Last week I started working on a new behaviour with Emmett (8 year old Australian Shepherd). We are working with a 6 foot ladder as a prop. I decided I wanted to use shaping to get Emmett to go under the ladder to start. Eventually I would like to make the behaviour a bit more complicated by adding a second behaviour to the “go under the ladder” part, but I haven’t decided on what yet. The picture below is from the video. I apologize for the poor quality of the photo.
Shaping is an interesting training method and if you want to do it correctly, there are a few rules to follow. Shaping is rewarding small behaviours that are not quite what you want the end behaviour to look like, but are on the way there. You keep rewarding closer and closer approximations of the behaviour you want, until the dog is doing the end behaviour.
While you are training, you don’t give the dog an extra cues using your voice or any body movements to let the dog know he has it right. Even moving your foot without thinking, or clicking with extra vigour can be a cue to the dog.
Why should you use shaping?
Why would you want to shape? I like using it because it allows the dog to think things through in order to figure out what gets rewarded. With luring or hand signal training the dog is, at least initially, relying on you to give directions about what you want. I think shaping is important to use as a part of a training program but not necessarily the only method. I use luring a lot, depending on the dog and the behaviour I am working on.
Even though there is a “proper” method to doing shaping, you will usually make a mistake or two. Sometimes I accidentally click when I shouldn’t have or throw a food reward (if I am throwing them on the floor) in the wrong direction which causes the dog to move out of the proper position to set up the next repetition. There are many different things that could happen accidentally that are not exactly how things should go, but it is really not that important. Most dogs will learn in spite of us humans.
In the picture below, AJ is being shaped to touch her nose to a scented playing card. In a past blog, I described training errors in working on other tricks with AJ. And again, with this trick I have to make a modification to the training process. Some of you may know that I was holding the card in my hand for AJ to find and she ended up looking for where my arm was instead of using her sniffer. By using the step chair, I can move the card around to different positions without her seeing (when she goes to get the food I throw on the floor), which requires her to use scent to search for it. Of course, this was the original idea of the trick to begin with.
I also did some shaping with our Boxer friend shown in the picture below. We have been working on teaching the nose target, and have progressed to having the dog touch the target stuck to the wall. During the training session in the picture, the Boxer was able to respond to the cue “touch” at the beginning of the session and then as I moved away from the target, she had more trouble. So we went back a step and did more work with me standing closer to the wall.
If you decide to try shaping it is a good idea to watch a few videos and/or read a few articles on how to do this correctly. One of the articles could be this one 😀
To shape a behaviour you need:
To decide on the end behaviour you want the dog to do.
Have a clicker and lots of small, tasty food rewards that your dog will eat.
Have your dog and the prop you will be using in an area of no distractions. The prop could be as big as a ladder or as small as the target we are using in the photo above. A target is particularly useful for many different applications, so consider teaching your dog to nose touch or paw touch a target in your first attempt at shaping.
Many dogs, when faced with no verbal cues or hand signals from their handler, will simply default to a sit and stare at you waiting for a reward. It doesn’t matter. Just start.
To shape you simply wait for your dog to make any kind of a move towards the object you are working with. This could be even just LOOKING at the target, or the dog moving his eyes away from you to something else. In shaping, you need to be watching your dog like a hawk for any little movement that can be rewarded that interacts with the object, or even just moving towards it with one paw. Reward the tiny movements the dog makes, the incremental movements towards the final behaviour that you want. If you think a movement is too small to reward but your dog is not offering anything else, reward those movements anyway.
Once your dog makes a movement that you want to reward, click, and throw a piece of food on the floor. I always throw the food reward on the floor away from where we are working so that the dog will have to return towards the object afterwards. This may be a problem for some of you if you don’t want your dog eating things off the floor. The dog will hopefully 😉 get up and move to get the food. This should set up an even better situation for rewarding a tiny behaviour you want. If your dog goes back to sitting and waiting don’t worry. It will take a little time for her to understand the process.
In the pictures above you can see Emmett going under the ladder after having been clicked, and then going for the food on the floor on the other side. When he turns around after eating the food he will be facing the ladder which is what I want hime to do so I click and throw the food to the other side of the ladder. This sets up the same situation. He will be facing the ladder and coming towards it, and will get clicked again.
Your dog will start to realize what is getting clicked and what is not. You will be click/rewarding the progression of small behaviours towards that final behaviour. It is likely that your dog won’t get to the final behaviour right away, but if he does click reward it like normal. It was probably luck that it happened so early in the training process, and if it is not repeated right away, keep trying. Make sure, though, that you aren’t waiting for the dog to do better and better approximations of the behaviour you want too soon.
I don’t add the cue until the dog is doing the final behaviour well. You don’t want to associate the cue with the wrong behaviour.
If your dog decides that sitting is better to get rewarded for, simply move yourself to a different position so your dog has to move towards you, or just throw a reward on the floor to get him going. This usually works.
Shaping is definitely a lesson in patience, but worth it as the final behaviour will be really well learned.
Happy Clicker Training!