Tuesday, June 09, 2009

When CU goes wrong, or, my nutty dog :-)

I put Shakespeare in a crate with a chew (which, offended, he did not deign to eat) and Inky in another with a chew (which she ate) and kept Laev out for a training session. I had two exercises in mind.

1) open her mouth, with a big "say Ahh" movement
2) relax on a mat in preparation for gunshot desensitization

I'd started the open mouth idea a full year ago, but I'd not worked very hard on it and hadn't kept it up. No reason to. Now, however, I suddenly need to fill more time at our demos this weekend, and Laev needs a cute trick. So back to the open mouth game. I'd done one session on it last night, just enough to remind her that jaw movement works for clicks. (It's very hard getting jaw movement with no vocalization!)

It would be very wrong to say that Laev doesn't do much with her mouth; Laev is quite oral. But she doesn't lick or kiss or pant like most other dogs on the planet. Seriously, I generally see her pant only during summer bitework or after mile 10 on the AD. It's not even a common stress signal for her. So it was bizarre when I sat down to start our open-mouth session and she was panting.

Not really panting, after a moment. Just sitting there with her mouth open. Did she actually remember the open mouth? I didn't think so; she was just "stuck" that way. This was not as good as it sounded -- I couldn't click her opening her mouth!

So I abandoned that project and went to the other side of the couch, where I set out a mat. Laev parked promptly, but I wanted to shape her into relaxing. I clicked for chin down, etc., but she was faking. She wasn't relaxed, she was working the click system.

It took a long while before I could click a hip flop. As I clicked, she immediately popped back into a sphinx down. Repeat. Repeat. Repeat. She tried the other direction. Cool! Mom will click either hip flop! Watch me work both of them!

Finally I got her stable for a few seconds. She looked at me, lying on one hip, and gave me a big open mouth. And another. Way better than before.

I don't want to click the wrong behavior in the wrong place. Back to the couch and clicking for open mouth.

Laev started getting the open mouth, offering it more regularly. (Never as good as on the mat!) She had some superstitious body movement too, but I can live with that; it's just a silly trick. I decide that we're not going to have time to finish the full open mouth and hold before Saturday, so I'll go with an open/close/open/close movement and call it something to do with "goldfish." :-)

Back to the mat, on the other side of the room. Laev starts working the hip flops, never actually relaxing, just trolling for clicks. I stop clicking hip flops and click only what can be accomplished with muscle extensions -- legs extending, head lowering, ears relaxing, etc. In theory, this should relax the dog.

I just had a FABULOUS session yesterday with a dog in this. This fear-aggressive boy used to aggress at dogs across the street; relaxed on his mat, he was able to lie quietly and calmly while Shakespeare did happy treat dances back and forth about 15' away. I was thrilled with his progress, and in theory I know the concept of shaping relaxation on the mat.

But not with Laev. Determined to make me click, she started throwing everything she could think of at me -- crossing and uncrossing her front paws, flopping from one hip to another, raising and lowering her chin, and opening her mouth repeatedly. ALL AT ONCE. She looked like some sort of demented Rube Goldberg device. I couldn't help laughing, but we were not getting relaxation on the mat.

Finally got an instant of stillness, clicked and threw the treat off the mat, and let her reset. Clicked and treated for stillness. Not relaxed, but at least less like a steam engine about to explode.

Back to the couch and the open mouth, where I started adding a cue. We don't have the behavior anywhere near stimulus control yet, but I think it'll be good enough to fake for Saturday's demos.

Back to the mat. I settled for clicking for a hip flop and chin rest, though she was faking. She wasn't really lying still, not in her brain. She was ready to launch if I asked!

So... yeah. She's not nervous on the mat, but she's not relaxed. We have a way to go. :)


Deborah Leão said...

That was really funny, I could easily imagine the look on her face as she tried every possible behavior to make you click.

Have you considered not using the clicker to teach relaxation? One of my Labs, Phoebe, is not able to relax around the clicker. She is too eager to play the game, and doesn't unfocus.

When I started the CU program, I realized the relaxation protocol would not work with the clicker, so I just started reinforcing without clicking. For that matter, I used lots of petting as reinforcement, too, stroking her gently and slowly caressing her behind the ears.

It seems like the clicker had become a cue for frantic, overfocused behavior. We managed to get relaxed behavior on the mat more easily without it.

Sara Reusche said...

Yes, what Deborah said!

Layla is like a little mini-Laev (with Issues). Her version of clicked-relaxation entails her lying with her head on her paws, tail held perfectly still, whole body vibrating. If that alone doesn't work, she offers deep breaths but rather than sloooow deep breaths she will fill her lungs up all the way in order to force all the air out in one quick, demanding exhalation. Not many dogs have her ability to be bossy just by breathing!

Normally I like using food for relaxation, but with her that still put her in Training Mode (which includes Laylaquaking and occasional demanding breaths). Instead, I focus on rewarding expression through social feedback. Her eyes soften slightly, I smile at her. She lets some tension out of her jaw, I briefly praise her. She's a "show me the money" kinda dog, so at first this was slow going. However, with the addition of an OCCASIONAL treat, and by making my smile into a conditioned reinforcer, she is now responding very well.

Last week I used her as the distraction dog in my CU Level 2 class. I had her hanging by her teeth from a tug rope, growling and jerking around wildly, and she was able to transition from that to calm, focused attention in a down-stay with pretty much zero feedback from me (I was discussing latent learning with the students). This would have been impossible even a couple months ago! Later that same night I had her do repeated crazy figure-8s around my legs, since I know this will cause her to bark loudly (the student dogs were working on classical conditioning: dogs barking=treats from mom). Two months ago, this many leg weaves (with the accompanying frustration) would have gotten me bitten. Not so anymore!

I think this would go faster with a dog who enjoyed touching/massage, but Layla doesn't yet. So I don't touch her unless she asks me to. Maybe Laev would be the same in a working context, but then again you want to disassociate relaxation from clicker games. So, maybe the massage would make that more clear. It would be nice to have touch as another reinforcer.

Reward expression. Don't let relaxation become a trick.

Laura said...

Thanks, guys! I have done a bit (not enough, obviously!) of associated relaxation with a verbal cue. I'm probably going to have to ditch the clicker here and mark verbally.

When she was a pup, I had to actually train Laev to let me pet her while working. :) Now, though, she'll even enjoy petting once she's in relaxed mode. (It's not a reinforcer when she's not really relaxed.) So I will be using that, too.

Sara Reusche said...

So please share, is relaxation going better sans clicker?

I initially used a verbal marker too, but finally ditched even that when I realized that Layla was just faking relaxation as a trick.

I have to say that relaxation has been by FAR the hardest thing I've ever worked on with my dog. Worth it, but very frustrating trying to get there. May you and Laev have better luck!