Criteria continued

Bud learned the jumping stand early. I like to teach it to small puppies because they are bouncy and you can often easily capture it by just standing and waiting for it to happen when the puppy has energy and is playful. I don’t care a lot about perfection at this point, I just click and reward any bounce where both front legs leave the ground until the puppy offers the behavior in order to get a reward. I also like to name the behavior early, because I think it’s good to work on stimulus control with my puppy, and they often learn to love this behavior quickly.

I have not planned for Bud to be an obedience dog, but I love to work on obedience behaviors and think they are great for anything I want to do in the future. I’ve used his jumping stand for cue discrimination, and I have added some staying in position after the jump too. But I haven’t really worked on it, and I realized recently that he always moved his back feet counter clockwise on landing. It wasn’t pretty, and he did it every time. And I had rewarded it every time. Now, I decided that I wanted him to stay straight infront of me when doing the behavior, maybe even be able to keep his back feet still and just bounce with his front feet.

If the dog is always performing the behavior in a way we don’t like, it’s not a good idea to just not reward it and wait for perfection. The dog will probably give up in the process. You could chose to reward the better responses, and not reward the ones where he moves his feet the most – but you’re still rewarding something you don’t really like. Another solution is to think about some other behaviors to train that might help with understanding and hopefully bleed over to the behavior in question. That’s what I started by doing. I worked on backing up straight, offering a stand (not a jump, just four feet on the floor) straight infront of me and some rear end awareness and finishes on my right side (moving back feet clockwise).

This helped enough for me to get much better behavior once I started the jumping stand again. There were a lot of offered jumping stands that I didn’t reward, but he did seem to remember the exercises we’d just done when I didn’t reward, and I could jackpot some jumping stands that were much straighter than before. I still rewarded back foot movement (which I don’t really see as a problem) as long as it wasn’t to his right. If I found that he reverted back to his old style, and rate of reinforcement got too low, I would remind him of one of the behaviors used before (often just standing in front of me, or some clockwise pivoting) and then go back to the jumping stand. But most important was criteria – not rewarding what I don’t want, and rewarding all the good attempts.

I made a little video of our training today. He was doing well and had some nice repetitions where he kept back feet still, and even could go from sit to stand with his back feet planted. When I found myself just rewarding many repetitions in a row, I decided to change something – challenge him. The behavior is nice, but it’s not ever finished. Just changing something small can make it a bit harder for the dog and increase his understanding of what we want. I chose to add an open hand with treats to the left. It made it harder, but he worked through it nicely and came out more confident in what I want. I usually don’t want to use a cue for the behavior when I add a challenge like this, so it would have been smarter if I’d let him offer it freely instead. Noted for the next challenge!

Criteria is King

Anytime you train your dog, you set rules that dictate what the dog needs to do to earn a reward – criteria. Without a criterion, you’re just feeding (or playing with) your dog. We’re all aware of this, yet it’s very easy to slip into training where we just reward without thinking about criteria. It seems easy enough in the beginning, when we’re just capturing all behaviors that can lead us towards our goal. We love to reward, and all we need to think about is setting the dog up for success and having good timing.

When the dog is on the right track and repeatedly offers the behavior we’re after, it’s very easy to get stuck on still rewarding all tries. I often get asked about what to do with training problems like the dog moving his feet in positions, chewing on the dumbbell or crooked sits. First thing you need to ask yourself is: Am I rewarding it? It might seem like a stupid question, but my experience is that in most cases, people are rewarding the very behavior they consider problematic. How on earth is the dog supposed to know that we don’t want it?

Of course, one reason why people chose to reward is because they want the dog to get rewarded for working and staying in the game, and they don’t really feel like they know what to do about the issue. In other cases, trainers are just so used to always rewarding that they’re not thinking about the possibility of actually choosing which responses to reinforce. The problem is: You get what you reward. And the longer you reward something you don’t want, the more frustrating it will be to get rid of it. In a perfect world, you set the dog up for success so that you never reward the behaviors you don’t like. In reality, you’ll often reinforce behaviors that are close to what you want, but not perfect, when you begin training. With a lot of behaviors, I don’t find this a problem if you raise criteria and move on quickly.

There are a lot of reasons why trainers get stuck with a behavior they’re not crazy about. There are a lot of ways to tweak and adjust training that will help even if you’re still rewarding all repetitions. But the bottom line is: many trainers need to get more comfortable with setting criteria and withholding reinforcement if the behavior isn’t good enough. In the words of Bob Bailey: If you want more, you have to ask for more. Of course, you can’t ask for everything at once. You need to raise criteria in small steps, but do it often. The longer you stay at one step, the more difficult it will be to get to the next one.

Why am I writing about this? Because I constantly make the mistake, and because I see how quickly I get results when I’m conscious of criteria. I’ll tell you about how I problem solved using stricter criteria in my next blogpost. Please feel free to ask any questions you might have on the subject, and I’ll answer here or in my next blog post.

I also wanted to share these beautiful pictures taken by Thomas yesterday. He’s in the north of Sweden, in the mountains, hunting grouse with old Pax (11 years old!) and Alot (8 months old). They seem to have great weather and some luck with hunting. And Alot is learning a lot about scent, as well as steadiness to birds and retrieving.

Alot on the mountain
Alot retrieving grouse

Learn Faster with More Mistakes?

Summer is coming to an end and things are starting to slow down a bit. It’s been a very busy summer with competitions both near and far away (European Open in Hungary), camps at home and abroad and not much time for anything else. We’ve had technical problems with the shop part of the website, so we had to postpone the start of our next Foundation Class Online. We promise to find a solution soon! I’ll also get back with more reports on what Wilco and I are doing. He just turned 6 months.

I came across an interesting article today regarding learning. Trainers often have the idea that error-less learning leads to faster and better results than learning based more on trial and error. While I like to keep my dogs successful, I’m never afraid of letting them try different things and try again when they’re not rewarded. I was therefore glad to come across these studies that suggest that off-target actions can be a key ingredient in quick learning. Even making mistakes happen (applying a force that pushes the action off target) can make learning go faster than striving for error-less repetitions.

Learn Faster with Messy Moves

Of course, we don’t know if this applies to dogs (although I would think so), and I don’t know about the quality of the studies or if there are other studies that suggest otherwise. Please let me know if you have any thoughts or added knowledge on this subject!

Why use Shaping #3 – Mastering no reward events

One of the hardest things to handle for our dogs in competition is the lack of reward. Not getting an anticipated reward can result in frustration, anxiety or reduced motivation. Those are not feelings that we want in training or competition, so it is critical that we teach the dog to master no reward events.

We want the dog’s reaction to no reward to be to focus and try harder, and we don’t want it to be a big deal to the dog. Shaping teaches the dog this from the start. It is normal in shaping for rewards to be withheld, and the dog has to focus and try again in order to get a reward. Good shaping makes intermittent reinforcement and training for competition easy later on!

Why use Shaping #2 – Division of Responsibility

I want my dogs to actively take responsibility for keeping focus and trying hard when we’re working together. I never want to have to ask them to work, but rather want them asking me for a new task at all times in training or competition. This is, I think, one of the key elements to success in the competition ring. Even energetic dogs that like to work will get lazy and lose focus if we constantly take responsibility for keeping them with us in the ring. Losing the dogs attention when training, or even worse – competing, is very aversive to most trainers and taking responsibility for the dogs focus (for example by asking for it, talking to the dog with a happy voice, moving in a fun way, commanding or threatening the dog) is negatively reinforced since it often works in the short term. Unfortunately, it only makes the problem worse in the long run.

What does this have to do with shaping? Everything! I believe that shaping is the perfect way of teaching the dog a great division of responsibility. I have great rewards, but the dog has to work to earn them. I will not cheerlead the dog to work, nor use the rewards as a lure to get behavior. I am quietly observing the dog and then rewarding good effort. This can be started as soon as I start training my puppy at 8 weeks, and the puppy will learn one of the most important lessons for success in competitions right away. Shaping teaches the dog to be the active part in training, always focused on working to get that great reward – especially when I am just quietly focused (just like in competition!).

Why use Shaping #1 – Instant feedback

One of the things I like most about using shaping (learning based on the dogs own initiative and no prompts or lures) to teach new things, is that I get instant feedback on the clarity of my teaching and the value of the rewards I am using. Dog trainers seem to focus way too much on getting the behavior and too little on actually reinforcing behavior. While luring might get you the behavior quickly, it is hard to tell if you’re actually building value for the behavior or for the prompts used. I think that using help often builds much more value for following our hand/the food/the target than for the actual behavior, and it is very hard to know what is what until you try to fade the prompt.

Using shaping gives you instant feedback on the value of your rewards (does the dog want to repeat the behavior that produces the reward). If you find that your dog loses interest in repeating the behavior, you should look at how you are rewarding rather than introducing a help (body language, encouraging chatter, cues, targets, lures etc.). This is harder than it sounds, it seems to be human to focus on prompting behavior rather than reinforcing it, but when you get in the habit of constantly evaluating value you will get addicted to the instant feedback that shaping provides in a much clearer way than luring does.

Heeling – Where’s the Value?

I taught the last weekend of four in the south of Sweden this weekend. It’s been a very nice experience, with great students that have a lot of fun together. The theme for the weekend was training for trials, so there was a lot of working on sequences and trial-like situations. We did also have time to work on details, and most chose to work on the heelwork. A common problem is that the dog choses the wrong position while heeling – walking ahead of the handler or drifting out to the side (dogs that hang to far back is usually just not motivaded enough). It’s common, and can also be hard to fix after it has become a habit. We did make a lot of progress this weekend though.

The first question we need to ask when the dog is not walking where we want him to is WHY? Why is the dog finding value somewhere else than where we want him to be? Our training is almost always the answer to why there is a conflict between where we wish the dog was and where he wants to be. It can either be an effect of us reinforcing behaviors that are incompatible with a good position – like teaching the dog that heeling is about watching our eyes. It can also be that we place the reward somewhere where we don’t want the dog to be. We might allow the dog to walk out of position to meet the reward coming from our right hand, or that we bring the reward to the dog (in a position we don’t want) instead of having him come to the reward (where we want him to be).

What you reward

It is of course important to reward when the dog is exactly where you want him. We often get annoyed when the dog does something we don’t like, but we keep rewarding it. How is the dog supposed to understand that we don’t want him there if we reward it? If the dog has found a position that we don’t like, we need to find a way to get him where we want him, so that we can reward that. You might have to start with the dog just standing in the right place, or place the dog on a platform next to you. You can also let the dog start behind you and then click and reward quickly when he is in the right spot – before he gets the chance to pass you and walk to far ahead. Re-training the behavior walking backwards can also be a great way to get a new start. Or maybe just reinforcing the correct position when you’re out on a walk.

Where you reward

Training a great heel is so much easier, and the result gets so much better, if the dog feels that the reward comes exactly where you want him. Placing the treat on the seam of your pant can be a good point for you to remember when training. I prefer to let the treat come from behind if the dog has a tendency to walk to far ahead. I usually keep a few treats in my right hand and then take one at a time with my left hand behind my back when I want to reward.

I used to reward in the opposite direction from where the dog had a tendency to go. Today, I’m more into rewarding exactly where I want the dog, but have some criteria for what the dog should do to get his reward. If I reward just by my leg, I won’t let the dog swing his rear end out when he is eating. I want him to keep his body parallel with me. I train this early, when I teach him rear end awareness with front feet on a platform, and when I start the backwards walking. It’s often a good idea to train your routine for rewarding before you start heeling training.

If the dog is good at ignoring rewards held in my hands, it’s so much easier to get good placement of reward. I don’t want the dog to think about the reward until it materializes in position. I can try to lure the dog out of position with treats in my right hand or behind my back, but the dog will only get them if he keeps a good position (I will then take the reward to the dog). I also want to test my dogs understanding of position by using external rewards like a bowl of food on the ground. I work with the dog very freely (I don’t start from halt and I don’t use a cue) and let the dog find his position by my side when I’m walking in order to get his “get it” cue.

Epic learning a big lesson

Having three girls in heat in the house can be a challenge with three intact males. This is the first time that I have noticed that Epic really found girls interesting, and a new world seemed to open for him. Instead of trying to survive these days, we’re making the most out of this high value reward. Bitches in heat will be allowed in Swedish agility trials from this year, so it’s good to be prepared. I will let the video speak for it self:

20 – Smart enough to know the difference?

This post is a bit connected to yesterday’s post about innate behavior. I said that if you don’t want a behavior in training, don’t allow it to be rehearsed in other places in your life. It could be that your obedience dog is making a lot of noise when waiting to get of of the front door, or that your dog is running around on his own with toys between repetitions in training, or that you’re rewarding a lot of blind crosses in freestyle, but don’t want them for agility. I’m not saying that these things have to be a problem. I know a lot of dogs that run around with their toys a lot, but still have a great retrieve. And I know a lot of dogs that make noise in some situations in life but never in obedience. You don’t need to create problems that aren’t there. But if you are having a problem with a behavior that seems hard to solve, it could be a great idea to look at what happens 24/7 with your dog. And with a puppy, you really don’t know what will become a problem, so I won’t allow my puppies to rehearse behaviors I don’t want to see (flanking me, eyeing rewards or equipment, blind crossing me, running with toys on their own in training etc.). I will rather build value for the behaviors I want in as many situations as possible.

I do have some problems with obedience behaviors that are contradicting what I’m trying to teach my agility dogs. In agility, I want the dog to come straight to the front of me if I’m facing the dog. In obedience, I want the dog to aim for my left side and go deep before turning around (all you guys doing front and finish for obedience have an advantage here 😉 ). This might sound like a small thing and people usually say “don’t you think dogs are smart enough to know the difference?”. I totally believe that dogs are smart enough to be able to tell the difference between different situations. It’s just that it takes a lot more training. And if I want the dog to make out the difference between agility and everything else, I will need to do a lot more training on equipment. I like to train my agility dogs a lot without equipment. It’s both easier (can be done anywhere, any time) and is more healthy for my dogs. Since I try to make agility handling rules part of our every day interaction, I don’t have to do a lot of training on jumps. If I reinforced a lot of contradicting behaviors in every day life and in training without agility equipment, I would have to spend a lot more time on the agility field. It’s of course possible, but I wouldn’t do it just because I’m to lazy to think about what my dogs are doing when we interact outside of agility.

We had a wonderful time herding this morning. A thin layer of snow and the rising sun. To my great relief, all the “naughty” behaviors we want in herding (stalking, flanking, blind crossing etc.) does not seem to affect other areas of our life and training with the dogs. 

Teaching heel

I have finally been able to make a little video to show you some of my thoughts on teaching a good heel for obedience trials. Heeling is difficult since it is a duration behavior and has many different parts that all need to be good. I split heeling into it’s parts and train one behavior at a time. This blog post is about the basics of heeling – the dog walking with me on my left side with perfect attention, attitude and position. Turns and straight sits is another part that I work on separately. Maybe I’ll write another post about that some time.

When teaching the dog to heel with me, I start by walking backwards and have the dog follow me. Doing this has many benefits, like:

  • It’s easier for a puppy to follow me when I walk backwards and I can get more rewards to the dog.
  • It’s easier for me to see what the dog is doing without having to turn and bend towards the dog.
  • Since it’s not the final setting for the behavior, I’m not worried about shaping one aspect at a time. I would not like to reward the dog for good attention but bad position if I was walking forward. By walking backwards, I can shape a good behavior and be pretty sure that I will get good behavior to reward once I start walking the right way.
  • I use the walking backwards-heeling for teaching stand, sit and down from heel.
  • It’s very good to have when training in a distracting environment, as the young or inexperienced dog has better chances of success.

When I start the training, I walk backwards away from the dog and mark and reward all attention the dog gives me. I’m not interested in the dog looking me in the eyes, as that is incompatible with a good heel position. As long as the dog is giving me attention, I will reward it. I keep the treats in my left pocket, or in my right hand (only if the dog can ignore treats in your hand). As i mark the behavior I want, I get one treat from the pocket (or right hand) with my left hand and reward at my left side. In the video, the rewards are pretty calm. This is because my dogs have great position by my left side and because they don’t need to get more active. For most dogs, I would recommend that the dog gets to chase the treat in your hand in a straight line and end up with his shoulder at your left knee. I show this with Squid at 1:05 in the video.

When the dog is giving good attention, I start to look at the dogs attitude. Some dogs need to get more active and with them I will mark and reward things like coming closer to my side, ears up, going from a walk to a trot, tail up, head up etc. I will also make sure that all rewards are active and that the dog has to chase the treat after the click. With some dogs, it’s a good idea to work on how to reward separately, so that the dog knows to come close and steal the treat from your hand on the click before you start training heel.

Other dogs have a little too much attitude. They might be jumping up and down, making noise or touching you too much. With them, you have to do the opposite. Walk a little slower and reward calm, quiet and rhythmic behavior. Stop when you reward and give a few treats for just standing still before you walk again.

When the dog shows great attention and attitude, I start shaping a good position. For a lot of dogs, this comes for free because of good placement of reward on the earlier stages. I want the dog close to my left side, straight in the body and far enough back. Watch the video to get an idea of the finished behavior.

When the dog can do a nice heel while I walk backwards for a few meters, I start to turn around. This is demonstrated by Squid in the video. It is an advantage to have worked on some rear end awareness before doing this, so that the dog is willing to swing his rear end in before walking forward.  In the beginning, I will reward the dog as soon as he finds position by my left side when I walk forward.

There is of course a lot more to be said about heeling, especially when thinking of all the problems that might arise. If you have questions or comments, please post them below.