Delta is almost a year old. She has a pretty good recall, but I think that there is room for improvement. I bought Total Recall and have read about half way through it (love it by the way). I have read many articles and books about variable schedule of reinforcement, but this is the first book that actually tells you how. I am still using the clicker. Let's say that I am to the point where I want to start incorporating the variable schedule of reward. On times that I do not plan to treat, do I still click and then no treat? Or does a click always need to be followed by a treat? So therefore, I would just not click at all and instead just say "good girl"?
The behaviour should be well established, and no need for the clicker by the time you are ready to fade rewards. The click should always be followed by a treat (or some other reinforcer, although I've found that problematic in practice and tends to always make the click a release which is not that convenient). Many people don't fade rewards at all for things like recall. I don't. My dog always gets some form of reinforcement for a recall when we are training.
I think most people try to fade far too quickly. Ask yourself, what is your dog's motivation to continue, if you are no longer treating? Like Julie, my dogs always get rewarded in one way or another for a recall. As they've got older and I've learnt their preferences more, I have been given more choice for what this reward is, but they still always get rewarded. And a click should always be followed by a reward, yes, otherwise you run the risk of losing the meaning of the click. I don't use a clicker for a general recall, but to capture certain events within the recall, such as a quick turn towards me, the point of acceleration etc. If I were adding a sit at arrival (I don't), I would click that to start with, too. I don't think a click is necessarily useful for recall other than specific instants within the recall, though. If you imagine the click as the noise a camera shutter makes, and you're trying to "photograph" the exact instant you're rewarding, then I would argue that it's rather meaningless to just click because your dog is running towards you. "Good girl", in itself, is also rather meaningless, unless you've associated it with something good, so she has an emotional response to it. When I say "good girl/boy" to my dogs, they lick their lips because it's been associated with treats! Pavlov in action.
10-4, always treat after a click. Don't fade until behavior is well established. Got it, thanks! I will have to read over the posts a few more times because there was some other good info that I will need to review.
You know, I don't really fade inasmuch as I increase my criteria. So, once they know how to sit, for example, then it can become the precursor to another behaviour, such as sit-stay. I'd still reward the stay. Introduce duration, distance and distraction. Then, that sit-stay becomes sit and stay while I walk out and toss a dummy and return to you. Reward for that. Then build in the retrieve chain. Eventually, you can have full and complex retrieves (or whatever it is you're interested in doing) which have a reward at the end. But the sit is still being rewarded, by the opportunity to do something fun. I can't think of any instances, day to day, where I don't reward a behaviour I've asked for, either by an immediate treat/game, or by a real-life reward (being released to sniff) or the opportunity to earn a reward in another way. You don't have to fade. It doesn't make you "better" to have your dog work for free. In fact, the chances are, it will make your dog less inclined to work for you. Why would she work for you for free, when she could do what she wants instead and be rewarded by her environment?
I think there is a lot of confusion about 'fading' treats - it is much better to think of the treats being replaced by something else. You use treats to get a dog to sit in front of a door instead of dashing through it as soon as it opens. Soon though the dog can do that, so can just have the reward of going through the door and the walk starting. Yes, you got rid of the treats, but you didn't get rid of the reinforcer. I work in this way when I can. Say I'm training my stop - I'll provide the reinforcer, the treat, the ball, whatever. If I've stopped my dog 'in real life' dashing to a river, I might then tell him he can go to the river. If I can't provide that as a reinforcer (say I stopped him because there was a barbed wire fence in the way), then he might not get a reinforcer that time but I'd still provide the reinforcers in training. I view that stop without a reinforcer as sort of 'spending' a bit of my training, because the time I needed it work for real, it worked.
Bruce had no interest in treats, so his reinforcers were a butt scratch a 'good boy' or a play with a toy depending on where we were training. His favourite was praise, he loved praise, the more the better. It was very interesting to train a non-foody dog! Mollie will do anything for food! We do fade treats but not until they are older, about six months old, and then never completely - we have to be as random as possible. I have to write down a random schedule or I get predicable in no time! ...
