One of the seminal books on positive reinforcement training with animals (and humans, too!) is ‘Don’t Shoot the Dog’, by Karen Pryor. When I was a zookeeper, this was required reading, and really has been the manual of this method of training for exotic animals since it was published. For clicker training with horses, of course, there is Shawna’s fantastic book, ‘You Can Train Your Horse To Do Anything’! The second chapter in Karen’s book is all about shaping; if you’ve been to a Shawna or Terra Nova clinic or workshop, you know exactly what shaping is. If not, here is an explanation; shaping is reinforcing behaviors that are closer and closer to the end goal behavior, also called ‘successive approximations’. For instance, if my end goal is to have my horse line up parallel to a mounting block and stand quietly, I would initially reinforce any behavior given by my horse that has her close to the mounting block. Once my horse understands this, I would then start reinforcing only when the behavior was closer to the block. Eventually, after reinforcing closer and closer movements to the block, in the alignment you desire, you will have shaped your end goal behavior. You may be thinking “but if my horse was reinforced for the initial standing close-ish to the block, how do you get them to come closer so you can reinforce that?” Shawna always reminds us that the learner will naturally raise their own criteria in positive reinforcement training, which gives us the opportunity to reinforce what we want when we see it. (When we talk about shaping, it’s good to remember that we have a couple of different options in how we shape- typically called free shaping or shaping with the target. We will talk more about those in future blog posts.) Sometimes my horse walks up to the block headfirst, in a perpendicular arrangement. Maybe that’s the only way I can initially get her to approach it, so I reinforce that. On the next approach, she does the same thing but shifts her body a little to the left after I don’t reinforce her for approaching and standing perpendicular, so I bridge and reinforce that tiny movement. As my horse is used to the game of training, she will understand the concept of offering something different if she is not immediately reinforced for an offered behavior.
Sounds simple enough, right? Effective shaping is one of the most critical components of good positive reinforcement training with horses, as well as any other learner, human or animal. Karen Pryor discusses ten guidelines for shaping in her book, which she calls ‘The Ten Laws of Shaping’ to help make R+ training more understandable for the learner, and more effective for the trainer. Because we at Terra Nova feel that understanding shaping is so important, we thought we would take each point and discuss them individually, on what we are calling ‘Training Tip Tuesday’!
The first guideline is “Raise criteria in increments small enough that the subject always has a realistic chance for reinforcement”. One of the challenges I see in many training sessions is the tendency for the trainer to move too fast and expect too much. We as trainers have in our heads exactly what we want our learner to do- but the animal DOESN’T know what that is. It is our responsibility as trainers to bring clarity to the training, and help the learner understand what we are asking. They are simply trying to figure out what will get them the thing they want (the reinforcer), and the last behavior they performed is what got it. Using the mounting block example above, what do you think would happen if, after the initial headfirst alignment with the mounting block, I decided to only reinforce parallel alignment? The chance my horse would randomly line up correctly is roughly equal to being struck by lightning (ok, that’s a bit extreme, but you get my point). I have changed the rules mid-stream, with no information to help the learner succeed. This sets the situation up for no chance of reinforcement for the horse- and if this non-reinforcement happens enough, the horse will a) get bored with the game and not participate anymore, or b) get really frustrated and start exhibiting unwanted and potentially aggressive behaviors. Here’s a human analogy to which most of us can relate; we’ve probably all played games on our phones like matching or ball popping games. We know that the difficulty of these games very gradually increases as we complete levels. Imagine going from easy in level 10 to expert in level 11- and the game developers had thrown in tactics that you had not yet learned. How fun would that be? Would you continue to play?
I hope this helps offer a bit more understanding about shaping and how you can use it in your positive reinforcement training. Next week we will talk about Karen Pryor’s second guideline and how adhering to it can make your training more successful!