[From Bill Powers (971224.0449 MST)]
Marc Abrams (972412.0200 EST)--
My interest in _applying_ PCT is in trying to understand .How and why
people _change_. From a PCT perspective there are 3 ways to change.
1) Change a reference level.
2) Change a CV
3) Change a perception.
In changing a reference level you are attempting to change a specific point
of magnitude (intensity ?) of a perception. (eg. Colder-Hotter, More-Less,
etc.).
In attempting to change a CV, your attempting to change the _type_ or
_kind_ of perception. I think the S-R line of thinking calls a CV a "goal"
or a "need".
In attempting to change a perception you are attempting to change the way
you "see" the world. This is probably the most difficult thing to do
because when you change one perception you _must_ change _all_ the ones
that are connected to that one in the hierarchy.
Since control systems have three components, there are really three ways to
change: change the form of the input function, so maintaining the same
perception means bringing the world to a different state; change the value
of the reference signal, so you maintain the same perception at a different
level (for example, zero instead of maximum); or change the output
function, so the same error leads to altering different lower-level
reference signals.
As you correctly point out, "changing the reference level" is ambiguous: it
could mean changing the level at which you maintain the same perception, or
changing to controlling a different perception. In any case, changing the
reference level of a control system is done by a higher-level system, not
by the system whose reference level changes. So that's really about the
higher-level system. If all the connections from the higher-level output
function to lower-level reference inputs stay the same, all that can change
is the amount of signal sent from the higher system to the lower one. But
if the higher output function is reorganized in a different way, the
connections can actually change, so different lower-level control systems
are used to achieve the same higher-level perceptual state.
It's possible that many more connections exist from higher to lower systems
than actually carry signals at any given time. Complex higher systems, for
example the systems at our hypothetical program level, operate by selecting
different output paths depending on the outcomes of rule-driven
computations. At one time you may get to work by driving, but if logic
tells you that is impossible (the car isn't working), you can select a
different lower-level behavior and take the bus or the train, or ride with
a neighbor. This kind of change in lower-level behavior isn't
reorganization; it's just the normal way a logic-level system operates. So
there are many ways in which behavior can change that don't need the
concept of reorganization as an explanation. Reorganization, at the program
level, happens only when all programs fail to correct an error. And even
then, a change in choice of principles can result in activating a whole new
set of learned programs, offering more possibilities for correcting errors
without reorganization. Provided, of course, that the new programs have
also already been learned.
A properly-organized hierarchy of control should make it unnecessary for
reorganization ever to occur. The more levels there are in a hierarchy, the
more kinds of situations there are in which a learned method of control is
available, and the less likely it is that intrinsic errors will occur to
turn on the reorganizing process. The entire purpose of the reorganizing
system is to turn itself off by building learned systems that can prevent
intrinsic error from occurring.
This is easiest to keep straight if you remember my basic principle in this
regard (and agree with it): no system ever changes itself. That's why there
is an explicit reorganizing system in my model; if a control system's
organization changes, something outside the system must have changed it.
That's also why there is a hierarchy: if a goal or reference signal
changes, something else at a higher level must have changed it. That's why
there is a Method of Levels: if you want to change something about your
organization, you must find the point of view (outside the system that is
to be changed) from which such changes can be made. In PCT, there is no
such thing as a "self-organizing system." If change occurs, we have to
consider both the system that is changed, and the system that is
responsible for making the changes.
All that the basic control system can do is convert inputs into
perceptions, compare the perceptions with reference signals provided from
elsewhere, and convert the resulting error signal into outputs. That is
ALL. As soon as you start talking about anything else, like changes in
functions or reference signals, you're talking about something other than
the basic control system. Even if there's a reorganizing process connected
with every individual control system, that process is different from the
basic control process; it does something that the basic control system
doesn't do, and so has to be considered as the action of a different
system. When that system has been successful, it ceases to act, while
control continues in the system that was modified. Even in Hans Blom's
model-based control system, there was that "Kalman Filter", which acted to
keep the model's simulated output matching the external plant's actual
output, by altering the parameters of the model that was part of the inner
loop. So there were really two different systems involved, doing different
things.
Creating change consciously, on purpose, is, as you suggest, very
difficult. Once organized, the hierarchy of control systems can't really
change its own organization (it can only degrade). One reason we make
models is to fill in the gaps that conscious experience leaves, and one of
those gaps is the process of change. A higher-level system can _appear_ to
alter the organization of behavior, but it can do so only by means that
have already been learned. When none of those ways works, all that is left
is reorganization, and reorganization itself can't be directed by any of
the hierarchical control systems, at any level. You can't reorganize just
by deciding logically that reorganizing would be a good idea. You can
reorganize only by suffering intrinsic error.
Since people don't like intrinsic error and will immediately start changing
themselves, quite automatically, whenever it occurs, it would seem that
there is nothing one can do to reorganize on purpose; either you do or you
don't. This would make it seem that there's no point in trying consciously
to change oneself. Yet it is possible to learn ways of allowing
reorganization to occur; for example, you can put yourself in painful
situations and, for perfectly logical reasons, _cause_ intrinsic error.
This is somewhat paradoxical, because according to theory this should
result in reorganizing the logical system that produced the intrinsic
error, so you stop doing that. But such situations can certainly exist
temporarily. In fact this is just another aspect of the idea of
reorganization driven by intrinsic error: if logic, or any other high-level
process, ends up creating intrinsic error, the result will eventually be a
change in that process until it stops creating intrinsic error. This is
what makes reorganization appear to be the highest level of control, even
though it's not a systematic control process and can operate anywhere in
the hierarchy.
But this still doesn't answer the question of how we can _consciously_
change our organizations. The only hint I have ever found in this regard is
the way conscious attention seems attracted to problems, to error signals
in the hierarchy. There are certain error signals that simply can't be
ignored. I remember watching a golf tournament in which there was a delay
at one tee, and it was discovered that the tournament leader had simply
disappeared. Nobody knew what had happened. Then, five minutes later, he
reappeared and took up his driver, and suddenly there was a lot of
embarrassment among the announcers who had been worrying publicly that
something had happened to him. It turned out that his bladder had made
demands that his tournament-winning systems could not deny.
That just shows that fear of wetting one's pants on TV can outweigh the
desire for winning money, but the point is that we all know what happens to
conscious attention under such circumstances: it goes to the problem. So I
speculated that reorganization follows conscious attention. This seemed
very convenient to me, because it can explain why, when we have some
intrinsic error that has no direct connection to a control problem,
reorganization doesn't just disrupt systems that are already working
perfectly well. If it can be directed to the cause of a problem,
reorganization will operate where it will do the most good, even if one
can't predict its outcome.
This ties in with what psychotherapists have found: whatever the method, it
must entail directing conscious attention to the real causes of the
problem. In the Method of Levels, this becomes the central issue: if
attention is directed to the right place in the hierarchy, nothing else is
required of the therapist. In fact there is nothing else the therapist can
do that will have any effects on reorganization. What actually works in
psychotherapy is redirecting attention; everything else that goes on is
irrelevant, or even detrimental.
Directing attention to a problem also seems to make it temporarily worse. I
suspect that there is side-effect of activating conflicts. Whatever the
explanation, this indicates why people do avoid becoming aware of their
real problems: doing so makes them feel worse, not better. This is known in
technical parlance as a local minimum -- the error has to get larger before
it can get smaller. This local-minimum effect is what gets people stuck in
bad organizations. If they start to change, the intrinsic error becomes
larger, and they will reorganize right back to where they were. Some kind
of external nudge is needed to get them over the hump. Maybe a drug can do
it, but redirection of attention is probably enough in most cases. The
problem with drugs, especially feel-good drugs, is that they may have a
side-effect of turning reorganization all the way off. Others can turn it
on, but not in a way that's related to any problem. ECT is obviously an
externally-imposed radical reorganization; I've likened it to the game of
52-pickup.
(For those in different cultures, 52-pickup is a card game that many
children learn. Want to play 52-pickup? Sure -- how does it work? I'll show
you. I'll be dealer. I take the deck of cards like this, shuffle it, and
throw it up in the air with a spin. Your turn: 52-pickup.)
All this is guesswork guided by a few bits of evidence. But that's what I
think about change, as of 971224.
Best,
Bill P.
ยทยทยท
So when we talk about helping people change. I think, practically we are
restricted to either 1 or 2 above. Thats not to say that at some future
time we will have the capability of understanding the interactions, right
now we don't.
We run into a similar problem with CV's, but not quite as inhibiting.
Knowing _a_ CV is a _starting_ point. I think the more sub CV's one knows
the better one _understands_ the characteristics of the situation. I'm not
sure how to "get there". I think the MOL provides a resonable way to help
someone reflect. The work of Chris Argyris and Action Science have a
number of techniques to help people help others "reflect". unfortunately
the reflection that they concentrate on is on the behavior and not the CV's
of individuals. but i do not believe that that is a major stumbling block
to utilizing their techniques for reflection on CV's. I am currently
writing a paper on it ( Utilizing PCT as the behavioral model for Action
Science) and I should be done shortly. I'd be happy to provide a copy to
anyone interested. (Privately e-mail me, don't waste the bandwidth on the
list )
Finally, I think Tim's C's questions are _extremely_ important, and when
anwered honestly (upon some _serious_ reflection) what is it about
"helping" others that matters most to us and to whose benefit are we working?
Thanks for your patience. With a final, 
OK, What am I missing
Marc
anyway as Dag would