Archive OLDT :: When Considering Using Punishment

A positive dog trainer is a trainer who uses positive reinforcement, the reward of good things for good behavior, and negative punishment, the removal of good things for bad behavior. A truly positive trainer would strictly use only positive reinforcement and negative punishment in training. So, though I am primarily a positive trainer, I must admit that I am not strictly so. I do believe in the use of positive punishment, the application of bad things for bad behavior. However, I feel that it is important that the dog (or whatever organism the learner is, humans included) needs to know the right answer before being punished for choosing the wrong one.

As an example, when dogs jump up in greeting, negative trainers might suggest using positive punishment such as catching hold of their front paws and not letting go, or kneeing them in the chest, in order to make jumping up less desirable. Well, for starters, I would never apply the latter technique – using a technique that involves inflicting pain will likely mar the fabric of the relationship I have with this dog, something I find an unacceptable casualty regardless of the end result. But, to return to my point, my approach would be to reinforce the dog for an alternative behavior that I would prefer. One such behavior might be sitting – a sitting dog is indeed offering a very polite greeting. With practice in many situations and with many people, I would train the dog to offer sits rather than jumping up when greeting people. Having equipped the dog with the right answer, only then might I decide that applying a punishment for choosing to jump up might be appropriate.

So, having decided to punish, now I must think long and hard about an appropriate approach. There are some basic principles that I consider when considering punishment:

  • Timing is absolutely critical: the punishment needs to occur within one second of the behavior to be effective. Punishment delivered later than this is inhumane, as it is punishment without cause. The typical example of this is punishing a dog for having peed in the house while you were at work. If you punish him upon returning home, he will assume he’s being punished for the last thing he did, which was most likely greeting you, rather than the potty accident he had hours ago. And you wonder why now he slinks away when you come home from work!
  • Ideally the punishment is impersonal. Punishment that is clearly delivered by the handler can damage the relationship, as I mentioned above. This is the problem with punishments such as using a spray bottle for counter-surfing. Rather than training him not to counter-surf, you’re just training him not to counter-surf when you’re around. It’s far better to use a booby trap, like a motion-activated Sscat air canister.
  • The punishment must be appropriate and sufficient to inhibit behavior. Too often, handlers will deliver either too harsh a correction or too mild a correction. As an example of the latter case: too mild a prong correction for pulling on leash may stop the dog from pulling at first but, as he gains tolerance for the mild correction, will just become a nuisance that he learns to ignore. A sufficient correction must be severe enough that pulling is extinguished within very few repetitions but no more than that.
  • Be aware that, depending on the dog, punishment may have unintended consequences, particularly in shy or fearful dogs. I’ve known dogs to become afraid of the entire kitchen when all the owner wanted was to dissuade the dog from counter-surfing. I’ve counseled owners with dogs who have become afraid of the backyard as an unintended consequence of their using an electronic fence.
So, given all of my warnings, I will come clean… I used positive punishment just last night. I was helping a handler work with a shelter dog who was leaping up and grabbing the leash, playing tug-of-war with it. This dog was putting a very dangerous twist on this fairly annoying behavior, though, as she was leaping high in the air and grabbing the leash as close to the handler’s hands as she could, often catching clothing or body parts in the process. I was afraid someone would get hurt, myself included. I had decided that we needed to use a Gentle Leader head collar to get better control of her head and had sent the handler to get one, when I discovered the perfect punisher! With the handler’s departure, the dog stopped the behavior and looked longingly in the direction of where he had disappeared. The game had lost its luster without her favorite person. What better punisher than for him to abandon her* whenever she played her dangerous game?!

So, upon the handler’s return, we tried a new approach. I asked him to waggle the leash handle a bit above her head. If she leapt to get it, he was to mark the moment with a stern “uh-uh! Too bad!” and turn and leave abruptly. He could return to her if she offered polite behavior. If she didn’t lunge at the leash when it was waggled, then she was rewarded with attention and a treat – don’t forget… I am a positive trainer! Wouldn’t you know that, within minutes, she was walking nicely on leash and not grabbing it in her mouth! It had only taken a couple of repetitions of the handler abandoning her for her to learn this valuable lesson – an example of an appropriate and sufficient punishment.

Punishment is my least favorite training technique, but it has its place. It is something I will absolutely consider using when necessary, particularly for behaviors that are dangerous in nature. But always remember:
Positive methods are the easiest to deliver, the most forgiving of mistakes, and the least likely to cause harm, to the dog and the relationship.

* Some might argue that abandoning a dog is an extreme example of negative punishment – removing a good thing for bad behavior – but I consider the severity of actual abandonment to be far beyond a time-out, which is the removal of one’s attention. This is a finer point that is better left for another time!