Clicker Training Edit
What is clicker training?Edit
There are many different ways to train animals. Clicker training is just one method. This article relates to how it may be used to train dogs, but it may be used for any animal which can learn (including mammals, birds, fish and even some invertebrates). It is based on the principles of Operant Conditioning, more specifically positive reinforcements. When the dog does something right, the action is marked with a "click" produced by a clicker. The click serves as a marker or conditioned reinforcer which lets the dog know it did something right at the moment it heard the sound and that a reinforcement will soon be delivered. The actual "click" is not important, and a different sound or even touch or visual signal can be used, so long as it is distinctive and consistent.
Operant conditioning may seem confusing at first, but it really just describes some pretty simple principles of learning.
There are four ‘quadrants’ of operant conditioning:
Positive Reinforcement: the primary tool of clicker trainers. If you rattled a soft drink machine and lots of money fell out, you would probably be more inclined to rattle soft drink machines in the future. i.e., If you do something and something good happens, you are more likely to repeat that action. In training, you can use this principle quite easily. For example, if your dog sits and you give him food, he’s more inclined to sit in the future.
Negative Reinforcement: Usually avoided by clicker trainers. It means that if you perform an action which causes something bad to stop or go away, you are more likely to repeat that action. For example, jabbing a horse in the side until it moves.
Positive Punishment: Also avoided by clicker trainers, and probably the thing you think of when you think ‘punishment’. If you perform an action and something bad happens, you are less likely to repeat that action in future. For example, if a bird comes near people and they chase it, it is less likely to go near people.
Negative Punishment: Sometimes used by positive trainers. If you do something and there is some penalty, something good being taken away, you are less likely to repeat that action. For example, if you are playing with your dog and she jumps up on you, and you walk out of the room and leave her alone, she will be less likely to jump up on you.
There are some other useful principles of learning which it may be helpful to know about:
Extinction: To use the soft drink machine example again, if you put money in it and nothing happens, you will be surprised. You might try feeding the even more money, or try different behaviours such as shaking it or kicking it or complaining to a shop keeper. This is called an extinction burst. If none of these worked, you would stop putting money into the machine.
Schedules of reinforcement The drink machine has a fixed schedule of reinforcement - you expect a drink every time you put your money in. In training, this is when you give your dog a treat every time she performs a behaviour. This is useful in teaching a new behaviour, but not for maintaining it, as the dog will go through an extinction burst almost immediately if rewards stop. The slot machine, on the other hand, has a variable schedule of reinforcement. If you put money in a slot machine, very often nothing happens. But people keep putting money into slot machines, because sometimes money comes out, and sometimes a LOT of money. Behaviours maintained like this are very resistant to extinction because those intermittent reinforcements make them think that if they play for long enough, good things must happen.
Classical conditioning: Associating a neutral thing with something else more meaningful. The clicker is an example. The animals associate the clicker with food (or other reward), and the click itself becomes a positive reward. Toilet training is another example. The dog, taken out often enough and at the right times, pairs ‘grass outside’ with ‘full bladder’, and soon the dog will pee on grass rather than on your carpet.
What is a reward?Edit
A reward is anything the dog wants (conversely, if the dog doesn't want it, it's not a reward. Some people thing they are rewarding their dog with praise or cuddles, etc., when the dog doesn't actually want them). Food is a very common reward because it is convenient - easy to carry, quick and quiet to give, and most dogs will do anything for a bit of food. However, some dogs are not very motivated by food, so another reward should be found. Some dogs really love to play, or cuddle. Once the dog has learned a behaviour, life rewards can also be used. For example, if your dog wants to go out, have them sit and then open the door; going out is the reward.
The first step in teaching a dog is, of course, to get the behaviour that you want. You can do this in many different ways:
Luring: Luring follows the saying 'control the head and the body will follow'. It is usually done by holding a piece of tasty food in front of the dogs nose. You can then move the food and the dog will follow so that it can keep sniffing or nibbling the food. The dog can be easily positioned into a sit or down or heel or other position and then clicked and given the food as a reward. The food must then be 'faded' so that the dog will respond to a signal when food is not visible. Some people train the dog to touch a 'target stick' and use that to guide the dog into position.
Shaping with shaping, the dog may be taught anything that she is physically capable of. When the dog performs even the slightest part of the desired behaviour, they are clicked and rewarded. For example, if you wanted the dog to lie down, you might reward the dog for looking at the ground, then for touching the ground with their nose and so on, until you had the behaviour you wanted. Then you could put it on cue. This method can seem slow at first but the dog really gets to understand what you want and leads to very neat, independent behaviours.
Positioning: The dog is physically manipulated into place and then rewarded.
Capturing: involves waiting for the dog to perform the action on their own and then clicking and rewarding. This can be slow and you may have to carry treats and a clicker on you all day, but it is the best way to get some behaviours.
Eliciting instinctive behaviours: If you want a retriever to fetch or a shepherd to herd, the best thing to do is put them in a position where their instincts come out on their own. You can also use it for every day things, for example if you run backwards your dog is very likely to chase you, and you can teach a recall in this way.
Distance, Duration, Distraction, DiscriminationEdit
Once you have a behaviour, you will want to improve it. A behaviour only becomes really useful when it can be performed on cue with distractions like other people around. Behaviours like stays and heeling are duration behaviours, which means that you want our dog to stay or heel for a long time, not just a second. Behaviours such as send-outs and retrieves involve distance. All these things should be introduced gradually, one tiny increment at a time until the dog really understands. Keep in mind that when working on one criteria, such as working around distractions, other criteria can be lowered (so you may accept a slower sit when there are noisy children around that your dog isn't used to yet).
Teach Your Dog to HeelEdit
You will need: Clicker, rewards, a reasonably large place where your dog can be off lead without too many distractions.
Step one: To begin with, it is easier to begin indoors. A passageway is great. Show Fido the treats (or toy) and start walking. Fido should be interested, and will probably start following you. At this stage it does not matter where he is in relation to you, just click and reward.
Troubleshooting: Fido doesn’t follow me, he just stands and stares! Try feeding him a treat then backing away quickly (but not too far) and calling to him. Once you can get him to follow as you walk backwards, try step one again.
Step two: Now start clicking when Fido is in the position you would like. Don’t wait for any duration, just click when he hits that spot. Talking to your dog can keep him interested so he keeps trying, and soon he will be in the right spot. If you would like him to be able to heel on both sides, click whenever he hits the right spot on either side.
Troubleshooting: He’s jumping up and mouthing my hands! Teach Fido some Zen. This will teach him that mouthing your hands will get him nowhere. If he still persists, whenever he jumps up just stop and wait until he’s just standing calmly by your side, and then click. It is a good idea to tire Fido out with a long walk before you start heelwork, and use slightly lower value treats. This will help him stay calm.
Step three: Go outside to your nice, reasonably large off-leash area. Start walking. Whenever he comes up to you, click and treat. Then start only clicking and treating when he hits that magic position.
Troubleshooting: He won’t stop sniffing the ground! Make sure he knows you have the treats. Try the first troubleshooting suggestion. If he still seems distracted, Go right back to step one, inside.
Step four: Increasing duration. Now you will only click if Fido stays in the magic position for one step. Once he is doing that very well, wait until he has stayed with you for two steps before you click. Slowly increase the number of steps, one at a time, only moving on once you are both confident. Start adding the cue. As you move off together, say ‘Heel!’ in a happy voice. If your dog heels on both sides, put a different cue on each.
Step five: When you can go about 6 steps together, start rewarding randomly. For example, reward after 2 steps, after 5, after 4, after 6, after 1, after 2, after 7. Try and reward the best performances. If you do this correctly, you should end up with an attentive heel, and also increase the distance you can go together. Now is the time to introduce the automatic sit (or stand or whatever position you want). Whenever you stop, say ‘Sit!’ Click and treat when he does, but don’t always halt before you c/t. Holding your hand down with a treat and then raising it up as a lure can really help some dogs. Keep up the cueing – start saying it as Fido comes to heel as well as when you move off together.
Troubleshooting: My dog sits crooked! Try walking next to a fence or wall so your dog has to walk and sit straight. Otherwise, teaching your dog to ‘finish’ can help improve his understanding of where heel is.
Step six: Speed changes. Try walking slightly faster than usual. If your dog keeps up, c/t. Then try walking a little slower than usual, and if your dog maintains heel position, c/t. Walking slower is generally harder for dogs than walking faster. Gradually increase the difference between the paces. When you are going to speed up, lean forward a little, and when you are going to slow down or stop lean back a little. This prepares your dog to make the change. Try out your stop-sits at the different paces.
Step seven: turns. Start curving slightly as you walk. Curving away from Fido will probably be easier for him than walking ‘into’ him. Then start gradually increasing the sharpness of the turns until you can do 180 degree turns either way.
Step eight: Finish, or go to heel. There are different ways of going to heel and many different ways of teaching them. (If your dog has learnt a right heel, reverse the directions.) The first way is for the dog to sit in front of the handler, then walk around the right side of the handler to sit in a left heel position. Another way is for the dog to walk left past the handler, turn around and sit in left heel. Then there is the flip finish, where the dog’s front feet remain fairly stationary and the dog swings its rear around to end up in heel position. And if you are not an obedience person, you just want the dog to walk into heel position from wherever they are. Method 1: With Fido following you, run backwards holding a treat in your left hand. Use that to ‘pull’ Fido out to your left. Lead him in a circle as you start to walk forward again, then come to a halt. Fido should now be in heel position. You can fade the lure into a hand gesture, and start taking fewer and fewer steps until Fido will go to heel with just a subtle hand gesture. This will result in either a flip or walk past then back finish. Method 2: Hold a treat in each hand. Lure Fido around to your right and then behind your back. Then attempt to switch his attention from your right hand to your left, so you can lure him around to your left side in heel position. Stop using lures as soon as possible and make the gestures less obvious until you just need one subtle gesture. This will result in a walk around finish. Method 3: Shape Fido to perform the desired finish. Moving your body, for example taking steps like in method one or just starting to walk forward, can help your dog catch on. This will result in a specific finish and probably also a dog able to find heel from anywhere. Method 4: Just repeat the cue enough times when Fido is coming to heel that he just catches on. This will result in a dog that finds heel from anywhere and not any specific finishes.
Step nine: different positions in heel. As you come to a halt, tell Fido to ‘drop’ or ‘stand’. At first he may sit or be confused. If so, release him from heel and do a few rapid fire drops and reward him. Now call him back to heel and say ‘Drop!’ If he does, click and treat. Try the same for stand.
Step ten: If you would like your dog to be able to heel in different ways – e.g., sidestep, heel backwards, now is the time to teach it. See individual instructions.
Step eleven: Start adding distractions. Have a helper come with you and just stand there. Once Fido is confident with that, have the helper move around, crouch down, make noises. If Fido breaks to go investigate, the helper should stand, turn away and ignore the dog until the dog goes back to the handler. No other punishment needs to be used. Once Fido is confident with that, have the helper play with a toy or hold out food. If the dog breaks, the helper should ignore the dog and not let the dog have the toy or food.
Step twelve: More duration. Now is the time to introduce 300 peck. Start off heeling and click after one step, then 2, then 3, then 4 and so on. If the dog makes a mistake or breaks at any time, go right back to one step and build your way up one step at a time. Try not to make these lessons too long or boring. Eventually you will be able to heel for long enough to do heeling in the obedience or heel work to music ring.
Step thirteen: More distractions. Start heeling in different places. Down the street, near shopping centres, near playgrounds, near other dogs, in class etc. With these non-controlled distractions, your dog should be on lead. If your dog breaks, stand still and pretend to be a tree until your dog looks at you. Then call him to heel and continue with either a higher rate of reinforcement or with fewer distractions. If you like, when the dog has done well you can use life rewards as well as or instead of treats. For example, heel 5 steps then release him to go sniff or say hello to that friendly dog or whatever. Once you are confident that your dog will heel until released, you can remove the lead.
You continue in this manner until you reach the level of proficiency that you want.
Teaching Go to BedEdit
This can be a very useful behaviour. You can teach your dog to get in their crate or settle down out of your way.
You will need: a blanket and/or crate, treats and a clicker. And your dog, of course.
Step one: Put the bed on the ground and click the dog for looking at it and any movement towards it. Throwing the treats onto the bed after the click can really help. The aim is to get the dog consistently touching or walking on the bed. If you do not want to free shape, you can lure the dog onto the bed, or walk around with the dog following you and as the dog steps on the bed stop and c/t.
Step two: Click only when the dog is on top of the bed (if it is a crate, inside it). I have found it really helps to sit next to the bed and c/t the dog while it is still on the mat, 5 times in a row. You can drop the treats onto the bed, which strengthens the association between bed and click. Then on the 6th click, throw the treat away from the bed so the dog has to get of the bed to get it. Then wait until the dog is back on the bed and c/t on the bed, etc. If you want the dog to do a contact behaviour on the bed – for example, go and sit or lie down – now is the time to introduce it. As they stand on the bed, say ‘drop’ or ‘sit’ and c/t when they do. After few repetitions, throw the treat off the bed and see if they do the behaviour when they find the bed again. If they do not, remind them ‘sit’ or ‘drop’ and c/t. Repeat until they can do it on their own. Alternatively, you can shape the dog to sit or drop once they are on the bed.
Step three: Start increasing the distance. Stand up and take a step back. Does the dog still go to the bed? If so, c/t and take another small step back. If the dog does not find the bed, go back to step one, but stay one step back. As the dog is successful, increase your distance one step at a time. Now is the time to add the cue. As the dog reaches the bed, say ‘bed’ or whatever you cue will be. Start saying it as the dog turns to go to the bed. After a lot of repetitions, you can try saying it when the dog is not about to go to bed, for example when you have been working on another exercise and have just released the dog.
Step four: Increase the time. Your dog should be waiting for the click before they move. Increase the time one second at a time. This is a great 300 peck exercise. This can be much easier if you have taught the dog to sit or drop on the bed.
Step five: If the bed always stays in the same place, keep increasing the distance until you can send the dog from another room, the other end of the house, the garden. If you want to move the bed, do so. First start picking it up and moving it about a metre and see if the dog can find it. If it doesn’t, either throw a treat onto it or c/t for any glances or movement towards it. When the dog can find it, move it a bit more. Start moving it all around the room, then in a different room, in the hallway, in the garden, in the car, in the park.
This can be a very useful behaviour, particularly when you are busy or when you go out. Imagine going to your kid’s sports event and being able to take your dog, safe in the knowledge that he will settle down on his mat and not bother anyone. It can be used in agility and obedience too, useful for teaching send-outs.