Autism Service Dogs

Yesterday was another good training day. Monica and I have been discussing ways to redress our dogs' reactivity, and decided to meet up in the afternoon to work on the problem, along with another lady who was having similar issues.

Normally, Glindy is so reactive around new dogs while on leash that I can't get her attention well enough to request a watch or a down. So, my personal goal was to work Glindy at the threshold of her reactivity so that we could reinforce these behaviors even under moderate arousal. My idea is that, even if we can't necessarily reduce the arousal level, at least we can improve her reactions so that she's more reliable.

Part of the challenge in shaping this behavior has been that Glindy is totally unmotivated by food when she's highly aroused. It really doesn't seem to matter what sorts of treats are involved; if she's really worked up, she'll still recognize the click itself as a behavior marker (she will even sometimes glance towards the clicker to acknowledge the event), but won't bother with the food reward. Here's how we managed to work around that, at least partially.

The three of us (and our dogs) wandered over to a small park, where we started out about 50 feet from one another. After a few minutes of yanking me around and marking everything in sight, Glindy settled into her normal whining routine.

So, as we stood around letting the arousal build, I took out my clicker. At first, I just waited for any break in Glindy's whining, and clicked for the momentary silence. After a few repetitions of this, Glindy would occasionally take a short break from whining to glance at me, and I clicked for the auto-watch behavior.

This actually worked really well. Normally, asking for a watch when Glindy is worked up is counter-productive. However, just out-waiting her seemed to work a lot better.

After a short while of encouraging the auto-watch, and seeing her arousal level drop a little bit, I began clicking for either the auto-watch or for glancing at the other dogs without whining. My idea here was two-fold: training an incompatible behavior like watch is a good way to take the focus off the other dog and reduce her arousal level, but it's essentially unnatural; trying to re-associate the sight of another dog with positive experiences, instead of trying to tear her attention away from something she wants to keep an eye on, seems a little bit more in-tune with her natural drives.

Anyway, I figured having two strategies is better than one. While I know that one should really only train for one behavior at a time while clicker-training, I decided that what I was really after was calm watchfulness, so it didn't matter to me whether Glindy was looking at me or the other dogs when she entered that state. Whether this is a technically-flawed rationalization or not, it certainly seemed to work.

After working her for a bit in this way, and occasionally moving closer whenever the dogs had settled a bit, we were eventually able to work the dogs within about a yard of each other. By the end of the session, we were even able to take turns passing between the other pair of dogs with just a few feet on either side, and making close circles around them.

Before resting on my laurels, though, I should point out that Glindy certainly got the idea that she was there to work, and settled in reasonably quickly. However, near the end of the session, the sight of a fourth dog playing ball on its front lawn a full block away from the park caught her attention and refused to let go.

While she continued to behave reasonably well around the other two dogs we were training with, every time she caught sight of this other dog she'd start whining, staring, and completely ignoring treats even when they were waved right under her nose.

Still, it's progress. I really like the idea of using free-shaping to make the sight of other dogs a positive experience, rather than a stressful or frustrating one. So, I think we'll continue working on this, and see how much progress we can make.