AKA: Positive Reinforcement, you’re doing it wrong.
I am a convert to positive reinforcement training. I started out using punitive based methods (though I didn’t really understand what that meant at the time) but came to realize that using pain to correct really wasn’t the answer. I am passionate and opinionated about this, and I defend my position all the time. Most people who share my views are the same way. However, there’s a problem. All the time I see other people go to defend this method, and they do a pretty terrible job of it. Because of this it seems like they don’t fully understand the method themselves, which just serves to give the more “traditional” minded trainers more ammo and more reason to think we’re all a bunch of wishy-washy, bleeding hearts who just don’t have the stomach to be mean to their dogs. And guess what? A lot of the time, they’re right.
This particular rant is inspired by this travesty of a post over at Dogster.com. Basically, it’s poorly written, sophomoric, and an insult to both sides. Even when the author is talking about Positive Reinforcement training (or “Modern Training, as she calls it, which I’ve never heard anyone around me call it and seems like a phrase chosen specifically to be antagonizing to anyone considering more “traditional” methods) she seems to just skim the surface without really understanding it. Two paragraphs in and I’m seeing questionable statements like this:
“Modern trainers believe that the leader is whoever controls access to desired resources.”
Controlling resources doesn’t make you a leader. You use resources to reinforce a behaviour, but leadership means being consistent (yes, in the way you dole out rewards), but also provide clear communication so that your dog can understand what is expected of him or her, and respect any communication your dog offers back. You also make sure that your dog feels safe with you and can trust you implicitly. If all you had to do to be a good leader is give out the best rewards than the highest paid employees would be the happiest in the workforce, yet studies have repeated found this to be untrue. In fact, taken out of context that sentence would better fit a description of dominance theory rather than +R (aka the leader of the pack eats first). The author also makes the claim that traditional trainers don’t use targeting or shaping, which is wrong and a gross generalization. Most of the old style trainers that I’ve dealt with will lure or target a behaviour (at least during the initial teaching phase) and then correct if the dog doesn’t get it. The point is the line between the dominance based training methods and the positive training methods isn’t so clean-cut as some would have you believe.
To be fair that article is pure, unapologetic fluff, which I would hope was something thrown together at the last minute in some attempt to fill Dogster’s daily blog post quota or something. If not then I feel for the author’s dog(s), since she seems to be sadly lacking in some of the fundamentals of the very methods she’s trying to argue for.
There’s also another point I’d like to make, which is a point that is sometimes hard for us who have switched teams to admit. That point is that traditional methods work–for training very specific behaviours. If you teach your dog to sit by luring or shaping them into a sit, or if you use a raised leash or some other manipulation to get there (which, I’d like to point out, can be as mild as simply backing the dog up until they sit), the vast majority of dogs will learn the command. Not only that, but they–on average–will learn it at about the same pace. There’s an article I read a while ago that actually tested this on horses and found the same results. I’m sure anyone who has ever tried to get a horse to load in a trailer when it didn’t want to knows exactly what I’m talking about (even before I understood the difference I always had more success with the techniques that were devoid of any force when it came to convincing a horse to do something it was initially frightened of). In retrospect I think it’s interesting to note that with horses, “conventional” methods suggest using positive reinforcement for things they don’t like (loading on a trailer–a dark, enclosed space that is loud and hot when the vehicle is in motion), and switching to punitive methods when a horse stops doing something they “shouldn’t” find distasteful (refusing a jump–which due to a bad experience or past injury they may well have a reason to dislike). But that’s a discussion for another time.
Getting back to my point–saying that traditional methods (though who is to say what is “traditional,” it’s not like +R training is new) don’t work is a lie. They work about the same for training a behaviour (if effectively applied, i.e. you met out the exact correct amount of punishment for the offense, which is an art large portions of the general public fail to perfect); the difference is in the side effects of each method. The long-term side effects of positive reinforcement methods are a strengthened bond with your pet (or your mount) that are persistent and increase over time if applied periodically. The side effects of punitive based training is a destruction of that bond and that trust, and in creating an animal that is essentially afraid to tell you what it is really thinking (or your misinterpret it), so that when your dog finally hits the breaking point it “happens out of the blue” (example: see Cesar Milan once again doing a superb job of failing at reading basic dog body language. I’d do a take down of this video myself if someone far more competent and experienced hadn’t already done such an awesome job).
My next issue is basically one of a poor choice of label (which I’ve been pointing out all throughout this article as you may have noticed). Calling the type of training I do “positive reinforcement” or just +R is a total misnomer. I remember talking with a woman in the park with a lovely goldendoodle who said she decided to stop using +R techniques because she needed a way to reprimand her dog’s bad behaviour, namely for not returning when off leash (she was using the dreaded e-collar, may the inventor of that particular invention rot in hell). The funny thing is, I reprimand my dog all the time–but I don’t do it without pain, fear, or physical manipulation. If he pulls on the leash we stop moving (a negative stimulus since he can no longer go towards whatever it is he wants to get to); when he stops pulling we move forward. Hence a behaviour is reinforced through something unpleasant. And yes, that does work. He also wears an Easy Walk harness, which turns his body back towards me when he pulls (not just his head like a gentle leader, again, whole other discussion) so rushing towards something doesn’t actually result in a reward.
However, any training regime’s success is dependent on the trainer keeping the criteria extremely strict. I still remember when the whole Monty Roberts craze was going on (thanks, Oprah!) and everyone was haphazardly trying to emulate him without understanding what positive based training really meant (or natural horsemanship in this case, again, another topic all its own. I’m sorry to glaze over so many things but this post is going to be long enough. In fact, if you’re still reading, I’m suitably amazed): anticipating problems through careful observation of your horse’s body language, and finding gentle ways to encourage desired behaviours and discourage undesirable ones. There was one girl in my barn whose horse had a habit of pawing and generally being obnoxious while on the cross-ties. At the time, if it was my horse I would have smacked it the moment I noticed it doing more than shifting its weight, curtailing the problem before the horse got to indulge in the undesired behaviour for too long. She decided that in order to be more “gentle” she would cup her hand and “lightly” smack it again the horse to make a loud popping noise, the idea being that she wasn’t hitting the horse hard enough to hurt it but the sound was startling enough to discourage the behaviour. It wasn’t long before the horse got used to the sound and went right back to being a jerk on the cross ties.
I will always remember something my very first riding instructor said to me, probably in my first or second lesson, “If you’re going to use that whip, I’d better hear it whistle on the way down.” Basically, if your horse has done something bad enough to warrant being hit, then hit them hard enough that they won’t think of doing it twice. Studies have shown that the more successful professional dog trainers that work with service dogs, or police dogs, etc, who use punitive methods keep to a similar philosophy. If you have to keep hitting the animal to make the point, they aren’t learning, and eventually pain thresholds rise and you have to shift to more and more severe methods. BOTH positive and punitive methods require that you are strict with what you expect, and have effective timing in your feedback. Given the same horse shifting around and pawing on the cross ties, I would reward when they stood still, and walk out to sight the moment they started pawing (prolonging the amount of time the horse is standing board in the aisle, and leaving them alone which most horses dislike). And if I was training that goldendoodle with a habit of running off, I would just stay in a fenced area or keep him on a long-line so that he couldn’t ignore me when I called, thus restricting the privilege (aka, reward) of freedom until they’ve earned it through progressive training steps. You will never convince me that the dog will suffer more from not being allowed to run free until it is trained than the pain caused by an e-collar. Also, training without punitive methods takes a bit more imagination, and seems slower because you don’t see immediate results. Another thing, the person who spoils the hell out of their pet without any criteria who then switches to punitive methods may think it’s better, but it is only better because you’ve started to your dog that there are consequences to their bad behaviour. But as I’ve said, those consequences don’t have to involve pain or intimidation. On the flip side to that positive reinforcement training emphasizes rewarding good behaviour, but trying to do it without any punishment will eventually come up against a wall.
Oh, and by the way, random lady with the goldendoodle who still probably has that e-collar on your dog, if you still occasionally have to zap your dog when it ignores your command (a “little” zap is still electrocuting your pet, by the by, the “I’m only electrocuting him slightly” response is the saddest rationalization I’ve ever heard in my life) or you have to leave the collar on so that the dog “knows it’s there” THAN YOU HAVEN’T FINISHED TRAINING THE BEHAVIOUR. A reliable behaviour is one you can give a command for and have a strong expectation that the dog, or horse, or whatever you’re training, will respond whether or not a reward or corrective tool is present.
People who defend punitive methods complain that positive methods only work because we’re “bribing” our dogs, and as soon as the bribe is out of sight the dog won’t listen. These are the same people who leave a choke chain or prong collar on their dog twenty-four seven, or whose horse won’t move off the leg if you’re not holding a crop–you might not need to use any of those tools but god help you if they’re not around. But here’s the funny thing about the way living things learn: intermittently rewarding a behaviour is just as reinforcing as rewarding every time, but the same is not true of punishment. Why? Because we only perform behaviours that are of some perceived benefit. Therefor, any time you don’t punish a behaviour you are indirectly rewarding it. Basically, the carrot is optional, but the stick ain’t. However, if the carrot is always present or easy obtained, you may have to rely on the stick. Best practice is to not let it get to that point.
Look, I could sit here all day and tell you all the reasons to not use punitive training methods or dominance theory, but if I can’t talk for an equal amount of time about the benefits of positive training (can we just start calling this “co-operative theory” or something? It seems more accurate), and how to counter the drawbacks to my preferred method, then I’m kind of useless to you.
I’m a strong believer in the idea that you should understand your training methodology from top to bottom–not just the “what”, but the why. The “why” is way more important than the “what”, since it can led you to the “what” without anyone ever having to tell you. I also don’t think you can convince anyone of anything that you don’t fully understand yourself (unless you have some wicked bull shitting abilities). Educate yourself, and you can share that knowledge with the people around you, otherwise you just start sounding like an unreasoning fanatic, and the only people who listen to fanatics are already fanatics themselves.