Clicker Training
written by Kerry Bradbury
Home
Introduction
Why Clicker
Getting Started
Using the Clicker
Teaching the Basics
Targeting
Some tricks
Links
Using the Clicker
Shaping
Much of clicker training, with the exception of very basic tasks, is achieved through shaping. When we shape a behaviour we are doing much the same as when we physically shape something - in pottery or dress making for example. Stage by stage the behaviour (or item) we are trying to create takes shape. Shaping does not involve forcing the dog to do anything, it simply involves picking out the behaviour, however small, which is heading in the right direction and rewarding it, whilst ignoring and therefore not reinforcing any unwanted behaviour.
When shaping a behaviour you should first know what you are ultimately trying to achieve. Once you know this you can have a good idea of stages that you will be able to reinforce along the way, although you do need to be flexible with this. For example - when shaping the behaviour ‘down’, looking towards the floor would be a good thing to click initially, bending down would be an interim stage and belly touching the floor would be the ultimate goal.
You should progress only as quickly as your dog is ready to do so. If he is really happy and working hard, then push on for achieving more, but if he begins to look fed up and stops trying, you have gone on for too long. Dogs, like people, learn better when they are having fun, so you should always try to stop when your dog wants to do more.
When shaping a behaviour, if you have not seen any progress for a long while you are probably aiming too high. Try making the next step a little easier, or completely changing tack and trying a different way of getting what you want. Some luring may be required when shaping behaviour, to get your dog thinking in the right direction. Try to only lure something a few times and then allow your dog to figure it out from there. If you get stuck and your dog is giving up, take it back a stage so that he can be rewarded a few times for the correct behaviour. This will help him to build confidence and begin trying harder again.
Back chaining
Back chaining is a method of teaching/learning which is very useful to know about, and employ when training your dog. When developing a more complex behaviour or a behaviour that involves a string of events it is often far simpler to teach the last part of the sequence first.
This is a method sometimes employed by actors learning lines or dancers learning routines. The theory behind this is that you are always moving from territory in which you are less confident, into an area you know, therefore making learning more rewarding. For example, when learning a poem, learn the last verse first. This way you can then have a try at the penultimate verse, swiftly moving on towards a verse you know. Traditionally you would know the first verse and then get less and less confident as the poem went along. This is very unrewarding for the person trying to learn the poem and saps confidence and keenness.
In the case of dog training a good example is the recall. In obedience this exercise consists of three parts. Sit and wait, run to owner, sit in front of owner. If you teach the dog the sit & wait part, walk away from the dog and then are nagging the dog because he runs in the wrong direction, doesn’t sit straight, etc it not only saps the dogs confidence in the last part, but also in the first.
If however you teach the last bit first i.e. the sit in front, the dog is always moving towards something that he has practised and knows will be rewarding. You can then teach the wait with a very short run to the sit in front, and then gradually build up the distance.
Retrieve is another good example. First teach the dog to take something from you and give it back to you. Next ask him to pick it up from the floor at your feet and give it to you. You can then ask him to go and get something, bring it back and give it to you.
Introducing the ‘command’
With clicker training the command (or signal) is always added once the behaviour has been established. This is something that many people find difficult at first. This is because we rely so strongly on verbal communication. If we want somebody to do something we tell them verbally what to do and if necessary, how to do it. If we need to we may then show them how. With dogs it is impossible to explain verbally what we want, as they do not understand, so offering a verbal cue is of no help to our dogs whatsoever. Because of the way we communicate with each other it seems obvious to us that saying a word should somehow make our dogs understand what we want. Since this is not the case, it makes sense to first get our dogs doing the thing that we want and then add the command - telling the dog what it is that they are now doing. We are then sure that we connect the correct word and behaviour together.
When should I add the command?
This should not be done until you are happy with the behaviour that you are getting from your dog. What you don’t want is to ask your dog to lie down and then have the dog sit, look confused, lean towards the floor etc after hearing the word ‘down’. This will result in the dog learning the wrong meaning of the word down. The time to add the command is when after each reward your dog quickly lies down again. You can then be sure that you can say the word ‘down’ and because your dog is about to do it anyway the command is then followed by the behaviour. The dog will soon learn that when he hears that particular sound and he lies down, he is rewarded for it. You should at this point ignore any down’s that you don’t ask for, to help the dog to realise what word he should be listening out for.
There are two different types of command that dogs can learn - visual and verbal. It is always useful to have a visual and a verbal command for each behaviour, as they both have different applications. For example - with your hands full of shopping it would be very difficult to ask your dog to lie down with a hand signal, so the word would be very useful. On the other hand if you lost your voice, or your dog was some distance away from you a verbal signal would be useless but a physical cue would be ideal.
To have very good control of your dog and ensure they offer behaviours as and when you ask for them, it is important to properly establish the signals for behaviours. Dogs tend to offer the behaviour they have most recently learned most frequently and it is in this situation that you can end up with the same behaviour, regardless of what you ask for. For example, if you have just taught the down to your dog, he may become so caught up in being rewarded for the down that when you ask for the sit he offers the down. In this case you should ignore the down and try again. Whenever your dog offers a behaviour that you haven’t asked for and that already has a signal attached to it, you should simply ignore this and perhaps turn your back so that he knows you aren’t going to give him attention for it. This is also the correct place to use the "wrong" command.
Praise
Although the click & treat is the signal to your dog that he has done the right thing it is necessary to have a way of letting him know that he is heading in the right direction, a 'keep going' signal. This is especially important when you begin to teach you dog more complex behaviours.
As with all kinds of training, telling your dog that he is doing well is important not only because he likes the encouragement from you but also because if you cut out the click & treat once the behaviour has been learned, it will be the praise that continues to let your dog know he has done it right.
The praise should mean to your dog ‘yes you’re doing well, keep trying’, followed by click and treat when he achieves what you want.
‘You got it wrong’ command
Some people when clicker training like to use a word to let their dog know that they are on the wrong track. This is a word such as ‘wrong’ or ‘uh-uh’ said in a neutral tone. ‘No’ is not a good idea because of the previous associations it may have had for the dog and because we find it hard not to put a negative tone into it. The word is simply meant to give the dog some extra guidance and to let him know that whatever he is trying is not going to get him a reward and so to try something else. This word is useful when teaching the dog to offer the correct behaviour for the correct signal. If you ask for a sit and get a down, you can say ‘wrong’ and then try again.
Proofing
This is the term for ensuring your dog will offer the behaviour you have taught in many different situations. Your dog needs to learn that the command applies wherever you are and whatever else is going on.
Once you have taught the behaviour in your initial training area you can go to different places. In each new place you should begin as you did in the first place - allowing your dog to offer the behaviour and clicking & treating for them without a verbal signal. In each new place you do this, you will find you need less repetitions before your dog knows what game you are playing. You can begin to the verbal signal more and more quickly. This is because dogs tend to learn to associate their training with a specific situation to start with but once their learning is varied they will begin to generalise and respond where ever they are. Depending on your dog, after three or more places you should be able to go straight along, ask for the behaviour and get an immediate response.
Proofing applies not only to places, but also to situations. For example - are there other dogs around, is the environment very noisy, is it raining. It also applies to the trainer and their behaviour. Many experienced dog trainers develop their own way of luring a dog and motivating a dog for certain things. They can get a dog to sit more quickly than the novice dog owner simply because they have done it a lot and have developed the knack. However because they give the dogs lots of physical cues, which they may themselves not be aware of, the dog may learn to rely on these rather than listen to the command. E.g. when teaching a sit, they may keep their hands at a certain level and not lean over the dog.
Proofing a dog means that whatever position the trainer may be in, the dog will still respond to the verbal signal. So, you ask the dog to offer the behaviour whilst you are standing, sitting, lying on the floor, stood on one leg etc. Again, as when you are proofing in new places, first of all go back to waiting for the behaviour and then using a verbal signal only when the dog is happily offering it. This avoids the need for the dog to hear the command at the wrong time - i.e. when he is not going to respond.
Improving performance
When trying to improve on a new behaviour it is important that we reward the dog in such a way that he can understand exactly why he is being rewarded. For this reason it is important to work on improving one aspect of the behaviour at a time.
If we are teaching a dog to lie down and we are asking him to lie down more quickly AND refine the position he lies in simultaneously it may be very difficult to be successful as it will be unlikely that the speed and the position will improve together. If a dog keeps trying and not succeeding for too long a period he will give up. This is why it is important to keep steps small and manageable. It is far easier to first refine the position and then, once he understands where he is supposed to lie, speed up the behaviour.
When you are asking for a better performance you should try to keep improvements small but once an improvement has been made, reward only the improved or better than the improved performance.
If you accept less than you know the dog is capable of, as a matter of course, he will not see the point of trying harder. Dogs are not stupid and they will go for the easiest option. (Why go over a jump if you can go under, or walk around?) If he realises that it is the improved performance that has earned him the reward and that you now expect that each time he will try hard to earn what is on offer. It is important however, that if a dog is giving up or just not offering that behaviour again, that you drop your standards a little. This will help to keep the dog wanting to work with you.
Release command
It is useful to introduce a release command, such as ‘OK’ which will replace the clicker as your dog's signal to stop offering the behaviour you have asked for. The clicker has the same purpose when you are using it but the OK can take over from that once a behaviour has been established. As with the clicker the OK to your dog should mean - well done, you’ve got it right, now you can stop. Instead of signal, praise, click & treat the sequence becomes signal, praise, release word (& treat.) If you are only going to praise your dog, and not reward this time, then follow the praise with the ‘OK’. If you are using a toy as a reward then follow the same sequence, only give the toy instead of food. It is important to use a release word to make sure that you can maintain behaviour for a length of time. This is something that you should have ensured, before you stop using the clicker.
|