These are my notes from the 2012 Art and Science of Animal Training Conference.
Click here for more notes from this conference.
If you’ve been around enough clicker trainers and positive animal trainers, you’ll find that everyone has an opinion about shaping, luring and targeting. And, trainers often don’t agree about which of these methods are best for training or if certain ones of these methods should even be used at all. Kay Laurence gave a very interesting lecture about this subject at the 4th Annual Art and Science of Animal Training Conference. This post is some of my notes and thoughts from her lecture.
Ways to get behavior: Shaping, luring and targeting
Shaping, luring, and targeting are three possible ways for teaching an animal a new behavior. Many clicker trainers are big fans of shaping, but have various opinions regarding how much a trainer should use luring or targeting. Here are a few definitions, just so we’re all on the same page.
Shaping: The trainer waits for the animal to do any approximation of a behavior and rewards these attempts. The trainer moves through a progression of steps, each step being closer to the final goal. This method is the most self-directed of the three methods that will be discussed in this post and the animal has the least guidance from the trainer. (Here’s a video of a rat being shaped to do a simple behavior.) Visit my article “What is shaping?” for even more info about this training technique.
Luring: The trainer uses a piece of food to guide the animal through the motions of the behavior. During this method, the animal is highly dependent on the trainer for guidance and direction. After the initial training, the lure is gradually removed. (This video shows a finished behavior that was originally taught using luring.)
Targeting: The trainer teaches the animal to touch a target. Then, the target is used to initiate the behavior or to guide the animal through the behavior. Depending on how the target is used, this method can be a more trainer-directed type of learning or a more self-directed type of learning. (Here are several video examples of using targets during training.)
The test: Which method is best?
Kay Laurence wanted to know which of these methods was the best for teaching a behavior. Would self-taught (free-shaped) behavior be more reliable or resilient? Would luring result in an animal who fixated on the food and didn’t actually learn the behavior? Could a trainer get the same end result no matter which method the trainer used to teach the behavior?
To test these ideas, Kay taught three behaviors to three of her border collies. The three behaviors were stepping into a box, going out and around a cone, and walking along a line on the ground. Each behavior was taught to one dog using luring, one dog using shaping, and one dog using targeting. So, for example, the first dog was taught to step in the box using luring, to go around the cone using shaping, and to walk along the line using targeting.
What were the results?
All three methods were successful for training and all the dogs were able to learn all of the behaviors, regardless of the teaching method used. However, the learning process and final outcome were slightly different for each dog and each behavior, depending on the teaching method.
Over all, Kay reported that the shaping method seemed more successful for teaching the step into the box behavior and the luring method seemed more successful for teaching the walking along the line. All three methods worked equally as well for teaching the dogs to go around a cone.
So, for a trainer who is skilled in all three methods, one of these three methods of teaching is not necessarily better than the other two. When deciding what training method to use, you should consider the particulars of the behavior, as well as other factors that might affect the teaching process and the outcome.
Choices, choices, choices
So, if all three of these methods can work, how does a trainer know which one to choose? When formulating a training plan, begin by considering all your options. Consider your skills, the animal’s skills, the particulars of the behavior, and your relationship with the animal.
These different methods require different skills on the part of the animal and on the part of the trainer. For example, for luring to be successful, the animal has to have learned a high degree of discipline around food and the trainer needs to understand how to fade out food lures. For shaping to be successful, the trainer must be able to anticipate behavior and avoid frustrating the learner. For targeting to be successful, the trainer must have the mechanical skills to handle the target, clicker, and food, all at the same time.
Kay gave a list of four considerations when choosing between these three training methods. A trainer must think about:
1) The trainer’s mechanical skills and the mechanics of the behavior
2) The confidence of the learner
3) The skill of the teacher
4) The future use of the behavior
Some behaviors logically need more guidance or direction from the trainer. Most of us would not want to be free-shaped if we were being taught to use a chain saw or to scuba dive! Likewise, when training animals, some types of errors can be costly. Luring or targeting can often be used to minimize certain types of errors, since the trainer can more carefully direct the animal. On the other hand, for some behaviors, it might be much better to shape the behavior.
Be careful about discounting any one of these methods. I’ve heard some people say that shaping doesn’t work, other people say that targeting doesn’t work, and still other people say that luring doesn’t work. One reason why people say these things is that each of these three methods has lots of different variations. And, some of these variations work better than others, or work better in some situations than other situations, or work better for some trainers than other trainers, depending on the trainer’s skill level and the trainer’s history with the animal.
Is luring evil?
I’m only half joking with the heading of this section. Of these three methods, luring, by far, has the worst reputation. I’ve met many trainers who seem to think that luring is awful and should never, ever be used when training.
Kay Laurence believes that most clicker trainers dismiss luring because they have not learned good luring skills. Most people try to “lump” when they lure and train a behavior in one big piece. (What is lumping?) To be good at luring, a trainer still needs to know how to split behavior down into small pieces and needs to know how to efficiently fade out the lure.
I’ve seen some brilliant video clips of Kay and her students teaching complex behaviors using luring and then quickly and quietly fading out the lure and adding in a cue. Also, in the experiment discussed above, Kay was able to quickly start to fade out the lure because of each dog’s history with training and shaping.
A few final thoughts
Don’t discount a teaching method too quickly because it has not worked for you (or others you have known). See first how expert trainers use the method. Are they able to get results? What do they do differently that makes the technique work well? Do you like how the final behavior looks?
Want to get better at training? Watch the experts. See how they get behavior and see how they use each of these three methods. Then go practice with your own animals. Videotape yourself (or have a friend watch you) and try to decide how you could improve your shaping, targeting, and/or luring skills. Do you already know areas you could be practicing?
Which of these three methods, shaping, targeting, and luring do you use most often? Do you ever combine two of these together? What influences you when you are deciding what method to use when training a new behavior?