Frustration Theory

[Mrs Rusty Bucket]

Teaching a dog to be persistant in training:

Note: this post could be more eloquent but then I might never share the good stuff. To paraphrase someone else - "I'd write a short post but I don't have time".

This is based on my interpretation of a Susan Garrett (SG) presentation. And her presentation was based on a book or thesis she read by Abram Amsel called "Frustration theory: an analysis of dispositional learning and memory"

http://books.google.com.au/books?id=-5YpdzQW6WgC&printsec=frontcover&dq=Abram+Amsel&hl=en&ei=JS2cTv3DKsbdiAK94LyyDQ&sa=X&oi=book_result&ct=result&resnum=1&ved=0CDcQ6AEwAA#v=onepage&q&f=false

(It is a very esoteric paper. Chapter 3 seems to contain most of the "good stuff").

So how to train your dog to be more persistant: SG suggests...

• Control your training environment.
  
• Only introduce small amounts of frustration ie only make the task a little bit harder than it was before.
  

Frustration Theory:

When the dog doesn't get their reward - pay attention to how they react... Most dogs (and animals and people) will default to one of these four (that's from the book)

Invigoration - Excited - hyper - zoomies

Suppression - Shut down, frozen, gone sniffing or scratching

Regression - repetitively tries what worked last time (don't reward this if you want something different - reward something different)

Persistance - keeps trying new things to get the reward

I think that's the four things SG mentioned.

You can condition the dog to default more often to persistance which is desireable in training.

When the dog gets something right after a period of frustration.

1. do not ask them to do it again straight away (or even in this training session).

2. give your dog his/her favourite bestest highest value reward and have a bit squeally joyful party with your dog. And a balance break (eg training break where you play tug or run round with your dog).

to develop persistance - this isn't part of the book - well it might be but I haven't read the all of book and had a hard time figuring out what he meant from what he wrote...

Do a lot of basic trick training with your dog so it is used to trying again if it doesn't get a reward. When you're starting out, reward every different behaviour, until you get ones that can lead to what you want and then only reward those and then only reward each step in the right direction.

And do not ask too much of your dog at once (lumping).

And try to get your training mechanics right. Ie good timing, clear criteria, and good preparation - everything you need is ready to hand before you start. And pick something it is possible and plausible for your dog to get right.

This info requested by a post in this thread

http://www.dolforums.com.au/topic/228291-love-those-training-epiphanies/

[corvus]

Interesting. I've recently been through this with Kivi. He started out throwing himself on the ground if he didn't understand what he was meant to be doing within about 5 seconds. Reward rate meant nothing to him if he didn't know what he was supposed to be doing. I started target training him, and that was very useful because it gave me a way to prompt behaviours and get the ball rolling. He knows he can always target and that will be right, which makes him more confident to try other new things. Before targeting, he was very reticent to try new behaviours, even though I'd done lots of free shaping and encouraging creativity from him when he was a puppy.

More recently I've been more sensitive to him. If he gets distracted and isn't looking at me, chances are he is starting to feel frustrated. I manage reward rate carefully to avoid this, and only do a very small number of reps on something we are just starting on and far outweigh it with old things he knows and is confident with. I keep my criteria very easy until the rough behaviour is on cue and he is clear on that cue. It is imperative for him that he be confident with the behaviour before I start shaping it. As much as possible, I set up the environment so he is more likely to get the behaviour I want early and is more likely to perform it as close to the finished criteria as possible.

Then I let him pester me and rewarded it. Heeling is his default behaviour to tell me he wants to train. I gradually increased my criteria for 'giving in' to his pestering me to train with heeling by cueing something and rewarding it. He must be offering a very nice heel with eye contact and in the sweet spot. He tried crowding me for a while to try to draw my attention to the fact that he was heeling. He still does that occasionally. Over time his heeling duration went up and up because I would always give in just when he was starting to wonder if he'd ever get rewarded, but before the behaviour suffered. You'd see the look in his eyes. His eyes get a little bright and his face tightens just a tiny bit, showing the anxiety. That's cutting it fine, but the relief when he gets rewarded a moment later is powerful. Then I started rewarding the offered heels with cues for other behaviours. I find the better he understands the behaviour, the more confident he is in performing it, and therefore the more persistent he is with it, which gives me the opportunity to shape and polish it. So I work on improving cue fluency a lot to help his confidence.

Anyway, that's what I've been doing. I'm really happy with how Kivi has come along. He's a different dog, and training him is a lot easier. I have to live with an obsessive heeler who sometimes refuses to be dismissed, but it's a nice problem to have, I guess. I'm happy that Kivi is so happy.

[Mrs Rusty Bucket]

i think I could be a lot better about consistent criteria when it comes to heeling...

Frosty head butts my hand from a beautiful tight heel - on the RHS if she wants some treats - not so sure about the training but she's willing to work. If that doesn't work - she tries the LHS... and if I'm a bit slow to reward at the LHS she zips back around to the RHS.

Nicking around behind me is really bad in Agility runs. Sigh.

But we're getting good perch work inside with high value rewards.

And I'm getting better weave poles outside with better persistance - this is where the frustration shows up the most with her. When I ask for the right entry from a less than obvious spot - like the wrong side of the poles. I guess it's like teaching a child that even though a d and a p are the same - the orientation and context matter. Pretty difficult concept. Especially given - she doesn't care about left and right when it comes to heeling - the best place for her to be as far as she's concerned is closest to the treat bag (eg RHS).

So that's a rambly way of saying that things started improving for me when I started "rewarding every different behaviour till we were on the path to what I wanted", and then when I got the "squeally party" bit happening * the first time she gets it right * instead of asking to run the test again (like a good computer programmer).

[megan_]

Thanks for this. I have done a lot of free shaping with my boy and in our early days rewarded just doing something different so he'd get the idea. Little sh*t still throws a tanty if he doesn't get rewarded often though, but I think it might be a bit of boredom too (been there, done that, next please).

[Mrs Rusty Bucket]

Yeah the tanty - is "invigoration" or going hyper. Frosty does that. Barkathon with her. Some dogs do zoomies.

And you want to work through the tanty to persistance. Which SG calls working through the "don't wanna don't hafta" - and then it really pays to party when they finally make an effort and try something instead of yelling at you.

I have entire conversations with Frosty going round the agility course at training. "Table up" - no this table is too short SHORT I TELL YA WHY WOULD I BOTHER SHORT SHORT, THE VIEW IS CRAP, NO WAY. Erm...

She does the weaves better ie will keep going if I keep telling her she's doing it right, so our weaves are like "go weave, yes yes yes yes yes yes yes yes... yes YAAAYYY". Still working on it but she's much more enthusiastic.

[ktig]

Interesting idea...things to try with my little guy who definately gets frustrated if he can't work out what I'm after. thanks!