Ep 53 How to Measure Learning
===
[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.
hey there everyone. Today I want to talk about how, as dog trainers specifically training for agility. How do we measure learning? How do we know when it's okay to move on to the next step? How do we know that it's okay to make something more challenging? How do we know? That the dog is ready [00:01:00] to increase the criteria in some way.
And so I did some research because that is who I am and the what I found is that researchers commonly use seven ways to measure learning. The first one is errors. So if we count how many errors from one session to the next, when we see a reduction in errors, we can assume learning. The next is topography.
This is the form a behavior takes. So when you set out to go train something, you have a picture of it in your mind of what you want that finished product to look like. That's topography intensity. So this can be more or less intense. And this. Also we have to define for our own dogs because I do believe that, um, temperament and [00:02:00] just their, their physical ability would play a role in what intense looks like for each dog.
Um, speed. So any change in speed can, um, indicate learning latency. Latency is the time that passes before a behavior is performed. So when a cue is present, could be a gesture, a physical cue, a prop, a verbal cue. But when a cue is present, the amount of time between the dog observing that cue and the dog beginning, the behavior is latency.
So high latency would be a lot of time passing and low latency. Would be very little time passing. The sixth measurement is rate. This is the number of occurrences per unit of time. So [00:03:00] this would be the number of reps in the session that you just did. And then the last is fluency. And fluency is the combination of errors and rate. Okay, so after that brief introduction to the seven different measurements of learning, how are we applying these in agility? So when I think about reduction in error, I immediately think of the 80 20 rule. This is how I was originally taught training. Uh, and while I might not completely stick to it or think that much about it nowadays, it is still how we can measure how many errors are happening within a session.
So the 80 20 rule is basically we want 80% successes and 20% errors. We don't [00:04:00] want any more than 20%. Errors before making something more challenging. So we're looking for at least 80% success rate before moving on. So this is very clear to me that if in your first session you had 70 30, and then in the next session you have something closer to 80 20, that's a clear reduction in error.
And you can, uh, measure learning in that case. So this can be applied to any of your training sessions, if that is a helpful tool for you. Like I said, I don't think so much about the percentage of errors as I do some of these other things, and that could be because I tend to have a very. Low rep rate anyways.
I don't do a lot of, uh, high rep training sessions. And also because I rely [00:05:00] heavily on setting up the session for success, I don't tend to have enough errors in a session to really concern myself with this, uh, at least not complete errors, which is why I think. S these other measures of learning can be more useful, especially when we're going back and reviewing our videos.
So the second one is topography. And I've only ever heard this talked about one time in the training, uh, things that I have observed and been a part of. And it wasn't talking specifically about agility. Um, the presenter was talking about. Building marker cues and what those behaviors look like. But topography is what we want the behavior to be, what form we want it to turn into as the finished product.
So what do we want it to look like? You know, what is the [00:06:00] dog doing? How is the dog moving? You know, where are they putting all those different feet? What are they doing with their head? All of those things go into the topography of the behavior and. Obviously when we're training something, this might change throughout the learning process.
And one thing that comes to mind, especially when I was training Sprint, was her running a-frame. She was not a hundred percent confident that gravity. Existed and that she wasn't just going to fall off if she added any sort of air between herself and the board. So she really did take a lot of additional strides over the A-frame before pouncing into her box and then hitting the ground and collecting her reinforcement.
But it was always very clear that she knew to. Pounce into the box, but how she got to the box [00:07:00] changed dramatically over several sessions. And so it was very easy for me to observe. Well, she's doing a lot of extra stuff with her feet. I don't wanna make it harder yet. I want to stay here until this looks more like what I want it to long term.
So by observing. That she was getting it correct. She was, you know, she was getting into the box. She was getting reinforcement every single time. So an 80 20 error rate rule wouldn't have helped me. Here it was looking at the topography of the behavior is how I measured whether I should move on or not.
Okay. So intensity is the third and is. A great measure, I think, especially when we're training agility dogs, because a lot of us are looking for those intense and flashy and snappy behaviors. So [00:08:00] what immediately comes to mind for me is thinking about those really forceful nose targets. And how we create dogs that go from really softly touching our nose to those ones that really drive their face into our hand and they press into you and they keep pressing into you.
That's a level of intensity that we can choose to look for, especially in certain behaviors like that. The next one is speed and you know, agility people. We know we want our dogs to be fast, and this one is tough for me personally to use as a measure of learning because I find that speed is so dependent on.
The dog's motivation, the dog's confidence, the dog's temperament, the dog's body type, the dog's innate [00:09:00] desire to move their bodies at top speed. So this one is mostly for me, a measure of how much do I break things down and how much do I play with the motivating operations so that I can make sure that the dog is capable of going as fast as they can.
For the behavior that I'm hoping to train. So again, this one is not as super important for me, but it's always something that I'm observing because obviously we want our agility dogs to go fast. The next one, however, is really important to me when I'm working with my behaviors and it's latency. So latency measures the time.
Between a cue happening and the dog starting the behavior that goes with that cue. So if my dog is sitting in front of me and they're in a sit and I ask her to lie down, so the cue here is lie down [00:10:00] from the moment she hears that cue, the time before she begins to move her elbows down and have her elbows hit the floor.
That is the latency. So low latency means it's happening immediately, and high latency means that it's taking longer for the dog to respond because verbal cues are becoming so much more important in our agility handling systems and just in our dogs' lives. In general, latency is one of the most important measurements of learning that I use this.
Means that if I give a verbal cue for my dog and they stare at me blankly, even for a moment, I am not ready to increase the challenge around that verbal cue. In fact, if my dog goes more than three seconds without [00:11:00] responding to that queue, I end that rep. I release them out of the position that they're in.
I go into the transport portion of my loop and I reset that loop and I try again. This is because I don't want to, um, bake in too much latency. I don't wanna, don't want to reinforce the dog taking too long to respond, and I also don't want to fall into the habit of re-queuing a behavior. Or then helping the dog too much, find the behavior.
Instead, I'm going to reset that, uh, loop for another rep and I might change something. I might go ahead and make that easier until I see the low latency that I'm looking for. And then in the next session or so, build back up. So latency is super important. I want my dogs to hear those cues and think of [00:12:00] them as green lights and they.
Speed off as soon as they hear or see that green light queue. The next one to measure learning is rate. And this is just how many times did they do it in a set amount of time. And so I don't think so much about using rate because I don't try to. Um, get as many reps done in say one minute or three minutes.
I tend to go out and train and pre-plan, then I'm gonna do five reps and take a break. Um, but if I do go out and just train for a certain set of time, I can use the rate to help me, um, define how many errors that I have and then that will help me determine. The fluency of the behavior because the [00:13:00] final form of measuring learning is fluency.
And fluency is defined as rate and errors combined. So if I, so this is very similar to the A 20. This is very similar to just noticing the reduction of errors. Um, but it is important to keep an eye on not just that we had. You know, three errors. Three errors out of how many. If you have three errors out of five, that's not great.
If you have three errors out of a hundred, who cares, right? So there's, there is some importance around those two numbers working together, and so hopefully this helps you think about your training and what you are paying attention to. And what you are looking for in terms of deciding whether it's time to move on.
I think the [00:14:00] 80 20 rule is fantastic and certainly useful, but it's not the whole picture. Those things like latency and intensity and the topography of the behavior also are really important here. And if we're only paying attention to success fail, we might be pushing on our dogs a little bit too hard, but we also might not be progressing soon enough, or we might be missing where we can make things easier or more challenging for the dog when they're ready for it.
So if this was super interesting to you or if you would like to continue talking about it. I am having these conversations inside my free community, fans of FX Agility. You can find a information on how to sign up for that free community on my website, fx agil school.com. I hope you will join us [00:15:00] and let's just keep talking about the science of training and let's get better at it.
All right. Until next time.
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 my next episode. In the meantime, you can find all of my offers on my website, fx agility.com. Happy training.
