Hi, Paul,
Thanks for forwarding my comments. I would like to know what you are doing
with emotions as control systems, etc. I think that most of the "fierce
resistance" is really towards people who are pushing their own agendas to
the point of distorting control theory. It's like saying "I think
arithmetic is so interesting, but I think that 2 + 2 should equal 5". Or
"control theory is fine except when it contradicts what I believe"
(reinforcement theory, complexity theory, the Bible, whatever).
I was happy to see the comment quoting Powers in Making Sense of Behavior:
"The higher systems, rather than telling the lower ones how to act, tell
the lower systems what to perceive." However, this is not new - it is what
Powers has always maintained. At the very lowest level, where the outputs
to the environment are, this means using action (behavior), to control
perception (...of behavioral consequences).
I don't think that top down vs bottom up is a particularly useful debate.
In Powers' model, higher levels are continually specifying reference
signals for lower levels, which are continually processing perceptual
signals and sending them on up, level by level, to be compared to reference
signals. Reference signals and perceptual signals are continually being
compared and the resultant error used to reset reference signals for lower
levels and ultimately at the lowest level produce behavioral outputs that
affect perceptions. All these processes are going on simultaneously at all
levels in both directions. None has priority, all are necessary.
The idea of perceptual levels levels encompasses sensation, perception, and
cognition. In Powers' theory, in consideration of (among other things) the
fact that neurons can only do so much, they are hypothesized to do just a
little at a time, so instead of 3 levels there are 11 (two more in Making
Sense of Behavior than there were in Behavior: the control of perception).
Some people seem to dislike the idea that the nervous system is what does
all this, and especially dislike the idea that someone may be getting close
to figuring out how it's done. And the ultimate unacceptable idea seems to
be that the people who got a handle on it first were working with
artificial devices. I'll say device instead of machine since it has less of
the big, clanking 19th century imagery and is more suited to the electronic
age of quiet machines with no moving parts that get smaller and smaller and
do more and more.
The artificial devices I'm talking about, like all machines, were devised
to imitate some characteristic of living things - it was, until the 1930's,
usually muscle power. The revolutionary novelty of control devices is that
they were designed to imitate a mental characteristic: to have goals - to
be purposive. To bring the speed, or temperature, or pressure, etc. of some
other machine to a desired state, and keep it there despite varying
conditions. And the required organization and function of these devices are
clearly within the capabilities of and consistent with the arrangement of
the nervous system.
So, while this is not inconsistent with what Kim James or Maturana have to
say about relational states and organisms defining domains of interaction
and so forth (per Kim's post of 06 October 1999 19:06), it is a good deal
more concrete. Alternatively, Kim's discussion is a great deal more vague.
A description of the problem, not an explanation.
Whether or not vagueness is inherent in non-mechanical systems as Mishtu
says, members of the Control System Group (or at least some of them) are
proceeding under the assumption that in living systems there is an
understandable mechanism that receives inputs from the environment and
generates the phenomena we call levels of perception; compares the state of
those perceptions with desired states, and produces corrective outputs
whose success is perceived and altered as needed. Lower levels are mundane
indeed - how can you pick up a cup a coffee and bring it to your mouth
without flinging over your shoulder? Higher levels are more complex - how
can I arrange my life in order to get this post finished and also fulfill
my promise to my daughter to look after her cats. Here I must develop a
strategy(level 9) which will satisfy various principles (I try to respond
to posts in a timely manner, I keep my promises, and I care about the
well-being of pets) - principles (level 10) which I maintain at a certain
level of fulfillment in order to be comfortable with that system concept
(level 11) called "me".
Of course these higher levels are more hypothetical than the lower ones,
but the principle we are going on is that they are organized in the same
way - and the lower levels, such as those involved in tracking a moving
spot, have been simulated on a computer. A computer which itself is NOT a
model for this type of system, but rather a digital computer that has been
programmed to act like an analog computer (which is the model).
Frank Wood: This tracking program actually _is_ what Stafford Beer
describes: multiple control systems acting at different levels. If there
are divergent goals, there is either a compromise solution arrived at by
analog means or there is an unresolved conflict. The resolution of
conflict involves going up levels to where divergent lower level goals are
seen as efforts to maintain the same higher level goal. This can be seen
in individuals and in organizations as well, and has been tried both in
therapy situations and in institutions.
If this list is primarily concerned with organizations, then I should say
that my interest in organizations is from the point of view of the
interaction of control systems, both in and between individuals as above,
and if that isn't relevant to the complex-m list, I'll bow out now.
Mary A. Powers
powers_@frontier.net
Oct. 9, 1999