Journal 9/20/10

Last week I got a job that involves working with a student in Arizona.  It's all going to be conducted over the internet, via Skype (or some such equivalent).  I have no idea what that will entail.  Nor do I fully understand why the mother (who lives in a large city) couldn't find an English tutor who lives in the same state.  Not that I'm complaining: a job's a job.  Still, it seems bizarre.  I guess maybe we are moving toward the day when no one needs to leave their house to learn or do anything. We can go into our holodecks and be taught by people on Pluto (we've also colonized Pluto, in this new world) and be the better for it.  Right?

* * *

All the hubbub about David Mitchell's new book got me interested in reading his last book.  For anyone who has the time, Cloud Atlas is, as reported, extremely excellent.  Because it's a book where the less you know going in the better, I'll say about it only that it manages to be both narratively compelling and formally complex.   Also, that it's main character is a wookie stevedore named Bomb.  (Okay, the last part's a lie.  But if there is a book with such a character, I want to read it).

Another recommendation: Out of Sheer Rage by Geoff Dyer.  A book about NOT writing--a topic you might think would get tiresome.  In fact, it ends up being at the same time funny and profound, insightful and engaging.

* * *

"The Problem With Incentives"

When we first got the Bink, my wife and I decided to try and train him to go to the bathroom indoors.  My wife ordered a large quantity of poster-sized absorbent medical pads from a hospital supply company, and we set about teaching him to direct his excretions.  Our first strategy involved negative reinforcement.  Any time he went to the bathroom somewhere other than his medical pad (which we taped down, each day, on the tile floor of a bathroom in our old apartment) we told him he was bad.  "No.  No!" we would say, over and over, while he looked up at us with a hopeful, fully uncomprehending expression.

Of course, this method worked not at all.  We neither of us could muster up all that much anger at his mistakes; even if we had, it wouldn't have mattered.  Dogs apparently do not usually respond to negative reinforcement.  Beyond that, it was hard to explain to him what, in fact, we were trying to negatively reinforce.  Even on the few occasions that we caught the Bink directly in the act of messing up our rug, it didn't obtain.  We told him he was bad, he looked confused, and that was it.  The mistakes continued.

Finally, one of us hit on a solution.  Instead of negatively reinforcing bad behavior, we decided to positively reinforce the actions that we wanted.  This entailed following the Binks around for most of the day, and giving him a treat whenever he went to the bathroom on his pad.   This worked almost at once.  Within a week we went from about a 65% success rate to one in the high nineties.  Even when we left off monitoring him, Binks figured out a way to come and tell us when he'd gone to the bathroom and deserved a treat.  (He walks up to where we are, stares into our eyes, and wags his tail until we get up and inspect the pee pad.  It looks a little like this:

Everything was wonderful; we'd just moved into our new house, and the newly finished hard wood floors were staying pristine.   To compensate for the fact that we were now giving our eight pound dog four or five extra snacks a day, my wife reduced the size of his meals.  We also reduced the size of the treats.  At this point, the usual reward is a maybe a single Cheerio.

What's interesting, though (and part of why I recount the story) is that without intending it, we've taught  the Bink that he should go to the bathroom as often as possible.   Every time he pees, he gets a Cheerio-so why not do it as often as he can?  Instead of holding it in, and waiting till he has to he goes whenever he has the slightest urge.  He also knows that it's better to go when one of us is at home: if no one's here, who will give him a treat?  (We have a rough two-hour statute of limitations on the bathroom reward process.)  So insofar as it's possible, when we're not not home, he waits--saving all his...activity up till one of us gets back.  Then, it all comes out.

It's not that I see our 'pay for play' treat formula as a mistake: I'd rather have him pee a few more times a day and have it all be in the right place than have him going on our rugs.  It does, however, illustrate one problem with incentives.  Inevitably, you'll get some of the behavior that you're trying to encourage, but you won't get all of it.  You'll get the surface of what you want, but not the underlying form.  There's a valuable, valuable lesson here for people currently trying to craft a new economic stimulus.  I don't know exactly what it is.  But the Bink, I'm fairly certain, does.