Yes, this has been drilled into me as well. I talked about it with our trainer yesterday and it's a long process of chaining and fade out. Eventually she said the work itself becomes the reward (which was certainly true with Brogan) but I'm not so sure that works on a food-starved Labrador?!
Carbon ALWAYS does better without treats. Always. However I'm afraid that if I'm always without treats I will 'break' his cues that he does now with the expectation of treats. Our trainer said that in addition to chaining, it's really key that I learn to treat intermittently, which I absolutely stink at for the moment. I either have treats on me and treat like mad or forget them at home and it's a treat desert. Bad ML!
@Selina27, I think @Emily_Babbelhund was replying to another question here... if you mark you HAVE to reinforce. Otherwise you're lying and it damages trust. But what you have to do (at least, in working situations like Emily's) is start fading the reinforcers so you reinforce at the end of a behaviour chain - and those can become long and complex, with each behaviour reinforcing the previous one. This is best achieved by a process called back-chaining, which is incredibly powerful.Yes, this has been drilled into me as well. I talked about it with our trainer yesterday and it's a long process of chaining and fade out. Eventually she said the work itself becomes the reward (which was certainly true with Brogan) but I'm not so sure that works on a food-starved Labrador?!
thanks for this, so for a non working dog, especially if you use a clicker a bit sporadically, the rule is always, always treat after a click, ( even if the behaviour was wrong and you clicked accidentally? )
For any dog, working or not: if you click, you reinforce, whether the click was a mistake or not. If you don’t want to reinforce, you don’t click. If you click by mistake, you reinforce and try to do better (but we all mis-click sometimes).
Thanks for the explanation and the link. I don't really understand the chaining myself, except the simple example I gave. Part of my homework is to go back and re-read the curriculum and then I'll go back to our trainer for more questions. The idea of lying and damaging trust is exactly why not rewarding with food or something else immediately makes me queasy. That's what I mean about "breaking" the behaviour that we've spent so long building up. I have a real fear about giving a cue - no matter how well learned - and then not marking/treating. I've still got a long way to go here in terms of learning!
Me too Emily. xxx
What about gambling/game theory? I remembering reading about not always treating, like when you gamble but you don’t always win. The chance of winning is what draws you back. The chance of a treat is what makes tour dog want to try again and again and again. If you treated every time, there’d be no chance, so motivation would drop.
Me too!GGJ just gets rewards every time she does what I ask! Oh dear! That's probably why she's GGJ! She doesn't (quite) continually try for rewards though, she settles well eventually and she's still young, and she does pubs, trains, cafés etc. She's a GoodGirl. Just not sure if I'm a good trainer, more like a trusty Monkey Butler!
I believe that has been largely discredited. I remember reading an article about it; I’ll see if I can dig it out.
Lovely article. And beautifully written! Loved this bit:
No, nothing to do with the camera, it's just an editing feature in iMovie.
