While writing up my last blog entry (Better Than Expected, Part II, http://www.clickertraining.com/node/1050), I mentioned that Laev has never appeared to find my "aart!" sound punishing of itself, but that I had apparently conditioned it by associating it with negative punishment -- specifically, by removing Laev from the helper after she'd made a mistake in the blind search or hold and bark. I make no sound as often as I do, but apparently it's been frequent enough that she knows what will come. (She has *never* appeared to be repressed by this noise alone, not in her lifetime.)
So, I thought, if an "anti-clicker" is conditioned to mean P-, does the marker itself become a P+?
I know this is getting dangerously close to the NRM argument, and I fully agree that most of the time, NRMs are a needless complication. I use one only rarely, and only in situations where I think it's necessary to avoid confusion -- it certainly shouldn't be showing up each time you don't click! But in tonight's example, Laev erred only a few feet away from blind #1; I didn't want her to continue the full distance to blind #2, perform a flawless hold and bark, and then receive a P- of removal for the next repetition without knowing specifically that she had lost R+ because of the original error, not because of the perfect hold and bark or for my arbitrary mood. And it apparently worked, because once the error was marked, Laev did not repeat it.
(Why an NRM at all? Why not just interrupt the repetition? In almost all cases, that would of course be the right answer! But here, I think that snatching the leash of a dog at full speed can't possibly be less aversive than giving a single verbal communication which she seems to understand fully. Right for every dog? No, of course not, nor even for every situation with this dog. But I'm comfortable with it here.)
So, NRMs aside.... I usually insist that a verbal "No" or "Ah-ah" is positive punishment. After all, if it didn't stop a behavior, the handler wouldn't be saying it, right? But it never bothered young Laev, never affected her at all, so I didn't really use it, certainly didn't rely on it. But I then consciously began using it as an "anti-clicker" to mark what would be negatively punished (removal of Laev from opportunity), and suddenly now it interrupts behavior.
Have I made a P+ where there wasn't one before?
I'll ask CS; somebody there will know, or at least have an opinion. :)
reply to previous comment
I *love* the antibiotics analogy!
I, too, have "pure data" words for my dogs, which is what a properly-used NRM is. (Most people flub it up as a punishment, which does cause problems!) I keep the pure NRM "try again" well separated from anything I use as a conditioned punisher ("oops"). One is to encourage more behavior, while the other is to give the dog very specific information on what should not be repeated.
It's an important distinction -- "try again" means that won't pay at the moment, but it's a perfectly valid behavior at other times. "Oops" is an anti-click; that behavior should not be repeated, it will never pay, and in fact the dog has just *lost* the opportunity for reinforcement. Offering a nose touch when I want a paw touch merits a "try again" (*if* there's a compelling reason to use an NRM instead of just letting the lack of click speak for itself); breaking a stay deserves an "oops."
For the most part, though, we humans babble too much and we'd get a lot more training done if we'd just shut up. ;-)
We have to remember, though, that "punishment" isn't about likes and dislikes, good and bad, etc. -- it's defined solely as to whether it affects the frequency of the recurrence of the behavior. So if the pup goes in the crate for chasing the cat, and the overall recorded incidences of cat-chasing do not decrease, it wasn't a punishment, no matter what the human thought of it. And if you sweetly pat the dog on the head mid-chase, and the overall recorded incidences of cat-chasing decrease, then it was a punishment, no matter what you intended. (Maybe a socially-sensitive dog? Who knows? That's why we have data!) My Laev finds collar pops actually *stimulating* rather than inhibiting -- reinforcement and punishment are defined by observed results, never intent!
I think, from other messages I received on this, that I bothered some people by "nitpicking" at vocabulary. I do like to keep my terminology precise, though, because that's really all we have when we try to communicate. The dog's (or horse's, or falcon's, or human's) wellness is obviously the first priority, but if I'm going to try to have meaningful conversations with another trainer about what I'm doing, we *must* have the same vocabulary or we talk in circles!
Laura &