By Sophia Yin, DVM, MS January 7, 2010
Hi Dr. Yin,
Thank you so much for the wealth of information that you provide on your site and through your youtube videos. I'm an aspiring trainer and am in the beginning stages of learning free shaping with clicker training. I watched your youtube video where you free shaped your dog to put her front two feet in a box. It was really helpful to see that done, as I know that 101 Things to do with a Box is a recommended beginner clicker training exercise, especially when working with a crossover dog.
I am, in fact, working with a crossover dog (a ten year old Springer Spaniel mix who I've had for a year and who I've trained with luring and clicking and other positive methods up until the last couple of months) and I noticed a pretty big difference in my dog's response to the exercise vs. your dog related to raising criteria. I noticed that when you raised criteria (didn't reward for something that you had rewarded for a second ago) your dog persisted. She offered additional behaviors until she got back on track.
I did the exercise for about five minutes. I try to keep a high rate of reinforcement, but my dog tends to get stuck in a rut (looking over at the box). When I try to raise criteria, he seems frustrated (whining, staring at me) and sits down rather than offering additional behavior. In five minutes, we got to the point where he stood next to the box and dipped his head into it. When I started to try to wait for him to do any foot movement, he laid down.
I don't want to stress him (or myself) but I am so thrilled by the benefits of shaping that I want both of us to be able to do it. However, I've been working on shaping in particular for about 6 weeks now and this is where we are…..is this normal? Are there specific things with regard to raising criteria that I could be doing differently? Any insight you could provide would be very much appreciated! Thank you so much!
Caroline Shriver, Portland OR
Good job sticking with the exercise. Free-shaping, meaning capturing a behavior you like by clicking to mark the correct behavior and following with a reward and then systematically rewarding behaviors closer and closer to your goal behavior, is always a good exercise whether you're a beginner or have been doing it for many years.
What's particularly interesting is that you find certain steps work really well for many individuals but many do better following a different path. It depends on the goal behavior you're looking for as well as the individual's past experience. I'm going to provide some tips for shaping and raising criteria at the end of this blog, but first I want to address some interesting traits of individual dogs.
Individual Dogs Have Different Learning Styles
I used to think that the behavior of just waiting for the human to provide a clue, the way your dog is, was a result of a dog being a cross-over from traditional force-based training methods. Or, that the dog had always been lured such that he never had to problem solve. While this is probably the case to some extent, sometimes it's more than that.
For instance, Zoe, the red Australian Cattledog from the shaping video, was actually a cross-over dog. When I first got her, I was still using some old-fashioned leash correction techniques. Yet she was pretty good at problem solving and offering new behaviors anyway. Then I thought, maybe it's because I didn't use that many corrections throughout her life but instead focused on rewards and careful training of the various steps in a systematic manner. Now that I have my Jack Russell Terrier, Jonesy, who's received about a thousand times more rewards than even verbal punishments, I can definitely say, sometimes the learning style of waiting for a hint is somewhat a characteristic of the dog and sometimes it has to do partly with what he's learned before. It's just like people. Some employees have to be told exactly what to do or they stand around and do nothing. Others can figure out what you want on their own. The first type of employee can be trained to work more independently, but it may require some work on the trainer's part!
Jonesy the JRT Goes into Auto-Mode
So, here's what Jonesy does that's similar to your dog. Often if I really just wait for Jonesy to offer behaviors that I can capture and shape I'll get stuck because he wants to perform behaviors he knows. One of the behaviors I reinforced about a million times is lie down and remain lying down. For the first several years of life, when we were outside, he had almost no ability to remain stationary if I was just standing next to him. I had to always have him sitting or lying down and on an appropriate reinforcement rate or he'd be pulling at the end of the leash. For instance, I would take him on “field trips” on a daily basis and practice rewarding sit or down-stay while I was randomly standing stationary—pretending to look in a store window or adjust my ipod. So it makes sense that, during free-shaping games where I just try to capture him offering a behavior, when frustrated, he'd try lying down. As a result, if owners are going to want to do lots of free-shaping for practice purposes, I tell them they should start before they train a really good stand, sit or down-stay. If your dog's really good at just waiting around for a minute, then you'll have to wait around much longer than a minute for him/ her to offer different behaviors! Don't worry, you don't need to delay the stay behavior for long. Maybe just a week.
Another interesting characteristic of Jonesy is that he has a tendency to get an idea of what I want in his head and then just repeat the behavior over and over with little ability to listen to the cue I'm giving him. For instance, if I'm practicing his cue discrimination by running him through a portion of his repertoire of tricks just on verbal cue-sit, spin, twist, right, left, down, play bow, he may do very well indicating that he clearly knows the verbal cues without visual signals too. But then if I add in something like “back-up” he may then just back-up over and over and now have problems doing the other behaviors. That is, he may back up a set amount of steps when I tell him to sit, and then when he's the distance he thinks he's supposed to back-up, he may sit. It's like his brain or ears shuts down and he just performs the exercise he thinks I want. My dad's 8-month old puppy, on the other hand, stops and listens and thinks about what I am asking for or want.
Even the Two Family Cattle Dogs had Vastly Different Styles
Jonesy's not the only dog who get's stuck in repeat-mode. While my last cattledog, Zoe, was great at thinking and trying new behaviors, my dad's first cattle dog was just like Jonesy in terms of performing the wrong behaviors over and over but in a very eager, dopey, happy way. Interestingly, once I shaped weave poles with Zoe and Rudy and took data. Despite their completely different learning styles, they learned in virtually the same number of steps.
Tips on Shaping Behavior and Raising Criteria
Don't worry if your dog's not a genius at free-shaping. You'll learn much more from him than you will from a dog that makes mental leaps and figures things out on his own. Here are some tips.
- Be aware that the location of the food reward is important. Animals want to orient to the direction of the treat. So if, every time your dog looks at the box you toss the kibble/treat into the box, he'll quickly learn that the box is something good and he should orient toward it. After that you can deliver the treat when you're sitting away from the box, but if he gets stuck you can reward a few more times by tossing the treat in the box.
- Reward a step enough time so he's sure of what he's doing is right. For instance, you might reward the same exact step 5 times before you increase your expectations. So if your dog reaches into the box with his head, reward this behavior 5 time and then raise the criterion to reaching into the box with his head and slightly lifting one foot. By rewarding something easy a bunch of times you're gaining what is called behavioral momentum. If he has lots of success doing something easy at first, he will be more resilient to giving up when it gets a little harder.
- If your dog starts to get stuck you may need to decrease your expectation before he gives up. Note that sometimes I wait quite a bit of time for Zoe. I can see her thinking and trying to solve the problem and I have first rewarded her quite a bit for the previous step. But when I get the feeling, based on my experience with her, that it's too hard, I reward something easier a few times.
- Take a break if needed. Free-shaping can be a big brain strain for animals. So if you stop after just a minute or two and then come back to it a even just a minute later, he may do much better.
- Have a plan or stop to revise your plan if your pooch does something unusual.
- Have a clear picture of what you're rewarding in your head. If you don't, you don't won't know what you want either. (Refer to the Indonesian moodle blog).
- Practice on people and have people practice with you. Humans have all the same problems playing the part of the pet as dogs do. You'll have a better appreciation of what the pet goes through if you subject yourself to someone else's training. (Stay tuned for upcoming blog on Saturday, January 9th).
- Sometimes behaviors don't have to be completely free-shaped. You can manipulate the environment to increase the likelihood the individual will do something close to what you want so that you can get started. Just be sure you only do it few times and then don't need it anymore. For instance, if you're trying to train your dog to head over to a certain part of the room, you can toss a treat in that direction a few times. Better yet, do something that makes them orient in that direction but requires no movement that the dog can see on your part. Or if you want him to turn, say 45°, you can place something he likes such as a toy at 90° to the direction he's facing. Then as he starts to orient to it but well before he's turned the 90°, you can click and treat. Once he orients several times, then remove the toy and see if he'll orient in that direction since you're rewarded it already.
- Stop while you and your dog are still having fun and experiencing success.