A variable schedule of reward though is not the same as negative punishment, not at all @Boogie. It's just not even remotely the same. Presumably Twiglet has never chased and killed a pigeon, and doing so is not a reward for ignoring them. So while she might be super interested in them, and you've had to work hard on desensitisation, the pigeon is not a reward she is fully expecting, and is whipped away as she goes to take it. The fact that you gradually fade out a continuous schedule of positive reinforcement, does not mean you are applying negative punishment.
Yes, it's perfectly possible that she was confused. Was this the reason for the noise? Maybe. If that's the case, I think confusion and frustration are similar enough, in the fact that they're both negative emotions, that they should both be avoided when recalling. I also don't believe that it will break her recall if she was indeed confused or frustrated on this one occasion, but, if indeed it was confusion or frustration, then it's something I have to plan for in the future because if I repeat the scenario enough times, it will have a negative impact on her recall, certainly. I'm going to say that I don't have enough evidence that it was a whine of frustration. Which means I'm going to keep it in my mind, but not react on it for now. I'll set her up for success, as I always try to do, and make sure she gets plenty of reinforcement for correct behaviours, but if she does deviate to a different target at some point in the future, I'll do the same and recall her and pay careful attention to her reaction. The thing I've been thinking about is whether I should recall immediately, or allow her to complete an incorrect sequence. The benefit to recalling is that I can set her up for success a lot more quickly and reinforce the correct behaviour, rather than letting her complete something incorrectly. The more times they do something wrong, the longer it can take to correct it. Also, if she did an award-winning return from the wrong target, I'd want to capture that on one hand, but on the other, wouldn't want to be reinforcing the earlier mistake at the beginning of the chain, so it would put me in a dilemma.
Yes, back to the main question. I'm doubtful that recalling a dog once it has made a mistake is all that helpful. Just like I don't believe taking a dog back to the exact spot it broke a sit is helpful (what's that suppose to do? punish the dog? and it took you far too long to get back to the spot, the dog long forgot he moved). The dog makes a mistake, and then gets a tertiary* reinforcer (another cue) instead of following the first cue and presumably failing to earn a primary reinforcer. Unless the first cue led to something intrinsically reinforcing I'd allow the sequence to pan out, implement LRT, and reset and start again. * changed from secondary, to save 3 pages of corrections....
You should always look at the wind first before you send the dog She may have scented the dummy at the 'wrong' post, so in her eyes she was doing the correct thing, even though it was incorrect in your eyes. If you are aware of the wind and the scent of the dummy is coming from the other post, give a 'leave it' command. However, maybe there was no dummy at the other post and I am talking rubbish! Maybe your hand command wasn't that clear, did you look down your arm and it was definitely pointing to the right post?
There were no dummies at any posts, so if it was a scent she was following, it wasn't from anything within my control. Yes, I was clear with my signal, it's something I am working on, ensuring the dog is looking clearly in the correct direction, well focussed and eager to run, before sending. Not to mention the posts were nowhere near each other - one was directly in front of us and the other was about 45 degrees to the right @JulieT, do you know of any articles, or even any anecdotal evidence, that compare using LRT against a recall? I don't suppose either is particularly helpful since you're ultimately trying to correct a behaviour in the dog that stems from a mistake you've made as a handler ( whether you're to blame for not having the olfactory capacity to know your dog is being distracted by something is one for another discussion ), so simply proceeding as normal (implementing LRT), bringing no attention to the failure and then setting them up for success, taking into account the new information you have sounds like it would make sense. But, again, what about the issue of letting the dog perform an undesired action "cementing" it into a behaviour pattern that needs more successful iterations to erase it? Is this simply one of the challenges we have to address as +R trainers? Vacillating, much?
I don't know of any references, no. I think preventing the mistake makes sense when it's something like running to other dogs, chasing a bird, or (in Charlie's case) picking up a dummy (the status of which has a special place in Charlie's reward value table). If the mistake doesn't access a primary reinforcer (play, chasing, food) though, it seems different. Sure, if the dog rehearses that mistake and it happens again and again, then that's not good. But why would it? If you don't reward it, and the action doesn't access a reward, the dog won't keep doing it. If I were training touch a target stick, and Charlie made a mistake, I'd just ignore it and not reward. I wouldn't try to prevent him, although of course would do everything to make sure he got it right next time (so I could reward him).
@JulieT I think you could write a book... you do have so much experience with Charlie, it would be really valuable to others. Well it would to me at least.... when you get a moment..
Ha! Yes. What would I call it? How about "Chocolate Clicker Gundog"? With the strapline "how to achieve very modest success in training a completely bonkers dog, possibly the least suitable you could find for gundog work". Would you buy it?
Well, yesterday Willow whined again, but this time it was just as I was lining her up. So maybe it's actually excitement?! It doesn't sound like an excited whine, but she doesn't do those, that's Shadow's domain, so maybe I just need to get used to it from her.
