Saturday, October 9, 2010

Tricks Workshop Part 2: Training is Hard

I'm constantly going between thinking that training is really easy, not all that complicated for the most part and thinking that training is so hard, so ridiculously complicated that HOW do ANY animals learn ANYTHING from humans?

Yesterday at the tricks workshop Griffin and I went to I had a really hard time. I'm feeling a little better about it today but I still don't know what I'm going to do about some of these training challenges.

For our third behavior, Griffin and I were working to stick his head in a wooden box thing. He has a good chin target to my lap and often will offer it to other objects. But primarily when it's at an 'average' height. This meant, that if he was lying down, he would easily and readily stick his head in the box. But if he was standing he would also paw at it (really wanting to do a 2 left (or right) feet position or a 2o2o). I wanted a standing chin touch without a foot touch. Again... we're having discrimination problems.

Part 4: Jumping over legs: Our next behavior was to teach our dogs to jump over an outstretched leg. The training plan the instructor recommended was sit with feet outstretched against a wall (so bottom of feet on the wall), wait for the dog to go over, feed. And repeat repeat repeat. And then raise your legs, sit in a chair, and then stand with foot against a wall and then move away from the wall. Griffin would lie down next to me and put his head in my lap. And do nothing else. After a couple minutes, I cheated, cued him with his "jump" cue, he jumped, I reinforced. And then he would offer it on occasion.

We talked about it. I was helping him too much...and for true shaping I should just wait.

But was that the right choice? If we go back to ROR, I should, for the initial parts of most easy behaviors, be getting 15-20+ opportunities to reinforce per minute. My dog was getting frustrated with so little information (waiting for big steps/jumping).

And then I started thinking about cues. I never do training while sitting on the floor or a chair. I have considered it in the past and decided there's no good reason for it and I want my dog to lie quietly while I am sitting. So is it really a surprise that's what he did?

Most dogs don't need a lot of superpowerful cues. Service dogs and working dogs sometimes have these... do X NO MATTER WHAT, even if cued for something else Luna was really great about this with her Stay cue. She would refuse to do other behaviors if I put her on a formal Stay. I taught her that the stay was the over-riding (ultimate?) response to the cue and NOTHING else was acceptable)

Should I have this "me sitting" cueing "dog lying down" as THE only acceptable response? Will I regret this later? Is it acceptable or should I work to get offered behaviors regardless of my position? How do I even make this decision?

And back to the workshop... I gave up on the sitting and went to a chair and cued it and was going to just try standing and seeing if he would offer, but we did not get a chance for the last step.

The next trick was a paw held on a muzzle. Most of the others were using a loop of tape. I want Griffin to ignore things on his face (muzzle, head halter) so I was going to shape the behavior (lowered head and higher paw lifts) but we were given another challenge and went on to work on paw lifts. Last spring Griffin started to learn to raise his two left or two right paws. He was great. He could do a hind leg lift and hold it for 5-10 seconds. He could do both against a wall for about 5 seconds. He offered the behavior a little last month in CGC class. But his duration is gone. We stopped this after his mysterious injuries in the summer (...Lyme!). So... I've underestimated my dog's current fitness level and we'll go back to more of these tricks and more varied exercises.

Our last trick, we worked on the trick without a name where a dog will do a half spin and then back up between the owner's leg. Luna does this but Griffin didn't. I quickly figured out it would be better for his spin to be clockwise. Counterclockwise turns resulted in him wanting to go into heel position.

Things went well. I was getting the turn. I would then reinforce in position (head away) and he would back up from there, he had that piece previously taught. And then, we had some really bad turns where he whapped me in the legs with his back end. I didn't reinforce. After three times... he lied down and gave his "I am a good dog" look. A few times I "reset" him with a hand target, he would stand and be able to respond a few times, but soon went back to lying on the floor.

The instructor gave an interesting comment at this point. "it is...interesting. I saw him giving good responses. And then he just stopped. It's .... interesting." I'm not sure what was implied by that or what her recommendations were other than to "just wait." Do i have the dreaded "helping addiction" that the Canis instructors mention?

I thought I was good about my shaping, to not have prompts, but do I help my dog too much? Do I need to alter my ROR or criteria more often? How do I effectively measure all of this?

At this point, or maybe a little sooner, it was mentioned that at the beginning of a session we might not know what our behavior is and where it's going. I'm a much more "have a plan" before starting. I have plans for all of my students. Sometimes we alter them after we get started. But I'm regularly feeling guilty for not always writing stuff down first and just "thinking" it. But maybe I need to be a little less careful?

There are so many little decisions that can be made in training. And so many variables!

I'd be interested to learn more about over-riding cues... any resources for this?

1 comment:

Laura, Lance, and Vito said...

I've never taught an overriding cue. I want my dogs responding to my cues, period. I do have default positions that tend to get rewarded a lot more, different ones depending on the context. Actually I do most of my shaping sitting on the floor so if my dogs sense that treats are forthcoming they are way more likely to offer a variety of behaviors with me sitting on the floor, and more likely to just offer nice manners when I'm standing. But in any position, I want them thinking.

I think you need to throw a lot more treats. When dogs get stuck, the tossed treats makes it so I can at least click some movement and hopefully get them going again. And personally while I love pure free shaping, most of my training is shaping plus reward placement, sometimes with the reward as a half lure, in the beginning steps. I would rather have the dog get a high ROR than stick to my gun to purely free shape. I also believe that the more confidence the dog starts to get in shaping the easier the next time will go, I don't think doing a little lure will hurt us much in the future unless that is all we do.