Tuesday, February 22, 2011

Warning Sounds and Non-Reward Marking

In my last post I discussed reward marking. How telling your dog with sound, word or body that what he does at the moment will be reinforced can expedite training success.
Do we have an opposite to that? An equally helpful… non-reward marker? The good news is that, indeed, we do. The bad news, that it is not as simple as merely vocalizing disapproval with our dog’s action.
Expressing disagreement with someone’s actions is a deep-seated behavior, and the human choice word in English is “no”. I have a T-shirt with a print of a smiling, goofy looking, tongue-lolling-out dog, with its voice bubble saying: “Hi, my name is No-No Bad Dog. What’s yours?” Sadly that comes close to what I hear regularly in my line of work: many “nos” and a lot of very popular “tssts”. Articulations meant to warn the pooch that his behavior will surely be disciplined, corrected, punished – you choose your word cause it is all the same, so he’d better knock it off. NOW.
Even though the “no” and “tsst” are clearly indicating that unruly Rover is certainly NOT getting a reward, they are not considered NRMs, non-reward markers, precisely because the intent is to threaten the dog with a punitive consequence. Unlike NRMs, such warning sounds, other than perhaps giving the owner the illusion that he’s the pack’s leader, are useless. Useless, because typically an action that impresses the dog does not follow the warning, and if there is no consequence the dog doesn't change his behavior, learns to ignore the warning, and eventually will tune you out. The fact that most people “no” or “tsst” all the time, day in and day out, for the same misbehaviors, for the lifetime of the dog, is evidence that they’re ineffectively nagging.
Warnings do work when backed up with something that hurts or frightens the dog in a big way. Then, indeed, a verbal sound or shock collar tone might be all it takes to remind him to toe the line. But aversive consequences intense enough that it stops an unwanted action permanently is abuse, and has fallouts most owners don’t bargain for. The dog avoids the misbehavior, and by extension you and where he lives, or where the punishment took place. Your companionship is in the ditch; your dog resists you, might attack you, or if too intimidated redirect aggression against someone weaker. A dog who fears the power of hands, who feels conflicted and uncertain when his person approaches him, reaches and touches him, or whenever he opens his mouth cause it announces possible imminent pain, will not want to be a human’s faithful friend and working partner, and the owner is missing out on something beautiful he could have had.

The purpose of a NRM is not to warn the dog that something unpleasant will happen, but like the reward marker is meant to strengthen your relationship. It does that by giving the dog clarity with a precise piece of sound information, many trainers like “oops”, that tells him that what he is doing at the moment will not yield a payoff and he’d better think of an alternate behavior to gain access to the resource he is after. So, the fundamental difference between “no” or “tsst” and “oops” is that the end result is not a punishment, but a reward, and with the NRM you’re helping your dog get it.
The Law of Operant Conditioning states that any behavior ignored will become extinct, so couldn’t we just ignore the dog’s unwanted behavior and wait till he comes up with the right one on his own? Patiently let him self-learn what pleases us; what causes us to release the resource he wants and we have control over? We could, and in fact some trainers argue that ignoring is better than non-reward marking, but in my opinion, a clear directional when the dog needs it can decrease frustration and increase attention. To understand when your dog really needs information and then give it, or when self-learning is better, can be a bit tricky and takes practice, but is the fine nuance between good, and fabulous dog training.
Ideally, be attuned to your pooch and apply both, with the caveat not to overuse reward and non-reward marking. Micromanaging can lead to a dog who becomes perpetually dependent on some sort of sound that tells him how to function; he operates purely mechanically and might disconnect from you.
In addition, non-stop noise is annoying. A friend and colleague shared with me a story she heard at a lecture with Certified Applied Animal Behaviorist Kathy Sdao. During a training session with dolphins, in which she used a NRM as information, one of the aquatic mammals became so irritated that he threw the speaker out of the water. He knew exactly where the constant clatter came from, and so does your dog, and you never want to be perceived as an irritant to be avoided or get rid of.
If the dolphin was that frustrated with simply getting information, can you imagine the impact constant warning sounds have on an animal? If you decide to give non-reward marking a go, ensure that you don’t use a word that was a threat in the past your dog has learned to ignore or fear. That word is spoiled and you need a new one. “Ah” works for me, but if you used “ah”, or a sound close in tone as a warning before, try “oops”, or “oh-oh”, or whatever falls naturally on your tongue. It doesn’t matter what you say, important is how your dog perceives it. How he should perceive an NRM is: “Mom says I’m barking up the wrong tree – better do something else to secure the loot.”
Trainer and author Gail Fisher, who calls the NRM Lost Opportunity Marker or LOM, uses a visual signal. She turns her head. Averting the head is also part of a dog’s natural communication and signals that he’d rather not deal with the situation; that he is worried, or not granting audience. Although I really like the subtleness of body language, the head turn obviously only works if the dog is watching you while being baaad. And I see another problem. Because it is dog-speak and expresses avoidance, I wonder how he’d perceive my head turn? Disengagement, displeasure, disinterest in further interactions? If that is the case, the LOM changes from informative to aversive – back to a warning. Even though no real punishment ever follows, but a reward once the dog changes his behavior, a sensitive pooch, for example one who was treated harshly and is shell-shocked, or a very timid one like our Will, could lose the courage to work and connect with that person right away. For that reason, I rather stick to my neutral “ah”, and if I want to add a visual signal I shake my head, cause that is not natural to dogs and my intention can’t be misconstrued.
After you tell your dog with an NRM that he’s strutting the wrong trail, should you use a command that tells him what he should be doing instead?
Gail Fisher gives the dog a second chance right away and cues the desired behavior. She explained that she does so because she aims for immediate command response, but also because she wants to keep the dog in the game; prevent that he gives up if he feels that the window of opportunity to earn the reward closes for good.
For the most part, I like to give my dog some 20-30 seconds to think of a way to get the booty, thereby combining information with self-learning.

You can play around with cuing or waiting, verbal or visual, as long as you remember the two fixed and crucial aspects: the information you give your dog should never be threatening but assisting, and your dog’s next best behavior has to be reinforced, rewarded, and ideally with what he wants at the moment. Then the NRM becomes one more tool that helps make you and your dog an awesome team.

No comments:

Post a Comment