This is a guest post by Aimee Davis. If you are interested in writing a guest post for the StaleCheerios blog, please check out the guest post guidelines.
My 15 month old German shepherd dog Shelby is a wonderful dog. Wonderful, but not extraordinary. She isn’t Rico the border collie or Einstein the parrot or Clever Hans the horse. She isn’t the stuff legends are made of. No one is beating on our door asking to do his/her dissertation on Shelby. She took over nine months to house train, she doesn’t know any “cool” tricks, and we struggle with her reactivity to strangers every day. She does exactly one thing exceptionally well – she herds.
That’s right, as in herding sheep.
When it comes to herding sheep, Shelby is a rock star. Some of her talent is instinctual, but mostly, it’s been shaped through training using operant and classical conditioning. My trainer and I have taken what came hardwired in my “ten in a million dog” and molded it into something that’s not only useful, but breathtaking. And we’ve done it all using positive reinforcement based methods.
Herding is one of the few dog sports that clicker training and positive reinforcement training have barely touched. This is a real shame, because herding really requires the dog to not only be bonded with the handler, but also requires the dog be able to make decisions based on the situation. Clicker training and free shaping readily prepare dogs to be able to do this.
However, it has become increasingly popular to use shock collars in herding training, probably because of the distance aspect of the training itself. Many trainers also say that you can’t train herding dogs without the use of aversives because it is such a high drive sport. I think the opposite is true. Because it is such a high drive sport, you need to be very careful about aversives, because the last thing you want is a frustrated, angry dog who wants to redirect his aggression, pent up frustration, etc. onto something weaker than him, like a…I dunno, sheep!
A short introduction to herding
To understand how and why positive reinforcement works in herding, you must first understand a little about herding. Herding can be divided into three main styles: driving, fetching and tending.
Driving is what most people think of when they think “herding”. Driving is where driving dogs (like blue heelers and Australian shepherds) get behind a flock of sheep and push them forward. These dogs are sometimes referred to as ‘heelers.’
Fetching is where fetching dogs (most notably the border collie) meet the sheep at the front and stare them down or use their bodies to turn the sheep and bring them to the handler. These dogs are sometimes referred to as ‘headers.’
Tending is where tending dogs (like German shepherds and Briards) act as a living fence to keep large flocks of sheep together while grazing. Tending dogs move the flocks, but typically at a slower pace than driving and fetching dogs.
Cross training is possible but uncommon due to the differences in necessary physicality. Where border collies may fetch by using their strong eyes and short, quick bursts of speed, the German shepherd must use his stature and endurance to tend at a calm trot for hours on end. Fetching and driving dogs are somewhat more interchangeable than driving and tending or fetching and tending. Australian shepherds can be taught to head and border collies can be taught to heel, but it’s much harder to get a German shepherd to work in bursts of speed (they’re simply not as quick) and as well, it is much more difficult to get a border collie to trot all day long (they’re not built for endurance).
Shelby shows us she’s ready to herd!
Shelby is a tending dog. She started her herding career by taking a herding instinct test when she was just a little over six months. A herding instinct test is a test used by all forms of herding instructors to check whether or not the dog has the right “drive” or “instinct” to excel at the work.
Shelby’s +R (positive reinforcement) herding instinct test consisted of two parts. During the first part, she went out to a pen with three sheep in it and sniffed around. Carolyn Wilki (my trainer) was looking to see if she would run around the sheep or try to head them off. This part of the test is designed for driving and fetching dogs, but Carolyn puts all the dogs that come for a test through it, just in case their instinct is non-traditional. She wants them to be involved in the kind of herding they will be best at, even if it isn’t what everyone else would expect. Shelby bombed. At the time, I was upset, because I thought that meant she wasn’t cut out for herding.
The second part of the test, however, required Shelby to approach a pen of about 30 sheep. As soon as she saw the larger flock, she took off over the hill. She was off leash for the entirety of the test. However, Carolyn has a 60 some acre fenced in paddock, so Shelby was going to have to come back eventually. When I was about to call for her, Carolyn told me to just wait. Pretty soon, Shelby reappeared. Before I knew it, she was running up and down the fence line. I looked at Carolyn who told me I had a “tending puppy”. Great! What’s next?
Shelby’s herding training begins
Well, first I had to learn the rules at Raspberry Ridge. I consider myself a positive reinforcement based trainer. Shelby has been clicker trained since she was 9 weeks old. However, according to Carolyn, I was not positive enough. She had some rules and some homework. First, I was to cease and desist with any use of a no reward marker. Second, I was no longer to tell my dog “no” in any capacity. Third, I was to stop counter commanding. (Counter commanding is when a trainer gives the dog a cue for a desirable behavior that will be incompatible with the unwanted behavior the dog is currently performing, such as saying “sit” when the dog jumps up on you.) Fourth, I was to work on one and only one behavior – “lie down” (which was to be retrained using that specific phrase and no hand signal).
Fifth, I was to purchase at least a 50 foot long line. Sixth, I was no longer allowed to feed my dog treats from my hand. Seventh, I was to stop doing any exercise that encouraged Shelby to look away from my hands and look at my eyes. (Instead, I was to reward any voluntary eye contact). Eighth, I was to reward all calm behaviors, at least fifty times a day, including eating, sleeping and settling into a sit or a lie down.
Some of those are specific for herding. For example, because you use a lot of directional signals with your hands in herding, you don’t want to play the “look away” game promoted by many clicker trainers. This is where the trainer holds out a treat in an open hand and the dog must look away from the hand and at the trainer’s eyes in order to earn a click and the treat. That may be very useful for sports like agility and rally, but in herding, it’s a no go.
Some of those rules are because Carolyn saw that Shelby was developing a slight anxiety around people. I really wish I had started rewarding calm behaviors with Shelby from the beginning, because as soon as I started doing this, Shelby became a much better behaved dog overall and she also started to cuddle more, an added bonus.
But mostly, the rules come from the $150 dollar approach to training that Carolyn espouses. Basically, it works like this: if the dog isn’t doing or going to do more than $150 worth of damage, ignore the behavior until it extinguishes, then reward an alternate behavior you can live with. If the dog is doing more than $150 worth of damage, go get the dog. (Please keep in mind that herding dogs are encouraged to begin training no sooner than six months of age and should already have basic obedience skills, so this isn’t going to be the same advice you’d give for something like puppy teething).
It’s crucial in herding that the dog exudes confidence. The dog needs confidence because she needs to work independently and make decisions based on her training. She needs to be confident in her handler, and she needs to be independent and strong willed in order to maintain hundreds of animals that are each about twice her weight, or even four times for a smaller dog such as a border collie. For that reason, Carolyn doesn’t like saying “no” or using no reward markers or counter commanding. Using these techniques is akin to nitpicking mistakes.
Say you’re out walking your dog off leash, and he goes to chase a squirrel. You scream “No!” or you counter command and say “Come.” Both of these things point out to your dog that “You’re doing something that I don’t like, now stop doing it.” If you use these techniques too frequently, it can make the dog anxious, like a child who has been corrected too often. It’s a self-conscious feeling, constantly wondering if you’re going to be told you’re getting it wrong again. Maybe that’s anthropomorphic, but when I stopped using these techniques around Shelby, she gained confidence I didn’t know she had, and she’s evolved into a much better dog as a consequence. That’s convincing enough for me.
Using management tools to set the dog up for success
In all forms of positive reinforcement based herding training utilized at Raspberry Ridge, there is a huge emphasis on management tools. You arrange the environment to set the dog up to succeed, not fail. In traditional herding, it’s very common to start the dog on the stock right away and simply punish (or “correct” if we want to put it nicely) everything the dog does “wrong”. The approach at Raspberry Ridge is to set up situations so the dog is unable or unlikely to do anything “wrong”.
That involves the use of a lot of fences. Sometimes, multiple layers of fences. The stock is far enough away that the dog isn’t overstimulated, but she is still interested. From there, we simply shape the behavior we want. You want the dog to tend calmly on a border? Easy, you set up a fence with the stock far away at first. Every time the dog tends calmly, you reward with a verbal marker or a click.
We use the clicker for a lot of foundation work, but we also use it in the field. However, the clicker works as a bridge in herding, not as an end marker. One of the reasons you don’t see a lot of clicker training in herding is because most people use the clicker as an end marker. The dog hears the click and the dog immediately turns to the handler to look for the treat. In herding, the dog shouldn’t be encouraged to turn away from the sheep. If my dog turns away from the sheep to look for a treat, one of the sheep might wander off. If the dog catches the wandering in time, he will most likely be forced to growl, snarl, snap or “grip” the sheep to move it across the border, which could have been prevented. Worst case scenario, the sheep wanders into the woods and is lost, or eaten. So, turning away isn’t something you want.
There are a couple ways we avoid this hurdle when using the clicker in herding training. The first is by “charging” the clicker with getting to work the sheep instead of food rewards. Sheep are intoxicating for working dogs, so most of the time, this is pretty easy. Click, when a cookie doesn’t come, the dog looks around, you put him in a place where he sees sheep first thing, that’s rewarding all in itself. The other thing we do is to ALWAYS throw the treat over the dog’s shoulder while he isn’t looking at you. So you click and the cookie kind of drops out of the sky. It takes some practice, but especially if you’ve done a lot of shaping or free shaping with the dog, it doesn’t take long for the dog to see the clicker not as a cookie maker, but as an indicator of moving in the right direction.
Later in training, the reward in herding is typically praise, plus the dog gets to keep working the stock. In the beginning, when the dog is unsure and still learning, we use treats as well. When the dog isn’t calmly trotting (i.e. he’s running, doing ‘helicopter tail’ or barking), you simply wait. If he offers the right behavior, you reward that. If he doesn’t and it continues for a while, you remove him from the stock without saying a word and let him run out some of his excess energy or move the stock further away, so the situation is not quite so stimulating.
As the dog gets better and better, you move the stock closer and closer. Then, you remove the fence and start at the same far distance on a long line. Next, you remove the long line and start at a far distance without a long line. You don’t add a cue to the behavior until it is absolutely solid. Shelby has been tending for over seven months now, and we still haven’t added the cue to the tending behavior called “border”. We’re close, but we’re not there (as you can imagine, patience on the part of the handler is crucial, and I’ve been developing it in spades).
Driving and especially fetching behaviors are sometimes taught using body pressure. This is essentially dancing with the dog. Without ever coming into physical contact with the dog, you use the angles of your body and a slight step into the dog’s flight zone to get him to turn. When the dog is going in the right direction, you ease off the pressure and reward the dog with a verbal marker or a click and getting to continue working the stock.
Could this be considered aversive? I suppose it could. But in terms of what other trainers are doing to teach the same behavior, I’d have to say we’re headed in the right direction. What I’m talking about consists of you turning your body at an angle into the dog while the dog is several feet away and then immediately backing off when the dog turns. For a well-trained dog and trainer, this is similar to two people dancing. One dance partner uses body movement and pressure to guide the second person in the correct direction. Although negative reinforcement is at work, for good dancers, the pressure becomes a form of communication, rather than something aversive. This is what we strive for with our herding dogs as well.
At the beginning of this article, I wrote that Shelby isn’t spectacular. I didn’t put my dog in that light to degrade her or put her down. I love her to death, and I think she’s the greatest dog on planet earth, but I also realize her limitations. I wrote that so that people would realize that this training doesn’t just work for “some” dogs. It’s not just for “smart” dogs or “soft” dogs. Shelby is an average, everyday German shepherd. She’s nothing special to anyone except for my family, and yet, we have done this amazing thing with her. She could be and was taught to tend sheep, live sheep, using positive reinforcement based methods. It can be done, and not just with extraordinary dogs. It can be done with regular dogs.
Herding has been trained using harsh, painful, coercive and sometimes just downright cruel methods for a long time now. Those methods obviously work. But these ones do too, and they take the same amount of time as the other ones with a lot less potential for rehab later. So given the alternatives, why wouldn’t you just try the +R way?
Aimee Davis blogs at mymegaedog about all things related to “living, loving and learning with her German Shepherds.” Check out her blog to learn more about her dogs, Shelby and Panzer, and to follow Shelby as she continues her herding training.