the right way of doing punishment-based training?

    • Gold Top Dog

    the right way of doing punishment-based training?

    No, I am not going to start using this method on my dog. Clicker training works very well for us; plus, she is quite timid and I don't want to shake her confidence. I just want to know about the "right" way of using things like chokes and ;punishment-based training. I remember reading in Pat Miller's "The Power of Positive Dog Training" that you were only supposed to issue two corrections or else you'd need to try another method -- a very different scenario from that typical scene of self-educated traditional trainers who correct the dog over and over. So what are the other ways of using such training tools right?
    • Gold Top Dog
    From what I understand, the rules of applying positive punishment or "corrections" (and they are laid out in Jean Donaldson's book The Culture Clash) are, as you state, being willing to apply it forcefully enough the first time to really make a major impression on the dog, and also the timing. The correction has to happen the moment the dog begins to engage in the unwanted behavior. A minute later or even a few seconds later is going to create confusion for the dog and way ups the possibility that your correction is going to have unwanted "fallout".

    Starting with wimpier corrections and then, when those don't work (which they probably won't if you have a pain-resistant independent-minded kind of dog) working up to stronger ones simply acclimates the dog over time to the feeling of the correction and they continue not to care. As with applying a reward, the timing for the correction has to be excellent in order to clearly communicate what the dog did that you didn't like. If your timing is off, you risk punishing something else, perhaps even something good.

    I think the other thing to remember about physical corrections is that sometimes they can escalate the problem instead of solve it, no matter how forceful the correction is. If the dog is keyed up, in an aggressive or fearful state, doing something to him that just proves him right, that this situation is dangeorus and scary, will only exacerbate the problem.
    • Gold Top Dog
    well, if you're going to use a punishment or correction, first you have to be 100% sure the dog is deliberately disobeying. It is rare for dogs to deliberately disobey. When they appear to disobey it is usually because they are distracted or have no idea what you want them to do. Humans often have trouble grasping the difficulty dogs have in generalizing their learning to different settings.
    Then you have to use perfect timing in inflicting your correction. Within milliseconds of the deliberate disobedience. Too late and all you're doing to beating on your dog, not teaching him anything. And you also have to be sure you using a punishment of sufficient power. You don't want to just "nag" at the dog, nor do you want to completely overwhelm the dog with your punishment, just strong enough to convince the dog to never do it again.
     
    It's really difficult to use corrections effectively.
    • Gold Top Dog
    From my understanding, "correct" P+  and R- training would involve the following:
     
    - Immaculate timing, within 1-3 seconds of the behaviour occurring
    - The appropriate punisher - I believe it was Gary Wilkes (I may be wrong here) that said in order for a punishment to be successful, you should only have to implement it once, at the most twice, in order for it to be truly effective. The gradual "Start gentle and get harder and harder when the dog is not listening" is a sign of an improper use of P+, as it is obviously not viewed as a punisher to the dog, but rather as an annoyance.
    - Consistency - every single act of the behaviour must be punished in order for it to bee seen as a punisher. For every successful occurrence of the behaviour, that behaviour is reinforced.
     
    In "most" punishment based training the dog learns through punishment of all behavoiurs "except" the appropriate one. The classic method of teaching a dog to heel on leash in the classic military obedience position is to physically correct any movement OUT of that area (whereas a person working in R+ would be reinforcing for position..note the difference in techniques), causing the behaviour of "obedience heeling" to be R-.
     
    There are a few trainers these days who use R+ to first teach the behaviour, and then once the "human" views it as learned, use ;P+ thereon out to ensure the dog complies, once again punishing any behaviour other than the one the dog was asked to do. It's really not much different at all than teaching by using P+ from the beginning, because the dog still experiences the same effects of pain/intimidation/fear/force/appeasement (different dogs respond differently to P+) that you would if you hadn't used R+ to begin with, and the dog, just like above, learns to do it to avoid punishment, rather than because it was reinforced. Often this problem occurs when the dog has not been proofed in different areas, with different distractions, or with built-up duration.
     
    Kim MacMillan
    • Gold Top Dog
    ORIGINAL: Kim_MacMillan

    From my understanding, "correct" P+  and R- training would involve the following:

    - Immaculate timing, within 1-3 seconds of the behaviour occurring
    - The appropriate punisher - I believe it was Gary Wilkes (I may be wrong here) that said in order for a punishment to be successful, you should only have to implement it once, at the most twice, in order for it to be truly effective. The gradual "Start gentle and get harder and harder when the dog is not listening" is a sign of an improper use of P+, as it is obviously not viewed as a punisher to the dog, but rather as an annoyance.
    - Consistency - every single act of the behaviour must be punished in order for it to bee seen as a punisher. For every successful occurrence of the behaviour, that behaviour is reinforced.

    In "most" punishment based training the dog learns through punishment of all behavoiurs "except" the appropriate one. The classic method of teaching a dog to heel on leash in the classic military obedience position is to physically correct any movement OUT of that area (whereas a person working in R+ would be reinforcing for position..note the difference in techniques), causing the behaviour of "obedience heeling" to be R-.

    There are a few trainers these days who use R+ to first teach the behaviour, and then once the "human" views it as learned, use ;P+ thereon out to ensure the dog complies, once again punishing any behaviour other than the one the dog was asked to do. It's really not much different at all than teaching by using P+ from the beginning, because the dog still experiences the same effects of pain/intimidation/fear/force/appeasement (different dogs respond differently to P+) that you would if you hadn't used R+ to begin with, and the dog, just like above, learns to do it to avoid punishment, rather than because it was reinforced. Often this problem occurs when the dog has not been proofed in different areas, with different distractions, or with built-up duration.

    Kim MacMillan


    I agree with your points, except that I think we need to differentiate between starting gently with equipment, versus starting gently with correction. 
    Bottom line, however, is that I prefer to reinforce the heel, instead of correct any position other than heel.  Not only do I get a dog that understands the heel position as a behavior chain (since it really is a position, not a single behavior)  and "gets it" faster, I get a happier dog who thinks it's cool to be in that position.
    • Gold Top Dog
    I agree with your points, except that I think we need to differentiate between starting gently with equipment, versus starting gently with correction. 

     
    Did I make mention somewhere of making the two seem similar? I don't believe I used "equipment", I think I only used the term "punisher", which would indicate punishing a behaviour regardless of equipment or lack of equipment? Or did I use an inappropriate word I can't see? [:D]
     
    Kim MacMillan