Clicker training, just starting out. Help!

[Camilla and Darwin]

I just bought a clicker and "charged" it for the past two days with Darwin. Now he understands that it is a reward, how can I get him to offer behaviors? He seems to only offer behaviors I have already taught him. Down, Sit, Say Hello, etc. If we are just barely starting, should I be rewarding him for these behaviors? (I heard you should start with completely new behaviors, not ones they already know) If not, how Do I get him to offer new ones? I understand you should reward little tiny behaviors, like shifting, etc. But all he does is sit and stare directly at me?

[Sue, Murphy and Bella ()*o*()]

I forget where I got this but you are welcome to use it. I started clicker training with some of these tips.
--------------------------------------------------------------------------------

21 Tips for Effective Clicker Training


1. Effective training requires effective reinforcers. Establish a large selection of effective reinforcers, preferably while your dog is still a puppy/adolescent. Take care not to ruin your dog’s appetite, and emphasize teaching your dog to play with many different objects. So-called ”high-drive” dogs and the classic ”food hounds” are the easiest dogs to (clicker) train.

If’s also great if you can condition an arsenal of reward substitutes (anything from clapping your hands to ruffling the dog’s fur, and so on, can be effective reward substitutes if you associate them repeatedly with food or play). And not least – teach you dog to work for things he wants in the environment. Then you will even be able to use things others see as distractions as effective reinforcers.

By working systematically on developing many different rewards and reward substitutes you won’t have to get stuck with only using click + treat.



2. Socialize your dog well, teach him how to be around other dogs, and make him used to a bit of rough, uncomfortable handling by both you and other people (he should learn to like this!). It doesn’t matter how well you clicker train your dog if he spends a lot of his attention on trying to control the environment.



3. When you want to improve your training or solve a problem you should first examine 1) timing, 2) criteria, 3) rate of reinforcement and 4) quality of reinforcement before you look for more advanced solutions.



4. Keep you training ”clean”. Avoid ”sloppy” training (luring, nagging, repeated cues, hands in your pocket before the click, treats in your hands, smacking your lips, bending over the dog, visible treat bag, reward hand held in front of your belly/chest, low rate of reinforcement, unplanned criteria and so on and so forth). Practice your mechanical skills. Video tape your training if possible, attend classes with a skilled clicker trainer – or join a Chicken Camp!



5. Start early with building a large file of offered foundation skills. you don’t necessarily have to train them ”perfectly” right away. You can do that later, once you have quickly touched on a large number of behaviors so that your dog has become ”creative”. THEN you can start establishing good fluency with the behaviors.



6. To develop the behaviors to ”fluency” is crucial to effective clicker training. Clicker trained behaviors without fluency are usually worthless in the real world (and in competition). Start training new behaviors in a distraction-free environment. Establish fluency for the behavior before you take the behavior out into the world and really generalize it.



7. Always use a high rate of reinforcementin early training stages. Ideally the dog should be earning a click at least every three to five seconds (depending on which behavior you’re working on). It is not good clicker training to stand around waiting for a long time before the dog is finally able to offer the correct behavior.



8. If the dog performs correctly for at least 80% of the repetitions you can increase the degree of difficulty. If the dog performs correctly for less than 80% of the repetitions you should consider temporarily lowering criteria. This yields a quicker progression and a more reliable behavior in the end (you avoid mistakes attaching to the behavior’s ”record”) But remember that the 80%-rule is only a guideline, not a law of nature!



9. Use a higher quality reinforcer when you’re training new behaviors, difficult behaviors, strenuous behaviors, or when there are a lot of distractions around.



10. The most important thing is to click accurately, but remember that the dog keeps learning after the click as well. With some behaviors you can achieve faster results by being conscious of how you deliver the reward (not only of when you click). This is called ”strategic delivery of reinforcement” or ”reward placement”. This can be particularly useful at the beginning of training. But don’t get dependent on a particular reward placement forever. On the contrary, once you’ve got the behavior, you should start to proof it by using e.g. reverse luring, distance rewards and so on so that the dog is able to perform the behavior correctly no matter where the rewards are coming from.



11. My ideal for training new behaviors is 1) as little help as possible 2) as few errors as possible 3) as quick a progression as possible 4) as little correction of mistakes afterwards as possible. Choose a training technique which will help you to achieve this. Use free-shaping/capturing as your plan A. Adapt the environment if necessary so that the shaping process will be easier. Consider using targeting if shaping is difficult and leads to a lot of errors or little progress. Luring is only to be used as a last measure, and should then be removed as quickly as possible.



12. Pavlov is always looking over your shoulder. Remember that e.g. the intensity level is conditioned along with the behavior whether you want it to be or not. It can therefore be a good idea to train typical high-intensity-behaviors (such as recalls, jumping and so on) when the dog is fresh and full of energy, while you train low-intensity-behaviors (long down, crawling and so on) towards the end of the session or after a long walk, when the dog is calm and relaxed.



13. As soon as the behavior is fluent you can put in on stimulus control (if you need it for that behavior). But don’t lose the offered behavior completely. It is always nice to be able to go back if you need to brush up the behavior / cue later.



14. Overtrain foundation skills and other behaviors that are important to you (this means establishing extreme fluency and training with more difficult distractions / variations than you really need. Then you will always have a margin of error in case you need it.



15. If you want to progress as quickly as possible, train frequently, every day if you can. If you think it’s more important to get the maximum benefit out of every hour you train, you can instead let a few days pass between each time you train a particular behavior. Dogs seem to learn during breaks, as well... And not least, you often get much more out of training when the dog is well rested. If you want to train a lot your dog should be allowed to get used to this gradually – and you should have a breed/individual which is suited for large amounts of training.




16. If you want to clicker train effectively, don’t mix punishment and luring with your training. Punishment and luring work if you use them correctly, but they always reduce the effect of clicker training if you mix them with the same dog.



17. Train a new behavior you haven’t trained before as often as possible (or train an old behavior in a new way, or with some other animal!) That’s the best way to keep evolving as a trainer. It’s not so important whether it’s a trick or a useful behavior, but that you progress as a trainer.



18. Clicker training is a young training method in the dog world, and continually developing. Knowledge gets old quickly, so keep up-to-date. There are good resources to be found both in books and on the Internet. Stay in touch with other clicker trainers. If you don’t, it’s easy to stagnate, because the development is rapid.



19. Clicker training can be made very advanced for those who wish it. Clicker training is based on science which takes years of full-time study to learn. But don’t let that scare you! The fact is that if you only master the essential main principles well, you can teach your dog practically anything! (leave the deep-down theory to the ”clicker training nerds” :-)



20. Don’t just clicker train your dog. Bring the principles of learning into your daily life as well. Use positive reinforcement with your children, your spouse, friends, colleagues, waiters and parking guards. It could change your life. And if you work a lot with teaching people – learn Tag Teach!



21. If you discover any new training method which is more effective than clicker training (and still ethically sound) – switch to that method the very same day, and never look back!



Acknowledgements:

Bob Bailey, Karen Pryor, Ken Ramirez, Kathy Sdao, Morgan Spector and Cecilie Køste - and MANY other good trainers I’ve met or had internet discussions with!


Morten Egtvedt

(Chief Instructor, Canis Clickertraining Academy)


--------------------------------------------------------------------------------

About the authors
This email course is made by Morten Egtvedt og Cecilie Koeste, chief instructors for Canis Clickertraining Academy and authors of the bestselling clicker training book Clickertraining: The Four Secrets of Becoming a Supertrainer.

The book is now available for instant download and comes with four excellent bonus videos where you can see how Morten and Ceci is training "live". Do you want to learn more about the book?

[Adina P]

Morten is the author of the book we used for our DK Clicker Training group....
http://www.doodlekisses.com/group/doodlespringtraining2009

[Adina P]

The clicker is NOT a reward. The clicker is a marker so the dog knows "Right this second [moment he heard the click] I did the PRECISE thing that gets me TREATS!" That is a very important thing to distinguish between...the click is a marker of a coming reward and of the fact the dog got "it" right. I think later on it can become a sort of reward when the dog is really clicker savvy..but it's not a reward in itself.

The way you start to get offered behaviors is to set aside a few minutes a day to reward random things...hard to explain. Let's say your ultimate goal is to get your dog to learn to touch a mouse pad and hold his paw there. Well...you toss the mouse pad on the floor and you reward APPROXIMATIONS to the ultimate task. So first just LOOKING in the direction of the mouse pad gets some clicks&treats...then as SOON as he demonstrates that he kind of gets that LOOKING in that direction is the right thing you don't click until he gets CLOSER to the mouse pad. Then for walking near it...etc.

So you can pick a goal to hold in the back of your mind to work toward in a session (you may or may not GET to the goal) or you can just kind of see what he does and mold that ... again just to get him TRYING stuff. You might even change your mind about your goal if going with the flow seems to lead in a different direction. The more stuff you can get Darwin to do without YOUR intervention, cues, clues, etc...the BETTER he'll be at clicker in the long run. IT is VERY tempting to try to "help" him or give him verbal clues or help him do the task physically, but in the long run your dog won't be as good at clicker stuff as he will be if you zip your lips and sit back and go with the flow as he learns a few things.

We had a clicker group training we did here on DK. If you go to the links provided you can buy the course for about $50--but try the FREE 7-day trial first to see what you think about the way this couple teaches via their ebook. Also clickertraining.com - good luck!

[Azure & Muppet]

Adina's response is great. She's totally correct: the click is a marker, not the reward. "loading" the clicker is accomplished through classical conditioning. Now that you've made the association by pairing the click sound with a reward, the next stage uses operant conditioning. Sometimes you might get lucky and the dog might offer the complete behavior, which is wonderful because you can "capture" the behavior in one step and it saves having to take lots of little steps to get there. But most of the time you will need to "shape" the behavior by reinforcing smaller approximations towards the goal. Don't give up too quickly. It's one of the best and fastest ways to train a dog (or any animal). It's a very scientifically sound method because it is based in the fundamental laws of behavior. It's also the foundation to more advanced work (such as agility). I first studied, learned, and applied these principles with lab rats in graduate school. There are many great books on the topic, and I can recommend some excellent resources if you're interested. There's also a group here on DoodleKisses that may help: http://www.doodlekisses.com/group/trickandtreatakashaping

[Azure & Muppet]

Of the dog isn't "offering" a behavior on it's own, you can use other techniques such as "luring" to elicit the behavior. Do something to help them perform the desired action, then click to mark it.

[Allyson, Peri & Taquito]

We use the clicker in the exact way Adina says - it marks a correct behavior so they know a treat is coming. Peri already knew sit, stay, etc...before we started clicker training. We used it with those commands, but made her start staying longer, doing longer sequences (like a sit-down-stay-sit-down, etc...).

Just so you understand, sometimes the clicker isn't handy and you can use your voice to mark also. We use a high-pitched "YES" when she follows a command - to mark it without a clicker. She knows a treat may/may not follow (now that we have done training for a while, you can vary the treat giving).

[Natasha]

Hmm, I have never heard of only using a clicker for new behavior. From our training perspective, the clicker is used to reinforce immediately that your dog has done what you asked. So, if he sat when you asked him to, then he gets a click followed by a reward (treat, "good boy" etc.) So, no need to hold off on clicking for existing behavior as it kind of reinforces the charging that you've been doing. My opinion, anyway.

I think I'm a little less of a click trainer than some, so I'm not a huge fan of offering. I find that doodles are quite intelligent and they work really hard at figuring out what you're looking for (via your body language, how you hold your eyes, etc.) so you can kind of plan your lessons around behavior that is natural and reflects the interraction between the two of you. I think about what behavior I am ultimately looking for and then I do break it down into a few steps, but I always keep the end-goal in my mind as we're training. I think that he actually picks up on the long-term goal if I'm focused on it so that we don't get caught up on two many steps (or offerings) in between. I did more offer training with other dogs that weren't quite as quick to pick things up as Rouser and it was a bit more necessary but also a bit more time-consuming (I gues I'm trying to say that I don't think doodles need as many steps to get to the end behavior).

For example: "Go get your bunny." Give him the bunny while saying "bunny" and click the second he gets it in his mouth. We he's got that, then put the bunny on the ground in front of you, say "bunny" and the second he picks it up, give him a reward. If he doesn't pick it up, pick it up for him and say "bunny" or "get your bunny" and click. Once he regularly is picking up the bunny, then throw the bunny a few feet away and say "go get your bunny" and reward him when he picks it up. Then, put bunny further and further away until you can sit on your couch and say "go get your bunny" and he'll search to find it. Using the "offer" technique, you'd have to catch him getting his bunny and, ideally, you would click and say "go get your bunny" every time so that he clearly understood this was a command. That seems pretty difficult. When I did this with Rouser he was getting his bunny from the other room the day we started the training (and, to be honest, I didn't use a clicker, just chicken, because getting a toy was its own treat - as was the game we were playing).

Where I have found offering extremely helpful has been on teaching and reinforcing "heel." The second he gets into the heel position, clicking and saying heel is definitely a benefit. It disrupts your dog's focus (which is on the walk) and gets him to look up at you and you can reinforce with a "good heel" or a treat. If you do that every time he gets himself into heel, after a few days, he'll get what that means. Then teach him "take a break" so he knows that he can go into sniff mode and your walks will be great.

[Allyson, Peri & Taquito]

This is how our trainer taught us Natasha.