ep 63: errorless learning procedures/coaching call
===
[00:00:00]
If you train, compete, or teach dog agility, you're in the right place. Hey there, I'm Megan Foster, creator of Fostering Excellence and Agility. Join me as I share key insights on all things related to dog agility and help find your team's path to excellence and unlock your best year yet. Let's get started.
Hi everyone. The following recording is an edited version of a recent coaching call that happened inside my free dog trainer community fans of FX agility. I'm holding these regular coaching calls, so if you are interested in joining in the next discussion, check the show notes for the link to sign up, and I hope to see you with the next one.
So [00:01:00] first, um, errorless learning procedures. This, um, this idea, this concept definitely gets tossed around in dog training discussions, and there's just a lot that we can clear up about using the term errorless learning or errorless learning procedures. Or errorless teaching procedures, things like that.
So I want to have a discussion and answer your questions and hopefully clear some things up. Um, first, the myth that gets passed along, like we can say that it is poorly named. Um, but. It's only poorly named if you don't know the history of how it was named, in my opinion. So the myth is that it is a requirement that you [00:02:00] produce no errors in order for it to be an errorless learning session, and that's not true, right?
Not when we're talking about specifically. Errorless learning procedures, right? We can say that a session without errors was errorless, and that would be true, but that doesn't mean that an errorless learning procedure was applied necessarily, right? Or it wasn't intentionally applied. So, um, errorless learning procedures, I posted about this in the community.
Uh, it was, uh, originally an experiment done with pigeons teaching them to peck a red disc and not peck any other colored disc. So this experiment took the red disc and made it very bright so there was a light behind it. So birds are more likely to peck [00:03:00] at something that is brightly colored rather than dimly colored.
So, uh, the researcher back lit the red disc and did not backlight the green disc. So obviously the, the pigeon was more interested in pecking at the red disc, and I'm sure in a one minute session, the rate for. Pecking the red disc was super high, uh, because it, the session was set up to put the animal in, in a successful state, right?
The, the context and the conditions were set up in the animal's favor. And then over time, the green disc was also lit up, but for shorter periods of time, so if for the entire minute session, the red disc was lit up the whole 60 seconds, the green disc was only lit up for [00:04:00] five. To start with I don't actually know the exact details, but I do know that. The, the amount of time that the green disc was lit up was different throughout, um, okay.
And we could assume that every session the green disc was lit up for longer and longer, but now we have a substantial reinforcement history for pecking the red disc. And therefore pecking the green disc has never paid off for this pigeon, and it probably didn't happen that often.
Okay? So that's where this term errorless learning procedures came from is just teaching a pigeon to discriminate between a red disc and a green disc. And you know a lot in dog training we hear about these. Uh, [00:05:00] learning and behavioral principles and, you know, through the grapevine, the definitions of these terms change, and I think this is kind of where the confusion and some of the misinformation and the myth behind things can happen.
Uh, so I don't think it's necessarily important for us to all, you know, run out and buy copies of learning and behavior, but if you do, I think it's kind of informative, um, of seeing how much it is different in the way we talk about dog training in comparison to research and behavior. And, uh, you know, what's, what we've kind of changed and morphed to kind of fit.
Our needs for dog training. So that is, that is the history of errorless learning procedures and how it came to [00:06:00] be. So how it impacts our training. And especially teaching our dogs things in, in my opinion, it is, the goal is to intentionally plan your training to favor the dog as much as we can to take into account their, who they are, their temperament, what they like, um, environments that they do best in. And their learning history and develop our training sessions intentionally around that with the idea in mind that the correct choice is the brightest one in the room.
Right? Is the one that is lit up the most for the dog, right? I am certain we have all heard be a splitter, not a lumper. Doesn't this sound [00:07:00] like that? When we're splitting finely enough, you are effectively lowering the intensity of the wrong choices in the training space, right? And when we lump, we are probably adding too many wrong choices to the context making too many.
Of those of the behaviors we don't want to see available, maybe we're even lighting them up either intentionally or unintentionally. Maybe we're making whatever we're doing when we're lumping we are making the wrong choices more obvious than the correct ones.
Where I think this gets tricky is that we can't just look at the thing that we're training [00:08:00] because I think we'll be surprised. I think we will miss something if we only look at the one behavior that we're attempting to train.
So, like I said before, the environment creates a lot of availability for the dog. When we're talking about behaviors, when we're thinking about these experiments done on pigeons and a laboratory, we're talking about a blank room with two discs available. The choices are all very controlled for.
In dog training, we are not training in a laboratory. This animal that we're working with has a complete learning history and worldview, not just the learning that was specifically chosen for them, for [00:09:00] the experiment. So I never expect. Uh, dog training to go as easily or seamlessly.
But my gosh, it's nice when it does, like, it's always like, yes, I nailed it. I'm so good at this. When it happens and when it doesn't go exactly as we planned, we take that information into the next one, into the next training session. So, like I was saying, the environment plays a big role here. In what choices are available to the dog and what choices, um, are most obvious to the dog?
Okay. The potential behaviors, both correct and incorrect are available here, right? So this is the dog's learning history. How many times have you started? You're [00:10:00] trying to train one thing, but your dog keeps offering a behavior that you were just training or that you've just started training. You might not have asked for it recently or even that day, but the dog is like, there are expectations of me.
What is the most recent file that's been opened? And they start offering that behavior. And that can derail our training that we've tried to set up the dog to offer, you know, going through a set of weave poles, but they are stuck offering spins to the left, right? We've seen it, we've experienced it. We have to take it into, uh, account when we're planning our training sessions of what possibly could the dog throw at me in this situation and what am I gonna do with if that happens?
We also have to be aware of the potential variations of the correct [00:11:00] behaviors, right? Because so much in agility isn't they pecked the disc or they didn't peck the disc. There's, well, they stopped on the teeter, but they stopped in two on, two off instead of four on. But so much of that was correct. They did all the weave poles, but they skipped one in the middle and they didn't notice, so they, you know, they kept weaving.
Gosh, so much of that was correct. What do we do about it? We have to at least be aware of the potential behaviors that could show up because it will always be more than red disc versus green disc. Because the more errors that we avoid, especially in the early stages of learning, the more likely we are to avoid [00:12:00] them long term.
Because every time a dog rehearses a wrong behavior, it's being essentially lit up via rehearsal. Even if we don't reinforce it, even if we, you know, respond to it perfectly every single time in the perfect way, when they rehearse it, it stays put. Um, I think of this of like, you know, when I open a document in on my computer, it shows up in the recent tab whether I looked at it for a second, whether I looked at it for.
10 minutes. Whether I immediately says that was the mistake, I didn't mean to open that file. The computer remembers and so does the dog. Okay? It will stay in the recently opened tab until enough other things have been opened that it falls off.[00:13:00]
So that means when a dog does make an error. I hope that you're reminded to get curious as to why was that choice the most obvious one for the dog,
and how can I communicate that? I would like a different response better, that I would like something different from them. How can I set that up to be more obvious to them? Is this something with the environment? Is it something with their learning history? Is it something in the way that I cued the behavior?
Is there a misunderstanding in what the behavior is? Right? So if we get curious about why was that choice the most obvious we can. Better set up [00:14:00] future training sessions for a higher success rate.
Most. Training projects out there, you know, if you're following some plan, some curriculum, we're pretty good at acquiring most agility behaviors in an errorless fashion or close to it, right? We are quite good nowadays about splitting things down and building them up. Piece by piece,
right? If we think about all the different ways to train weave poles, many of them are, you start at a place [00:15:00] that reduces the number of behaviors that could possibly happen, right? And over time, the the weave pole environment shifts so that more and more behaviors are available. But the reinforcement history li lies with the correct behavior of going through the gaps in the correct way.
Um, Melissa asks, so if a dog makes an error, do you change your own handling to ensure the correct behavior for the next repetition?
Maybe, um, if I'm talking about sequencing. This is again, where we are not in a laboratory. There are so many factors when we're, when we are focused on handling and sequencing and getting the dog to follow, uh, any [00:16:00] number of cues. So my kind of go-to is if, if the dog goes off course, it was. 99% it was me that there are so few times that the dog goes off course and it's because they were ignoring the handling.
Something about the handling queued them to go wrong. So I, in that situation, absolutely. I'm watching the video, I'm getting information from whoever I'm training with and I'm making some changes. Right. But in that case, I'm changing something because I don't think that they actually made a mistake. I did.
If we're talking about something like refusals I, I think it is more fair to. Teach the dog a high level of commitment so that the handler doesn't have to be perfect in their handling in [00:17:00] order to get the dog to commit and stay committed to obstacles. So I tend to rearrange the environment, maybe use a pre placed reward or some other prop in order to make sure that the dog can maintain a, uh, maintain commitment without me having to.
Changed my handling a whole lot, assuming the handling was adequate, right? It doesn't have to be perfect. And then when we're talking about like obstacle skills and maybe my handling distracted them from doing the correct thing, I might ask the dog to do it again if I thought that that challenge was fair.
Because remember,, I'm not afraid that the dog made a mistake, and so if, if I am always changing something to make it easier, I might not progress. But if I feel like the education and the learning history [00:18:00] supports this question, I will ask it again and as close to the same way as possible. Most of the time, if my assumption is right and the dog really did know it, they get it right and we can move on if they get it wrong.
Again, I am now curious about, oh, why? Why can't you? And it might mean that I am changing something to make it a little bit easier, so making my distraction a little less intense or. Um, setting it aside for another day, going back and watching the video and trying to sort out which piece of the environment was actually keeping the dog from performing the behavior.
And I know that was a lot for one little question, but it's a big question actually.
I also kind of [00:19:00] like, especially in coursework, I like to lean in the direction of. If I'm not sure that the dog can get something correct on the first try, I am very likely to pull that sequence out, you know, and at and test it without so much speed. So if I'm not sure that the dog will choose the correct side of the jump or choose the weave poles over the tunnel, I will ask them that question separate from the coursework.
So then I can at least observe, you know, what is the dog thinking about in this context? You know, how interested in the, in all of the choices, are they, because, you know, most dogs on the start line are, they're, you know, or if they're in a stationary position, they're kind of scanning [00:20:00] what could possibly be next.
So this is also information because if I set Sprint up in a, in a sit, and she can see weave poles and she can see a tunnel, if she's staring at the tunnel, I know that that one is more brightly lit up for her. And if I say weave in the moment that she's staring at the tunnel, I know that I'm taking a risk.
So I can look at where she's looking in this moment and go, oh, you, you could not possibly choose the weave poles from a stationary position, let alone at speed in the middle of a course.
But if I'm in the habit of asking these test questions over time, I will eventually learn. That she either needs more individual training on that type of question [00:21:00] or that she's nailing it every time and that I can ask her to do it at speed eventually.
So being more intentional with the questions that we ask and thinking about. What is most likely for them to do is the whole goal, right? And I have taken this kind of to the point where if I ask my dog a question and she gets it wrong the first time, I immediately take a break right there. And I think about what I've just done.
I think about, did I ask her a fair question? Do I think that she's, you know, [00:22:00] what is she? Is something else happening in the environment that I don't know about? Was this a poor choice? Do I need to change anything? And then I can either choose to go back to that training session as is. I can choose to adjust that training session and then start over.
Or I can just choose to move on. And this has served me a lot, uh, because I, especially when I'm going back to a setup that's been on my field for a few days, and, you know, I might train through things on day one and then come back on day two and try to piece some things together and the dog suddenly can't do something that they were just doing 24 hours earlier.
Rather than dig in and be frustrated that they're not doing it, I lean in the direction of, you know what, [00:23:00] she did this yesterday. She does know how I'm going to assume the, the area just needs to be saged and you know, maybe it just needs a cleanse and we're gonna move on.
Almost always, the next time I come back to that situation, it has been solved. There was just something that I could not perceive that day that was blocking her ability to do the behavior. And it can be things like, what did we do that day? Is she in an exercise deficit? Because that's bad news for my high drive, high arousal border collie.
If she's in an exercise deficit, almost everything is harder than if her exercise needs are met. [00:24:00] Um,
is she feeling a hundred? She happens to be very clear and non stoic. So usually if she's not feeling well, I, I definitely know, but so many dogs are stoic and at random they might not be able to do something because they just might not feel well. Um, how are you feeling? Are you training and you're in a good head space or are you only, are you training because you think you should be training?
You know, so almost always when it starts off poorly, I pause and ask, do I really want to continue down this path? Because I'm ver, I'm very aware of what's in that recent tabs folder, not only for what they're rehearsing, but also how we're both feeling about [00:25:00] that behavior in that training context and also.
Like how I respond because if, if I dig in and into the frustration of, Ugh, I really, I need to just make this happen, then that's likely what I'm to remember. And when I go to the trial and I see that challenge, I will remember the, ugh, the one time she couldn't do it and I had to take all day and I had to make it so easy and I had to break it down.
Into preschooler talks, there's no way she's gonna be able to do this at the trial versus leaning on the evidence that I do have of her ability to do it. So not only is it going to impact their mindset and how they feel about training, it's really going to impact your mindset about what you think that they know and what you think [00:26:00] that they're capable of doing. Is the one I was just talking about of they made an error, so you immediately start making it easier, way easier than they logically should need. For the skill level and reinforcement history that they have, and this can be problematic because then you're stuck always kind of starting over, right?
I would much rather skip a training session and assume it's not a problem than start over and make something really, really easy when I have evidence that they can do. A lot more different variations, right? So there's a lot of nuance and every situation is going to be a little bit different and a little bit unique and bring a whole lot of [00:27:00] information to the table.
But I definitely want to avoid starting over all the time or feeling like I have to start at step one or two to build up to five or six. Every single time I train, I wanna avoid that if I can.
The other thing that we can get into in trying to be, you know, whatever label you wanna use, like a positive reinforcement based trainer or a reward-based trainer, or kind of leaning into the myth that we have to reward everything. We, and I'm very good at this too, is that sometimes we can lean too heavily in the rewarding effort space [00:28:00] because, and, and again, this is where Nuance and the individual needs to come into play because we are doing both.
We're teaching them how to do something. And teaching them how to feel about something. So a lot of times if the dog is giving their 100%, but their 100% is for in that moment, you know, it's less than they have to give. Right? Their motivation could be higher. Their desire to do the thing could be higher, their enthusiasm could be higher.
Then I need to pivot my training sessions so that I can feel good about rewarding all of the effort that they have to give, right? , Like with my terrier, I never told him if he skipped a weave [00:29:00] pole, it wasn't worth it because it wasn't worth telling him. He made that one tiny error. Because it was only gonna happen if his drive and enthusiasm were up, so there's no way I was going to tell him, oh, I love your enthusiasm, but you skipped one.
So no cookies for you. I would never because most of the time he was, his motivation was kind of on the lower side, so he would never skip a poll when his motivation was lower, because you can be more accurate when you're going slower. Right. So all of my rewarding was kind of focused on maintaining his motivation for the thing versus my border Collie sprint.
She got to do that maybe a handful of times when she was first learning how to weave, but beyond that, she was like, if I skip polls, I get finished more quickly and I get to keep running more quickly. That's fantastic. [00:30:00] Me telling her, Hey, you actually need to be precise about this. Did not diminish her drive nor motivation one bit.
She was like, oh, well you could have just said so from the beginning. Fine. Right. So there there is a lot of things to discuss and things to consider when we are. Choosing what to reward and how to reward.
And sometimes I think that we can lean a little too heavily in the, oh, I'm rewarding the effort, or, that was a really good try, but you might not need to in that situation. So that is the other pitfall. Um, and I see that. I see it happen most, and it certainly happened to me, um, with running Dog Walk training where the dog [00:31:00] gives an incredible effort but actually doesn't meet the criteria and we reward it.
And when I say we, I am a thousand percent in this category. I've done this. I'm trying to avoid it. Now. I'm trying to fix the error of my ways, especially with the dog that I have, because the same thing applies. If I wanted to reward more to maintain, drive and motivation, I would change the criteria in order to see the dog meet criteria more frequently, if that makes sense.
So with my running dog walk example, if I'm trying to.
Maintain drive and motivation, but they're just barely missing their target. I would either make the target bigger or move the target so that their [00:32:00] effort and the criteria match
versus if I don't have to worry so much about maintaining their drive and motivation, I can maintain the criteria that is clear for them. We have to make sure that we're navigating that space of, I need you to do this accurately. I need you to be motivated to do this and not muddying the waters between those two.
Thanks for listening. If you enjoyed this episode, I really appreciate some feedback. You can leave me a review, engage on social media or share this with a friend. I hope you'll be back to listen to [00:33:00] my next episode. In the meantime, you can find all of my offers on my website, fx agility.com. Happy training.
