Originally posted June 22, 2013
What do these five phrases have in common and what do they have to do with dog training? Many trainers use them to tell the dog when he has done the wrong thing, and they can be either very helpful or totally discouraging to the dog, depending on when and how they are used and how they are taught. The technical name for them is the “No Reward Marker”.
Like the term implies these pieces of verbal information tell the dog to try something different because no reward will be offered for the marked behaviour. In a savvy trained dog, this can be very helpful information, but in most cases, the dog has no idea what the words mean, what the consequence is or how they apply to the situation at hand.
I have watched for years as my students quack at their dogs (try saying “uh uh”, eight to ten times in a row. Then try doing it in a group; it works out to what sounds roughly like a flock of quacking ducks), while their dogs continue the behaviours they are doing without change.
Training is the art of changing behaviours. When you are working on a specific behaviour and your dog doesn’t do what you expect him to do, you need to have a way to communicate that to him. One way is to with hold the treat or the click and the treat. In an experienced dog this is valuable information and they will continue to try again, in a binary fashion. If you click then the dog guessed right, if you don’t click, then the dog guesses something different. Yes/no, zero/one, on/off, click/absence of click. As long as the dog understands that click means he got the right answer and no click means to try something else, then all goes well. If the dog is unsure, or new to the game, it is not as efficient. The difficulty comes in when the trainer has little experience teaching a binary system to an animal and they are not having quick success. In that situation, the trainer reaches into his toolbox to look for something different. Often the new trainer starts trying to give the dog information in the form of a no reward marker; the trainer quacks and the dog not knowing what it means doesn’t do anything differently. Eventually, the dog does what is wanted and gets his reward and the trainer thinks that this all works out well. But did it? Let’s look at an example for clarity.
The trainer is moving forward with the dog on leash. The dog is pulling and the trainer wants the dog to walk nicely beside her. The trainer stops and says “uh, uh”. The dog continues to pull for a period of time. Eventually, the dog stops pulling and looks back and the trainer clicks and treats. The dog bolts forward and pulls again, and the trainer again says “uh uh” and the dog eventually stops pulling and looks back and the trainer clicks and treats. After many repetitions of this, the pair make it around the block. What has the dog really learned?
Pulling this apart, the dog hears the word “uh uh” while he is pulling. He may learn that “uh uh” means “pull hard on the leash and then stop pulling to get a click and treat.” Hang onto this image because we will come back to it. Not only that the dog is learning that he must pull and then stop pulling in order to get the click and the treat. Not quite what most of us have in mind when we are talking about loose leash walking skills. Finally, consider what might have happened had the trainer kept silent. Would there be any difference between what happened in my example above and what would happen if the trainer had stayed silent? In the short run, probably not. Keep the scenario in mind though.
Coming back to the point regarding turning uh uh into a cue of “pull on leash”, consider what happens when you are standing with your dog on a loose leash and you are working on eye contact. This time, when the dog looks away and you want to prompt the behaviour you are training, eye contact, and you say “uh uh” the might think “that is the word you use to indicate that you want me to pull really hard on leash and then slack off”. You have in effect turned the warning of “there is no reward available for the pulling behaviour” into the cue “this is the time you can get a click and treat if you pull hard on leash and then back off”.
Learning is a really sophisticated process. When we train dogs to do things, we start out with some simple principles and along the way, we may get confused and side tracked. Behaviours that result in desired outcomes will increase. Behaviours that result in undesired outcomes will decrease. And when the trainer gets frustrated because things aren’t going the way they expect, then the trainer will often stray from the basics and add in embellishments in ways that may or may not be productive. If you have gotten this far in the article, you may be thinking that I am firmly in the camp to not use no reward markers. In fact, I do use them, but carefully and with thought and I start out by teaching the dog what they mean.
In fact I use several different categories of information to help my dog to understand what I want. And let’s start from that point; I want to make it as easy as possible for my dog to understand what it is that I want. I start with using a marker to identify when the dog is doing what I want. I use a clicker, and the clicker predicts treats. When I am starting a new puppy, then I might start out by clicking and treating and not asking the dog to DO anything. I just want the dog to understand that the click predicts the treat. Once the dog understands that, I give the dog ways to make the click happen. When the dog understands that he controls the click, and the click predicts the treat, then I work on teaching the dog about shaping; the process of capturing successive approximations towards an end goal behaviour. When I get the whole behaviour, I name it by saying a word while the dog is doing the behaviour. In this way the dog learns a binary system; if you get a click, you did it right, and if you don’t get a click, you should try something else.
During this phase of training I also teach the dog something called a delta signal. Delta signals are signals that indicate that if the dog changes his behaviour, a reinforcer might be available, but if he doesn’t change his behaviour, a punisher WILL be available. Punishment is anything that happens that decreases behaviour. I start with puppies in my very messy kitchen. I live on a farm and our main door opens into the kitchen. If the dog chooses to empty the toy box or manipulate the toys, then I say nothing. If the dog chooses to pick up a shoe or a boot, or climbs me or picks up a dropped glove or sock, I say “That’s Enough”. If the dog drops the item or gets off me, I will give him a treat or redirect him onto a toy. If the dog chooses to keep doing the undesired behaviour, then I say “Too Bad” and the dog loses his turn and goes back to his crate or goes onto a leash for a short period of time. The “Too Bad” signal is technically a conditioned punisher. In this case it is a conditioned negative punisher because we take the dog’s freedom away. Once the dog understands the delta signal and the conditioned negative punisher, I apply those to a variety of situations where I want to change the dog’s behaviour.
|Using a delta signal such as “That’s enough” tells the puppy to make a different choice, and it also tells the puppy that if he doesn’t choose to do something else, there will be an unpleasant consequence. Saying “uh uh” and then not providing any other information is just confusing and is not actually going to change behaviour. Image credit: isselee / 123RF Stock Photo|
The no reward marker I use is “try again” and I don’t use it until I am certain that my dog understands the process of shaping, the conditioned positive reinforcer (the click producing the treat), the delta signal (That’s Enough), the conditioned negative punisher (too bad) and the dog has at least twenty behaviours on cue. Furthermore, I want to actually carefully teach the dog what the no reward marker means. The way that I do this is to play a game called try again. I get out my clicker and treats, and I say to the dog “try again”. No matter what he does, I click and treat. Then I say “try again” a second time and whatever he does OTHER THAN what he did the first time, I click and treat. I keep repeating this, only clicking for new behaviours, so that the dog learns to offer something different than what he did the last time when I say “try again”. As a trainer, you have to be very liberal in your definition of “new behaviours”. I might click for a head turn the first time, and a paw movement the next time and a sit the time after that. Different behaviours don’t have to be different cued behaviours; we are teaching the dog to do something other than what they are currently doing so that they don’t repeat undesired behaviours within the training session. Once the dog understands the concept, I will insert “try again” into training scenarios when the dog is struggling with an advanced shaping technique. It takes a little bit of practice, but with advanced dogs and experienced trainers, they catch on quickly enough to integrate this technique into their learning repertoire, and it can make learning much more efficient.
Once the dog understands the no reward marker, I will play a separate game to encourage the dog to think creatively. When the dog is able to offer different behaviours when I say “try again”, I change the cue to “try something else”, and practice this as a separate exercise. The “try something else game” challenges the dog to try a variety of behaviours just for fun, but I don’t practice this until my dog has fifty or more behaviours on cue. The dog has to have a large bank of behaviours to draw from in order to be successful at this game, and it isn’t a game that is easy for trainers to master until they are pretty good at other behaviours.
When the dog is efficient at understanding the no reward marker, the positive reinforcement marker, the delta signal and the negative punishment marker, then the trainer has some powerful tools in his pocket, but only if they are used properly. The positive reinforcement marker is used to mark desired behaviours. The delta signal is used to mark behaviours that are undesired and should change; but this usually happens outside of the training realm, where the dog is working on a behaviour. I use the delta signal for “naughty” behaviours. The negative punishment marker is used when the dog does not cease doing behaviours that he is not permitted to do and it comes with an undesired consequence. The no reward marker is used to indicate that a different behaviour is desired in a training context.
As a final note, I chose a no reward marker that was two words for a couple of reasons. Quacking is just way too easy; you don’t really have to think about it. I intentionally teach the behaviour, and I want the dog to actually listen for what I am telling him, so a verbal cue is important. “Try Again” is a cue. “uh uh” is a gut level reaction. And once the dog understands that “Try Again” means to try something else, I can modify the cue to “Try Something Else” and modify the activity to be a game of offering a variety of different behaviours.
No reward markers are often contentious in the field of training. There are some studies that seem to indicate that no reward markers don’t work, and yet many trainers like me still use them. I think that there is room for much more research in the field, and we need to start to look not only what they are, but how they are taught and how they are used. Quacking at your dog is usually not efficient because it doesn’t tell the dog anything helpful. Training a meaningful no reward marker can lead to a deeper way of communicating with your dog.