[From Rick Marken (920326 9:00)]
I see this morning that there is a long post about open loop behavior
from Bill Powers. I haven't read it yet. I decided to post something
I wrote last night and make a fool of myself before reading what Bill
has to say:
···
--------
After finishing a personal reply to Martin Taylor I started to
ruminate about open loop control (again). The ruminations began
with an obvious observation (that I made in my note to Martin). Open
loop means NO LOOP. There is a problem with the expression OPEN
LOOP: it implies that LOOPS are common and sometimes they are
OPEN. I don't like that. I think what is common (in science and in the
universe) is cause-effect. That is, NO LOOP. This is what the term
OPEN LOOP really refers to; situations where one or more variables
have an effect on one or more other variables etc. where none of the
variables have an effect on themselves (via any of the other
variables). What is less common is where a variable is the start of a
chain of cause effect where the last effect in the chain is the cause of
variations in the variable in the first part of the chain -- a causal
LOOP. When this happens (and in the universe it is apparently rare)
the circle of causes and effects has very low gain; cause1 = f(cause1)
where f makes the effect of cause 1 on itself quite small. What was
amazing was the development of LOOPS that not only have VERY
HIGH GAIN but also have NEGATIVE GAIN. So far the only loops of
this kind that we know about exist on earth. The search for
intelligent life in the universe is a serch for the existence of other
high gain, negative feedback loops.
So OPEN LOOP behavior is really a verbal trick. It is an attempt to
give a LIFE SCIENCES- like name to processes that have been studied
in the natural sciences for decades. The idea that OPEN LOOP
processes can be responsible for the behavior of living systems (a
behavior we refer to as CONTROL) is exactly the same as saying that
causal variables (like gravitational force) can be responsible for the
"purposeful behavior" of a leaf as it wafts to the ground. Aristotle
was laughed out of the gym for claiming that leafs "seek their natural
place" on the ground. But now, 2500 yrs later, life scientists are
being celebrated for explaining purposeful behavior with cause -
effect models -- ie. physics models. Go figure.
William T. Powers noticed a "fact of life" that should, perhaps, have been
obvious to others, given the maturity of control theory as a discipline at the
time of his observation. But no one else in the life sciences noticed (or
was willing to notice) this fact: when there is high gain, negative feedback
from the output to the input of a system, then cause-effect models are no
longer appropriate. Powers also showed why negative feedback
systems would look like cause-effect systems -- the kind that people
assumed they were all along.
There was a BIG PROBLEM with this apparently simple discovery. It
was made in the 1960s, about the time that the life sciences in
general, and the behavioral sciences in particular, were settled
comfortably into a life of studying cause-effect relationships using a
statistical/experimental paradigm bequeathed to psychology by R. A.
Fisher. What Powers found was that negative feedback made this
entire approach irrelevant; it was, quite frankly, ALL WRONG. Who
would believe this? It turned out that nobody would (except for a
few degenerates from the midwest--like me). Some people did like Powers'
language -- hearing the strains of a new approach to that good ol'
cybernetics. But many (most?) of those who liked Powers' control
theory (other than the crazy degenerates) kept thinking that it
cound not really mean what it meant -- that the whole cause-effect
kit and kaboodle of psychology had to go. That was too rough. There
had to be something good left in all that 100 plus years of work by
so many smart, ambitious people. So they tried to preserve what they
could -- and, of course, that meant that they could not get the point
of control theory. The prime example of this is the Carver/Scheier
approach to control theory.
But the fact of the matter is that when there is negative feedback
involved in the relationship between an organism and its environment
then cause effect laws are no longer applicable, They are not partically
applicable, sometimes applicable or sort of applicable. They are
WRONG. EVERYTHING that psychology thought it knew would have to GO because
it was all based on the assumption that behavior was the last step in
a cause effect process. The fact that outputs effect inputs WAS noticed
by classical psychology but its existence was thought of as a minor nit.
It was just an extra relationship to be considered -- sure, said the
psychologists, we know that "feedback" is important. But a lot of
behavior is OPEN LOOP, they said -- thus inventing a term that was to
haunt control theorists to this day. It suggested that psychologists
knew there were LOOPS in behavior -- but it said the existence of these
LOOPS was not really THAT important, and, besides, the LOOPS are usually
OPEN anyway.
The fact is, there is no other way to go. I don't want to be a radical or
"religious zealot" as it appears I am perceived to be. But the uncomfortable
reality is that it just can't work both ways -- if organisms are negative
feedback control systems, controlling perceptual variables relative to
internally specified reference signals, then all the "facts" in your
intro psych book are not facts at all. They are, well, illusions that
result from looking at a control systems as cause effect systems. They
are illusions in the same sense that it is an illusion to look at a falling
leaf or rising steam as examples of purposeful behavior.
Sorry, but that's the way it is.
Regards
Rick
**************************************************************
Richard S. Marken USMail: 10459 Holman Ave
The Aerospace Corporation Los Angeles, CA 90024
Internet:marken@aerospace.aero.org
(310) 336-6214 (day)
(310) 474-0313 (evening)