So a combination of a comment from a reader (detailing a similar experience with a lured down being picked up quite quickly while a lured spin required the lure in some form for much MUCH longer) and a 2am puppy potty break led me to a Eureka moment with regards to luring.
Most of the books I've been reading lately point out that with Luring, you are showing the dog what you want it to do. It really takes a jump in thought for the dog to realize that you want them to do it ON THEIR OWN. In addition, the dog is simply thinking about the food, not about what their body is doing. With Shaping, the dog is making conscience decisions trying to figure out how to get the treat. It's like when you give someone a ride home who is telling you which turns to make. You get to their house and they say "So, do you think you can find your way back here to pick me up tomorrow?", you say "Crap, I totally forgot to pay attention". Now, that's all fine and good, but why then did Luring work so well for the sit in the back of the crate in crate games (or reader Pacco's down), but fails so miserably for our spin?
Here's my theory. With both the down and the sit at the back of the crate, the behavior you want is a static position. You might lure the dog into position, and they may not pay attention to how they got there (they were lured), but once they get the treat and the glazed look leaves their eyes - they are in the position you are trying to teach. They are also getting the treat in that position. So even if they aren't paying attention during the luring, they may start to pick up on the common ending position.
Now look at a spin. The action they are taking during the luring (when they aren't paying attention) IS the action you are trying to teach! In this case, when that glazed look leaves their eyes and they finally get the treat, they are simply right back where they started.
I believe this is the key difference between how well Luring works for some behaviors and how hard it is to fade the lure for others. Based on this, if I could get better at shaping, I would endeavour to use shaping for any behavior that is an action; and would consider luring for any behavior that is assuming a position.
Which leads us to part 2 of this blog post. Turns out I might not be as horrible a shaper as I thought. I just need to learn to have faith in the process and faith in my training. After giving my dog a head twitch yesterday, I decided to try one more session on shaping a spin. To my delight, she immediately started her head twitch in a more vigorous way than I had left it, often ending up with her body following her head a bit! After a few clicks of this, she actually turned so much that after getting her treat, she just finished the turn rather than turning back the same way to me. JACKPOT. While she never intentionally completed the spin, it did happen at least twice on it's own. I'm very interested to see what today brings. It will greatly boost MY own confidence if I am able to successfully shape a spin.
Friday, September 05, 2008
Subscribe to:
Post Comments (Atom)
10 comments:
Could the initial issue have been the placement of the reward or the timing of the click? If I were teaching my dog to spin by shaping, I'd try very hard to click when the dog's head is turning away from me and hasn't yet started to turn back. And then the reward would follow as fast as I possibly could do it, several inches further along the path of where the spin would take the dog's head--not where the dog's head ended up on its own, and not back towards me.
-ellen
Oh, which is to say, you're on the right track--if you reward the spin like it's a static position (e.g., reward where the dog's head ends up and not further along the path), it becomes a static position, whereas putting the reward further makes the result a moving position.
I essentially shaped it as you pointed out. Click for intent - feed in position. And you could do that with Luring, but spinning especially seems very difficult to fade the lure. They view spinning as "I follow your hand to get the treat". This morning she basically got it. She's kindof walking in a circle, now I need to tighten the circle and speed up the spin!
I was at a Susan Garrett seminar a couple of months ago and she went into great detail on the intial stages of crate games. She does not consider the treat at the back of the crate to be a lure. A lure would be holding the treat up to the dog's nose and leading him to the back of the crate whereas for crate games you're supposed to hold the treat at the back of the crate and reward when the dog chooses to go there. Now, you can argue whether or not that is indeed luring (and some people I talked to after the seminar certainly thought it was) but that is her explanation. In my mind it's something in between shaping and luring but there you go.
Also I found it interesting that she said she rarely uses a clicker anymore because your timing has to be so precise or you will click the wrong behavior. She didn't use one for most of the shaping demos.
There's a lot of art involved in shaping and it takes some practice and a lot of patience but I think it's fun.
Yup, I love it! And I'm finding that, over time, I'm being more selective about when I use the clicker, too (sometimes not at all, as you say).
Elayne -
What was the focus of the seminar you attended? I'm serious considering the Puppy Class 3 day workshop at Say Yes this winter. Quite a large chunk of change though, esp with airfare!
It was her 'Advances in Dog Training' seminar. I think she covers many of the same concepts in her puppy workshops (she referred to them many times during the seminar). I thought it was absolutely worth the money but I only had a 1 3/4 hour drive. It was really interesting though and I learned a lot. My brain was melting by the end of it, lots of info. to absorb.
Reading Elayne's post made me realize that it's been quite some time since I watched my Crate Games DVD. I had forgotten the exact mechanics in the early stages. The reward is there in the back corner, but it's hidden in your hand (that's very important--the sight of food is very reinforcing), and the dog has to make a choice to earn it. I think that's the key: what does the dog have to do to earn the reward? In another game, Susan teaches self-control by hiding a treat in her closed fist and keeping it closed until the dog backs away. Then she opens her hand, but the dog can't have the food unless it holds position or backs away again. The fist closes if any other response is offered. That's not luring. In the same way, I can understand her arguing that a closed fist in the back of a crate isn't luring either; it's just that the response that earns the reward happens to be the dog walking into the crate toward that fist. If the first example isn't luring, then I don't think the second one is either.
Anyway, one last tidbit on luring vs. shaping. In Shaping Success Susan has a section in the beginning of the book called Pavlov on Your Shoulder (I think--I can't find my copy right now). As we all know, the bell predicted the reward and caused the dogs to salivate. If the reward was given first and then the bell was rung, the dogs didn't salivate. And duh, why would they? There's no value in the bell being rung if it doesn't predict earning a reward. Susan says lufing has the same effect. It doesn't help the dog think or build value for responses.
In general, shaping takes time and patience, but it's a whole lot more fun and, perhaps most importantly, it creates a dog who isn't afraid to be wrong. This is especially important if you have a soft dog.
Like the others who posted here, I'm also using my clicker less and just "chucking food" (as Bob Bailey says) to reward. Or toys!
I think your insight into the possible reason why luring would work for a static behavior and not so well for an active one is brilliant! I also have stopped using a clicker. It's too hard to hold it, food, leash and I hate when I click for something I shouldn't have.
I'm curious why you seem have your puppy in an x-pen while you're doing the shaping. It seems like a really good idea--did you come up with that, too?
Aren't you the flatterer! Necessity is the mother of invention. I thought to shape in the xpen, because she was a VERY mouthy pup! She would just spend her time biting my toes, feet, pants cuffs, etcetc. Finally I put her in the xpen so she'd do something I could click!!!
Post a Comment