Wed Aug 30 17:49:55 PDT 2006

Reinforcing Positive-Reinforcement

Okay, I'll admit it: while I use clicker training with Glindy, deep in my soul I'm not a whole-hearted convert. I still believe that there are times that corrections are warranted, and I'm product enough of our punishment-based society that I still harbor the occasional suspicion that reward-based training will fail me at some critical moment. However, I had an experience this weekend that served to reinforce my belief in positive-reinforcement training.

As I've no doubt mentioned before, one of the bad habits that Glindy has always had is arousal-whining. Whenever she's anxious or excited, she whines like a puppy. It's not a demand bark, and she's not in pain, although strangers who don't know her the way I do have occasionally asked "Why is she crying?" It's just her way of expressing too much pent-up emotion; it is most definitely annoying, though.

We were day-tripping to Carnelian Bay to provide some Asperger's Syndrome advocacy to a group of parents and school-aged children; that's something one associate of mine chooses to call being a "self-narrating zoo exhibit," but which I prefer to call providing hope and insight. Anyway, it was a long drive through unfamiliar terrain, and Glindy started her "Ooh, ooh! Something strange is going on!" whining almost right away.

An hour of ineffective shushing, useless muzzle-gripping, and unproductive "look" commands was enough to convince me that I needed a different strategy on the way home. I'd tried the cued silences, the corrective touches, and incompatible behaviors; now it was time to give pure shaping a chance.

On the way back, I propped a treat bag filled with Grizzly NuTreats on the flat surface just in front of my gear-shift, and kept an iClick in my left hand as I drove along the mountain roads. Every time Glindy was quiet for even a fraction of a second, I'd click and treat.

Of course, in the beginning, I made all the usual mistakes that I warn other people against: being afraid to dole out a lot of treats at the start; putting her on too thin of a reinforcement schedule at first; being afraid to click if she wasn't quiet long enough; stretching the time between clicks past her breaking point; missing clickable opportunities; and of course, occasionally clicking just as she started to whine again.

Nevertheless, Glindy and I made the most of that hour to refine our individual techniques. After about 15 minutes of click/treat, she started to get the idea--or maybe my timing just improved. Either way, I started to up the ante.

First I stretched the time between clicks. Then I dropped the time requirement, and only clicked for quiet downs in the back seat no matter how short. Then I started lengthening the time again, and eventually worked up to a three-minute fixed-ratio schedule.

By the time we got home, Glindy was lying quietly in the back seat, and hadn't whined for more than a split-second in well over half an hour. I'm sure we'll have to repeat the process again many times in the future before it's second nature to her, but I just knew by the end of that ride that Glindy had figured out that lying quietly in the back was a good way to make salmon pieces appear like manna from heaven, and that for whatever reason I was no longer moved to do unpleasant things like grabbing her muzzle for no reason at all (from a canine perspective).

Even if we're still reliant on treats at this point, and even though I'm not convinced that I can rely on this behavior in higher-stress environments, the important thing is that I got the behavior that I wanted with minimal fuss. The greatest value, of course, is that I elicited a desirable behavior while strengthening my bond with my dog.

I'm still a product of a punishment-based society, but it's training sessions like this that encourage me to have more faith in the clicker. In short, this time the clicker reinforced me.

Posted by Todd A. Jacobs | Permalink