Tuesday, September 27, 2011

Milking Stools and Operant Conditioning

Operant conditioning has three legs.

You know what's wrong with four legs?

A four legged chair is unstable unless all four legs are the same length (i.e. have the same value) and the floor is perfectly flat.

A three legged chair, however, is always stable, even on rough ground.

That's why milking stools have three legs.

The three legs of operant conditioning are:

  1. Reinforcement (treats, play, etc.);
  2. Punishment (voice corrections, leash corrections, etc.), and;
  3. Extinction (no reaction from behavior, desensitization).

Can you add legs by splitting things and listing them as separate legs?  Oh sure.  You can make reinforcement "positive" and "negative," for example, and by so doing make it completely opaque to a lay audience. 

In fact, with very little effort at all, you can make operant conditioning so incredibly complicated no one will understand it.  But let's start small and simple first.  After all, we are training humans, not a higher life form like a Beluga Whale.

Training humans?

Yes, we are training humans to train animals . . . Which is to say we are training humans

Now here's something important about the shaved ape that is called a human:  It has a hard time remembering numbers greater than three. 

Humans prefer three ideas, three rules, three characters, and three examples.

This is a fact.

It's why we we pray to a trinity called the Father-Son-and-Holy Ghost and why we call the alphabet the "ABCs" and the musical scale "Do-Re-Me."  

It's why we have the Three Stooges, the Three Little Pigs, the Three Blind Mice, and Goldilocks and the Three Bears. 

It's why the Nazi's talked about "Ein Volk, Ein Reich, and Ein F├╝hrer" and why our Founding Fathers referred to our national purpose as "Life, Liberty and the Pursuit of Happiness."

When a speech writer is crafting a speech, if he really knows his business, he will make only three points, whether that speech is three minutes long or three hours long.  After all, the audience is only going to remember three points -- pick them and frame them tight!

The Ancients, of course, knew this.  The Latin phrase, omne trium perfectum, conveys the idea in just three words.

THREE is teachable and memorable for humans.  

In fact, when scientists talk about memory, they use the rule of three, noting that memory has three parts (encoding, storage and retrieval) and that there are three types of memory (sensory, short-term and long-term).

You want to teach humans to teach dogs? 

My suggestion is to start with the three parts of operant conditioning:  Reinforcement, Punishment, Extinction.  

That's a conceptual milking stool that will work even if one leg is longer than the other, and even when it is being tested on the rocky ground of real people and real dogs in real situations!



PaulAndMuttley said...

I think you have a good point. Most people get confused with the + and - of OC, since positive and negative have emotional connotations that conflict with the actual meaning of "added" and "removed". For an electronics engineer, there is positive and negative feedback, which corresponds to reward (reinforcement) and punishment. But I'm not sure "extinction" is actually a method, as it is more a condition that has been obtained by OC methods, or which has simply happened because of other influences (such as maturity, aging, or environment).

PBurns said...

Operant conditioning has been clearly defined since the days of Skinner, and extinction is the third leg. See the first link if you want to read more at wiki, but the fact that operant conditioning is found in nature (the apple tree provides a reward to returning deer, the porcupine walks around with punishment on his back, the rock is always practicing extinction) does not make it something other than operant conditioing.