It’s good news for dogs and people that more folks are realizing that science-based training using primarily rewards, rather than punishment and force, is not only fairer and kinder to dogs but works.
That said,common misunderstandings linger about “positive reinforcement” training. The first is “He only does it if I have food.”
That’s sometimes true for an individual dog: He will sit or come when called, but only if you “show him the money” first. Is it because he is stubborn or is it because using positive reinforcement means you will have to carry cookies forever?
The answer to both questions is no.
Dogs that perform what is asked only when they are shown the food do it for one reason: They have been taught to do that. It’s not a conscious choice on the part of their human. Like many behaviors, this one is taught accidentally.
Here’s how. Often we begin teaching a behavior by using food as a “lure” along with a hand signal. With most dogs that lure should be used only a couple of times before we can employ just the hand signal, without the lure, as the dog begins to recognize what it means.
At that point we must take the lure/reward and put it out of sight. In our other hand and behind our back will do. That means the dog cannot see the food. Doing that helps her begin to trust there will still be a reward even if she cannot see it.
Of course, once dogs perform the behavior we still reward them, because we want to build their faith and because their behavior is not yet well established. The reward can be out of sight anywhere — in a pocket, held behind our backs, on a nearby table.
Marking the moment we see the correct behavior with the word “yes” or a clicker means we have a few moments to present the reward. That gives us a little more wiggle room time-wise when it comes to producing the reward. But having it out of sight is key.
Unfortunately, when we skip that stage, as some people inadvertently do, dogs will come to learn exactly what you have taught them: that when they see the food they will be rewarded. When they don’t they will not. So they learn that executing a behavior is rewarded only when they see they food.
Instead of using food as a lure to begin teaching a behavior, it has been used as a bribe to perform a behavior each and every time. Dogs are clever and observant and will quickly learn what we have taught. And voila! You now have dogs that do what is expected only when they see the food. Oops. (This can also be the case with wearing a treat bag, so putting rewards directly into pockets is recommended instead.)
And, no, you don’t have to carry cookies forever. Generally, it’s a good idea to reward a new behavior the dog is just learning a lot — every time or close to it. Once a behavior is reliable in a certain setting you can start rewarding intermittently — only for the best responses, like the fastest recalls or the sharpest sits.
Remember, dogs don’t generalize, so there will be some places where your dog is good at some behaviors, like sitting in the kitchen, and other settings where the same behavior is still not solid — maybe sitting at a crosswalk downtown. It’s a good idea, by the way, to gauge the value of the reward to the difficulty of the behavior.
Habits are neural pathways formed in the brain through repetition. Eventually certain rewarded behaviors will become habits, and the tasty tidbits can go away.
However, we have to work to be sure those behaviors become reliable in multiple settings and with increasing levels of distraction before weaning off rewards in those settings. You will then reward rehearsals of these skills occasionally with life rewards: going for walks, car rides, access to other dogs.
When is a dog’s skill reliable enough to start reducing rewards? Ah. Dog training is both art and science, and herein lies the art of it. A good baseline to determine a dog’s skill is ready for the next stage is an 80% or so success rate in a certain setting. The problem is that many people stop rewarding too soon, whereas others reward for far too long. Both are errors.
It’s up to us to understand how dogs learn and how to apply that correctly to teach our canine companions. The method of positive reinforcement training is not to blame: Both of these common pitfalls are simply due to nothing more than misuse of it.
Because dogs are such clever creatures you can undo any mistakes you’ve made using the learning theory on which positive reinforcement is based — the right way.