Question About Reinforcement...

[Kavik]

Well dogs could do a number of things - they could still go the bottom and stop but self release and not wait for the release command, or still get in the contact zone but jump off before the bottom and keep going or jump off above the contact zone.

[Aidan]

Well dogs could do a number of things - they could still go the bottom and stop but self release and not wait for the release command, or still get in the contact zone but jump off before the bottom and keep going or jump off above the contact zone.

In any case they don't usually go back to the car and have a snooze? They usually run off and do something that offers more reinforcement, which is a Matching Law problem.

[Staranais]

Thanks for your thoughts, everyone. Will read through in more detail later!

Just wanted to say that I do not think the phenomenon that Kavik describes is necessarily created only by an extinction burst.

I can see what Aidan is saying, and think he is entirely correct in some ways. I guess I would think of it as following: a dog that does not get a reward when expected at his first trial may run faster in frustration (the extinction burst). If he finishes the course & earns the reward when he is still running faster, then he may eventually begin to believe that he has to work extra hard on trial day in order to get the reward. Voila = a dog that runs faster in trials than in training. :rolleyes:

But I think this is not necessarily the whole story for what Kavik describes. I think the dog can feed off the excitement of the owner, becoming more aroused themselves simply because the owner is aroused, and that this increased level of excitement can by itself cause the dog to give a more animated, better performance.

Kavik is not the only person who has noticed this phenomenon. I was watching a Michael Ellis clip the other day on Youtube & he described a similar thing. He was talking about sharpening up sits for advanced schutzhund obedience. He said he had very little success in getting really lightening-fast sits on cue by using differential reinforcement of the fastest sits, although he knew that method should work according to theory. (His theory about why it didn't work was that the dogs were already performing so well; he thought that the difference between a sit that took 0.8 seconds and one that took 0.6 seconds was just too small for the dog - or perhaps the handler - to be able to distinguish effectively). Anyway, he said that one thing that worked very well to create the very fastest sits was to increase the dog's arousal level before they were even asked to sit. They then consistently sat faster, without him having to differentially reinforce anything.

This seems like a similar thing to what Kavik is describing. There's probably some sort of physiological mechanism to explain it. For example a dog does naturally become more aware, more alert, more switched on, when adrenalised & aroused, & I think it is reasonable that this could automatically translate into faster running or faster sitting.

[Staranais]

But wait... there's more!

About the original question, I don't think the behaviour has become self reinforcing as Vickie suggests since it's rather boring stuff I'm talking about (lie down, or wait, or come over here and give me whatever you're holding, etc). Not an exciting event in itself, like an agility trial.

Possibly the reward is still somewhat unexpected, as Aidan suggests, but I doubt it - she's a smart cookie, she knows what she gets if she does good (usually an excited sounding owner, sometimes a life reward, occasionally a kibble piece). On average, she knows how "good" the reward is that she can expect for these things. It's never spectacularly exciting. :rolleyes:

Thinking it through, though, I have realised that I do punish mildly if she decides she's not going to do something I ask that I KNOW she knows. So perhaps that is my answer right there. The rewards get less once she knows a behaviour, but she learns she gets a cranky sounding owner if she refuses.

I also wonder if there is something here to do with amount of effort vs level of reinforcement. As in, the rewards get less when she knows the behaviour. But from her point of view, perhaps the behaviour is easier, since she now knows it and has done it many times? So even though the reinforcer gets crappier, perhaps the effort:reward ratio doesn't get less? Does anyone know if this is possible?

Also, could you expand on this, Aidan?

Matching law quite reliably predicts (at least in animals) what happens when the dog has choices about what he can do, what happens when you have lots of little reinforcers vs one big reinforcer, what happens when a reinforcer for one response is delivered immediately vs a reinforcer for a different response being delivered after a delay and that sort of thing.

I have a basic understanding of matching law & I've read it in Lindsay again, but would appreciate you explaining your above quote, if you have time. The more versions I hear, the better I tend to understand.

Sorry, that rambled.

[Aidan3]

I'm actually writing a piece on matching law as it applies to recalls as we speak, and in the course of my research I uncovered this gem which is a good introduction:

http://users.ipfw.edu/abbott/314/MatchingLaw.html

[corvus]

I still don't get why a dog should care if their owner is happy or cranky. It weirds me out. For happy sounds, they are usually paired with affection, which would be a reinforcer for most dogs, but what if cranky owners never do anything other than grumble and look cranky? Erik responds to a cranky owner by putting his tail down and his ears down and lots of displacement behaviour. But if I'm cranky with anything other than him he doesn't respond at all. I've never followed up looking cranky with an aversive towards him, but he still acts like it is an aversive. Kivi despite being a softer dog in general is much less sensitive to human emotional states.

Sorry, that was off topic.

I used to show a cow in high school that was a beast to handle anywhere but in a show ring. For some reason she would turn into an angel the moment she went into the ring. There were no rewards involved in training the cows, really. It was odd that her whole demeanor would change in the ring. It's not like she didn't push you around at the show outside the ring.

[bedazzledx2]

I know a brilliant trainer who's 'emercency recall' is in a really cranky voice! She knows when she needs it she wont be happy and she pairs it with click/treat. Dog responds first time, every time

I still don't get why a dog should care if their owner is happy or cranky. It weirds me out. For happy sounds, they are usually paired with affection, which would be a reinforcer for most dogs, but what if cranky owners never do anything other than grumble and look cranky? Erik responds to a cranky owner by putting his tail down and his ears down and lots of displacement behaviour. But if I'm cranky with anything other than him he doesn't respond at all. I've never followed up looking cranky with an aversive towards him, but he still acts like it is an aversive. Kivi despite being a softer dog in general is much less sensitive to human emotional states.

Sorry, that was off topic.

I used to show a cow in high school that was a beast to handle anywhere but in a show ring. For some reason she would turn into an angel the moment she went into the ring. There were no rewards involved in training the cows, really. It was odd that her whole demeanor would change in the ring. It's not like she didn't push you around at the show outside the ring.

[corvus]

There's probably some sort of physiological mechanism to explain it. For example a dog does naturally become more aware, more alert, more switched on, when adrenalised & aroused, & I think it is reasonable that this could automatically translate into faster running or faster sitting.

Afaik arousal is intrinsically linked to increased physical activity. But there are a lot of things that could cause that. For example, frustration if reward rate is suddenly lower than usual, or just the sensual stimulation of being in a different environment, drastic change in routine (at a trial instead of snoozing??), as well as the arousal level of the handler and other dogs in the vicinity.

You know, thinking about it, I can't think of any behaviour my dogs have been taught that are simply 'maintained'. I do find that if I don't practise Kivi's recall regularly it weakens very quickly, but if I practise it more it becomes stronger. It doesn't ever seem to just sit there at the same level.