[From Bill Powers (920915.2000)]
Greg Williams (920915-2) --
There are common threads in both arguments, but they ARE different.
In >the former argument, I am mainly concerned with purposive
influence; in >the latter, with more general sorts of environmental
influence >(including influence due to non- living parts of the
environment).
Well, good. Yes, these are different arguments. Let's dispose of one
of them so we can get to the other.
With respect to "more general sorts" of environment influence, I think
we need to draw a sharp line separating non-purposive from purposive
influence. Purposive influence is not just a subset of non-purposive
influence: it is a totally different phenomenon. Let's put it aside
for a while.
Non-purposive influence results, in the final analysis, from the non-
sentient environment applying undirected and goalless disturbances to
variables that are critical to a living system's survival. These
disturbances are not aimed specifically at an organism; they fall
equally on any object or process, living or nonliving, that gets in
their way. They are simply the normal unfolding of physical processes
as entropy winds up toward its maximum. As the environment does not
care whether such influences have any effect or not, the organism can
win; but it can win only if it is predisposed to acquiring control
systems, or comes into individual existence already controlling (as
most of the organisms on this Earth do).
Organisms must learn to control the variables important to them by
learning how to act in opposition to these undirected disturbances.
They must learn what actions are needed to prevent disturbances from
affecting the variables critical to their existence, which means
simply that they must learn to affect those variables themselves.
That's the genesis of control: learning what actions have an effect on
critical variables, then then building organized systems that can act
on those variables to maintain them in specific states. The organism
does not need to know the source of disturbances; it acts directly on
the consequence that matters to it. Learning that connection closes
the loop and creates a control system.
Clearly, it's the environment that determines what the organism must
learn, because the effects of the organism's actions on its own
critical variables is mediated by the environment. But the mere fact
that the environment makes learning necessary and possible does not
mean that the environment quides learning. The environment doesn't
care; it makes no difference to the environment whether the organism
learns or not. Whether the organism learns or not depends on the
existence inside it of a mechanism for learning, which has to predate
the first occasion on which it's needed.
This means that this mechanism for learning can't take advantage of
any of the organized methods of control that eventually will result
from its actions. Given an error signal, it has no a priori way of
determining what action will have an effect on the error. Neither does
it have any sophisticated hierarchy of perceptions and computations to
aid it; that ultimate product of its action lies far in the future. As
far as I can tell, the MINIMUM machinery needed to acquire control
from this primitive start is an error-driven output that affects
action in a random way.
With respect to food-seeking, that is as far as the evolution of E.
coli got. Every time it seeks food (or avoids noxious substances) it
has to reorganize. It starts from scratch every time it tumbles. This,
apparently, is all it needs to find enough food and to avoid poisons.
What lies behind E. coli's behavior? It is an inherited ability to
sense substances that are good for it and substances that are bad for
it. Each E. coli individual comes into existence knowing these things.
It knows what to sense (has the equipment for sensing) and it knows
whether the time rate of change of each input signal should be
positive or negative. The current environment did not teach it that:
evolution did. In fact, the current environment never does teach E.
coli anything about seeking food and avoiding poisons. E. coli never
learns to control these variables any better during its whole life as
an individual.
Yet we can certainly say that the environment influences E. coli. What
it influences is the direction in which E. coli, on the average,
progresses through 3-space in relation to sources of diffusing
molecules. From an outside point of view, we can use substances of
various kinds to push E. coli around in any way we please. We have, it
would seem, total control over E. coli. Even the nonpurposive
environment has total determinism of its actions. E. coli must move
according to where the molecules are coming from, every random shift
in the locations and concentrations of various substances rotting away
causing E. coli's path to change.
But have any of these influences actually changed E. coli in any way?
No. Through all of these effects on its direction of locomotion, E.
coli continues to maintain a positive time rate of change of
beneficial substances at its sensors, and a negative time rate of
change of harmful ones. From inside E. coli, these external effects,
its own movements in space, don't even exist.
And any external influence that DID materially affect, for any length
of time, the variables E. coli senses and controls -- would kill it.
E. coli can't use its reorganizing capacity to build control systems.
But higher organisms can. The particular systems they build depend on
the environment that happens to exist outside. But those nonpurposive
environments still don't care what organized systems result; whether
the result is to seek or to avoid creating any particular state of the
environment. The criterion for what is learned is still the inherited
one: whatever is learned, it must preserve the critical variables in
their inherited reference states. To an external observer, it may seem
that the environment has enormous effects on the organism, because the
organism behaves differently with every change in the environment. But
the net result of all these apparent effects is NO effect -- on what
matters to the organism.
Of course the organism may fail to deal with the challenges of natural
disturbances. If so, it will die. Then the environment can't even
influence its behavior. Its behavior has stopped. No external event
can get it going again. Only autonomous systems can irrevocably die.
···
----------------------------------------------------------------------
For well over 300 years, science has looked at all the things
organisms do and all the environmental events that seem to cause these
doings. Common sense has gone along with science, or perhaps it was
the other way around. Our social interactions are perceived and
understood from the same point of view. Both science and common sense
have become obsessed with what the world can do to organisms, from
evolution to everyday stimuli. Most of our concerns revolve around
this concept of behavior-as-effect. When we want to change our
behavior or someone else's, we look for something in the world around
us to change. It's impossible to overestimate the pervasive effects of
this mind-set on the way each individual sees and understands all
living systems, including the observer.
It is very difficult to find that point of view from which it is
obvious that the environment's influences on an organism are side-
effects of what is actually going on. Even the behavior of organisms
is a side-effect, misleading in the extreme. What we see organisms
doing is of only passing interest; it will change with every change in
the environment. All that matters is what the organism is
accomplishing, and to know what that is, we will have to become very
much better acquainted with organisms than anyone is today -- and from
a radically different point of view.
---------------------------------------------------------------------
I don't know if that "disposes" of non-purposive influences, but it's
a start.
---------------------------------------------------------------------
Purposive influence is entirely different from non-purposive
influence, because the causal influence is no longer of interest. The
causal influence is just like any other; its effect on the organism is
the same as if non-purposive sources were generating it. What the
influencer is concerned with is not influence-as-cause, but influence-
as-accomplished-result, as effect. Not "applying an influence" but
"having an influence." These two meanings of influence, which our
language mashes together into a single fuzzy concept, are different
phenomena.
When we are talking about influence-as-effect, only the reasons for
which influence-as-cause is applied matter. An influencer who wants to
have a particular effect on another person must attend to the effect,
not the cause. Because the environment, including the total system of
the person to be affected, is always in flux, producing a given
disturbing action will never have the same effect twice in a row. So
there can be no set formula of action for having a desired effect. All
that can be done is to monitor the aspect of the other's behavior that
you want to affect and vary your actions until you see what you want
or fail.
Having an effect is most easily done when the effect you want to see
is of no consequence to the influenced person. You don't have to
understand control theory to find such effects. All you have to do is
experiment and remember. Some kinds of actions will elicit particular
actions from the other person. You may have to vary your action under
different circumstances to continue having the same effect, but you
can learn that, too, by experimenting. Applying the Test is a
perfectly natural procedure that people use all the time. They just
don't put their conclusions into a theoretical framework, so they
usually miss seeing what the other person is controlling for. What
they see is action and reaction. That is quite valid as an empirical
observation, even if it does give a false picture of cause and effect.
This is done all the time. Almost invariably, it leads to conflict. It
is rare, I think, that any protracted use of this method on an
individual can avoid creating situations in which the controllee is
caused an inconvenience or worse by these attempts at control. The
seething violence that we see all around the world is the direct
result of failures of this method of controlling people. Most of these
attempts at control are done because of what the controller sees as
high-minded reasons, for the good of the other or of the society. Of
course there are other motives too, but I would guess that most of the
violence-producing control is done in service of an ideal. But what
causes trouble is that the controller's high-minded goals are not
those of the controllee.
What makes this method of control fail is that it is invariably done
for the good of the controller, not for the good of the controlled. It
is done to satisfy the controller's concept of the right way for other
people to behave, the best foods for them to eat, the substances they
should avoid, the way they should interact with others, what they
should do for a living, how they should dress and speak and appear,
what they should know. The controllers are trying to arrange the world
and the people in it to satisfy their picture of how things should be.
Things can be rearranged without any problem. But not people.
Theoretically, it's possible for one person with unlimited resources
and time to learn enough about another person's control structure to
elicit any desired behavior from that person without inducing errors
that would create resistance. The controller, in effect, would become
the means by which all the controllee's wants and needs are satisfied.
So in theory, half of the human race could devote its time to
controlling the other half's actions. The controlled half would find
that all of its goals were met while each person performed an unending
series of inexplicable acts that had no bearing on their own goals.
Their actions would satisfy the whims, prejudices, beliefs, and
convictions of their controllers, while the controllers made sure that
the controllees' goals were not violated, whatever the effort or
expense.
This is a totally unrealistic scenario. It has never been even
remotely approached. What has happened is that every attempt at
controlling another eventually arrives at the point where avoiding
inconvenience to the other person is just too inconvenient. The
controller, after all, has a life to live, too. Unending solicitude
for the welfare of others is impractical, not to mention that it
demands more knowledge than anyone has.
Pat implores me at this point to make the point that most day-to-day
interactions involve purposive influence which works without the
influencer having a very sophisticated, detailed model of the
influencee. Simple instances like "pass the salt, please." Give-and-
take. The success of the influence is presumed -- such behavior is
"civilized." Won't work if the influencee doesn't know English, true.
But it works almost always, otherwise.
Yes. But no model at all is needed for this. All you need to know is
that if you ask someone for something trivial, you will probably get
it. It isn't necessary to realize that the person is controlling for
something, and that NOT complying with the request would cause that
person an error. You may think that your request was what caused the
other to pass the salt. You may come to think that the request should
be SUFFICIENT to get you the salt. That's the pervasive mind-set I'm
talking about. All that the request did was set up a situation. The
salt would not have been passed if the other didn't have a sense of
error until it was passed. The result you see, which you interpret as
an influence and its effect, was really generated by the other in
order to prevent an error, not in order that you get the salt.
This will not work for requests that demand anything important of the
influencee. Please change your job. Please hand me all your money.
Please cut your hair off. Please speak respectfully to me. Please let
me ahead of you in line. Please don't tell my insurance company that I
got drunk and ran into your car. Civilized give-and-take works only
for trivial things, things that are really optional, that matter only
in a sort of hypothetical way, like manners.
----------------------------------------------------------------------
So how do we take care of our young, get along with our friends, teach
people things that will be useful to them, help people in trouble,
persuade others to do things our way, all without starting down that
road that ends in violent antagonism and chaos? I think there are
ways. I think that PCT can help us find them. But to find those ways
we have to break with the old mind-set, and we have to stop looking
for justifications of the way we're already doing things. The way
we're already doing things, no matter now nice and logical and right
it may seem, has given us the world we live in, which is approaching a
state of total failure. This is why I have so little motive to
establish links between PCT and more traditional approaches. My
attitude toward the traditional approaches is that the smartest people
in the world have tried them, and look what it got us.
----------------------------------------------------------------------
Best,
Bill P.