There is a great amount of science thrown around in dog training. It’s becoming a real problem. Don’t get me wrong. Science in dog training is a good thing. Understanding what works and why it works is the key to success. But the two camps of dog training, that go back as long as dogs have been trained are claiming all science as proof that their camp has always been right. Since this isn’t possible, given they disagree on some pretty important points, let’s discuss.
Punishment. Effective punishment is hard to come by. But when it can be arranged it has a greater effect on behavior than rewards. Yep, it’s science. But is a “greater effect” what we are looking for? During the learning phase of new behavior, it often is not what we want.
Reinforcement. Reinforcing a behavior comes naturally. The idea that a clicker and some treats teaches a dog “how to learn” is a ridiculous notion. If your dog eats food, he learns. If he breathes, he has learned. If a dog is hungry or oxygen deprived, he corrects those feelings with behavior which is rewarded. Much science has been offered about the schedule of reinforcement—what it takes for a behavior to get linked with it’s consequence. It turns out that the strongest rewards are 1. earned 2. on a VARIABLE ratio and 3. valuable to the subject. This is where dog trainers get into to trouble. The reward based trainers insist that the only primary (most useful rewards) are food and access to chasing things, that they should be given every time a behavior is offered (FIXED ratio of 1:1). This flies in the face of science. Trainers say punishment doesn’t work unless it happens every time the behavior is offered and in the real world, that’s impossible. But that rewards must happen every time a behavior occurs and that is the trainer’s job. Having it both ways is something science doesn’t allow.
Rewards. One of the reasons dogs make such amazing companions is that relationships are a primary reinforcer for them. I know, I know. Pats on the head are what people believe signify a relationship. Dogs don’t like that, just so you know. But an actual relationship is motivating to dogs and a relationship is not simply praise, especially verbal which is really a dog’s last interest in you. How you smell and what you do are way more important. And again, the reward is up to the subject, not the trainer. No matter how much you think your dog is motivated by food, hugs, pats, your voice…..if the dog is not working to get those things, they are not rewards.
Capturing behavior. This is a great tool in training and requires immense patience and creativity on the part of the trainer. If you can arrange for a dog to do something you want to reinforce and you can be ready with that reinforcer at the time, you can make a leap in training that can’t be arranged any other way. So setting up the conditions and then waiting, are the best ways to show a dog the consequence for his behavior such that the behavior will be repeated often enough to become cued by you. However, so many behaviors we want are not going to just happen. So shaping a behavior in baby steps is a secondary method that works. The big problem is humans don’t shape behavior correctly. We start with part 1 of a long line of steps a dog will have to make to complete a task that has no meaning to the dog now or when it’s finished. Back Chaining is teaching the last part first. Do this instead. If you want a dog to run out into a field and jump over a stick, start with teaching him to jump the stick. Then move it farther and farther away. See, it even sounds easier when you describe it.
Practice. Science gives us a break here. It turns out dogs learn a new task just as well by taking a lesson three days per week as when they attend class seven days per week. So no need to badger and tire them. It’s happening. Just wait. This also means you don’t have the excuse that you don’t have time to train. If you can’t spend three sessions per week with your dog, rethink dog ownership.
Learning v. Performance. Another misconception we have is that a high ratio of reinforcement in learning a new task (science says this is good) is required for the rest of the dog’s life or the behavior will slowly extinguish. If this were true, all dog problems would be solved with management. If the dog is prevented from a rewarding consequence he would stop…..everything. But this is not the case. Excepting behaviors that have an inherent reward, such as stealing trash or eliminating indoors, dogs still offer behaviors that have even been rewarded one time, if the reward was directly associated with the behavior, and the reward was good. This is the way superstitions are born. The one time a broom slipped down to the floor when the dog walked past makes the dog walk wide around all brooms propped up. No other broom ever needs to fall anywhere for the dog to continue to give the broom some room. Never again reinforced yet behavior persistent for life. So it’s just not true that constant reinforcement is needed for learning. Performance? Same deal. But a slight variation comes when a request to perform coincides with opportunities to perform a conflicting behavior that is promising a greater reward.
Premack. The Premack principle says that a dog can be reinforced for sitting with the offer to chase a squirrel. But we get this wrong all the time. If you show a dog a squirrel and then ask him to sit, you are probably not going to get compliance instantly. The idea is that a dog will work for rewards. He will sit which is not that fun if he knows that you will offer access to good things when he complies with your wishes. But a dog will not sit when asked while squirrels run about unless you have proofed against distractions.
Distractions, Distance and Duration. After you use a high rate of reinforcement to capture a behavior you want to put on cue and practiced three days per week until your dog will respond correctly to the cure 8 times out of every 10, you need to set up distractions with ramped up rewards before you can know if your dog is fully trained to respond in an emergency. Once you know that distractions won’t dissuade your dog from responding to a cue, add distance. Once you know that is no longer a factor, require longer periods of control before you request another behavior. This is one of the biggest failings in horse training. Horses often learn to trot when kicked but they fail to learn to keep trotting until a cue to walk is given because their trainers don’t require duration. Same for sitting or staying dogs. We teach them to stay but not until another cue is given. This is probably where the greatest argument between the two camps occurs. One group says the three Ds can be accomplished without punishing and the other says they can’t. Science says they can and both require laboratory conditions that rarely occur in life.
Alfie Kohn. This is probably the most important point to make to both camps of trainers. If you reward a behavior by paying for it, you reduce the value of the behavior and increase the value of the reward. Remember Premack? A dog will perform a less desirable behavior to gain access to a more desirable behavior. Yes, but when you offer a more desirable consequence in exchange for a behavior, that behavior becomes LESS desirable. In other words, a dog expects to get paid and won’t perform unless the paycheck is a foregone conclusion. This is how dogs who only respond to cues when treat bags are in view, are made. If a reward a dog currently wants seems to happen as a result of a behavior, that is going to make the behavior more likely to re occur–so that it can be reinforced and put on cue. Once a behavior is on cue, the request is completely different. Asking a fried to go dancing is based on whether you like dancing AND whether you like the friend but it also requires that you both know what “dancing” means. But if the behavior is requested in order for a treat to be dispensed, then the dog has to want that particular treat at that particular time to be engaged.
Earning it. It turns out the most important part in the value of reinforcement/rewards is earning them. When a person or dog gets random gifts he did not earn or cannot tie to his own behavior, they are not worth working for. It is only when we feel it’s within our control to get the rewards that we actively try to make them happen. Do we still like a surprise? Yes. But we don’t work for them because there is no point. We work for things we know are related to our own behavior–things we earn. Once again excepting superstition, which results when we make a mistake about whether we caused a thing, earning it or control is a powerful aspect of reinforcement of behavior. Why do so many dogs poke their nose in your work enough times to get scolded then go away happy? Just checking whether the rules have changed. Nope. Good. We often create really bothersome dogs by allowing them to ask 15 times before they are told no. Just distract him, right? If you distract him with something he likes then the consequence for his bothering you is actually a reward. And what happens when a behavior is captured and rewarded? Right.
So, if you want to use science to back up your favorite form of dog training, go ahead. But know the science or just admit you do what you like whether science says so or not. That’s fine. Science doesn’t care whether you use it or not.