Wednesday, June 11, 2014

A "Pavlov Poke" to End Self-rewarding Behavior?

Suppose a human was engaged in destructive, but self-rewarding, behavior such as compulsively going to Facebook when they should be working or doing something more productive?

How would you change that behavior?
In an earlier post, I noted that a few mild shocks, right at the beginning of teen drinking, might change a life.

Imagine if, when you were sixteen years old, you had been shocked just as you reached out to touch that first can of purloined beer.

Would you have reached for a second? A third? Would you have ever drunk a six pack?

And if you had not, would you have done better in school? Would you have married a different girl or gotten a different job?

Would three or four well-timed shocks have changed the entire trajectory of your life?


Now, two MIT researchers, Robert Morris and Dan McDuff, have turned that idea to combating compulsive Facebook use, another self-rewarding behavior that can border on addiction.

Using monitoring software and a capacitor, a computer was created that would give the user a mild shock whenever Facebook was checked too often or for too long -- something the researchers call a "Pavlov Poke."

Did the shocks work?

Yes, of course. In fact, they worked so well that the MIT post-doc lab rats that devised this experiment, and who did not really want to quit Facebook, decided to change the experiment!

Sadly, we found the shocks so aversive, we removed the device pretty quickly after installing it. Anecdotally, however, I did notice a significant, though temporary, reduction in my Facebook usage.

They then programmed their machines to respond to Facebook overuse by posting a job to Amazon’s Mechanical Turk service, paying a random stranger to call them up and brow-beat them. They then recorded some of the results.

So do well-timed
aversive consequences work?


The difference between an annoying phone call and a mild electric shock, of course, is that the shock is physical and immediate and cannot be ignored by sending the message to voice mail. 

"Phone call" training is likely to cost both the trainee and trainer more time, and also result in more "slippage" as the truly Facebook-addicted will find more ways to manipulate the system rather than actually change their behavior. To escape all consequences for "Facebook cheating," after all, all you had to do was take the phone off the hook!

This experiment would have been a little more interesting if the computer had been programmed to send a warning beep 15 seconds before the screen went blank, or if had been programmed to build up "credits" for time away from Facebook that could later be cashed in for a longer session.

Of course, the simplest way to quite Facebook entirely is to fence it off -- to close your account.  And if the goal is to limit time spent on Facebook, the simplest fix is a computer-enforced routine that only allows Facebook to be activated for a set session, say 30 minutes a day, between a set time period, say 7-10 pm.  While "Edison medicine" can work well, and quickly, to train behavior, it is best used when the rules are well understood, and when more signaling is going on than appears to have been built into this little exercise.

1 comment:

PBurns said...

Yes, yes, I realize Pavlov is the wrong name to use for operant conditioning. That just underscores the fact that this experiment could have been designed much better with a little input from others. That said, an improved version of this would make for a terrific science experiment for some high school kid.