The clicker and behavior chains
August 20th, 2008 Posted in life with poodlesWe don’t get a lot of thunderstorms around here, and in fact, we’ve only had one or two in Elly’s life. We had a LONG thunderstorm, with a lot of noise and light, just before Independence Day. By the end of the storm, Elly had started to flinch when she saw the lightning, not when she heard the boom. Times from lightning flashes to booms were, at that point, more than ten seconds apart.
This got me thinking about behavior chains and clicker training. Both Elly and Dancer are clicker trained from a very young age. It would appear from the thunderstorm that Elly, at least, can associate two events quite far apart in time. That got me thinking about rewarding behavior chains with a click or similar event. (In fact, you could think of the CLICK as the lightning flash (predictor) and the BOOM as the reward (treat or toy).)
I’ve been working on contact training with Dancer on the dog walk. Once that’s fairly well established, I’ll move on to the A-frame.
Contact training seems to me to be one such long behavior chain, so I’ve started training it that way. I say “GOOD!” when Dancer gets on the dog walk; I keep saying “GO GO” as she runs across it; I say “PLATE” to tell her to stop at the contact; I click (and say “GOOD”) AS SHE STOPS. Three flashes of lightning to predict one event (the click) and then a reward.
I think the biggest issue has to be the definition of the behavior chain. It needs to be obvious TO THE DOG (not to the person) what the dog must do to earn the reward (and of course the DOG has to perceive it as a reward). So I have to be very clear: where will you mark the behavior? When and where will you deliver the reward? What behavior will earn the reward?
Dancer’s dog walk has improved enormously. Yesterday, in fact, she took the dog walk after a jump at a canter, for the first time. She slowed to a trot at the top of the up ramp, of course, as any sensible poodle would do. But it was a good fast trot with no signs of stress.
Another few weeks and we’ll move on to the A-frame.
2 Responses to “The clicker and behavior chains”
By Amy on Aug 20, 2008
Yahoo! Congrats on the progress…want to train MacGyver since your doing so well?
By admin on Aug 20, 2008
I think you’re doing a great job with Gyver. The only thing I’d suggest is the “less is more” principle that you mentioned with regard to Elly some months ago. If she runs off at 15 minutes… walk for 14 minutes and reward and quit, I believe you said.
You were right, by the way. I’ve been upping the reward rate with Elly, and being VERY clear about the behavior I expect, and I’m getting great results. This morning at our lesson, in fact, I got half a run that was so fantastic I gave her all the treats I had and declared it finished, even though the entire lesson was only about 2 minutes as a result.
It seems to be all about clarity and reward rate. Click for clarity and reward when you get what you want.