[From Rick Marken (930930.2100)]
I read over the Kelso, Tuller, Bateson, Fowler paper. Wow; talk
about fear of feedback. The basic point was that a coordinative
structure exists when a "challenge experienced by one (or more)
[potentially independent] members of a group of muscles [is]
responded to by other members of the group at a site remote from
the challenge" (p. 813). The "remoteness" of the responding site
is, I believe, supposed to rule out a feedback control model --
when, in fact, it rules it right in. Their demonstration of a
coordinative structure is evidenced by the change in the mean
position of the upper lip that occurs when there is a disturbance
to the jaw while a subject says the /b/ sound; there is a remote
compensation (lip movement) for the effects of a remote disturbance
(load on jaw).
Kelso, et al could have saved a lot of research dollars with this
simple demontration of the same phenomenon using far less apparatus.
Just get a subject (I used my bewildered but dutiful daughter) and
ask her to hold her arms wide apart and slowly move them together
until the index fingers of each hand are about 2 inches apart. The
subject will probably do just what you ask, moving the arms together
until the index fingers are two inches apart and directly in front
of the torso (though there are no instructions about the final location
of the two inch gap). Now have the subject do the same thing but add
a "challenge" to the movement of the subject's right (or left) arm.
Just reach in while the arms are moving together and hold one of
the subject's arms stationary. Lo and behold, the other arm compensates
for the challenge, moving the index finger two inches from the index
finger of the seized arm. What we have here, folks, is a coordinative
struture in action. The two inch gap (like the /b/ sound) gets produced
despite a disturbance to one of the muscle systems that produced it. A
remote challenge (seizing one arm) is compensated for by a remote muscle
system (producing extra movement of the unseized arm).
If the analogy to the articulatory "coordinative structure" is
made tenuous for you by the fact that the subject can see the
"articulators" then just have the subject do it again with her
eyes closed; it still works when the subject is controlling a
proprioceptive perception.
It is trivially easy to model this "coordinative struture"
using PCT. I leave the creation of this model as an exercise.
But I will give some hints: The location of the arms on the shoulders
obviously "affords" making a two inch gap between the fingers
(among other things, of course). This affordance is created, to
some extent, by the contrast between two inch gaps and non-
two inch gaps. The non-linearity of the the laws of motion
takes care of the rest, resulting in a strange attractor that
can apparently be quite powerful (my daughter is still trying to
get her fingers apart; I promised to type her homework).
Best
Rick