Clicker Training

What is a Clicker?
 

A clicker is a small box with a metal strip in it. Pressing the metal makes a click-click noise (it clicks once when pressed and once when released). Clicker training is used to train all kinds of animals, from dogs to cats, killer whales to birds. The actual sound used may vary, but the technique is always the same. Even fish can be trained using a light as a marking signal instead of a click! See this video for example: http://www.youtube.com/watch?v=b3JFmrlgWAk&feature=related

The Fundamentals

1.       Every click is followed by a food reward (within about 2 seconds). Sometimes you’ll press the clicker by accident; you’ve still got to give the dog a reward because the point of the clicker is that it signals three things: ‘That behaviour was correct, the behaviour is finished, a food reward is on its way’.

You only need enough food for the dog to taste it. To give you an idea, with my German Shepherd Dog I use liver treats that are about the size of a piece of kibble. My Chihuahua gets a piece of cheese that’s roughly 2mm cubed. The reasons are simple: you don’t want your dog to get fat and if you’re training with large pieces of food the dog gets full faster and you can’t do as much training!

It’s important to work out what your dog will work best for. Cut up hot dogs, cheese and chicken are common favourites. Kibble isn’t particularly exciting to most dogs, Labradors often being the exception. Training before mealtimes is beneficial for obvious reasons.

2.       Timing is everything. The reason clicker training is so successful is that the marker (be it a click, whistle or light) is a very obvious and fast marker that the animal doesn’t see/hear at any time other than when training. How many times do people say ‘good dog’ in a day when your dog isn’t doing anything in particular? And if you don’t say it any time other than when training, think how much your tone changes each time you say it, and how long it takes to say. In the time you’ve said ‘good dog’ your dog could have performed several behaviours.

Of course your timing with the clicker has to be good. If you’re trying to click when a dog sits, you have to click the moment his bum touches the floor. If you click when he’s got up, moved a paw, or twitched an ear you haven’t rewarded sitting (yes, you can reward behaviours that precise). You can practice your timing by throwing a ball in the air and trying to click when the ball reaches its highest point (I suggest you don’t try this around your dog however!).

3.       Ignore all behaviours that you don’t want. There is absolutely no need for punishment. One of the best things about clicker training is that it encourages the dog to be creative, and to try different behaviours. Start punishing things you didn’t want and the dog’s going to shut down and give up pretty fast because he’ll be worried about making mistakes. Punishment is never conducive to learning. Animals are smart; they don’t expend energy if it doesn’t obtain a reward so behaviours that you don’t want will disappear if you don’t pay any attention to them.

4.       Introducing the cue word. Once the dog is performing the behaviour you want reliably, you can introduce a cue word just before you know the dog will perform the behaviour. Remember to just say the word once or the cue will become something like ‘sit, sit, sit, sit’!

After a few repetitions of using the cue before the behaviour, and the dog getting the behaviour right, you can stop using the clicker altogether and just use praise/toys/treats as rewards each time. If your dog stops getting the behaviour right, you’ve gone too fast and you’ll need to revert to using the clicker.

5.       Reduce the rewards. When the dog is consistently performing the behaviour on cue, you can stop rewarding every time and instead randomly reward the behaviour. It’s important to remember that all behaviours you want should always be randomly rewarded from time to time with different types of reward (praise/toys/treats). It’s essentially the same way that gambling works; you don’t always get a reward but occasionally you do so you keep trying. If you got rewarded constantly it’d be like a vending machine and pretty boring; so you won’t have to have treats on you in order to get your dog to do something, but randomly rewarding ensures reliability and keeps the dog interested.

6.       Train little and often. Several five minute sessions a day are far more productive than one hour-long session. There are many opportunities in the day to train. E.g. when waiting for a kettle to boil or during an ad break on TV.

7.       Start all training in a non-distracting environment like the house, then move to the garden, a quiet area of the park, and finally somewhere with a lot of distractions. You’ll probably have to go back to basics each time you increase the distractions because dogs don’t generalise well but once you’ve trained in lots of places the dog will reliably perform the behaviour anywhere.

8.       ‘Shaping’ Behaviours. Shaping is when instead of clicking the final behaviour you want (e.g. sitting) you reward each tiny step towards what you want. This is useful for behaviours that are unlikely to be performed naturally. For example, to teach a dog to get your slippers you can click/treat a head turn in the direction of the slippers, a paw being moved, a step in the right direction, any interest in a slipper, nudging the slipper, picking up the slipper, holding the slipper, moving it towards you, and finally actually getting it back to you. This can take some time initially, but dogs love these games and once they understand shaping it’s surprising how quickly you can teach a new behaviour.

To see shaping at work, watch this video from a great trainer I know, Mark Sanderson: http://www.youtube.com/watch?v=dSVo_ccqpPw&playnext=1&videos=M-iAy1ImX9U&feature=mfu_in_order

Here’s a lovely video of one of Mark’s other dogs playing butler: http://www.youtube.com/watch?v=vlIJNfQe8Vc

If you’d like to know more about clicker training, please visit http://www.clickertraining.com/ or leave me a message. I also recommend Karen Pryor's book: Getting Started: Clicker Training for Dogs

Happy training!

The Dominance Myth

I have dominant dogs. As a dog behaviourist and trainer should I be admitting that? Well, if the definition of dominant includes behaviours such as being allowed on my sofa, walking through doors before me and eating before I do, then yes. However, my dogs are also willing to move if I ask them to and are obedient and friendly. So why do some people believe that in giving dogs such privileges you create problems?

Dominance theory has been around for a very long time. I grew up with it so it’s even taken me a while to get my head around the new concepts. The idea is that dogs are domesticated wolves, so we should treat them like wolves. However, that’s like us looking to chimpanzees for parenting advice! Dogs have evolved. Studies show that if we’re going to relate dogs to their ancestors at all, we should be looking at juvenile wolves, which don’t form hierarchies and therefore could not be classed as dominant. Interestingly the studies that led to these theories were also done on captive wolf packs; they didn’t form naturally and didn’t behave like wild packs.

Even if dogs could be dominant, I’m not convinced that behaviours such as walking ahead of an owner could be determined to increase their rank in the household (and that assumes that we are seen by our pets as strange-looking dogs to be competed with!). If it were, why do dogs pull on their way to the park, but walk behind us on the way home? Anyone else think that maybe they’re just excited about getting to the park? And I’m sure dogs want to lie on sofas because they’re comfortable, not because they think they can be leaders if they do.

Traditional dog training for ‘dominant’ dogs usually involves punishment to ‘show the dog who’s boss’. This kind of training was on the decline, but has seen a resurgence with shows such as Cesar Millan’s The Dog Whisperer becoming popular. The trick is to ignore what the trainer is saying, and watch the dog during training; is the dog happy or does the dog look miserable? Even aggressive dogs can be retrained using rewards. Punishment can actually cause further aggression as dogs try to defend themselves (Millan has been bitten several times on his show). Additionally, dogs that are punished for something like pulling may associate something else, like a passing dog, with the punishment and then develop a fear of that.

So, if a dog that is aggressive or disobedient is not ‘dominant’, what is he? Simply untrained! These are dogs that haven’t been taught manners and it is our job to teach them. The dog who growls because he is told to move from his favourite sleeping place has just learned that this behaviour works – the person leaves him alone, and he has no incentive to move. Behaviours such as this are easy to retrain with positive training methods.

If your dog is aggressive please consult a professional and ensure you are happy with their training methods before you let them handle your dog. It is also worth consulting your vet for any sudden behaviour changes, including aggression; the problem may be medical.