Humans are notoriously verbal creatures. We love to talk, and we do so automatically, even when the person we're talking to doesn't speak our language, can't hear what we're saying, or even when the "person" isn't a person at all.
We're so used to verbal instructions that the ideas of stilling the chatter during early learning and adding the cue after our dogs are offering the behavior seem ridiculous. The clicker, however, we understand. The clicker may not be verbal, per se, but it gives clear, audible feedback that the dog did what we want. It seems only logical then that we need another signal to tell our dogs to try something else when they offer something we don't want.
This signal is called a "No Reward Marker," or NRM. NRMs are intended to be a verbal cue for extinction, not a punisher, so people attempt to say them in the most neutral tone of voice possible. "Uh-uh," said quietly and calmly, is a common NRM. In a training session, the trainer would either click or use the NRM after each repetition to let the dog know whether his behavior was correct or not. Although they sound logical, NRMs are not without problems—and controversy. The biggest problem is that NRMs may not be as neutral as we want them to be. Imagine yourself on a game show...
You know that every right choice will get you closer to the grand prize. After a right choice, a bell sounds. After a wrong choice, a buzzer sounds. How do you feel about that buzzer? Is it neutral? Is it simply telling you "wrong choice," or is it increasing your stress level? Would it be any different if the sound were a beautiful harp chord instead of a buzzer? I doubt it. Now imagine that you weren't given any feedback for wrong choices except the lack of the "right choice" bell. Would you be as frazzled by the puzzle-solving process?
Some dogs will take NRMs neutrally, simply as information offered. Others will view it as a mild punisher. In the short term you might not be able to tell how an individual dog views the NRM. The effects may be clear only over the long-term, as rate of emitted behavior falls off and the dog becomes reluctant to experiment with new behavior.
You may be lucky enough to have a dog that takes NRMs neutrally, but that may not be true of your next dog. Another problem with NRMs is that they're habitual. Once you, as trainer, are in the habit of giving an NRM, you will do so nearly automatically. It can be very, very difficult to break that habit if you need to work with a dog who finds them punishing. NRMs are also habitual for the dogs. They learn to rely on them, expect them. If you get into a situation where you can't give that expected feedback, the dog can become confused and anxious.
Fortunately, NRMs aren't required for training. Though they make sense to us verbal humans, the reality is that you can communicate the same information simply by withholding the click. The click means "You did what I want!" No click, then, means "Try something else." By clicking and reinforcing the choices you like and ignoring and not reinforcing the choices you don't like, you allow positive reinforcement and extinction to work together in a powerful way. Reinforcement makes those behaviors stronger and more likely to occur. Extinction makes the other behaviors weaker and less likely to occur.
The process for training without an NRM is simple. At the beginning of the session, set your criteria—decide exactly what will make you click. Each and every time the dog achieves that criteria, click and reinforce. If the dog offers anything else, including a sub-par version of the goal behavior, simply do nothing but reset for another repetition.
The bottom line is that NRMs, though logical, add an unnecessary level of complexity to training. Keep training simple for you and for your dogs: forego NRMs and stick to basic positive reinforcement and extinction.
Learning About Dogs - ©2003 Learning About Dogs/reprinted here with permission.
I agree, the vast majority of the time that people want to use an NRM, it's unnecessary. It's a valuable tool in a very few situations that far too many people use in all the other situations. :-)
Diane, I answer your question with another question -- how do you determine if an NRM is truly necessary? Why are you using it? If you have a very good reason as to you need an NRM rather than simply a lack of R+, then you next need to determine whether this individual dog is aversely affected by it or not. But if you're using an NRM because the dog didn't do the correct behavior, well, maybe you don't really need one; you can simply withhold R+.
I do use an NRM -- "try again" -- but it's used in rare and very specific circumstances. It *does* obviously motivate fresh behavior in my dogs, rather than inhibit or even keep the same rate of emitted behavior; a friend called it their "reset button." But overuse would kill that effect very, very quickly!
How best do I determine if my dog sees the NRM as a punisher or just a chance to try something else?
Post new comment