Tuesday, January 22, 2008

Record Keeping for Dog Training by Jane

I suck at record keeping for dog training so I asked my friend Jane how she does it. This is the comment she left for me.

Jane said...

A simple way to track progress:

1. Decide how you want to count the treats. I make piles of 15 treats. Other people use a cup to count, and put one treat in the cup for each treat they give to the dog.
2. Set the timer to 3 minutes.
3. Toss a treat for each click.
4. At the end of your 3-minute session, count how many treats you gave out.
5. Write it down, along with any comments about your session.
6. Divide the number by 3, and that’s your rate of reinforcement. If you want to do shorter sessions, that’s fine too. Just do the math.
7. Chart it out on a spreadsheet or piece of graph paper.

What’s interesting is that you can actually see whether you’re making progress or not. When I was doing this with one of Finn’s exercises, I thought we were making progress, but then I looked at my graph and could see that we weren’t. So then I had to change something. (I changed the session length and that seemed to work. In a duh moment, I realized he couldn’t sustain the behavior for 3 minutes!)

So that's simplified record keeping. For two dogs you'd need to be sure to figure out a counting system so that you can keep track of the treats given to each dog.

Thank you, jane. I will try counting during today's training.


Holly said...

I'm confused.

I thought you would count out 15 treats and put the ones he missed in the cup.

Jane said...

I think everybody finds a way that works for them. I like counting out piles and afterwards I count out how many are left in each pile. For the exercise I was working on, I did sets of three one-minute trials. So I started out with three little piles of 15 treats.

With clicker training, you give a treat each time you click. And you are only clicking the behavior you want to increase. Since you're often rewarding small incremental steps in a behavior (for example, stepping on the mat), you aren't really thinking in terms of wrong "answers" by the dog.

If you aren't rewarding the dog a lot, then you have your criteria too high and you need to lower it. So if Blue isn't stepping on the mat at all, Robin would reward her for glancing at the mat. After Blue has glanced at the mat a couple of times, Robin starts rewarding her only for sniffing it.

Because Blue has played the clicker game before, she knows to keep experimenting to figure out what it will take to make Robin click the clicker and give her a treat.

Robin won't tell give her the "command" (or cue) to go to her mat until Blue is already doing it reliably. At that point Robin will start using a word or gesture that means "go to your mat and lie down now."

So . . . there's nothing to miss.

Does that clear up your confusion at all?


Anonymous said...

This is very interesting stuff. And I don't even have a dog.

-- Cloth