RickB wrote:As for learning new tasks quickly using R+/N-: I'd have to say that that is technically true, but it comes with some real world caveats....
Yes, quite true. I believe any of the research showing that reward & denial is faster has always been dealing with simple tasks.
When I say goofing off is a different training problem, the point is that we are no longer learning a new, simple task, so that research doesn't really apply anymore. Whether a sharp correction, a more valuable reward, or moving back down the learning chain and reinforcing is best seems to be an open question.
RickB wrote:As for the dog you are steadying, I respectfully disagree with your premise that no conditioning is happening. The reward is the bird (if you hit the target), or the privilege of continuing the hunt if you miss (Premack principle). To break means a cessation of the hunt and denied access to the bird. To me, clearly N- and R+.
This is true. The routine of steadying has always involved reward and denial, with the bird and the hunt being the reward. But what I was getting at was the question of "putting pressure on the dog" by moving up on it, which I don't think actually has any punishment meaning unless of course the dog has learned to expect punishment -- just as the command "No" has no inherent meaning.
By the way, in that particular scenario, we could see stopping the dog as denying the retrieve, P-. But let's look at what actually happened:
1st run:
- dog breaks at shot, stopped. (P-)
- dog breaks at shot, stopped. (P-)
2nd run:
- dog is steady, gunner misses. (P-, punishing
correct behaviour)
- dog is steady, given retrieve (R+)
In a third run, the same scenario repeated.
So there is no real conditioning pattern here. While the first two trials in effect punish incorrect behaviour, two of the remaining four trials actually punish correct behaviour! And the dog gets contradictory patterns of punishment and reward in each run. Yet the dog actually improved consistently throughout, even as I backed off the "pressure" by moving up less and less. I'd suggest the dog is applying more brain power to the situation than the high priests of OC would have us believe.
Don't get me wrong. Obviously operant conditioning is very effective and is the basis of most dog training whether the trainer realizes it or not. I'm just saying it's a mistake to try to pigeonhole every dog-handler interaction into one of the four quadrants.
