Training your dog using operant conditioning and positive reinforcement involves a process known as ‘shaping‘.
This means that you teach a complete behaviour (e.g. touching a target) by breaking it down into several small stages that are easy for the dog to understand and learn.
Positive reinforcement (PR) involves rewarding the dog for performing a behaviour correctly.
But in many cases, the dog won’t be at your side, close by for the reward, when he performs the desired behaviour.
By the time the dog has returned to you for the treat, the association of the reward with the behaviour has been lost.
So you need to have some way of ‘marking’ the point at which the behaviour you want happens, a ‘bridge’ between you and the dog.
For most PR trainers, the marker of choice is the clicker. It produces a short, clear sound which, unlike the human voice, does not alter in emotion or intensity.
‘Clicker words’ such as “yes!” and “good!” can also be used, but are best kept as an occasional alternative to the clicker, not a replacement.
Turning the Clicker into a Marker
The clicker is a useless toy unless its sound is made meaningful to the dog. So how does the clicker become a marker that signals to the dog that he’s performed the right behaviour?
You need to start by associating the clicker with a reward. The easiest reward to start with is food. Choose something soft and very tasty, such as cheese or sausage.
Cut the food up into very small pieces, so that your dog gets a brief taste, not a meal. You’ll use dozens of treats in a typical clicker training session, so the pieces need to be very small.
Start by scattering a small handful of treats on the floor, and click every time your dog eats a piece. If your dog is sensitive to the sound of the clicker, try holding it in your pocket.
Next, throw down one tidbit at a time and click just before the dog gets it. The dog should now start associating the clicker with the food. Test the association by clicking once. Does the dog look at you for a reward? If so, throw the tidbit down. If not, return to an earlier stage where you were successful.
Shaping a Nose Touch on a Target
You can use your clicker to mark existing well-established behaviours that you are going to reward. For instance, you ask your dog to sit; you click as he lowers his bum to the ground; and you give or throw him a reward.
When you decide to shape a new behaviour, it is vital to define accurately in your own mind (or on paper if it helps) what the complete, end behaviour is that you want. Then, you need to consider the stages that the shaping process can be broken down into, so that the dog learns step by step.
Here, the behaviour we are going to teach is a nose touch on a small transparent target (e.g. piece of perspex, yogurt tub lid). This is a core skill that we teach all our new agility handlers and dogs. The target can be used for teaching sendaways, and at the end of contact equipment and weaves.
If you like, you can ‘free shape’ this behaviour, which involves putting the target on the floor, and waiting for the dog to show some interest (even just glancing at the target). Click/treat (CT) each gradual step closer towards the target.
Alternatively, use a ‘food lure’ to encourage the dog to investigate the target. Place a small tidbit on the target as a lure.
Click at the precise moment that the dog bends down and takes the food, and throw a piece of reward food to one side. Remember – the lure is not the reward – you still need to reward the dog after you click to keep the association.
Throwing the reward to one side (vary where each time) makes the dog move away from the target. You now need to watch carefully what he does next.
He may understand the connection between the target action, click and reward very quickly and go straight back to the target.
But he is more likely to sniff around for other food treats. CT him every time he’s heading towards the target. Eventually he’ll end up back at the target, and at that point you can give him a ‘jackpot’ – a handful of treats to make that moment more memorable.
When shaping, remember to be prepared to reward very small steps towards the desired behaviour.
And if the dog becomes confused, go back a stage or two to a point where he can be successful again – you want far more successes than failures.
Use your voice to encourage the dog to keep trying. Take your time and the end behaviour will be all the stronger.
Over the course of several training sessions, you should end up with a dog that is keen to go and nose touch your target as soon as you put it on the ground. Remember, at this point you are still not using a verbal cue. The behaviour is cued by sight of the target, and your main feedback is the click/treat.
Variable Reinforcement, Cues and Generalising
The next stage is ‘variable reinforcement’. This means that the dog does not receive a CT every time it performs the target touch. A bit like a slot machine for a gambler, variable reinforcement (VR) keeps the dog guessing – will it receive a treat this time?
It also enables you to start raising your criteria – to select and reward the better versions of the behaviour you’re offered, rather than every instance of the behaviour.
You are ready to introduce VR at the point when your dog is confident in the behaviour – in this case, is repeatedly doing the nose touch. Instead of a CT every time you get a nose touch, occasionally just give the dog praise.
This enables you to start raising your criteria. Select and CT the better versions of the behaviour you’re offered (aim for about 50% CT), rather than every instance of the behaviour.
Raise your criteria by small stages. If you have occasionally been rewarding paw touches (to get the dog interacting with the target), start to CT nose touches only.
Raise the criteria further by selecting (i.e. CT) for greater accuracy – those occasions when the dog’s nose is actually touching the target, rather than hovering a couple of inches above it.
Select for greater speed – CT those occasions when the dog is faster getting to the target. Work towards greater independence – vary the position of the target, and your distance from it.
The next step is to introduce the cue – the word you are going to use to tell the dog to perform the nose touch.
Choose a word (such as “Touch!”) that you do not already use in some other context with your dog. As the dog is about to nose touch the target, clearly say your chosen cue word. Then CT as usual. Repeat this several times, so that the dog starts to associate the cue word with the nose touch.
To put the behaviour on cue, start to fade out rewards given for nose touches that you hadn’t asked for (by giving the cue word).
Click and treat when you’ve said “Touch!” just before the dog touches the target. Don’t click and treat if you’ve said nothing. Be patient and your dog will learn to target on cue.
So far you have been working in the same area, probably a room in your house. This is the point where you need to start ‘generalising’ the behaviour.
You do this by moving to different rooms, outside into the garden or drive, then a quiet park or field, finally busier places with more distractions. Be prepared if necessary to retrain the behaviour almost from scratch.
The purpose of generalising is to show the dog that the same behaviour is desired, whatever the working environment.
This is the stage that is most often neglected. We’ve all heard someone at training say: “But he does it perfectly at home!”
