Tuesday, January 23, 2007

Better Than Expected

Schutzhund training tonight, and we have mostly good news to report.

I started with retrieve practice, because I've made a silly error with that; I began training the retrieve from scratch with my plastic AKC dumbbell. Now, you'll hear people say that's a horrible idea because then the dog has bad associations with the dumbbell, but of course that's only if you're training with aversives. Laev loves the retrieve and in fact has a full formal retrieve with that same dumbbell, but I hadn't tried the behavior with any other objects, and I discovered one day to my horror that she thinks "Take it" means "bring the white plastic AKC dumbbell and nothing else." Oops!

So tonight I took a Schutzhund 1 dumbbell (650 g or 1.5 lbs) and asked Laev to take it from my hand. She did, but messily, as I've rarely asked her to take things from my hand. I wised up quickly and put it on the barn floor. (We were training indoors due to frigid temperatures and snow outside.) Laev picked it up a couple of times and then decided that the wood chip footing was *very* interesting. She had a tough time leaving it alone, so I abandoned the retrieve for a couple of minutes and did some heeling, just focusing on movement so she could work some wiggles out. Then I came back to the retrieve.

Yes, she could do it, but the wood chip buffet of smells was just too tempting. It was taking her a long time to perform the retrieve, and she was just missing the dumbbell and going to sniff the chips as often as not. She'd now been reinforced for bringing the new dumbbell a number of times and I could tell that she'd entered The Question Phase.

The Question Phase is something most clever dogs go through with a new behavior, but Laev has perfected it to an art form I've rarely seen before! She will come up with several -- many -- dozens of variations on a behavior and offer them all, testing what the *exact* parameters are. For a simple nose target, for example, it might look something like this:

  • touch the target with the nose

  • touch the target with the left side of the nose

  • touch the target with the right side of the nose

  • circle the target, face the handler over the target, and touch it

  • nudge the target

  • rest the chin on the target

  • circle the target completely, returning to the original spot, and then touch it

  • touch the target with front paws in various positions


Laev doesn't offer a new variation after a failed repetition which did not result in reinforcement; it's not lack of reinforcement producing variable behavior. It's as if she's just trying to imagine all the ways it could be done and testing them. It's funny, interesting, usually amusing, and slightly creepy.

This time, however, it was mostly annoying, because it was clear that Laev wanted to know if we could just possibly include wood-chip-sniffing as part of the dumbbell retrieve. I went to an NRM, marking the sniffing as the end of potential reinforcement. (The No Reward Marker is something that I recommend rarely, but which I can use occasionally with this particular dog; she understands it, is not suppressed or punished by it, and it usually works well.) The NRM didn't do much, though, as it was clearly a lie -- sniffing was providing its own reward as Laev sifted through the scents of barn cats, mice, birds, horses, other dogs, etc. So I escalated to some mild P+, taking a handful of loose scruff skin when she wandered away from the dumbbell to sniff. Again, I very, very rarely recommend or use positive punishment, but this was not a new behavior, it was pretty mild P+, and I'd already reduced criteria to a level below which she had already been performing in the same session. I didn't want to end by reinforcing backsliding. I couldn't reduce the distraction -- removing the floor was not an option -- and I judged that mild P+ was ultimately better than continuous nagging, a sudden lack of reinforcement for retrieves during environmental reinforcement for sniffing, and my own frustration which would certainly affect her.

Laev finally gave me some eye contact and appeared to think through the math involved. She went and collected the dumbbell, and I reinforced with cheerful praise, petting and several treats in succession. One more repetition, through which I coached her, and then similar reinforcement. Ah! Okay, Mom, I get it; no sniffing. We did 2-3 more clean retrieves and I quit. Huzzah!

On to bitework. Actually, this is getting long, so let me make a second post....

3 comments:

Shawndra Miller said...

Hi Laura, I like the new look and the many posts! Question on the NRM: Is that what I'm doing when I say "oops" if Marley doesn't (for example) sit when I say sit? Curious why you don't recommend it?

Laura said...

Yes, an "oops" when the dog does not respond is an NRM. However, it's really unnecessary in that case; the dog can clearly tell that reinforcement is not coming without you saying anything.

NRMs can very quickly become a crutch for trainers and can very quickly degrade into a P+. We like to use them to "help" a dog, when our babble might be interfering with his ability to think ;-) and when it might actually be demoralizing or punishing to be told "you're wrong!" so often.

A true NRM simply gives data to the dog about a multiple option scenario. But in most cases, the lack of R+ is just as much information and without the risk of addiction or fallout.

Clear as mud?

Shawndra Miller said...

Interesting - I never thought of it as addictive or punitive. But I could see that, in a situation where he's learning something new. What about for cues that he "should" know? I think I've been using "oops" to try to get Marley's attention back as he's lollygagging on things we've already been working on for months, like the recall. It hasn't been particularly effective, so maybe I'll cut it out.