300 Peck Method

Kopiert fra Sue Ailsbys mailinglistes arkiv under “Duration and distance”. Mange andre ideer til artikler finnes i dette arkivet:

http://groups.yahoo.com/group/traininglevels/files/

I have had enormous success with the “300-peck” method. Ask her to Sit,
count ONE, click, treat. BTW, when you click, the behaviour’s over, so
if she gets up to get the treat, that’s perfectly OK. Ask her to Sit,
count TWO, c/t. Ask her to Sit, count THREE (you’re counting quietly,
BTW), c/t. Ask her to Sit, count FOUR, c/t. etc. When she breaks before
you hit your count, you start right over again back at the beginning:
Ask her to Sit, count ONE, c/t. Ask her to Sit, count TWO, c/t.

What this does is force you to work at the dog’s level of
understanding. If she understands up to 3, you’re working 1-2-3. If she
understands up to 50, you’re working 1-50. And the DOG tells you when
she’s understanding more.
Otherwise what we ALL tend to do is just haphazardly hope she’ll stay,
build in those nasty Sit-Down-Sit chains, etc.

Sue Eh?

“Sue, do you think this works better than the technique (which I
“think” you used to recommend) of figuring out the longest time the dog
could go, then working variable times around that time frame?”

J—, I’m not sure it works BETTER, but it WORKS, and it works without
a lot of complicated figuring. And I think it is a lot easier for
people and therefore MUCH more likely to get compliance. I found with
working on the variable times that people gave me a lot of “yeah, well,
this is a long one, the next one will be shorter” but when I asked HOW
long, it was just a guess, and they frequently forgot to do the next
one shorter. 300-peck comes so easily, I find myself counting when I
didn’t even think I was thinking of it.

Sue Eh?

“Does the principle hold up equally well for building distance?”

Yes, it works very well for distance as well. Alex Kurland was the
first to describe it, and that in relation to trying to get a horse to
lead without grabbing his lead in his teeth. She started with 1 step,
no grab, c/t, then 2 steps, no grab, c/t, etc. It’s a forehead-whapper,
as far as I’m concerned.

Sue Eh?

This was an experiment by Alexandra Kurland with one of her horses.
http://www.theclickercenter.com/news/news03b.html is the URL. 

Helix

“moving on to new criteria when you
have had an 80% success rate”

An excellent question. 80% is Bob Bailey’s criteria for asking for 
more, but that’s not 80% once (how do you measure 80% one time? I 
dunno) – say I’m going to try a 3′ SitStay. I’m going to do it 10 
times. If she fails 3 times, we’re only at 70%, we’ll have to do it 
another 10 times. With that next 10 times, even at 70%, she’s being 
rewarded 7 times for doing it right, and failing 3 times. Having been 
rewarded 7 times, in the next set of 10 trials, she does it right 9 
times, that’s 90%, now I can move on to 4′ SitStays. The first set of 
10 she gets 4 right – 40%. Yuck. The next set of 10 she gets 6 right - 
60%. Better but not good enough. The next set of ten she gets 8 right - 
80%, I can move on to 5′ SitStays. She was rewarded for her 4′ SitStay 
18 times before she appeared to understand it and we moved on to 5′. 41 
reps, but more failures than I want lay people to have, because most of 
what I’m trying to teach them is to NOT have failures, and you know 
that if I tell them it’s OK to have failures, suddenly the dog has to 
do 40′ 4-hour SitStays for a “good boy”. And it requires more figuring 
and keeping track than most people can give it. These are LAY PEOPLE 
complaints, please note, and I am, essentially, a lay person, not a 
scientist.

So for years I said this way of doing it: Ten times right, one time 
wrong. So I ask for a 3′ SitStay, she gets it right 4 times, then she 
makes her first mistake. That’s 4 right, one wrong. Having one wrong, I 
go back to a 2′ SitStay. She gives me 10 right, so the next one is a 3′ 
again. She gives me 8 right in a row and then one wrong. Having one 
wrong, I go back to 2′, and she gives me 10 right, so back to 3′. She 
gives me 10 right, so the next one is a 4′. The first set she goes 4 
right, one wrong, so we go back to 3′, 10 right. On to 4 again. This 
time she gives me 6 right and one wrong. Back to 3 again, 10 right. At 
4 again, this time she gives me 8 right and one wrong. She’s now been 
reinforced 18 times for the 4′ SitStay, plus 20 times for the 3′ 
SitStay in between. Easier once you understand it, but still way too 
complicated. And seems slow. 80 reps.

Then 300-Peck. She does a 1′ Sitstay, a 2′ Sit-Stay, a 3′ Sit-Stay, 
breaks on the 4′. 1′, 2′, 3′, breaks on the 4′. 1′, 2′, 3′, breaks on 
the 4′. 1′, 2′, 3′, 4′, breaks on 5′. 1-2-3-breaks on 4. 
1-2-3-4-5-breaks on 6. 1-2-3-4-breaks on 5. 1-2-3-4-5. At this break, 
the dog has been reinforced 8 times for a 1′ SitStay, 8 times for a 2′, 
8 times for a 3′, 4 times for a 4′, and twice for a 5′. 30 reps. Each 
failure means the dog is describing her threshold of understanding, and 
at EACH failure, we step back below her threshold and explain it to her 
again. And each time she’s successful, we change the criteria and ask 
for a little more. I LOVE this method, and I don’t see that it is at 
odds in any way with Bob’s.

And since every time you change something about your general criteria, 
such as location, you start again from scratch, the dog is ALWAYS 
working around her threshold with almost no thought on your part at 
all. (I really like that part).

Sue Eh?
http://www.dragonflyllama.com
> So its not directly a part of the Training Levels then right?

That’s right. This was something that Alexandra Kurland came up with when 
training her horse to solve a particular problem. 

It’s hard to grasp at first but Alex’s criterion was “do one MORE than you 
did the last time”. So in her version, if the horse made his mistake when 
she was workign on 27 steps, his criterion was still 27 steps. She started 
counting over again from 1 to 27 but she did not go back to 1 CT, 2 CT, 3 
CT.

Many of us dog trainers mis-interpreted that and went with resetting back 
to 1 CT, 2 CT, 3 CT and so on. This in its own right was a cool thing to 
do and very productive for many people. I used it to solve a weave pole 
problem in agility with my younger Havanese. Just for the heck of it I 
worked my older Havanese on it and serendipitously solved a YEAR LONG 
weave pole problem of his! (strictly as a result of building muscle memory 
in ME and how I cued the weaves)

Alex recently posted a bit about this on ClickSport and she noted a few 
other variations that people have been doing. One of those is to stay on 
your number for as many reps as that number is. That would go like this:

1 CT
2 CT 2 CT
3 CT 3 CT 3 CT

I’ve been doing that with Joey at times and that’s kind of fun too.

We need a 300 Peck Pigeon FAQ someplace that keeps track of all the 
variations!!!

I’m also doing it as intermittent reinforcement, as follows:

Down, Stay – 1 sec, T, 2 sec, T, 3 sec T, 4 sec T, 5 sec T

Or as separate Down-Stays – 1 sec, OK/C/T; cue, 2 sec, OK/C/T; cue, 3 sec, 
OK/C/T and so on

And either of those could be Reset the Count and CT upwards again or Reset 
the Count and work for ONE MORE. So there are four variations right 
there!!!

I need anything to make training stays less predictable.

I have another method, too. I count seconds and reinforce on each prime 
numbers. Math geek. But at least it’s variable and hopefully 
unpredictable. 

Helix Fairweather