A response to a previous post accused me of being “painfully obvious in your contempt for any form of training outside of rewards or ( +R ) only” and ill-informed and biased to boot. I feel compelled to respond.
Anyone who thinks that any training (or life) can be wholly positive is wholly mistaken. Life delivers consequences, sometimes for actions taken intentionally and sometimes simply by chance. Some of the consequences are pleasant, some are not. There is no way to ensure that my life, my dog’s life, or the life of any dog I am training will be aversive-free. And, I believe that we all learn from the consequences of our actions, whether those consequences are pleasant or not. So, yes, I do believe that we, and our dogs, can learn from aversives. If we get burned, we are more careful around a hot stove. That said, I believe that, as a teaching tool, using aversives is unnecessary and cruel — and a lot less effective than other methods.
Behaviorism identifies four “paths” of learning through consequences, or four types of behavior-consequence pairs. What my critic referred to as “+R” is positive reinforcement, rewarding a behavior in hopes of encouraging the dog (or person or goldfish …) to repeat it. This is the foundation of motivational training. But some of the other three behavior-consequence pairs are useful too. Few trainers use only one; I would go so far as to argue that it is impossible.
What are the other three?
Next up is negative reinforcement.
Since this is often confused with another type, negative punishment, I’ll offer an easy way to sort out the four types. Dr. Pamela Reid, an outstanding trainer and behaviorist, and one of my teachers, explains it this way: “Positive” means you add something; “negative” means you take something away. “Reinforcement” makes a behavior more likely to be repeated; “punishment” makes it less likely.
Thus negative reinforcement (like positive reinforcement) is likely to cause the dog to offer a behavior again, but instead of giving the dog a reward (positive reinforcement), you are taking away something that the dog doesn’t like. Our dogs use this on us all the time; they get us to play with or feed them by bugging us until we do. By performing the desired behavior (feeding) we end the negative (heads jostling our hands when we’re trying to type and sad-eyed faces resting in our laps).
So what is negative punishment? Well, it’s punishment, so it makes a behavior less likely to recur. And it’s negative, so it means taking something away. Here’s an example: when our shepherd gets mouthy and rough during play with a ball or Frisbee, I warn him once not to touch me. The next time his mouth touches my hand, the game ends. Ending play early is a punishment to him; I am taking away something he wants. It does make the mouthy behavior less likely to recur; he is always more careful the next several times we play.
The fourth pair is positive punishment. This is adding something that will make behavior less likely to be repeated. Kneeing a dog in the chest for jumping up on you is a common example. This is the one pair I try hard not to use in teaching or training.
I have studied learning theory and behaviorism, and I have applied this knowledge in training hundreds of puppies. I have also studied the effects of punishment. Based on both my research and my experience, I firmly believe that the best way to teach new skills or encourage desired behaviors is through motivation and reinforcing those behaviors. But this is not always practical, and, in some situations, negative punishment can also be useful.
I also believe that anyone who chooses to use positive punishment must think carefully about the consequences and the alternatives. There might be situations in dog training where it is justified, but these are few and only, I think, when a dog’s or a person’s life or safety is in jeopardy. I cannot justify the routine use of positive punishment as a part of my training toolkit.
Motivating and encouraging a dog to do something tells the dog what you want — and rewards him when he gets it right. This is knowledge he can apply in the future and generalize to learning new behaviors and skills. Simply punishing him for getting it wrong tells him what not to do — but for every one behavior we are looking for, there might be hundreds of wrong guesses. Punishing each wrong guess is not going to give the dog information that steers him toward the right answer, but it might shut him down and discourage him from trying.
Kneeing your dog in the chest for jumping up in greeting, the example given above, can be harmful to the dog. It almost certainly harms your relationship with your dog. It might teach the dog not to jump on you in greeting. What it does not do is teach him how you want him to greet you. He might try barking wildly as his next guess. Not an improvement.
Your dog jumps because it is a normal doggy way to greet, because he’s happy to see you, because he wants attention. But, since you are the human and, at least in theory, you are in charge, you can decide that you’d rather he bring you a toy or sit next to you or walk around you in a circle when you get home. You can use positive reinforcement to teach the dog to do that, rewarding him, let’s say, with treats and attention each time he sits quietly to greet you. You can add in some negative punishment by ignoring him when he jumps. Soon enough, the dog will learn that to get your attention, he needs to sit next to you, not jump. You’ve not only reduced or eliminated the bad behavior, you’ve replaced it with a good one, all without hurting your dog. In the process, you have probably had fun with your dog, and you have improved communication with him and probably strengthened your bond as well — not a bad return on your investment of a little of your time and a handful of dog biscuits.