It’s a sunny spring afternoon in New England at the annual meeting of a group of animal behaviorists. I’m sitting on a folding chair alongside a small corral, watching some demonstrations with horses. My friend Tim Sullivan, curator of behavioral husbandry at the Brookfield Zoo in Chicago, is sitting on my left.
We’re watching a demonstration of clicker training. A calm old police horse is led into the corral and turned loose. The trainer stands on our side of the fence. She has a clicker, a bucket of feed, and a huge target stick that looks like a toilet plunger, with a big padded lump on one end.
The horse comes over to her. Click, treat. The trainer picks up the target stick, swings it over the fence, and jabs it at the horse. The horse has seen a lot of things in his life, but he’s never seen this. Startled, he throws up his head and backs away. I laugh.
Then I’m embarrassed. I don’t mean to humiliate the trainer; I should have kept my laugh to myself. The trainer coaxes the horse over again, shoves the target stick at him again, and startles him again, but less. In a few more tries she’s got him touching the target, and the fear, though still visible, is almost gone.
Tim Sullivan leans over to me and says, “You can always tell when a trainer is coming from method instead of principle.” Boy, can you ever! Golden words.
Method vs. principle
This trainer knew the method: present the target, click for touching the target, then treat. Repeat.
But, she was not relying on principles. One principle of shaping is that it is necessary to begin with something the animal is already doing—in this case, approaching the trainer. Nosing the end of a suddenly appearing large unknown object was definitely not in the existing repertoire. One would shape that event from a much simpler starting point.
Here’s another fundamental principle of training: fear decreases existing behavior and increases avoidance behavior. Did the trainer want to slow down the horse’s learning of a new behavior? No. Did she want to develop avoidance behavior, such as throwing up the head and backing away? Not really. She had a method, but not an understanding.
Watching, my take was, “Well, she’s not really one of us, a clicker trainer as we mean the term. Or she’s not one of us yet.” Tim Sullivan’s remark was more profound. You CAN always tell, but what, exactly, is the difference?
The good news is that now that so many principles about behavior, and how and why it changes, are understood, you don’t need natural talent plus 40 years of life lessons to be a highly effective trainer. Principles-based trainers keep the reinforcement contingencies clean and in effect from the start. Principles-based trainers observe their learners and keep emotional signals in mind as useful information about the training progress.
That’s what we aim to teach through ClickerExpo and through Karen Pryor Academy.
With principles-based training, you can be as creative and effective as only a few geniuses could be in the past, and not just at training animals—at helping with all kinds of human behavior, including your own.
Methods are fun too
Of course I’m still interested in methods, though, aren’t you? The widening circle of people training from principles leads to invention of more and more new methods, which we gladly share with each other. This week on the KPA Alumni list I saw some new methods for teaching the obedience scent articles exercise, in which the dog selects, from a pile of predetermined objects, the only one that has its owner’s fresh scent on it. (I liked one of the new ideas enough to print it out for evening fun with my elderly poodle.)
Knowledge of principles also lets you weigh one method against another realistically, and include methods with a long tradition of use. (Quick: What’s contrary to shaping principles in the grand old method of scenting one article in the pile and tying down all the others so they can’t be picked up? Answer below.)
But methods alone aren’t enough
Meanwhile, back at the ranch, a new horse is in the corral. A TTouch practitioner, trained by Linda Tellington-Jones, is massaging the front of the horse’s shoulder. It’s apparent that the horse likes this experience. He stretches out his neck as if she’s hit a particularly itchy spot. He waits for more when she stops; in fact he solicits more, by positioning his shoulder closer to her. I don’t know the neurological principles that explain this phenomenon. But I can see that the result is very reinforcing, and I wish I knew exactly what she was doing! It’s obviously a fine method for winning the confidence of a strange horse.
Thank goodness, I already know how to do that. And not just with horses—with anything with a nervous system. All you need is a few principles—and a click and a treat.
*Answer: The shaping procedure benefits from maintaining a high rate of reinforcement for correct responses. The tie-down method offers one correct response opportunity among many incorrect response opportunities, a ratio that is more likely to lead to extinction than to acquisition.
Training scent articles
I know there are multiple ways of training scent articles, but just going with the:
Q: "What’s contrary to shaping principles in the grand old method of scenting one article in the pile and tying down all the others so they can’t be picked up?"
A: "The shaping procedure benefits from maintaining a high rate of reinforcement for correct responses. The tie-down method offers one correct response opportunity among many incorrect response opportunities, a ratio that is more likely to lead to extinction than to acquisition."
Would a better variation be to have multiple articles scented, and 1 unscented, get >=80% success rate, then have 2 unscented, get >=80% success and continue until you're down to 1 scented article?
Post new comment