In several areas of my life, coincidentally, people are discussing training and teaching (of dogs, children, grad students, you name it). It’s been very interesting to me and I’ve been thinking a lot about how brains take in and process and then remember information.
There’s a great temptation, in dog training, to make big sweeping statements that are not exactly motivated by love. How many times have you heard that clicker training is “just bribing the dog” or that using a leash correction is “jerk and puke” training? How about “there’s no such thing as alpha” or “dominance downs are abusive” or, from the other camp, “I’ve never seen a truly reliable dog that was trained with food” or “fruity reward trainers are ruining this sport!”
As should be obvious to anyone who has read this blog longer than a day, I don’t exactly like sweeping statements. They drive me nuts, in fact. And these are no different.
One of the first huge mistakes that people make when they think about or talk about dog training is confusing PRACTICE with PRACTITIONER.
For example, there are (a lot of) trainers who do use food to create a dog who can’t think beyond the next cheese flake. Bad practitioner. Does that mean that reward-based training is an invalid process? No, of course not–reward-based training is not only extremely well described scientifically, it’s the predominant method used to elicit some very complex and self-motivated actions on some of the best-trained dogs in the world.
Conversely, there are (a lot of) trainers who use physical corrections in a way that is ineffective and bordering on abusive. Some cross even that line in a horrible way. Bad, bad practitioners. But does that mean that using physical signals to indicate to a dog when he or she is doing something wrong is invalid? I would argue no. Dogs communicate between themselves using physical touch; they understand it, perhaps even more easily than working for food. And wise touch-based training is by far the dominant method used to create incredibly confident and effective dogs in a huge range of disciplines.
It’s common to criticise a method by criticising a practitioner. For example, there’s a study of behavioral problems that has a good dose of science and a HEAVY dose of biased interpretation. The study showed that people who were bringing their dogs to UPenn’s dog behavior center had used a bunch of aversive methods and those methods had not succeeded. Now before you even BEGIN to interpret those results you have to look at the sample population–people who were having such serious issues with their dogs that they had escalated way beyond their trainer, probably beyond multiple trainers, beyond their regular vet, all the way to a teaching university’s behavior center. We have a behavior center here, too, at Tufts, and the dogs that end up there are not just digging in the trash. These are the ones that are going to be euthanized if the behavior center doesn’t fix them. So there’s a huge bias in the sample right away. Second, the questions asked whether these techniques had been tried and whether any of them had prompted the dog to respond “aggressively” (there’s that word I hate again). The only thing I find shocking in that study is the fact that only 25% of the dogs that had had aversives applied tried to bite back–these are the very hardest cases and still over 75% of them had never reacted badly to being confronted.
What the UPenn study says to me is exactly how difficult it is to apply some of these techniques when you are not a very, very savvy trainer. It’s also a picture of owners who have major communication problems with their dogs, who have very little idea how to effectively ask the dog to do anything (or they would not be having such giant issues with their dogs), desperately trying everything they can think of and (almost certainly) doing the aversive exercises just as incorrectly as they had been doing other stuff. The study described a PRACTITIONER problem. It did not establish whether or not the theory or practice is invalid.
It’s also tempting to latch on to a particular philosophy or method because you are fooled by the OTHER practitioner problem–one who is really amazingly good at it. If you go to a Pryor clicker-train seminar or look at Parson’s Click to Calm, you can come away from that feeling like this is the ONLY way ANY HUMAN should EVER train any dog for the REST of EXISTENCE. Or you can watch Cesar Millan and conclude that physical touch and submission are the key to all that is holy. But you need to step back from that and realize a couple of things.
First, the truly brilliant trainers are not just using one method. They THINK they are; they’ll write passionately about how they are. But if you watch a fabulous clicker trainer, he or she is also adept at using body language, energy, timing, and physical cues. They are actually telling the dog to do something with their entire body, and with the energy “aura” that surrounds exciting, naturally dominant people. I don’t mean this in any New Age-y way. Look at what happens when a dominant NFL player stands next to an interviewer. The person holding the microphone naturally leans backward, while the player stands completely relaxed and upright. It’s not the height difference–it’s the energy. If the same NFL player was in a bridge engineering conference, unless he himself was an engineer, he’d probably be leaning back slightly while the 90-pound guy in glasses and a pocket protector stood relaxed and confident. Someone in complete command of a situation, with confidence and expectation of success, can literally push you–or any dog–around just by moving their body. For a dog, that is as much a physical signal as reaching out and yanking a collar is.
Similarly, an incredibly dog-savvy physical trainer, like Cesar is, is NOT just using physical cues. He is (probably just as unconsciously as the clicker trainers are using physical cues) extremely good at timing potent rewards. Dogs ADORE Cesar, and usually want to be in his lap at every second. So when he relaxes, which is what he does when they behave well, they feel all warm and googly inside because they can feel that from him. He controls his body very well, and is always aware of where every part of him is; he is always centered and calm. So both negative and positive attention from him are instantly felt and instantly interpreted by the dog, and positive attention is like the stinkiest cheese-and-salmon pate he could hand them.
Similarly, training methods often work really well even when they have completely different motivations (as described by the humans) behind them. Dogs who display anti-dog behaviors while on leash but are perfectly friendly when off-leash have usually been (inadvertently) trained to lunge and roar by a human who tightens up the leash every time they see another dog coming. The normal clicker training solution to this problem is to focus the dog on you while quickly dispensing treats, so the dog does not go after another dog. The classic physical-correction solution is a quick leash correction every time the dog looks at the other dog. Proponents of both methods will tell you that they are TOTALLY different, couldn’t be more different, that the other method is crazy, etc. But the fact is that both methods get the leash tension off the dog’s neck and so the dog’s drive is radically reduced. The dog goes from being told to attack other dogs to being told that it’s less attractive to attack other dogs than it is to stay focused on the handler. Both methods are reducing drive by ensuring a loose leash; both are demotivating the forward movement. The window dressing is honestly a lot less important than just getting the leash loose.
The second truth you have to realize is that you are probably not a brilliant trainer. Most of us aren’t. So the key is figuring out which method, or combination of methods, allow you to come the closest to being a good trainer, and are the least dangerous in terms of wrecking your dog. For different people that will be VERY different things. You may be a person who is brilliant at popping those treats in but always fail at timing corrections; you may be a person who is terrible at timing treats but knows how to deliver a sane and non-painful physical signal.
I’m always willing to admit that I am a CRAPPY behavior or classic obedience trainer. I don’t hear the angels sing when I get a perfect heel. A crooked front doesn’t look all that different from a straight one, as far as I’m concerned. I don’t have the motivation or the skills to achieve the kind of perfection that others find easy and natural. But I think I’m pretty decent at making dogs happy, fixing neurotic or badly behaved dogs, and keeping a calm and functional pack. So I tend to glom on to methods that are based on using pack dynamics to make dogs feel normal, on methods that are big on consistency, on methods that lower the energy in the room rather than heighten it. An owner who wants to create the world’s fastest flyball dog (which is no less legitimate than my goal of creating a constant environment of calm) might use completely different tools.
If you look at the whole canon of dog training, the books and lectures and gurus tend to basically fall into one of three groups, with a lot of overlap. I want to explore the practice and the good/bad practitioners of each.
This is the granddaddy of training and is still practiced by a huge number of trainers. Basically, the dog is corrected for doing something wrong and is praised for doing something right. Sometimes the praise takes the form of cheese or another food reward. Sometimes it takes the form of nothing at all–in other words, the absence of correction is the reward.
Best-case practitioner: Creates a very secure, confident dog with great consistency of behavior because the dog has clear signals about exactly how it is expected to behave at all times. Excellent physical trainers never hurt dogs; they give aversive cues that are signals but not painful.
Worst-case practitioner: Abuses dogs, causing pain or physical damage in the name of punishing an unwanted behavior.
Critics of this theory say that it’s useless to punish behavior, that the dog becomes joyless and afraid, that the methods don’t work. When presented with a joyful, eager dog who is freaking out with happiness because he gets to go retrieve, and then is told that the trainer uses an ear pinch to proof retrieves, these critics say that the retrieve would be even better (somehow) if no aversives had ever been used. Proponets of the methods argue that dogs who are never given any signal that means “no” often become frustrated and frantic because they don’t know when or where to stop. They point out that self-rewarding behaviors (like digging, barking, chewing, running away) are often so pleasurable to the dog that no amount of reward will overcome their lure.
This is “mother knows best” or “be the pack leader” training. Various pack-language-based signals are given to the dog to indicate that it is lower in status than the handler and therefore should not behave in certain ways. This is usually focused on preventing poor behavior like jumping, mounting, rushing the door, using punishment bites on humans, etc.
Best-case practitioner: Because you’re talking to the dog in a language that is very natural to the dog, this can get amazing results almost immediately. The dog becomes much more secure and behaviors cease on their own once the dog doesn’t feel that it needs to control the actions of the household. This is the method that can look like miracle-working, and with the best practitioners it really does work miracles.
Worst-case practitioner: Gets his face ripped off because he confronts a dog without having control of the situation. This is BY FAR the most difficult method to do correctly. If you don’t “speak dog” you can do FAR more harm than good with these techniques. Difficult to teach–people who do this well are very instinctive in their timing and movements.
Critics of this method tend to say that there’s no such thing as the “alpha dog” and that dominance is a fluid thing in a dog pack. They say that pack ranking cannot be scientifically proven. They’re right… and wrong. Pack dynamics do shift, and the dog who is best suited to a particular task often takes the lead on that task. For example, in our pack Bronte was “alarm giver” and scout, and she was definitely in charge of those pack functions, even though she was nowhere even close to in charge of the whole pack. And it’s correct that dominance is difficult to pin down scientifically; designing an experiment to prove it would be extremely difficult. However, anyone who has watched dogs for any length of time knows that dominance is real, and that status-seeking is something that is extremely important to dogs in a pack. We know that dogs physically punish other dogs for bad behavior, and that even when puppies are weeks old they are sorting out who is in charge in the litter. I always wonder if behaviorists who insist that there’s no validity to dominance theory have ever been around a pack of dogs for any length of time.
So the question is not whether these techniques work… the question is whether your’re enough of a dog to implement them in a way the dog understands.
REWARD-BASED (operant conditioning, luring, postive reinforcement, etc.)
This theory says that dogs, like the rest of us, like to work for a paycheck. If they are given a small reward for a behavior, they will repeat that behavior. If they are given no reward for a behavior, the behavior will gradually extinguish.
Best-case practitioner: Someone who has taught the dog not that sit = treat, but that it is a good thing to experiment with and offer many behaviors because one will be rewarded. This method can produce amazing creativity and self-motivation on the part of the dog. It tends to rev up energy levels, which is why it has come to the fore in an incredible way in agility training. If you present one of these dogs with a jump, he will be eager to go investigate it and will immediately offer the handler multiple behaviors (touching it, standing on it, moving it, straddling it, jumping over it) that the handler can selectively reward and thereby shape the behavior the dog always offers when presented with that object or a similar one. It’s a method that truly encourages dogs to think.
Worst-case practitioner: I suppose the absolute worst case is someone I heard about who just pointed the clicker at the dog and clicked when the dog was doing something wrong (!) but I think the more common misapplication of this technique is to create a dog who stands there looking at the owner waiting for cheese. A second, unfortunately common, set of mistakes creates a dog who is horribly behaved unless and until he is distracted by a human pez dispenser of treats. As soon as the cheese is gone the dog is back to its out of control ways.
Critics of this method usually call it “bribing” the dog, or say that dogs should behave just for love, or that food creates too much excitement, or that there’s no way a treat-trained dog will be consistent. A healthy subset call it fruity or “new,” with heavy sarcasm, the implication being that “real” trainers, those in the trenches, don’t use these methods. However, there’s really no other method that is as effective in encouraging the dog to offer behaviors. If you want to train your dog to jump up on the arm of a chair and then stand on his hind legs, how are you going to do that with physical correction? How are you going to use “mother knows best” to train a dog to fetch a hot dog but not eat it? It’s telling that some of the most flashy and complex behaviors seen in dogdom are the direct result of positive conditioning training.
It should be obvious that I don’t think that any one training method is the “right” one. In fact, I encourage puppy buyers to read the whole spectrum, from Monks of New Skete and Barbara Woodehouse all the way to Donaldson and Pryor. What I want them to do is find the passages in each book where they say “Hey, I’ve seen that happen–that’s exactly right!” and highlight them. If they’ve seen it happen, it’s likely that they can replicate it successfully. So maybe they’ll choose to pop treats for walking on a loose lead but physically correct for jumping on counters; maybe the exact opposite. What matters is whether they can make it happen, consistently and predictably, and whether they are aware of what they are doing and understand what their dog is thinking and feeling.
Where this intersects with human learning is simple–and I do mean simple. Humans learn best when they are presented with information in SMALL, REHEARSABLE bits in their OWN LANGUAGE. What this means is if I say to you (and I’m totally making this up), “The intersection of the etymology and Miller’s heuristic here is characteristic of late 20th-century Akkadian scholars.” that means NOTHING to you. But If I say “Oh, yeah, see how Miller wants you to look everything up in a dictionary, even if you already know it? That’s the way virtually all the early-Babylon experts who are alive today want you to work, because they were all trained by the same guy at Harvard who was an absolute nut about word origins and was convinced that nobody was smart enough to just remember them.”
One uses virtually no words you instinctively know; one gives the exact same information but uses words you DO know. One gives you four facts in one sentence (a large bit) and one gives you the facts spread out in digestible units (lots of small bits). A good teacher will now stop and make sure that there are no questions, and will usually call on someone to repeat back to her what was just said, ensuring that the student really did understand it (the unit of information was rehearsable).
Dog brains are no different. They learn the same way. You can’t teach “Stay off my legs unless I invite you or I am wearing jeans” in one session. You break it down into small bits (“paws down” “paws up”). You rehearse each discrete bit as close to exactly the same way as you did before, until you are absolutely sure the dog understands it no matter what room you’re in or how excited they are. You use the dog’s own language (don’t teach “paws down” before you teach the dog that moving his or her body in a deliberate way is something they need to learn how to do). This last part is more important than you know–many dogs really do think that their body ends at their neck and you have a long period of just teaching them to pay attention to where their feet are and where their butt is before you can ask for ANY behavior. THEN you can start associating jeans with paws up, or white pants with paws down. And so on.
Whatever method or training philosophy allows you to predictably and coherently and consistently teach the dog in small repeatable bits in its own language, that’s the one you need to use. Don’t listen to the haters; hating never did anyone any good. Listen to your dog and do what works.