[Martin Taylor 961023 14:40]
Bill Powers (961022.1430 MDT) to ?Bill Benzon ?somedate
Why would single
cells congregate and like to be in each other's close neighborhood?
Because that provides a selective advantage. Because they are able tocontrol better that way. For mutual advantage. What other mechanism
is even imaginable?These things don't happen BECAUSE they confer a selective advantage. They
happen AND they convey a selective advantage.
Yes, but lots of things happen and never are noticed--ships that pass in
the night. Bill and Bill are talking about things that happen and stick.
It is the sticking that is important, not the happening.
In my old draft taxonomy of helping, system number 8 is called "mutuality."
It is a situation in which System A is controlling a perception PA, and
system B is controlling a perception PB. Neither need know about the
other's existence, but either directly or through side-effects, A's
control of PA reduces the strength of the disturbing influences on the
CEV corresponding to PB, and vice-versa.
To see this, consider a simple example in which A and B do know about
each other and are deliberately "helping". A wants a heavy stake to be
driven into the ground, and his actions are to hit the top of the stake with
a mallet. B wants a stake to be vertical, and his actions are to hold
the stake with his hand. If B doesn't hold the stake, A has great trouble
hitting it. If A does hit it, B finds it easier to hold the heavy
piece of wood upright. B's actions do not affect A's perception of the
stake being driven into the ground. A's actions do not affect B's perception
of whether the stake is vertical. But each finds his own control job easier
if the other successfully controls his perception.
Now, one can readily imagine situations in which the chemical effluents of
one cell are useful in easing the control problems of another (to shift
the level of metaphor, oil on troubled waters), and that, perhaps in a
roundabout way, the control side effects (effluents, perhaps) of the second
are useful to the first. So long as the two cells stay together, they
both work better (lower error values) than when they are apart. Slow
reorganization when they are together, faster when they are apart. If
a mechanism develops that makes them tend to stick together, they continue
to work better and the mechanism doesn't reorganize away.
I don't like the implications of saying that a cell organizes other cells
to do what is good for its own welfare. But it does seem reasonable that
mutually beneficial structures will tend to be more robust against
potentially destructive influences than will structures that have just
come together by happenstance and don't convey any mutual benefit to the
elements of the structure. Reorganizations of the elements that enhance
the "stickiness" of the mutually beneficial structure will tend to be
kept; reorganizations that don't will tend to be lost.
The argument is much the same as for the "structures" of social convention.
Their existence is not controlled, but it is because of control that they
exist. Their elements control better because of the structure than they
would in isolation or in a random different structure.
Martin