It seems everyone has a slightly different slant on their application of Non Reward Markers (NRMs) in training. Here is a little more on my approach. First of all, I never use them in training during a value building session (shaping). Actually, I rarely use them at all while teaching anything. I can think of only one particular use I have for a NRM and currently, I can only think of two different sceanarios which would prompt me to use one while training my dog.
Don’t get me wrong, I think NRMs are a valuable tool, but for me in my training, they have a very specific use. Today I will explore more of what they are not.
As I mentioned in a “value building” stage of learning (shaping) I will never use one. I know many of you who have written in believe it will speed up learning and there is no harm in using a NRM. Thank you for your input, I do value your opinion but let me share why my differs.
A NRM, although pretty benign, is still punishment. All punishment has fallout. Even if that fallout is mild frustration, over time that frustration will have the opportunity to grow and express itself in ways you may not be able to predict and may or may not be equipped to deal with.
With all punishment I stand by my ongoing mantra;
“You must earn the right to use punishment”
Earning the right means the behaviour you desire has a history rich in reinforcement prior to the application of the punishment.
For example; what if I asked you to shape your friend to pick up the yellow ball from the group of three balls below. What approach would you take? Linear thinking may have you look at that group of balls and say “phew, no big deal, if my friend reaches for the red ball, I just give me NRM and they will reach for another . . . eventually getting the yellow ball and viola within seconds the correct behaviour identified . . . fast and painless”
Although that line of thought may appear “painless” and as a one off approach to training you can argue that it is painless. But I challenge you, we are not just training the subject to select the correct ball here, we are actually doing something far more important. With every task we attempt to train a dog to do, the bigger picture is we are actually building our training relationship.
I encourage you to stop thinking of training individual responses as a isloated training experiences and start thinking that we are not just “building behaviour XYZ” we are establishing, in the mind of the dog, what life is like while working with you. Every session counts; particularly the early ones, you are growing your working relationship.
So back to our balls. Lets have your session now take a less linear and more lateral approach. We won’t consider introducing a NRM to the shaping session because it should be a value building session, we have not “earned the right to use punishment.” We first must establish value for the our friend selecting the yellow ball then, if required and absolutely necessary punishment may be introduced during a “value testing” phase of training.
So how do we value build? In this case, the easiest answer is to remove the yellow ball from the group of three balls. You now have one yellow ball on the table setting up easy reinforcement for your friend to choose to pick it up. Do this several times and value building for picking up the yellow ball (without the distraction of any other balls) has been established at the same time an enthusiastic working relationship has been started (due to the reinforcement given during the session of success).
Life is grand, training is fun, when can we do this again? Thinks your friend.
Now introduce a blue ball –our session is now a “value testing session” as we are evaluating the value of our yellow ball in the presence of a blue one. Now, if the new blue ball gets picked up and NOT rewarded (the mildest and, in my opinion, only form of “punishment” that is required) it will quickly be discarded for the yellow, previously highly rewarded action.
Session ends quickly, with success, no frustration and with both parties itching to get back to training with each other again. Now when you introduce a third colour it is unlikely it would even be looked at as there is so much value for the yellow ball!
Success in your targeted behaviour (yellow ball selection) and, more importantly, tons of value dumped into your working relationship!
Next time I will address the two situations where I feel a NRM is necessary in my own training and would welcome the input from those of you that can see a way I can do without it!
Today I am grateful for the wonderful, candid discussion we have going here. Non judgement, just exploring possibilities. Thanks to all for contributing your thoughts. All opinions welcome!
I absolutely hate the idea of punishment in any form. Positive reinforcement and relationship-building is the absolutely most important way to communicate. I’m in total support of what you’re doing Susan and I appreciate that I am learning more through your experience and expertise.
Excellent explanation and example. Now that’s what I call making the right thing easy and the wrong thing difficult !
Thank you, Susan for this excellent explanation of NRMs. This makes sense to me, and as I am going through Recallers as an alumni, I think I have a pretty good sense of what I have, and I can journal about this while playing the fun alumni games. This gives me good baseline info – on what I have.
I’m sorry but this just doesn’t make sense to me. The example involves human beings – not dogs. So I just don’t see the relationship between the two.
Hello Susan, I recently bought your book, “Shaping Success and have found it a big help. My dogs are only beginners, not up to trialing stage as yet and like many, I’m having trouble with weavers. I am attempting to teach the 2×2 weavers and I don’t seem to be able to get pass step one. They don’t seem to want to approach the poles without some incentive.They look at me, look away and look back at me again and get tired of waiting for some response from me and sit. They will go through with a cue or lure, but other than that, I don’t know what to do. Can you help me?
To keep it simple, we use a NRM in our training specifically for a behaviour that if the dog carried on the behaviour, the behaviour itself is self rewarding and we need them to stop and try again. The best example where we use it is in our 2×2 weave training. If the dog tries a entry and gets it wrong and continues on attempting to weave the behaviour of moving through the weaves continues to be reinforcing to the dog (because they LOVE to weave). Our NRM is trained to halt the dog in a trained but fun and playful way it brings them back enthusiastically and we can try again. “Oops” is a word we use in this case and we can set the training to try again. As explained above we do not use this word in every day life but specifically for behaviours we train in agility or obedience.
Greetings, Thank you for the plethora of educational articles on your site.
I wonder if you could define Non Reward Markers (NRMs) so I can understand this post. Thanks.
The one thing I am learing from Susan Garrett’s method is the joy of training! It’s no longer work, for Kizzy or I, it’s FUN! I look forward to our sessions and get excited when I see such amazing progress! I do find, if I fail to plan properly for a session, that my frustration level goes up and I will revert back to “reward/punishment” training. The dog is fine, I’m the one struggling to change. However, I am learning, changing and expecting more from myself for the better.
unluckily shortly after I got my last puppy my life changed a lot and this meant a lot of stress and very little time. When I was thinking about my relationship to this dog it very well fits to this mail, as in ment to be dog training my relationship is less keen than with my previouse dogs and in daily life he is ok but far more stressed in many ways due to using a lot of non consciouse NRMs, in my opinion.
The best life long relationship and training and real life working I have (now 15) with the dog I followed this much better and it is like engraved and very stress resistant.
Example: Train “new” verbal commands consistently in a given controlled area/setting until the dog realises that doing something in that given area/place means doing/demonstrating something “new” (where by, eventually getting a cue command out of it).
One could then use this cue command in conjunction with a release from activity/done command and reward value to elicit the action.
Low value for, yes, you did something new.
High value for, yes, you did something new that I want.
Positive re-enforcement, but you get even more if you show me something that I want.
From a dog’s color perspective, I think the above 3 colors look similar vs. the potential human induced/obliviousness to the stronger smell differences. Texture might work better.
@ craig, your comment helped me think through a shaping challenge I’m having over here. With successful results. Thanks!
My first thought with the ball problem where you were supposed to use a NRM was, what if the person first picked up the blue ball, you gave your NRM, then they picked up the red ball and again you gave the NRM. They might very well conclude from that, that picking up a ball in general is not what you want, and don’t even bother trying to pick up the yellow one, but instead move onto a totally different sort of behaviour. It might then take you a while to convince them that picking up a ball IS what you want, they just picked the wrong colour.
Great post – this is EXACTLY what happened to me back in the 1980’s when a highly regarded OTCH trainer told me to correct my Sheltie for choosing the wrong scent article in Utility. This was a highly motivated, great working dog. After the correction (which was not particularly harsh), he did not trust himself to make a choice anymore. He would trot to the articles and stand there waiting for me to tell him which one I wanted. It took almost a year of playing fetch with one article before he was willing to choose and risk being wrong again. I learned a LOT from that mistake and that dog.
But now, I’m having a problem with doing this exercise of Susan’s – after several weeks of trying, my current dog still can’t seem to tell the difference between the blue, red and yellow balls… (Just kidding! Thanks to Susan for having such great websites and sharing her knowledge with the world!!!!)
Do you have a cue that tells the dog “don’t wait for my signal, offer me something”?
The “ball” game example sounds like the “hot target” game we learned a while back ago with the dogs!
On the 2×2 video, you introduce the NRM at a certain point. Do you still find it necessary to use it for teaching weave skills?
This exploration of the NRM is sooooooo interesting. I have always thought the “uh-uh” was kind of wrong in many situations. And I was always so surprised it was used in clicker training… I thought it was a contradictory message. Or a counterproductive cue.
(Ok, sorry for my English, I’m a French Frog!)
And I still have many questions about that. But I’ll keep reading you and see… 😉
Thanks for your input and your attitude in doing so is a winner!
p.s. URL field will not accept http://www.jeanlessard.com neither http://www.chienposium.com
re: Maria’s dilema, I may be totally off the wall here, but I think I would go back to behaviour B (touching something with the hind leg) and put it on stimulus control, or if you think you have it under stimulus control, test that and be sure that you do. If a behaviour is under stim control, the dog will perform it on cue, will not perform any other behaviour when you give this specific cue, and most importantly in this situation, will not perform behaviour B when being asked to perform another behaviour, in this case, the steps you are taking to shape behaviour A. Sounds like your dog is just offering a behaviour without the completion of training that beaviour, i.e the cue. Does that make sense to anyone but me?
Just want to chime in an your mantra Susan 🙂
When I wirked for a rescue that was using ALL punishment methods and was looking for a new direction to help their dogs, I drillled the following sentence into the heads of the handlers :
You cannot punish the dogs for what is wrong, if you do not first show him what is right.
Once they understood this concept we moved on to :
If you are consistently showing the dogs what is correct and he choses what is wrong first, have you made the correct choice valuable enough?
These lessons allowed the handlers to start looking for the small little things that the dogs were doing correct and the amount of punishment went down nearly 90% and the relationship between the dogs and handlers went from “poop-picker-upper” , to companioins and working partners.
I donlt like to use punishment at all but am guilty of using it from time to time. But every time I find myself using it a little voice in the back of my head says ” Does the dog even know what the correct choice is???”
Good mantra for those of us that were raised on the punishment system 🙂
I think this post is especially interesting in light of Patricia McConnell’s recent post:
Where she was discussing that BC who knows 1000 words or something because my first thought with the Scorch/Pony conundrum is that Pony has not had equal value built to that of Scorch 😉
Expand the ball question: what if you build value for both the blue ball and the yellow ball separately, how would you set up trials to have the dog differentiate without NRMs? Blue/Red, Yellow/Red a million times then immediately after a successful Blue/Red discrimination, ask for a Blue/Yellow discrimination?
I love the following excerpt from your post: “I encourage you to stop thinking of training individual responses as a isolated training experiences and start thinking that we are not just “building behaviour XYZ” we are establishing, in the mind of the dog, what life is like while working with you.”
This is really something for me to keep in the front of my mind when working/playing with my dogs, I can get too serious and worried about the end result instead of thinking of how we are working together….I was actually thinking about this tonight while I was shaping a trick, I ended the shaping session feeling “over the top” myself, sort of buzzing….wonder how that translated for my dogs….something to work on for me!
I think I would wait until the dog offers any other behavior besides touching the table leg with her hind foot and reinforce it to end the session. Then I would make the adjustments that have already been mentioned (carpet runner/move chair) and start a new session.
@Maria – Would if you used a carpet runner (or a yoga mat that is cut to fit). It would fit between the legs of the chair and be long enough to stick out the front and the back of the chair. The training sessions would build up to the dog being able to crawl on the mat without the chair around. Then put the mat under the chair. Or possibly under a table first and have the dog crawl on the mat under the table.
Or try something different like having the dog back up under a chair. Reframe it and do a similar, yet different behavior.
Or, if you want to use the foot targeting then use a target for the hind feet that keeps their feet under them and not off to the side.
Ha! Your right 🙂
But in the flat we were living then in every possible direction of moving the chair there was something what could be touched with the hind legs (wardrobe, wall, sofa). Maybe just MOVING the chair would have helped, but in this moment I just don’t know. I know she decided that she had to touch something with her hind legs.
Susan, so what would you do in such a situation, when the dog is completely stuck on wrong behavior, there is just nothing to be clicked and you know that for this particular dog stopping the session in such a moment would be extremely frustrating? Of curse such a situation is rare (it happens once to me and my dog in more than six years), but sometimes it could happen (maybe because of some mistakes we made).
@Maria, how about I turn it back to you and others . . . what else could be done in Maria’s situation? I know it is one I have come across . . . and if the rest of you haven’t yet, you are bound to at some point!
I would just wait the dog out. Go to a happy place in my mind and wait for the dog to offer a different behaviour. If you’re looking for a tricky behaviour like targeting with hind leg and then I’m assuming that dog has had heaps of experience with shaping and therefore knows how to fail and recover without becoming overly stressed. If dog doesn’t know how to fail and recover he/she needs to learn that in the context of “meaningless” shaping games like trick training.
Oh, is my English so bad? 🙁 I have an impression that you misunderstood me completely 🙁 I didn’t shape hind leg targeting, it was already done.
Imagine that you are shaping easy behavior A, everything goes right, you are really close and suddenly dog decides that you want her to do behavior B (which is far more difficult ;)). “Yes! you want me to do B! I’m doing B! I’m doing it!! Can you see? B! B! B! B!”. Dog, who usually knows that lack of click means “try something different”, in this moment just doesn’t react, just is doing B again and again and is becoming more and more and more frustrated. What the handler should do in such a situation?
I think I know what you mean. Dog is providing a behavior that has been rewarded in the past in hopes of getting the reward. The dog is going back to default behaviors. Mine does that often. He will do a nice down when he can’t figure out how else to get me to throw the ball. Hmm. In some cases, we break away (physically move away) from that training without reward and try shaping again. I guess this is a type of NRM. How long would the dog try behavior B before trying something new? Wait him out, reward anything new or different from B.
Answer is the same…. wait the dog out, give her time to think and don’t be so quick to jump in and “help”. I’m sorry I muddled up my first reply a little – I know you were shaping under the chair and the dog offered a hind leg target.
Was also thinking that sessions don’t ALWAYS have to end on a high. If things are going bad sometimes its best to cut your losses and stop the session. Then immediately go record keep and ask yourself lots of questions. Was the session too long? Did the dog get enough balance breaks? Was the rate of reinforcement appropriate for the level of training? Did I lump the behaviour or split it? Make whatever adjustments are required and try again later.
But wait, isn’t Susan saying in the case of the chair and the table she would have moved the chair away from the table to prevent the foot targeting rather than use a NRM to stop it?
Either way, thanks for challenging us to think outside the box! I was actually really conscience during this morning’s training session of every time I used a NRM just by habit. Having self awareness and working my dog all the while practicing my front crosses,made for quite the complex morning.
@Linda absolutely the line of thinking I am suggesting. Create value before adding complexity to decrease the need for “more information” while shaping.
@Susan To me, this is what is so brilliant about using your approach with NRMs. It teaches the dog what to do rather than NRMs teaching it what not to do! Back to your “live in the land of DO”. Rather than the “land of DON’T”
When we get stuck in shaping I will wait out the behaviors I am getting but not looking for, find something to reward and then break off the session quickly, to re-think my plan, because usually something I am asking is not being communicated clearly to the dog.
Usually this communication breakdown is related to my rate of reinforcement not being high enough in the environment or for adding value to the correct behavior, or I have not broken the behavior down into small enough parts for my dog.
This is a great discussion I am loving hearing how others handle this, as well!
As usual – very interesting post 🙂 and very interesting problem 🙂
Personally I think that there is one situation, when it is better to use no reward signal in clicker shaping – when the dog is completely stuck on wrong behavior.
I had such a situation some time ago, when I was shaping going under the chair. Everything was going right, we got head under the chair and suddenly my dog dog recalled that we had been shaping hind foot targeting and she started to touch leg of the table with her hind leg. I have no idea why, but she was completely stuck. I think in such a situation, when I have working relationship with my dog already built, it was better (=less frustrating) for her to say “ee-ee” (which is very neutral and means “noo, you are on completely wrong path”) than everything else (like stopping the session, it would be very frustrating for her, we usually are finishing on a high mark).
I think it is also “dog-depended” – different dog can react different.
Except such a situation I think that no reward signal is also useful in weave training, when dog misses right entry (because doing the rest of weave poles is rewarding, I don’t want to reward incorrect entry).
This is exactly where I use the NRM. With my active little dachshund and at times she gets stuck like this and I have found that if I let her go at it, it only gets more intense. I might be doing something wrong ofcourse but even the tries to click the initial behaviour from the start again, hasn´t had the desired effect and she keeps at trying to grab the yellow ball (a behaviour she thinks I am asking for that has been reinforced before) from different angles when I am trying to get her to the blue (a new behaviour we are trying to still figure out that has no previous history of reinforcement).
In this case, I would simply change environments. If you were working on hind leg targeting in that room, try to train the new behavior in a different room and slowly transfer it to the original room. Clear the area around the new training object (chair) and, use really high value treats to reward anything close to the new behavior, and simply wait her out. If the targeting is not on cue, it is only sensible for her to go back to a behavior she was previously working on if she gets stuck.
I’ve found that in training new behaviors, NRMs simply don’t work for me.
In the recallers course, you used a NRM to make the distinction that after an error your body movement towards the dog was NOT a predictor of reinforcement. What else could you do. You could have someone else reset the dog so you wouldn’t have to move towards him.
“We are establishing what life is like working with you”-exactly-the outcome is not so important as the process. I want my dogs animated and involved in ALL training sessions-they have to be okay with making mistakes. I just don’t reward the wrong choice. That is enough of a marker.