Originally Posted by
sixpacklabs
You touch on something extremely important to training mainly with +R and -P that's ignored by many or glossed over, and that is the necessity of preventing--to the extent possible--your dog from being reinforced from behaviors you don't like or that don't meet criteria for the behavior you're training. You structure training so that the dog is allowed to make a choice, but you control the consequences so the dog isn't reinforced for making a bad choice. For example, if I'm training a sit stay with distractions with a very young pup, I might drop a piece of kibble by my foot. If the pup stays put, I say "good" to mark the correct choice and he gets a tasty little morsel. +R. If he starts to move, I just put my food over the kibble. -P. Different consequences for different behaviors. That's what operant conditioning is all about. In this example, I'm not controlling the dog. I'm controlling the consequences he experiences. He learns that self control is the route to what he wants. Being impulsive gets him nothing. I think preventing the dog from being reinforced from bad behavioral choices is important even if you're using all four quadrants.
If you study the methods of people who are training effectively using mainly +R and -P, you'll find that they're obsessive about knowing what's reinforcing to their dog in any given situation, and about preventing--to the extent possible--their dogs from being reinforced from bad behavioral choices. You'll also find the vast majority of these folks in sports like agility and freestyle, where the competition environment is relatively sterile compared to what dogs experience in the field and the handler is working close to the dog, making it is far easier to manage the dog's access to reinforcement. Also, in those sports, you are to a large extent building novel behaviors for which there's no inherent reinforcement value for the dog, and must supply reinforcement to build the dog's motivation for the behavior. There's some of this in field work, of course...but you're also working with behaviors that have high reinforcement value in of themselves, like chasing prey.
I think there's a tendency among people who come to field work from sports where they've been successful using mainly +R and -P to assume they can just export their methods wholesale without thinking about the many things that are different between their old sport and field work. IMO it's better to think about where the methods might be effective, and where they might fall short. In my experience, tools like markers are really helpful in the initial training of obedience behaviors like heeling where the behavior itself is not inherently reinforcing, and where it's helpful to be able to communicate precisely when the dog is doing the correct behavior. Once you start moving basic trained behaviors into the field, you're dealing more with behavior chains where one behavior reinforces another and I don't see much use, if any, for markers.
In the case of the dog scooting at the line, I don't think it matters whether you've done your training with a clicker or an e-collar. If you allow the dog to scoot, and then re-heel, and then send him, he's just learned something you didn't want him to learn. He's learned that the standards at tests or trials are different than they are in training. Maybe next time he'll creep further to see if that works. If you let him retrieve, he learns it does. Maybe the next time he'll experiment with breaking. So yes, if the dog does not meet criteria, I'd thank the judges, leash the dog, and walk back to the truck. I've done that. I don't want to start my dog down the slippery slope of learning that standards are different in competition than they are in training. You're not throwing your entry fee away. You're making in investment in future success.
It's positive reinforcement and negative punishment. The example you gave of the rat getting his nose whacked would be positive punishment (positive because you added something, the nose whack) if in the future the rat was less likely to touch the lever because he got his nose whacked. Negative punishment involves removing something the animal desires or expects. For example, you toss a bumper, the puppy scoots forward and you go pick up the bumper. If the behavior of scooting forward decreases in the future, denying the retrieve was negative punishment.
Posting Permissions
Reply With Quote