Yesterday evening I left work (I have a part-time job at a pet supply store) wondering how we have so few serious dog bite injuries. I was regaled all day yesterday with horrific stories from pet owners about their “problem” dogs and the awful training techniques the owners were using to try and fix common problems like pulling on the leash and shyness around people. I marveled that dogs put up with so much crap from humans and don’t often resort to biting. It really is a testament to the temperament and bite inhibition of most family dogs. Humans are lucky.
Why do people think it’s acceptable to hit, kick, slap, scream at or physically intimidate their dogs? I want to say I don’t understand it, but a quiet voice inside of me says I do. It’s about power and control. It’s a little dark piece inside all of us – larger in some than in others, but always there in some way. I think it’s part of being human, and unless we actually dedicate ourselves to using our brains and trying to understand and outthink our animal inner selves, we succumb.
I try not to be preachy about my training methods because I know what it’s like to feel like you don’t have options and you are being judged for just trying to cope. I was there for a long time with Cerb. His behavior seemed impossible to understand or control. I wanted to scream at him that I was just trying to help him, that I wanted so badly for him to be a Good Dog, and why couldn’t he just understand that if he could do this simple thing for me — sit on command, stop hauling me down the sidewalk, refrain from trying to eat other dogs — then everything would be better and we could be Happy, capital H. Of course, he didn’t understand. He looked at me in bewilderment when I screamed and then resumed whatever he had been doing before.
Why did I think that teaching a dog would be any different than teaching any other animal, including humans? Why did I think that dogs just worked differently and that using force would somehow make Cerberus understand what I wanted? I look back now and I’m ashamed and embarrassed. I was illogical. Drawing analogies helps me explain this to others: Think of your dog as your student, a young child who needs to learn something “unnatural” (because, really, a lot of the things we want our dogs to do are unnatural for them) — how to read and write, how to do long division, how to kick a ball into a goal. At first, the child won’t understand. He will make mistakes. He’ll occasionally get it right but mostly get it wrong. How will you teach him? Will you scream at him when he makes mistakes, loom over him to frighten him into getting it right, or smack his backside when he’s not sure what to do? That doesn’t seem like an effective teaching method. Instead, you would — I hope — investigate which part of the task he doesn’t understand and where he is struggling. You might break the task down into smaller steps (first work on kicking the ball in a straight line, then add the extra difficulty of getting it into the goal). You might practice again and again until he gets it right, drilling long division problems until he has a breakthrough in understanding. You might change your approach — if a child isn’t learning to read well by sounding out words, maybe you’ll switch to a different technique. You might ask for professional help or turn to recent studies on childhood learning to see if there’s something different you could try.
There’s science behind education and there’s science behind dog training. They’re not very different.
Why don’t we offer our dogs the same options? I’m really not sure. I think there is a tendency to just give up on dogs in a way that we can’t do with children. We make excuses for dogs — “He’s stupid”, “Oh, he only listens to me at home” — or we employ different tools like prong collars and shock collars so we can just get by.
This is a problem. Let’s go back to the analogy of a child who is trying to learn a new task. There are lots of things children (and dogs) are asked to learn, so we repeat our teaching process again and again. If we use force and intimidate our student whenever he makes a mistake, we teach him that mistakes are unacceptable – mistakes bring pain and fear! Some students are resilient and draw power from within themselves to survive the pain and fear. Perhaps they harden themselves towards the teacher and become more and more difficult to effectively punish, forcing the teacher to become more scary and more forceful with every lesson. Eventually, the teacher will probably hit some sort of ceiling where the punishment can not possibly be increased. At that point, the teacher’s power is lost and the student is “free”, although it seems pretty damn unlikely he’ll ever want to perform the task ever again. In the worst situations, the student may get sick of being intimidated and threatened and decided to strike back at the teacher.
Other students aren’t as resilient. In response to threats and force, they shut down. When mistakes mean punishment and the student hasn’t yet perfected a task (meaning mistakes are inevitable), the fear of committing a mistake can become so powerful that the student is paralyzed. He offers nothing and makes no progress because it’s better to remain in limbo than take a risk that could result in punishment. This happens to people and it happens to dogs, and it’s really hard to bring a “student” back from a place where he is paralyzed by fear. The student may stop making mistakes, but he can’t move forward in perfecting the desirable behavior because he refuses to engage.
These are the outcomes from force-based education/training. Sometimes people get lucky and force works for them — I am not the kind of person to say that teaching through force never works, because I know there are trainers (especially in protection sports) who succeed through these methods — but it is a risky, risky game to play and playing it well is beyond the reach of most pet dog owners.
Let’s change up our scenario now and consider educating/training through positive reinforcement rather than “positive punishment” (the term used when a negative stimulus — like pain or fear — is inflicted in order to discourage a behavior). Our student is trying to learn long division. Every time he gets it right, we reward him in some way. Maybe he gets a gold star and after X gold stars, he gets to pick a toy from a bin. Maybe he gets a piece of candy for every right answer. When he makes mistakes, we reteach the correct behavior (e.g., he accidentally transposed two numbers so we point this out and have him repeat the problem) or we just ignore the wrong answers (no candy for incorrect problems). If we have chosen our reward carefully and it is very motivating for this individual student, he will attempt to perfect his technique so that he gets as many correct answers as possible — and, in return, as many pieces of candy or opportunities to get a toy! In the future, even when we’re not actively trying to teach our student how to do long division, he will associate this task with positive experiences and may be more willing to try to puzzle through a more demanding activity (like more complicated mathematics), because his early experience was so rewarding. By teaching him this way, we are also empowering him — we are saying “Mistakes are no big deal, what matters is that you keep trying for the correct answer.” There is no risk of pushing the student into a shut-down situation where he becomes too afraid to try novel approaches. This sounds much better to me than the force-training outcomes!
Positive training translates into an eagerness and happiness to work.
Beyond the simple question of “Does force-based training work?” is “Is force-based training ethical?” When presented with multiple training options, I believe we, as dog owners, have an ethical obligation to choose the method that achieves our goals while inflicting the least amount of harm. When you can teach a “sit” by (a) yanking forcefully on the dog’s neck until he sits or (b) luring the dog with a treat and then giving the treat once he sits, I truly believe (b) is the ethical choice. It achieves the same goal — a “sit” — while minimizing pain/fear and the risk of harm.
So while I try not to get preachy and judgmental, I admit that I struggle with people who defend force-based training because I do truly find it unethical to use this approach when positive training both achieves the results and minimizes risk. Sometimes the people I encounter truly don’t know there are alternatives to force-based training, and I love talking with these dog owners about my history with force training and how things have changed for me since discovering positive training. Other times, people have tried positive training and feel it hasn’t worked for them, so I try to figure out where the system broke down – wrong reward for their individual dog? Rate of reinforcement not high enough in the beginning? Attempting to teach a new behavior in an environment that’s too distracting? Other times I can tell that the discussion is not worth having, as the owner seems convinced that positive reinforcement training is for sissies, that positive trainers are “cookie dispensers”, and that dogs who make mistakes “deserve” to be punished — and hard. At that point, I find I have to walk away or risk losing my cool, because I don’t know how to combat someone who defends a practice I find highly unethical.
Okay, now that that’s out of the way…
Cerb and I had fun last night playing with his new mat, a Molly Mutt “dog bed duvet.” I freaking love duvets. I don’t know why we don’t cover EVERY BLANKET with a duvet. They just make it so much easier to keep comforters looking nice and clean. I was excited to see these dog bed duvets because Cerb frequently gets his beds really gross. He’s a dog, it happens. Sticking the whole bed in the washing machine usually results in a lumpy, misshapen bed, though, so a duvet seems like the perfect solution.
Of course, that’s not how I plan to use this thing. I’m using it as a mat. Cerb and I are taking a Control Unleashed (CU) class on Thursday nights and a huge part of the Control Unleashed curriculum is “mat work.” We teach the dogs that their mats are safe places, places where treats rain from the sky and nothing bad can happen to them. Why? Because mats are portable. CU is an awesome approach to helping reactive dogs relax in new places. The mat becomes a place of retreat where an anxious dog can find comfort. Through clicking and rewarding when the dog relaxes on the mat, we form a connection between being on the mat and feelings of relaxation and pleasure – then, when you’re in a stressful place and you roll out the mat, your dog can sit on it and return to those calm feelings. It’s extremely effective and deceptively simply, really, and I think it will be a valuable tool for managing Cerb at trials and competitions. He is so willing to work and so easily excited that it can be really difficult to make him relax and keep his head on straight, and when he’s not relaxed, he can’t brain. When he can’t brain, he reacts without thinking and he can’t make decisions that will earn him rewards. It’s not a good situation for him (or for me).
So last night I whipped out the new mat and we started some foundation work, because I’m anxious about starting the new class and it makes me feel better to prepare ahead of time. I unfurled the mat and threw some treats on it to get Cerb’s interest, then sat down with my clicker and waited. Whenever he touched the mat, I clicked and threw a handful of treats in front of him. He caught on to that pretty quickly and was soon bounding over to his mat whenever we moved away from it. Then I increased the difficulty – I didn’t want him to just touch the mat, I wanted him to lie down on it. At first, he didn’t realize I’d changed my requirements, so he would touch the mat and then look at me expectantly. I ignored him, because that wasn’t what I wanted. Frustrated, he touched the mat again. Nothing. He sat down next to me. Nothing. He flopped down on his side. Nothing. With an exasperated grunt, he offered me a “roll over” — and rolled onto the mat. CLICK! Treats rained from the sky. He caught on immediately and, from that point, would go to his mat and lie down. I moved the mat to a few different locations in our house to make sure he was aware that the mat was important, not just its location (the behavior I wanted was “lie down on the mat”, not “lie down next to the bookshelf, ignoring the funny-colored mat that happens to be there”) — dogs are often poor generalizers and I wanted to be sure he was associating good things with the mat, not with the room where we were training. After a few false starts (fewer every time!), I’m pretty confident he thinks the mat is a Very Awesome Game. I think he’ll enjoy taking the CU class and puzzling out what I want from him.
If only I had a mat.
I get very frustrated by people who whip out stories of truly bad dog trainers who are force free as evidence that force free training doesn’t work. Of course it doesn’t work if you do it wrong. Neither does force-based training. A lousy trainer is still going to be a lousy trainer, just potentially a less damaging one.
Yep. As a public-health-y type person, I’m all about risk management and harm reduction. Bad trainers are bad trainers and I wish they weren’t out there, but a bad positive trainer is going to have little to no impact on a dog, while a bad force trainer can do serious damage. That’s where I stand on the issue, anyway.
I’m curious about your take on positive reinforcement training with consequences? Thing is, I find far too many people who use positive reinforcement (or PR) tend to spend the rest of their dog’s life tossing cookies at them and walking around with the equivalent of a five-course dinner in their pockets. In other words, they’re never taught how to “graduate” their dogs to the next stage. IMO, the idea is to train the dog toward the goal of performing a specific behavior when asked, but without the “reward” of the cookie (or toy or what have you). If they don’t perform as asked, then there is some sort of consequence. Obviously, the consequence isn’t a beating or getting jerked around, but something that strikes a chord with the dog. For some of my gang, that might mean not getting to go outside with me if they can’t sit nicely at the door and wait for the release command to exit. While this consequence doesn’t “hurt” my dog, it gets the point across. Since they fully know what I’m asking, they should offer the appropriate response. The “reward” is that they get to come out and be with me (not a cookie), and the consequence is that they don’t come out unless they respond appropriately. PR is a good thing … it’s humane, easy to teach and easy to learn, but unless you move to the next level you become little more than a Pez dispenser for your pet. Like kids who are constantly told how great they are and given stars, treats and “rewards” for any show of effort …. no matter how small … many dogs soon learn how to get that reward and do as little as possible or worse yet, will do nothing at all unless they know their “reward” is eminent.
Absolutely. Thanks for commenting.
You’ve actually got it reversed. The first thing you described, the “cookie dispenser” part, isn’t positive reinforcement training. It’s positive reinforcement, but not training! Giving a dog rewards for every little thing doesn’t train *anything*, as you’ve observed. At best, it’s bribery and luring. At its worst, it’s actually reinforcing bad behaviors.
The second thing you describe, training with consequences, *is* positive reinforcement training. Not every reward has to be food. My trainer is big on what she calls “life rewards” – things the dog wants to do that can be used as rewards for things he doesn’t particularly want to do. This is known scientifically as the Premack Principle and it means you can use things like “going outside to pee” as a reward. The dog wants to do it, so you ask for something first (sitting quietly by the door), and once he does that, he can go outside! That is absolutely positive training and that’s how it should be done. You don’t want to feel like you must ALWAYS carry treats everywhere, of course. Dogs can also be rewarded with games of tug, playing fetch, getting to pee on THAT particular bush, getting to greet another person/dog (calmly) — my dog LOVES the well water out at Karen’s farm, so we use a bucket of water as a reinforcer for him! Instead of just letting him haul me over to the bucket to drink, I demand that he do something for me first, like walk up to it in heel position or do a figure-8 around it. Once he does that, I release him to drink. That is a WAY more powerful reward for him, at that time, than a cookie!
It’s very important to set standards in positive training, too. Rewarding EVERYTHING a dog does means you never teach him exactly what you want. For example, you’re trying to teach him to “sit”. If you reward every sit, no matter how slow or how crooked, the dog will never know what to offer you. You can start out with lower standards to shape the behavior, but after the dog understands what “sit” means, you have to raise the bar. You can reward ONLY for a “sit” that happens within 2 seconds of saying the command, for example, or ONLY straight sits in heel position (not crooked ones). You have to show the dog what the One Right Answer is so he doesn’t bother offering all the wrong ones!
You can even go so far, in my opinion, as “negative punishment.” Negative punishment is the removal of a pleasant stimuli. When people think about “punishment” they’re normally thinking of “positive punishment”, which means you are ADDING (hence the “positive”) a negative stimuli in order to decrease a problem behavior – for example, jerking the dog’s collar to get him to stop pulling you down the sidewalk. Negative punishment means you take away something the dog likes as punishment. Maybe your dog refuses to release the tug toy when you ask for it, so you stop the game of tug and walk away – this is punishment, but it’s not painful or forceful, you’re just stating your terms clearly: if you want me to play tug with you, you will drop the tug when I say so or the game is over. The dog loses out on the enjoyment of a game of tug. This is why it’s important, when you’re training a “release” or “drop it” command, to not STOP THE GAME immediately after the dog does what you ask (at least in the early training stages) – if the dog thinks that every time you ask him to “drop” the ball, the game of fetch ends, he’s never gonna wanna drop that thing! So when I’m teaching a “release” or “drop it”, I give the command, get the behavior I want, and then reward my dog by playing with him again.
Hope this answers your question!
Yes, but unfortunately, this: (“The first thing you described, the “cookie dispenser” part, isn’t positive reinforcement training. It’s positive reinforcement, but not training! Giving a dog rewards for every little thing doesn’t train *anything*, as you’ve observed. At best, it’s bribery and luring. At its worst, it’s actually reinforcing bad behaviors.”) is what the vast majority of most PR classes teach, and nothing else! What you call “negative punishment” is what I like to call “consequential training.” There are consequences for non-compliance, the selection of which relate to the dog’s overall sensitivity and level of training. That said, there are dogs who really don’t give a hoot if you take things away. Privileges, toys, even movement … these things mean absolutely nothing to them. I have one of those dogs. They’re an enormous challenge to train! Finding what trips their trigger from day to day can be a huge job!
Well, of course there are bad trainers out there – more than there are good trainers, perhaps. This doesn’t change the meaning of what “positive reinforcement training” actually is, it just means that those trainers are misrepresenting PR training
There will always be bad trainers or bad representatives of ANY activity – it’s important to judge their training ability and not take their LACK of ability as a condemnation of the entire method. A trainer’s failure to properly use PR training doesn’t mean PR training doesn’t work or that it’s insufficient – it means that trainer sucks!
And yes, you definitely have to tailor PR training to the individual dog. That’s another mistake a lot of people make. Not all dogs are sufficiently food-motivated to be rewarded by treats in a distracting environment. That’s why I think that the perception of PR training as “cookie dispensing” is dangerous – food is not the only reward, it’s not always the strongest reward, and trying to use it as a reward for a dog who doesn’t care is just going to set you up for failure. When my dog is over threshold, he couldn’t care less about the food in my hand – at that point, I just have to manage him, remove him from the situation, and get him to a point where he can think again.