Errors in Training: Part 1
I've had a dog (dogs!) make most of the possible errors in the obedience world, and yeah in agility too. Errors of anticipation such as dropping early on the drop on recall or finishing on the judge's cue. Errors of doing the wrong behavior such as sitting instead of lying down on signals and of course tunnel sucking in agility. And all sorts of other errors stemming from anxiety, or over arousal and of course even poor training.
I wish I was that perfect trainer that always had my sessions setup towards errorless learning! But the reality is that I've had a lot of time to think about how to handle errors both in training and in trials!
What do you do when your dog is making a mistake in the moment? Do you reward the error if you use a cookie? Do you need to make it easier for the next time? There are so many questions that I still am thinking about!
As I started to write down my thoughts about errors, it ended up being quite a bit more than I anticipated! So I ended up splitting this post into 2 sections. This first part looks at errors where the dog loses focus, my general frame for keeping the dog's attitude up, and errors of anticipation.
Errors of Engagement
Before I go into how I typically handle errors in training, I wanted to make sure that we are all on the same page. When I'm talking about mistakes, I'm talking about pretty much all errors except errors due to being not ready to actually focus and train.
Whether this is due to anxiety issues or due to some sudden distraction the dog is really focused on, I handle the issue differently than a dog who is wanting to work but for some reason is struggling with what we're working on.
I never want to use my energy or rewards to try and force a dog to stay in the game. Been there, done that! Spoiler alert, it doesn't work. At least not in the long run where you need your dog to work in the ring with my formality and no cookies! How to deal with these issues needs to be a whole separate blog post. I recommend checking out Denise Fenzi's work on acclimation and engagement on her blog, or the self-study course on FDSA.
The only part I want to briefly talk about in this section are errors where the dog checks out due to curiosity versus truly being unsure. I do not want to train my dog that they can check out whenever they want and that I will quietly wait for them to be ready while they go off exploring. For my dogs, opting out of training is a valid option, but it does have consequences in terms of it either meaning I will calmly end the training session or that it will take them more effort to get me started again.
Here is an example I caught of Zumi being distracted during a training session. In this case I know exactly why she was distracted (she was expecting Grandma to arrive at the training building around this time and that is a very exciting thought!!!). And it ended up showing up during her reward time of playing with the toy. If you don't know our playstyle you may not see the difference so you'll have to trust me that while I do encourage "victory laps" with her toy at times, her circling here did not have that vibe to it! She was air sniffing and thinking about the upcoming visit she would have!
I put Zumi in a down stay and kneel next to her. This gives me time to evaluate the distraction and whether I really can resume training with her or not, makes it so that she isn't on a free for all, and the break in work is a small punisher for the act of leaving. When I see that she is able to give me nice focus and I think the risk of her going off again is low, I stand up and ask Zumi if she's ready. Her response is to do her "ready to work" routine of jumping up in the air and then falling into heel position. I then resume work before giving her the first reward.
Ok, back to how I handle most errors in training! When my dog makes an error, my first priority is to make sure that the dog stays in the game. My response to an error has ranged from actively praising it and even rewarding it, to happily interrupting the dog and resetting. That's not a big range!
Rarely am I even "neutral" about errors and that's because I find it super hard to be neutral in real life. When a mistake happens, our body language almost always shows subtle cues in our puzzlement or even disappointment. I find that actively encouraging and supporting the dog through an error helps me to prevent the dog from focusing on any of the brief
What you need to keep that attitude up varies from dog to dog. A cookie before restarting? Sure! Just personal play?!
Here is an example of how I interrupt and reset Grace's heeling for a minor error.
She happened to do a super brief look away and missed my hand touch cue so I talk to her and quickly back up to reset. She also has a history of being reinforced for following me as I walk backward so it's not a huge deal! And I give her a cookie for lining back up at my side. When I go forward again, she drops her head so I stop and try to connect with some play/petting. In reviewing the video I see that while she does a great job on the next attempt, I should have actually paused longer before restarting as she does miss a playful attempt (21sec) and may not have been fully in the game yet.
Do treats reinforce the mistake?!
First, whether the cookie you're giving is truly reinforcing the behavior depends on whether the wrong behavior actually increases in the future. If it doesn't, then by definition it's not a reinforcer!
But more importantly, I use what I sometimes refer to as pity cookies to just keep the dog in the game. I care about my dog's emotional state. If my dog is getting too high and squealing, treats can help calm them down and make the next rep a lot more successful! And if my dog wasn't sure how to win, then the cookie can keep him playing the game and allow me to adjust my strategy.
If you're handing over a lot of cookies for mistakes, then your training plan needs to be adjusted. Giving treats for errors is not a substitute for a good plan that keeps your dog's success rate high!
Errors of Anticipation
A very common error in training is the dog doing what you want, but doing it before you actually cue it! This is especially common in obedience where dogs learn the predictable routines, even if you try to vary the chain in practice!
I try to handle anticipation errors by immediately resetting the dog. And if the behavior is set up for it, emphasize those pauses!
I don't mask errors by not asking for the chain! For example, a common solution for dog's anticipating the drop on recall is to only do straight recalls in practice. I see this as a quick bandaid, not a real solution to the problem! The dogs might not show the error in training but the error almost always shows up in a trial where the dog knows the full chain of come followed by down is going to be there!
I handle this issue by first trying to take the issue out of the formal presentation, but yet keeping the same elements that lead to the predictability of the offered behavior. For our drop on recall example, I might skip the formal stay while I walk away and start from a cookie toss or sending the dog around a cone. I still ask for downs, trying to vary when I cue it so the dog doesn't focus on a specific location. Down right away? down at my feet. And of course down in the middle. Lots and lots of rewards. And sure, sometimes I'll keep them on their toes by waiting until their close and turning to toss a treat behind me or send to a cone behind me. But my goal with that is to keep the game fun vs actively avoiding cuing the drop.
And if you find that your setup is n back to reset the dog with a hand touch and the thrown cookie/cone send again. For others, that's too much pressure and it goes back to my number 1 goal to keep the dog in the game. So I might clap and call them to me when they make an error and play with them before trying again.
And if you find that your setup is not having a high enough success rate? Can you split the exercise down even more to an easier version? Another way to practice anticipation of the drop is to do what I call "follows" where the dog is heavily rewarded for following right in front of me as I jog backward and then randomly cue the drops. Working on the skill of keep coming until I cue the drop, but very little formality attached!
Eventually, I do start rebuilding it back into the formal chain. But I've already taught the dog the skill of knowing to keep running until the down cue. And more importantly, I've already taught them how I'll respond if they make a mistake and that it's not a big deal!
Some behaviors are a little more straight forward with anticipation and allow you to focus on actively rewarding the PAUSE.
Another common example are dogs who anticipate the finish after coming into front. Some are responding to the judge's cue, but many others are just anticipating the next step in the predictable chain!
Focus on teaching your dog that your pause is a normal part of the chain. Same thing with taking a deep breath. Or even looking up at the judge for the next cue. All should be actively trained as part of your chain!
Reward lots for pausing. Gradually building up to at least 2 full seconds. And then, if the dog anticipates, quickly get your dog back into their start position!
Here is an example of Grace learning to wait for the stand cue. This is very early on and she was in the process of learning that waiting is important even if she knew what behavior was going to happen next. Each time she stands I instantly get her back into a sit and talk to her! On the first rep I reset a ton and end up getting her off the target without ever cuing the goal behavior. But that's not a long term strategy as it avoids the issue of the dog learning to wait even if they know what the next behavior is! So the next 2 reps I reward a ton for the pauses again but do end up cuing the stand.
Next time I will be writing about 2 other common mistakes in training and trials. The dog freezing and not responding to their cue, or the dog offering the wrong behavior!