Shaping

In order to teach a behaviour, you have to “get” it first.

Shaping is one method of teaching a skilled behaviour. Learning moves in a step-by-step progression from not knowing something at all to becoming highly skilled. Behaviour is changed incrementally, through many intermediate steps.

Shaping behaviour is a very effective way of teaching complex actions. The complex action or series of actions is broken down into small steps, with those steps rewarded along the way.

I like to think of shaping as a pyramid. You begin at the base, which represents every thing the dog doesn’t know. Your goal is the very tip of the pyramid, which represents what the dog has learned:  the final, skilled behaviour.

Any action which even remotely looks like it might lead to the target is rewarded. As you move up the pyramid, the actions gradually start to look more skilled.

In the beginning, almost anything goes as long as the dog is working with you and trying things. Try to reward any action that even suggests what you’re looking for. It’s like the childhood game of “warmer/colder”. The child doesn’t know where an object is hidden; simply walking into the room is reinforced with “warmer”. As the child takes a step towards the hiding place she is again reinforced “warmer”.  If she turns around, it’s “colder”.

Similarly, in shaping the trainer starts with a broad base of acceptable actions. Simply looking in the right direction might be reinforced. Each tiny step is rewarded. The behaviour is refined as you go along, with each step leading a little closer to the target behaviour. As the steps progress, you no longer reward the same things you did in the beginning, but ask for more.

After a dog has performed an intermediate step correctly several times in a row, withhold reinforcement. This encourages the dog to try something different. If something the dog tries  looks like it might lead in the right direction, reinforce.

Look for only slight improvements. Don’t take big leaps. Small steps are better than large ones. Overall, the dog will progress more rapidly if you reinforce smaller increments.

But don’t get stuck at a low level of performance — raise the bar! Making progress means moving along.

Rules for shaping:
•    Define the goal or training objective. Have a mental picture of the desired end behaviour — know what you want to teach.
•     Design a plan. Break the training down into small steps. Organize and teach the steps in the most easily learned order. Have a good idea of what you plan to reward and what will you ignore.
•     Start easy — reward early successes frequently.
•    Reinforce success and ignore failure.
•    Teach each step well enough that it becomes a solid foundation for the next step.
•    But raise the bar! Gradually raise the criterion for success. Don’t get stuck at a low performance level.
•    Be flexible. If it’s not working, try something different or go back to an easier step.

As you move through the steps, you may occasionally get stuck. If the dog fails to offer something new, looks confused, etc., reinforce the previous step again.
If the dog repeats a step several times in a row, withhold the reinforcer until he tries something new.

Common mistakes:

• Saying NO or correcting a wrong action. Ignore failure!
• Failing to reinforce a desired action.
• Reinforcing something you don’t want, either through poor timing or an unclear picture. Does the action lead towards the desired final action? Reinforce it! If it doesn’t, ignore it!
• Taking too big of steps.