Luring is probably the most widely used method. It means that you, by holding a treat (or toy) in your hand and moving your hand, manipulate the dog into doing the wanted behavior. For example sit, see video below.
Shaping is not really a new method, but it is relatively new in the dog world. Dolphins are trained largely by shaping and if you've seen a dolphin show you know that this is a powerful method of training. The basic idea is that the animal should "offer" you a behavior that you can reward it for. A dog will for example do a lot of desired things during a day, and simply by rewarding it when it does something you like, you increase the likelihood that the dog will do that again. In a training situation you wait for the dog to offer you the particular behavior you want and then reward it. You can see an example of how to shape "sit" in the video below.
As you can see in neither method the command "sit" is given. When teaching an animal a new behavior you should not start using the command until the animal knows the behavior. In this example, you should not start saying "sit" until the animal already knows that sitting is what you're after. When luring a dog the dog might not even realize that it is sitting down, it just wants the treat. So if you start saying "sit" while training, the dog thinks "sit" means "take the treat". This is a problem I've had while training Scott to stand up and still. We usually start with him sitting or lying down, so he thinks that since he gets the treat when he stands up that "stå" (which is the word I use) means "stop sitting/lying and come and get the treat". He has not realized that what he is rewarded for is the change in posture. So the command looks like it's working just fine, as long as you keep the treat in your hand just in front of him, but as soon as you are standing farther away it becomes clear that he has not understood it, because when you say "stå" he comes running to get his treat.
This is not a problem you will encounter when teaching the dog a new command by shaping but the rule of not saying the command before the dog knows what to do, is still true. When shaping the sit, the dog will not know if the command "sit" means "look at me", "step with your paws" or "to sit down" or what ever other behavior the dog might do like blinking or licking it's nose. Since in the beginning it will try different things to get the reward, continually hearing you say "sit" will only make it confused, it has nothing to connect the word to.
So is there a preferred method here? Well for the dog it doesn't really matter as long as you don't correct/punish it for not doing what you are trying to teach it. Punishment will only make the dog insecure and worst case scenario is that is doesn't want to do any training at all with you, it is to afraid to dare to offer you a behavior. For mental stimulation the shaping technique is better, since the dog actually has to think. Luring is in a way easier since the dog doesn't have to learn anything before you can apply this technique. Shaping on the other hand demands that the dog learns that a certain sound (like in the video, the sound of the clicker) means that it gets a reward. After that it has to learn that in order to get "the click", it must offer behaviors. When this is done the dog is ready to start learning new behavior by shaping.
The click is a so called conditioned or secondary reinforcer, that is the click will only reinforce a behavior if the dog thinks the click means it will get something it wants. The treat or toy that it get after hearing the click is the primary reinforcer. This is actually quite similar to the command rule (that is, don't say the command until the dog knows what behavior you want), but you teach the dog in a somewhat different manner. To teach the dog that the click means there is a treat coming, you just have to get some treats that you know your dog loves and start clicking every now and then and immediately give it a treat. Preferably you click when the dog does something you like, but don't chose just one behavior at this point. (Examples: when it sits or lie down, when not barking at someone walking by, when walking nicely on the leash etc.) When the dog starts to respond to the click by going to you to get its reward, you can start the training. You can start having short training sessions where you decide before you start what behavior you want, and only click when the dog does this behavior. It is important that the dog doesn't get stuck to one behavior at this point, so change what behavior you click for the next time you are training the dog. Don't practice too many behaviors at the same time though, but you could for example train sit, down and standing in different sessions, and vary which one you're doing in one particular session.This way your dog will start learning that offering different behaviors will give it rewards. When the dog starts to get the hang of a specific behavior, you can start adding the command and within a short while you will see that the dog knows the commands. At this point you don't really need the clicker anymore, the clicker is for LEARNING the behavior, to maintain it you just have to give the dog a treat every now and then after preforming the right behavior after the command. This goes for luring too, you don't have to keep luring the dog forever, just until it knows what to do.
Just to be clear here, there is no need to use a clicker to be able to shape behavior, but the sound (or if you prefer light flash, hand movement or other visual or audible cues, even smells can work but are hard to do in practice) has to be the same every time, so it is actually a bad idea to use the voice since your own state of mind is easily understood by the dog by the tone of your voice, and if you are not genuinely happy when the dog did the correct thing it will pick up on that and not be as motivated to do it again.
No comments:
Post a Comment