Friday, May 09, 2008

Don't Reward the Correction

Reward the correct behavior and not the correction. Something I have been thinking about. I think we do reward the correction a lot and need to be very careful about that. Example is when Wyatt pops up on his down stays (like this morning - frustrating!).

If I put him back in a down and finish the down stay and then reward him, I am probably rewarding the correction and may be unintentionally rewarding the incorrect behavior - exactly what I don't want. That is pretty clear. So what is a poor trainer to do instead?

I can finish the down stay and praise but not treat him. That is what I usually do. Or I can stop, do something else, and try again later.

How would you handle this situation?


Stephanie and the Dogs, Zoie & Quiz! said...

I don't think that's rewarding the correction at all. What is the dog doing at the time of the reward? If he's in a down, you're rewarding the down. The only thing to watch, IMO, is that you're putting a gap of time between the correction and the reward so you don't create a chain of get up > dad puts me down > get a cookie b/c then getting up leads to a cookie.

Stephanie and the Dogs, Zoie & Quiz! said...

Also, I think that even if you finish the down and praise but not treat, you've still rewarded the down anyway with the praise.

My guess is that the biggest problem is that Wyatt has learned that getting up on the downs is a sure fire way to get you back from out of sight... which is what he wants, and a dog will always continue to do what works. That's why I think it's so important to train OOS groups with a helper so that it's only remaining in position that brings the owner back.

John Heffernan said...

Your point about there being time between the correction and the reward makes me wonder. I will have to think about that. Thanks.

In terms of me coming back, I don't come back. He pops up in sight too. Actually, he most often does it when I am leaving and have my back to him. Also, I don't go back to him but give him the down signal at a distance. Good suggestion but I don't think it applies in this particular case.

Katrin said...

Hi John

I acutally just had a discussion with a student in agility class about this recently. I had noticed the her green dog would do equipment very confidently, then start hesitating more and more needing more and more 'help'. the dog had figured out that if she does it 'right' she doesn't get 'help' (and therefore less rewards and cookies) but if she falters and does it 'wrong' the handler was unconsiously rewarding more. So the handler was actually rewarding the dog for needing help. If that makes any sense?

What I talked to her about doing was break things down a bit more and reward the dog when she was Right the 1st time, thereby not reinforcing the 'needing help' inadvertantly. (this particular dog is super smart and I am fairly confident that this is what she has learned- if I hesitate and falter, handler will 'help' me and that gets me more attention and rewards in the long run than just doing what the handler is asking the 1st time)

What I would probably do, in your situation with one of my dogs (I don't know your dog, so not sure it would work) would break things down and actually get out a stop watch and see if I could figure out if there was any sort of consistency and pattern to when he broke the stay. Then I would be making a point to go in BEFORE he broke to reward. And over time, make the time I went back to him more and more delayed by very small amounts of time until I had the dog staying for how ever long in 1 position which to me in the perfect world would be- you don't get up until I come back and release you, could be 3min could be 4hrs.

Another idea would be, if he breaks, heel him out of the stay and just reapproach it as an entirely new 'stay'. So he breaks, rather than recue a down from a distance as you are currently doing. I would probably walk back to him, quietly, no praise no cookies, release him quietly as unemotionaly as I could. Heel him forward then back to the line, redown him, recue stay and walk away. There by doing a 'reset' in his head that 'oh, breaking doesn't get you anything fun and you have to do the ENTIRE thing all over again.' And I would probably not go back when the rest of the group went back if you were in a training class, but stay out of sight for a bit longer, then go back and reward my dog (I'm pretty sure this is what I did with Niche when he was having an issue with the out of sight down, and he really started to get the full concept once I did this a few times)

Good luck!

Pacco de Mongrel said...

i'm not sure did i consider as rewarding d correction.

what i did was, when my dog break d 3 minutes stay, i pop on her leash and then made her repeat the full 3 mins again... when i return to her, i reward her while she's still maintaining the down position. never made her sit up and then reward.