NOTICE: Posting schedule is irregular. I hope to get back to a regular schedule as the day-job allows.

Saturday, October 8, 2011

The Value of Repetition... [Full link to blog for email clients.][FT:C44]

In my guest Blog for Sarah Hoyt last Thursday (Oct. 6 - ) I discussed the value of looking toward the future as a Lab Rat - knowing that as long as conditions remain the same, repeating our past actions is valuable - but the moment conditions change, too much "memory" in the form of repeating past actions, can be detrimental. 

I stated:
"A popular saying among some groups such as the Baen Barflies is that insanity is doing the same thing and expecting different results.  Likewise, insanity is doing different things and expecting the same results.  Yet sometimes we have to change strategies in order to get a better result, as in the transformation from traditional to indie publishing.  I’m not an “insider” to this phenomenon, but I certainly  recognize the signs and symptoms from behavioral psychology."

There were many good follow-up comments to the blog, and I wish I had been able to engage in the responses and replies.  Unfortunately, most took place while I was at work, and it was a particularly busy day.  However, one comment merits a response here.

quote: Likewise, insanity is doing different things and expecting the same results

I take issue with this, there are frequently many different ways to get to one result. some ways may be more efficient than others, and it’s usually true that if you are picky enough the results are not ‘identical’ (after all, the process you go through has side effects), but if you are looking for ‘equivalent’ results there are frequently many ways to get to the result.
David and I have had some give and take on other issues in other forums, but that is immaterial.  However, I take issue with David's comment.  Granted, equivalent results can be obtained, but the same results never are.  Most importantly, how we learn and behave is significantly altered by the approach we take to obtain a result.

I will forever remember the example taught in my first lab psychology/animal behavior course:

A rat is placed in a cage with a levers.  If the animal pressed the lever, it heard a tone and received a food or liquid reward.  My professor, as a student, was training his rats and had one that would never purposely press the reward lever.  The rat would bump up against it, rear up and sniff the cage top and lean against the lever, latch onto the cage top and fall on the lever, but never reach and press the lever.  Now this is normally a very easy behavior to train to the rat.  The instructor for that course suggested a barrier be placed in the cage, such that the lever was at the end of a corridor, and the rat could not "accidentally" bump the lever. Rats are naturally curious and will usually explore under those conditions.  Unfortunately the animal chose to climb over the barrier, stepping on the lever on its way down.  If the barrier was removed, the rat would not press the lever.  My professor told us he tried for weeks to teach the rat the appropriate response - his other rats learned it in 5 days, but this rat could not learn it in 5 weeks.

So, why did that happen, and what's the "punch-line" for this column?  Well, to teach a rat, you need consistent conditions so that there can be an association between the action and the outcome.  In this case, the rat obtained an equivalent result - the food reward - but it never learned to associate deliberately pressing the lever with the outcome.  To that rat, it was the act of climbing over the barrier that resulted in reward - it obtained food mysteriously whenever it explored, but never learned that it could produce a reward as a consequence of its own actions.

A corollary to this example is something I see quite often when teaching students:  They train a rat to perform appropriately when the time delay between action and reward is negligible, but as soon as the delay is increased, the animal performance deteriorates.  To fix this, the student decreases the delay, then makes the trial easier by eliminating one or more choices, then they try adding cues, then changing the reward - often all at once or on successive days.  What this means to the rat is that it never has a stable context in which to learn the appropriate responses.  We can teach animals and humans to do many things as long as there is a stable context - Action A results in B - during the learning phase.  We can even teach them to be flexible and learn new mazes, as long as we introduce consistency in the learning/relearning phase.

Alvin Toffler called it "Future Shock" - the inability to cope with rapid change.  Deprived of enough stability to learn the rules, some people never learn how to deal with change.  What distinguishes the lab rats - and people! - who can cope with change is that they have in fact learned  a few basic rules of how to respond to changing context.  Most importantly, they learn just how important those "picky" differences are.

So, I argue that there are in fact a few ways to get equivalent results.  However, each different method teaches different rules - and if we keep changing the rules, nothing can be learned.  Its one reason why rote recitation used to be taught in school - and still is in professional schools.  Perhaps it would be more appropriate to say "... insanity is: constantly changing the rules, but nevertheless expecting the same results."  On the other hand, I stand by my original statement.

We can learn a lot form Lab Rats.

No comments:

Post a Comment

Please add comment - no links, spammers will be banned.