Free shaping

There are at least a couple different ways to teach a dog a new behavior. The most commonly used one is "lured" shaping, where you move a treat in front of a dog's nose, the dog follows the lure and does the behavior you want as a result. Then you reward the completed behavior. You gradually remove the lure and add a word or hand gesture to cue the behavior. For example, to teach a dog to sit, you hold a treat in front of the dog's nose and raise your hand upward and slightly behind his head, and he'll sit to be able to see the treat in your hand; when he sits, you give him the treat; then you add the command and voila the dog can sit on cue. Adding a marker like a clicker to let the dog know the instant he's done the right thing speeds progress. Lured shaping is an efficient method of training, but it is pretty dull and unimaginative for the dog.

"Free" shaping is done without any lure and is much more stimulating for the dog. You have a behavior in mind that you want to teach. You break it into tiny increments in your mind; when the dog offers the tiniest increment of its own free will without any prompting, you mark the behavior (e.g. with a clicker) and reward your dog. The dog figures out what behavior got the reward and will try it again. After a few rewards for the first increment, you then withhold the reward until your dog achieves the next increment. The dog "discovers" what the next increment is by trying things. For example, to teach a dog to back up, you might click/reward when the dog stands in front of you; once he's doing that, you might click/reward for staying standing; then you might click/reward when he moves any paw backwards; then you click/reward when he moves two paws backward; then you click/reward when he moves all paws backward; then you click/reward when he takes a couple full steps backward. In the midst of all this, the dog is offering all kinds of behaviors that you don't want, and you just ignore those. The dog keeps trying things to find out what works and keeps stringing the increments together to build on what worked in the previous increments.

The very first free-shaping training is very slow going. The dog just doesn't understand what the whole process is about. He stares at you wanting some direction and can't figure out what to do to get a treat. But I've seen this demonstrated with service dogs who were accustomed to being trained with free-shaping and it was astonishing how fast they would figure out really complicated behaviors. I saw a dog learn to flip a light switch in about 10 minutes. And the kind of exploration and problem-solving that dogs must do in free-shaping is ideal for service dogs who need to think and act, not merely respond, to do their jobs.

Karma and I tried free-shaping for the first time this past week as part of her agility training. I was supposed to teach her to back up. We accomplished that, but somehow I've trained her to walk in a circle around me before backing up. At first she sat and stared at me for long spells. It took her a long time to figure out that trying things was the only route to doing the right thing and getting the treat. But once she got the idea of that, it was fun to watch her mind work to come up with what she thought she was supposed to do. "I'll try lying down; I'll try rolling over; I'll try getting on the wobbly board; I'll try getting one piece of mail off the mail stack and taking it to the crazy lady; I'll try barking; I'll try sitting down with more determination; I'll try getting my food dish". Next up, teaching her to back up onto an inclined plank. I think it'll be easy now that she's got the idea.