Los Tres Rubber Bandidos

[From Rick Marken (950103.1345)]

Gary Cziko (950102.1450 GMT) --

While PCTers like Marken would say that this [cause-effect analysis] is an
incorrect and useless explanation

Who, me?

I really like your "rubber band on a string" demo, Gary. And I'm really glad
to hear that you will be presenting it at the Durango meeting in July; that
means I'll be coming to the meeting for sure.

Here is a rubber band demo that illustrates the point I've been trying to
make about IV- DV research. I'll call it the "three rubber bands demo" or
"los tres rubber bandidos" (porque non?). You set it up like this:

DV-

ยทยทยท

-
        A
           -
             -
               K- - - -B- - - ->IV
             -
          -
        c
     -
DV-

Rubber bands A and B, of approximately equal size, are attached, as usual,
making a knot, K. Rubber band c, much thinner than A (notice it's in lower
case;-)), is also connected at K. Start by doing the rubber band demo, as
usual, using bands A and B (with band c just hanging off of the knot). The
subject uses band A to keep the knot, K, on some stationary target. The
position of the subject's finger is the DV; the position of the
experimenter's finger is the IV.

Experiment 1: Measure DV (distance of subject's finger from K) for various
settings of the IV (distance of experimenter's finger from K). Plot it on a
graph. It might look something like this:

                   *
DV *
         *
    *
  -----------------
          IV

Now repeat the experiment but this time with the subject's finger in rubber
band c. Now the results look like this:

                    *
                  *
DV *
              *
            *
          *
        *
      *
    *
  -----------------
          IV

The DV changes much more abruptly per unit change in the IV. It looks like
the subject has suddenly become much more "responsive" to the experimenter's
tugs on the rubber band. A conventional researcher would conclude from this
experiment that the subject behaved differently in the two experiments; in
fact, the subject behaved in exactly the same way in the two experiments;
she kept the knot on target. All that changed in the two experiments was
the environmental connection from the subject's output to the controlled
variable. Although this change in the feedback function is visible in the
"three rubber bands" demo it is invisible under most ordinary circumstances.

Now I can hardly wait to hear the reasons why this could never be a problem
in "real" behavioral research.

Happy banding

Rick