Monday, May 24, 2010

Bribes/Lures and the Great Oreo Question

OK, so nothing gets dogs trainers as up in arms as does the question of punishment. But you can get some pretty good (if somewhat less snarky) disagreement going over the question of just how you use rewards. For one group of people (clicker trainers) the reward always comes AFTER the behavior. It's, like, the definition of a reward. You do what I want, and I give you a reward. But how does the animal figure out what you want? Long story. We'll get there some other time. Suffice it to say that there is another group of people who find it useful to use the goody to show the animal what you want. For instance, if you have a puppy that you want to teach to sit, you can hold a treat just above its nose. The nose goes up and the bottom goes down. Puppy physics. Then the puppy gets the cookie.

Works just fine. But...you've basically just bribed the dog. Rather than cheering after the fact, you've established the treat as the basis for getting the behavior. The puppy has lost the opportunity to use its own little brain to figure out the problem itself, and might not try it again until you pull out the treat to follow. Or, if you're on the other side of the argument, you've saved both yourself and the puppy the frustration of it wondering what the heck you want, by giving it a nice, clear, fun explanation. I told you there was disagreement.

But what, you ask, does this have to do with the wonderful world of people? And, yes, Oreos, which we were promised in the title? Well, it goes like this. Mattea (the kid) does a lot of dancing. Like ten classes a week. She and the girls she dances with compete in dance competitions and all. And the studio just added three new classes for kids her age with amazing teachers. To our surprise, however, Mattea didn't want to take the tap class with the ultra-cool, extremely hip and up-to-date tap teacher.

Mattea loves tap -- tap and hip hop are her favorite kinds of dance. However, unlike, seemingly, most of the girls she dances with, while Tea loves dancing, she has a limit. She likes to dance, but not if it takes away from her playing time. And this class was happening on a Tuesday evening, and she wanted her Tuesday evenings free. Now, from a parental perspective, we could see a variety of ways that Tea was going to get left in the dust if she blew off this class. But our policy has always been that dancing is a recreational activity, and recreational activities are things about which you get to choose (within the limits of the rules of the studio, which explains why she is taking ballet). So...we encouraged and we reasoned and we explained our concerns. And Mattea said no, she didn't want to.

And so Kelsey went to pick her up from hip hop, which immediately precedes this new tap class. But Kelsey brought with her a small package of Oreos. And she told Mattea that if, on the off chance, she were staying for the tap class, well then she'd probably need the cookies for extra energy. In other words, Kelsey bribed Mattea with cookies to take the class. Have I mentioned that Tea, like the dogs, is highly motivated by cookies? As in, pictures of Oreos on her computer background, just because it makes her happy to look at them?

Yeah, she took the class -- and the cookies. So, have we now ruined her intrinsic motivation to learn for the sake of learning? Or have we helped her over the hump to realize that the rewards for taking the class will exceed the costs? Will she take the class again? Will she demand cookies? Stay tuned...the class comes up again tomorrow.

4 comments:

  1. Not sure how this will help you think about the current situation, but the key concept to making positive reinforcement really work is Intermittent Reinforcement. Start with the puppy. If you offer the treat every single time you ask it to sit, but later stop offering the reward, it will soon stop sitting. If you offer the reward at first to get the behavior going, then start rewarding at random and ever-increasing intervals, soon you'll be able to stop altogether or only reward very sporadically. Or you can shape the behavior. If "sit" is a waypoint towards "sit, down, stay", eventually you should only be rewarding the stay part, and then only for longer and longer intervals. Regardless of your goal, if you do it right, the behavior itself gets associated with the good feelings the reward initially induced, and will happen all by itself. So maybe an experiment worth trying with Tea is, after a couple weeks, "forget" to bring the cookies, or substitue a more nutritious snack you "just happen" to have. Mix it up, so she knows that *sometimes* she'll get the cookies. That might help her over the hump until the dance class is so rewarding of itself that the cookies don't matter.

    ReplyDelete
  2. Well, I asked Mattea whether she would be taking the tap class this evening. Her response: (pause) "...Yeah. Will there be cookies?" So, yes, the cookies got her into the class and might make it stick (along with a variety of other factors, including the fact that it's a great class). But it also created the expectation that there would be cookies before class. This often happens with lured/bribed dogs -- if no treat is in evidence, they're not interested in working. The trick with luring dogs is to remove the lure as soon as possible. Maybe there will be cookies tonight, maybe not.

    Joseph, to your points, you're totally correct about intermittent reinforcement. But bear in mind that the problem of the original oreos is that they were a bribe (before the behavior) rather than a reinforcement (after the behavior). So the best approach might be to show up with a cookie after the class, and randomize from there. I'm hoping the class itself will be sufficiently reinforcing, though, and it's very easy to create the expectation of cookies....

    ReplyDelete
  3. ooh, good point re: bribe vs. reinforcement. Come to think of it, if the cookies come before, *she's* training *you*. Give her cookies, you get the reinforcing stimulus you desire (her going to class). Then you're more likely to give her the cookies again.

    ReplyDelete
  4. Well color me abashed. Now I read the rest of your blog, it's obvious I'm preaching to the preacher when it comes to training. Sorry for being a know-it-all.

    ReplyDelete