When I was teaching, I remember being given advice at the beginning of the year that I should be really strict/clear with the rules to start. That if rules were broken or behavior was not what was expected that I needed to point out the mistakes and set very high standards. If I was strict/clear during the first few months, I would be able to relax the rules once I had good compliance, but if I were loosey-goosey in the beginning, I would struggle to pull them back once the kids gained bad habits.
I’ve sort of taken this rule with me as I transitioned to training. When I’m first teaching a behavior I am very strict with my criteria. Ultimately, the stricter I am, the clearer I am in communicating what is being rewarded. I try not to allow the dogs to be rewarded for being “close” (when I’m not shaping a behavior)… I do not want to fall into the “if you give a mouse a cookie, he’ll ask for a glass of milk” type trap.
So what does this mean in terms of dog training?!
Well, let me give you an example. I was teaching class the other night and I had everyone working on hand targeting–this behavior is not entirely new but is far from perfected. I was watching one of the teams and watched the dog give two good hand targets and he got rewarded but then he made the motion like he was about to touch the hand but never did and he still got rewarded. He did this a few times, getting very close to the hand but not making contact, and each time he was rewarded. I pointed out to his handler and what do you know, on the very next attempt, he made an even smaller move to hit the hand. She gave the mouse (her dog) a cookie by rewarding him for almost doing the hand target and he began to ask for a glass of milk by making less and less of an effort to ‘fake’ the hand target.
So, to prevent the further degradation of the behavior, I instructed her to hold to her criteria. What is “touch” for her… does it mean “hit my hand with your nose” or “get close my hand with your nose.” Once she decided what she wanted “touch” to mean, she had to hold her dog to that criteria. So she cued the hand target and the dog fake-touched her hand with his nose, I instructed her to wait him out. Sure enough, 10 seconds after not being rewarded for his fake touch, he relented and touched her hand with his nose. Her dog was really starting to see what he could get away with… and frankly, I don’t blame him, she was paying him regardless of his performance.
The stricter she is with her criteria at the beginning, the clearer and stronger the behavior will ultimately become. Once a behavior is really well known and is proofed it’s not the end of the world if you occasionally reward for a behavior that doesn’t entirely meet the criteria. Just the other day, I was doing some hand targeting in the store–we were moving at a rapid pace and one of the “touches” didn’t quite hit my hand but I had marked it and had to reward. With a dog who knows the cue very well, it’s not as detrimental to reward a behavior that hasn’t met the criteria, Shayne isn’t going to take that one incident as a cookie and begin to ask for a glass of milk.
One of the cues this is most important for is “stay.” When I’m teaching a “stay” behavior, I consider any change in position a break in the stay and not rewardable and a sign to reset. If you reward your dog for a stay and he has scooted two inches toward you, over time, that two inch scoot that you rewarded could easily become two full steps toward you. If, however, you keep your criteria strict/clear and only reward for stays where the dog hasn’t moved a tall, you set yourself up for having a solid stay.
If you keep your criteria clear and consistent, you will end up with more reliable behaviors than if you let things slide, especially early on in the training. You may have to be strict in the beginning and not just reward the “almost” attempts (when they are capable of completing the behavior) but you will be rewarded in the end with a dog who reliably completes the criteria you set and doesn’t necessarily try to push buttons to see how much they can get away with.
Great post, but the comment that the student’s dog was “starting to see what he could get away with” needs more of an explanation. Could it be that the dog, after being reinforced for not touching, thought that that was what he needed to do?
The great thing about clicker training is that even if we confuse our dog with bad timing , or reinforcing the “wrong” behavior, is that it’s really easy to reset once we have determined what we really want, and our dog isn’t afraid to try again. Sounds to me like the dog was a little confused by the change in “rules” and was just experimenting a bit.
Thanks-I really love this blog and look forward to reading every post.
Thanks Pamela, to clarify, I do think the dog was testing the boundaries of what he could be reinforced for…but i think it’s a difference in semantics… i think you were correct in your use of the word “experimenting.” Once the rules changed (that an almost touch was rewarded) he began to see what he could offer but still be reinforced for… just how serious was his handler that he had to hit the hand. Given his response once she held to the original criteria, and considering his previous skills with this cue, it does lead me to believe that he had learned a lazy habit of “almost” touching and was seeing/experimenting with what he could get away with and still be rewarded (even less movement toward the hand)–he had a pretty clear understanding of the cue once she stopped rewarding for the ‘almost’ touching.
Great point. Karen Pryor makes the same note herself in one of her books. If you do competitive events, such as obedieance, with the AKC, they are very strict and even if your dog is a little off they will take away points. I’m a pretty demanding person, (lol) but even sometimes I fall into the mistake of rewarding half-efforts.