time to bust out those treats. And when your dog has all four paws on the floor, pet more, pet longer, and pet in the way your individual dog enjoys the most, whip out a favored toy, and/or scatter a handful of food on the floor.
* Alberto & Troutman, 2003 (as cited in “Schedules of Reinforcement and Differential Reinforcement for Reducing Problem Behaviors,” S.G Friedman, 2013)