There has been a bit of a tempest brewing in the small southern city where I train dogs. There’s quite an active dog community here, and it seems that my employer made a statement several years ago that keeps coming back to haunt us. The statement, which follows, is paraphrased, only because I am not the one who heard him, though he does not deny saying it. “There are four quadrants to operant conditioning for a reason.” I would have thought that this was a reasonable statement—after all, for every black there is a white, every yin a yang, and so forth. Instead, it appeared to be the shot that was heard around the world (or, at least, our little world). Now, I believe I know this man reasonably well—as well as one can know one’s employer, and do not consider him to be a bad person. I have watched him interact with dogs, both his own and countless others. I have had long rambling discussions with him on the topic of dog training (along with many other things—the depth and breadth of his knowledge on a wide variety of topics is nothing short of impressive). I have personally never seen him bully a dog, nor do I think he would. If anything, I think his biggest failing (assuming this IS a failing, and I am not remotely convinced that it is) is that he is extremely intelligent, well educated, and tends to be very literal in his communications. He says exactly what he means and he doesn’t dumb it down to appeal to the least common denominator. To converse with this man, you have to be able to keep up, because he’s not going to slow down to wait for you!
Let’s talk specifics here. There ARE, in fact, four quadrants to operant conditioning—positive reinforcement, positive punishment, negative reinforcement, and negative punishment. And, technically speaking, all four are effective methods in training. The ones we choose to use depend mainly on our own moral compass and the situation at hand. I personally use 99% positive reinforcement, a smattering of negative punishment, and, only in the rarest of occasions, positive punishment. I do not use (or, at least, I can’t think of an incident where I might have used) negative reinforcement. I have a fantastic working partnership with my dogs. But am I am all positive trainer? Hell no! And here’s why.
In the pure Skinnerian terms (and these are the terms we are using here amongst professionals), positive means only one thing. To ADD something. That’s right, just that. It means to add, or introduce, something to the training equation. Thus, if I am ADDING a treat, that’s positive! But guess what? If I ADD a sharp kick to the ribs, that’s still positive. In pure scientific terms, positive does not mean good or bad. It means positive, nothing more. And negative? It means to SUBTRACT. Are you following me here? Right, so, we move on to the second step in our equation—punishment versus reinforcement. Again, I am going to simplify this down. Reinforcement is used to INCREASE the frequency of a behavior. Punishment is used to DECREASE the frequency of a behavior. That’s it! Punishment does not equate to pain or fear. It is simply something that the dog does not enjoy and, thus, the dog will be less likely to continue that behavior. Reinforcement is something that the dog does enjoy, and hence, will be more likely to increase that behavior. Still with me? So this is what we’re looking at:
Positive Reinforcement = Adding something to increase the frequency of a behavior
Positive Punishment = Adding something to decrease the frequency of a behavior
Negative Reinforcement = Removing something to increase the frequency of a behavior
Negative Punishment = Removing something to decrease the frequency of a behavior
Yes, all work. Look, I know most of us do not like positive punishment and negative reinforcement, but the ugly truth is, they DO work. And, at times, they are very very effective. I’m not advocating their use; I’m just stating a fact. If I put an e-collar on Cherry and gave her a good zap every time I cued her to do something and she refused, I would be willing to bet that I would get some unbelievable response times the next time I cued her. But would she really want to work for me? Or would she just be avoiding the zap? Would she continue to work for me if I took the collar off? Maybe. But would she continue to do so once she realized that the threat of being zapped for noncompliance was gone? She’s a terrier, and a pretty independent one at that, so I’m going to say no. That means that I am going to have to have that collar on her pretty much every time we practice and only take it off for competition, thus keeping the idea of the punishing stimulus fresh in her mind. She probably wouldn’t enjoy working with me and would learn to dread training sessions. By the way, the above example would be considered positive punishment (yes, there’s that “positive” word again.).
What about teaching Cherry a retrieve? She’s not a breed for which retrieving comes naturally, so, following the older school of training, I could just pinch her ear (just enough to at least cause her some discomfort), and the instant she opens her mouth, I would pop that dummy in, and reward her by releasing pressure on her ear. (This would be an example of negative reinforcement, btw, as I would be subtracting something–in this case, discomfort or pain–in order to increase the frequency of the desired behavior—opening her mouth to accept the dummy. Another example would be the nagging buzzer in your car that tries to remind you to put your seatbelt on, turn off your headlights, or to remove your keys from the ignition after you’ve parked.) I have no doubt that the ear-pinch method would work—countless trainers have used it with successful results. Personally, I’m not going to use that method. Instead, I have been methodically back-chaining the retrieve, and have successfully gotten her to the point where she will routinely bring me the dummy because she knows that, as often as not, there’s a pretty decent reward in it for her. It’s worth her effort to bring it back to me. She has also figured out that bringing it directly back to me with a nice straight sit in front and not dropping it directly onto my toes generally nets her a higher value reward. She’s not 100% on that last part though, as my poor battered toes can attest to. Nonetheless, the retrieve is worth her while and is rapidly becoming one of her favorite games. Positive reinforcement for the win!
What about negative punishment? That one sounds like the scariest of the lot, doesn’t it? However, if we just break it back down, it simply means that we would be subtracting something to decrease the frequency of a behavior. In our house, we use this quite a bit at feeding time. We have about a dozen dogs here, all varying from one year to ten years of age. If we allowed it, feeding time would be an exercise in chaos. Instead, once all of the dogs are in their respective crates (yes, every one of our dogs eats dinner in his/her crate), they wait quietly for dinner. Only the youngest or the occasional guest dog barks and screams for his dinner. The rest just quietly wait. Why? It’s simple—we wait until a particular dog stops barking and carrying on before we give him his dinner. In other words, the dog that is barking and pitching a fit gets ignored. The one that is patiently waiting gets a delicious bowl of fancy holistic dog food with some kind of gravy, some supplements, and a cookie on top. The one that is screaming has to wait until he stops to get his bowl. This, my friends, is negative punishment. I am taking something away (in this case, immediate gratification in the form of dinner or attention) to decrease the frequency of an undesired behavior (caterwauling like a banshee at the injustice of having to wait an extra 5 seconds for his dinner). You can hear a pin drop in the crate room during feeding time.
I will also throw this out there: I firmly believe that any trainer that tells you that they are a 100% positive reinforcement only trainer is flat lying to you—either willfully, or simply because they have never truly looked at themselves in the mirror. We are all human here, and we all do things without thinking about it. Anyone that has ever yanked a misbehaving dog out of a busy street by their leash and collar to avoid them being hit by a speeding bus has used positive punishment. Anyone that has ever gotten frustrated and yelled at their dog in exasperation has also just used a positive punishment. The other day I had a beautiful corn muffin sitting on a plate on the coffee table, and one of my dogs decided to try to snatch it. It was the last of the corn muffins, and I had lovingly heated it up, slathered it with butter and a layer of my watermelon jelly, and there was my dog going in for the steal. I didn’t even fully register what was happening before the sharp “NO!!! MINE!!!!” left my mouth. My poor dog dropped the muffin on the plate and instantly backed off looking very shocked. (You’d best believe I still ate that muffin too, but that’s a whole different blog post!) Dr. Patricia McConnell takes on this very topic on her blog with an entry entitled, “Positives of Negatives & Negatives of Positives”. In that post, she outlines her occasional use of the less popular quadrants. And if someone as well respected as she is can admit that she is human and that there is a time and a place for everything, then perhaps us mere mortals can stand to cut each other some slack!
Am I a “positive” (as in the colloquial sense) trainer? Yes, I am. I do not bully dogs. I always seek a training methodology that invites my dogs to work for me because we both enjoy it. Nothing gives me more pleasure than that moment when my dogs suddenly get it. I am fascinated and thrilled when I watch them problem solve and offer behaviors to try to get the correct answer. Sometimes there isn’t even a correct answer, and we will just be engaging in a dialogue (coming up with new tricks or playing the famous box game). I wouldn’t get to enjoy any of that if I simply demanded that my dog comply and punished him or her for not doing so. But would I ever use positive punishment? Yes. Have I? Yes. Would I do it again? Yes. Why? Because there are 4 quadrants to operant conditioning for a reason, and as a trainer I am going to pick the best one for the task at hand, and my preference leans heavily towards positive reinforcement/negative punishment. I will never use a positive punishment to perfect a heel, proof a stay, or make a sit or down snappier. There are so many more dog friendly ways to teach those behaviors that I just cannot see the value in the prong collar or e-collar approach. But I apparently will enthusiastically use positive punishment (in that case, yelling sharply) to defend a corn muffin (or, more realistically, much as one might slap a child’s hand away from a hot stove, in a situation where the immediacy of the action requires it, such as car chasing). Being forced to use that methodology, in my opinion, is not only a short term fix, but is also a great big warning flag that long term training needs to be revisited (perhaps proofing the recall, etc). In the case of the corn muffin, perhaps I could spend a little time on the “leave it” command! From a purely scientific point of view, I am not an “all positive” trainer, but I am a dog-friendly trainer, and I hope that this article has made that distinction clear. I will never intentionally hurt a dog, and I will always encourage that dog to be an intelligent and creative partner in our training journey.
In dog training, much like every other ethical decision in life, I remind myself of the writings of Niccolo Machiavelli and ask myself this: does the end truly justify the means? It’s certainly worth considering.