Sunday, April 17, 2011

Clicker Expo 2011 (Chicago): Kathy Sdao- The All-Seeing I

I almost didn't go to Kathy Sdao's presentation on observation. I've been told- and generally believe- that I'm a pretty observant person, but I enjoyed her other presentations so much that I went because I wanted to see her one more time before the weekend was over. I'm glad I did. I learned a lot... mostly that my skills, while good, aren't as good as I thought. I will definitely be working on improving them.

But if I'm already pretty good, why bother? Well, as Kathy said, the most important training skill is timing, especially that of consequences. Behavior flows continuously, which means that consequences- the click and treat- can just as easily attach to the wrong behavior as to the right one. Easier, maybe, since fractions of a second can matter. I realized this recently when I began capturing a full body shake with Maisy. Once she started to offer the behavior, it became incredible clear exactly the point at which I had been clicking. The shake stops at the same part of her body- about halfway down her back- every time. It looks ridiculous, and absolutely proves that old adage “you get what you click.”

The goal, of course, is to click at the exact moment the dog is doing the behavior. Unfortunately, there will always be a lag time between when you see the clickable behavior and when your thumb reacts. Pure reflexes have a lag of about .120 to .130 seconds; it simply takes that long for your nerve impulses to go from your eye to your brain to your clicker thumb. It gets worse once you add in processing time; an average person's click is going to be .215 seconds late. If you want to know your lag time, you can test it out here.

Although you can improve your timing, the only way you'll ever get that number down to zero is if you can predict the behavior and decide to click just a fraction of a second before the behavior actually happens. You will sometimes guess wrong, of course, but Kathy says that's okay- the results for clicking too early are usually better than for clicking too late. And to do this, you need to learn to see the precursors to the behavior.

This, of course, means that you need to learn to see small behaviors. The very best clicker trainers are the ones who can break down the goal behavior into small, attainable increments. This is widely referred to as “splitting.” The opposite- called “lumping”- happens because the trainer literally doesn't notice the smaller behaviors. The so-called lumpers need to learn to see their dogs, not just look at them.

So what influences your ability to see? Kathy identified four broad areas:

1. Your labels and preconceptions. Research* has shown you don't see what you don't expect. So, guess what you won't see your “bad” dog doing? If I expect Maisy to behave reactively, doesn't it seem likely that I will see and fixate on any example of reactive behavior, even to the point of ignoring her calm, relaxed behavior? Have you labeled your dog? Is he stubborn? Was he rescued from an abusive situation? What are those labels making you see? You need to move beyond the backstory. Move beyond your expectations. See who your dog is now. Expect good things.

2. Judgments and analysis. As important as it is to analyze your training, don't do it while you're training. Just do it.. or rather, just see it. If a click is late, don't berate yourself, don't even think about it, just pay up and keep going. Constantly thinking about and evaluating what you're doing is going to interfere with what the process, and you'll miss seeing some pretty awesome things from your dog as a result.

3. Talking and prompting. Like thinking, talking and prompting gets in the way. The truth is, people just aren't that good at multitasking. On average, people won't see unexpected stimuli 30% of the time. When asked to multitask, this rockets up to 90%! Avoid having this happen to you by remaining silent while training. This includes avoiding the use of a marker word if possible- they engage your “talking brain.”

4. The audience effect. Being watched changes our performance- ask anyone who competes in dog sports! To move beyond this, I really think we need to go back to Kathy's second point. Don't judge yourself- or worry about how others might be judging you. Simply be in the moment. Harder said than done, I know.

Although you can't significantly alter your attention span or ability to multitask (those things are hardwired into your brain), you can change how you attend to things. In other words, you can practice paying attention. Kathy showed us tons of videos, and while I couldn't find most of them online, a Youtube search for “observation test” led to a lot of interesting clips. (This channel has some good stuff, including some of the things she showed.) Along those same lines, take videos of your dog and then watch them. Watch them at normal speed, then slow them down. Learn how your dog moves.

Do an ethogram on your dog. The next time you're watching television, reading a book, or surfing the web, look at your dog once every five minutes. Write down what he's doing. After an hour (or more, if you're patient), look at your log. Do the results surprise you?

Finally, become a “choice architect” while training. Manipulate the environment so that your dog can choose correctly most of the time. This helps cut down on visual distractions (for you). Then decide what to look for, and pay attention! Give your dog your undivided attention.

I hope these suggestions help you. This was a very participatory session, and it's hard to describe the videos in writing without totally wrecking the experience for you in case you are lucky enough to make it to one of her seminars. And even if I did write about them, I think the impact is lost if you aren't experiencing it yourself. Anyway, please let me know how these suggestions work for you! Comment with videos you watched and enjoyed, the results of your dog's ethogram, or any other ideas about observation skills that you might have!

* Much of the research and facts cited by Kathy came from the book The Invisible Gorilla, by Christopher Chabris and Daniel Simons. I haven't read it, but I certainly want to!


Kirby, CGC said...

Hi Crystal, glad you were able to go to the seminar, knowledge is power is what I always say and we can always learn something new.

I went through your blog and watched your rally videos, thank you for posting those. I learned alot from watching you. Kirby and I have a chance to attend a fun rally match in June, I'm just nervous about how he will be with the other dogs but excited to try this with him.

Cheryl and Kirby

Crystal Thompson said...

Aw, thanks. We're not even really that great at rally- Maisy's a good dog, but I've never bothered to train with much precision. It's all about fun, after all, right? If you want to see some REALLY NICE working dogs, go here: Laura's amazing. :)

Laura, Lance, and Vito said...

I know I'm a late clicker! Maybe my dogs just learn to adjust since they know how horrible I am. That timing link you posted to made me feel even worse about it, thanks Crystal!

Interesting that once again she is in disagreement with Mr. Dunbar. I like talking while I train. I tell the dog that they suck all the time :)

I like the #2 suggestion though. No over thinking anything! For once I do something on the good side :P

Joanna said...

This reminds me that I've been thinking a little bit about teaching a clicker class, and the idea of getting "average" dog owners to become good enough clicker trainers that their dogs make significant progress during the class is overwhelming.

Sophie said...

I find that when I'm training a behaviour, sometimes I miss the subtle little event with the clicker - but I laugh and that seems to work as a delayed-reaction clicker for Lola. :) Of course, our slow reactions are mostly the reason we need to split a behaviour into so many small aspects and then click progressively toward your goal; then if you miss the most recent offering, you won't be too far down the list, and you'll already know exactly what to look out for!

This is probably why I am pretty crappy at capturing natural behaviours, because I never expect them (even when I'm watching for them!).

Crystal Thompson said...

LAURA, I do think dogs can figure out what you MEANT to click. Still... I try to be as good with my timing as possible because it only seems fair. I supposedly have the bigger brain and am smarter, so shouldn't I make the effort to figure out how to do it instead of making Maisy figure out what I MEANT to click? Plus, like I said, we've had those moments where I very clearly got what I clicked. Her half-body shake is RIDICULOUS.

I also talk (a lot) while training, to the point that I often "YES!" and click at the same time. That's just... special.

JOANNA- Kathy talked about "back chaining training skills" in people, but I just didn't have room to write about it. Basically, she starts by teaching people to get good at using a high rate of reinforcement, independent of the clicker. Then she teaches them to use the clicker+treat combo. Finally she adds in stuff about timing and observation.

SOPHIE- your comment about capturing is right on, I think! It definitely takes a skilled observer to be able to capture a behavior.