"Alignment" of references

Martin_Taylor · January 28, 1999, 6:07am

[Martin Taylor 990128 0:16]

Kenny Kitzke (990127.2100 EST) replying to Rick Marken (990126.1830)

... Bill Powers M SOB page 98 says, "And
people are hierarchies of control systems, in which reference perceptions
at one level are adjusted so at to make more general, higher-level goal
states."

Call it match, call it align or call it made internally consistent, I fail
to see where my understanding is wrong. Sorry.

It's quite important to an understanding of HPCT to see where your
understanding is wrong--to see how the notion of "alignment of references"
differs from the intention of the Powers quote.

What Powers is saying is that reference perceptions (i.e. the values of
the reference signals) vary _because a higher level system is bringing
its own perception to its reference level_. A higher-level system
output contributes to the value of several reference signals at the lower
level, and the higher-level system's output may change with time.

The higher system adjusts its output according to the difference between
its own perception and its own reference. Changes in the output of _any_
of the higher-level systems that contribute to the reference signal of
a lower level system can, and usually will, alter the level of the
lower level reference signal.

The perceptions at different levels are all simply strengths of signals,
but the things they correspond to in the world are of entirely different
kinds. One cannot "align" an event perception with, say, a configuration
perception, or a principle perception with a program perception.

Quoting Rick, you say

<Perceptions align with references (via operation of
the control loop which varies lower level referecnes); references
don't align with references.>

If my Jewish reference belief principle is to "eat no pork," I would set my
program level reference as "buy no pork." These are aligned references at
two levels for I can satisfy both of them easily by walking right past the
bacon display at the super market as dictated by lower level references.

What you are saying is that certain values for low-level perceptual signals
contribute to bringing values of higher-level perceptual signals to their
reference value. It is true that occasionally to achieve a particular
reference value of a high-level perception, there exists a lower-level
component perception that _must_ be at a specific value. Usually this
is not the case, and there are many different possible patterns of low-
level perceptions that result in the desired high-level one. One of the
aphorisms of PCT is "many means to the same end, many ends by the same
means."

You cannot satisy your reference belief "I am a good Jew" simply by not
eating pork. You have to perceive other things to have certain values
as well. But part of the output of a control system maintaining the
perception of "your goodness of Jewishness" at "high" will be trying to
set the reference value for the perception of eating pork at "zero."
However, it is not necessary that it also set a reference value for the
perception of buying pork at "zero." As someone (perhaps you) brought up
as an example, you could be a good Jew and nevertheless buy pork for a
Gentile friend. All the same, "many ends by the same means." Both
perceptions, "eating pork" and "buying pork" are more easily kept at
a level near zero by the action of "walking right past the bacon display."

If you call all the "ends" that are aided by a particular "means" "aligned
references" then I guess you have the right to use that as a definition.
But I don't think it is helpful when trying to understand HPCT. And it
will confuse other people trying to understand PCT.

Again quoting RIck...

What seems inconsistent is your interpretations of the way things work.
Here is an example:

A: Then that hypothesis is resting on a very shaky assumption since,
in an HPCT hierarchy, it makes no sense to say that "the
references we set at all levels would be aligned or internally
consistent".

B: But, based on my understanding of what you are asking, I think the
answer is
"yes" if you mean that references at two different levels of the hierarchy
cannot create a conflict; this is true because references at two different
levels cannot specify states of the _same_ (or even a similar) perceptual
variable.

Indeed, which is it.

It is both.

When reference levels we set cannot create a
conflict, I would say they are mutually consistent and aligned.

You would then necessarily say that all references at one level are
"mutually consistent and aligned" with every reference at every other
level, because you cannot have a conflict between a reference value to
perceive a particular kind of sequence and a reference value to perceive a
particular kind of colour or a reference value to perceive a particular
kind of program. Note that I'm not talking about the _values_ of those
references. It is quite possible that to achieve a particular value of
sequence, one might have to perceive a particular value of colour that
is different from the one set by the other reference signal. But the
conflict is between the need to perceive the two colours in the same
place, not between the sequence and the colour.

···

-------------------

When we reorganize, our references are changed to align under our
perceptions of the world.

Oooops! What do you think is meant by "reorganize?"

Reference values change all the time as a consequence of changing outputs
from higher levels. Reorganization doesn't change references. It changes
the _structure_ of the hierarchy. Reorganization can change _what_ we
perceive (by changing the connections to perceptual functions), it can
change what perceptions we _can_ try to control (by creating new
control loops to influence things we can perceive), and it can change
what actions we may have available to control those perceptions (i.e.
how lower level reference inputs are linked to higher-level outputs),
and the distribution weights that refine all these things. Not much else
(actually, years ago I posted a taxonomy of the possibilities. There are
12, but most are probably not useful).

IOW, at any time t, all the references we have
set are aligned and mutually consistent. Until there is a disturbance that
requires a reorganization they will stay that way.

Disturbances don't induce reorganization, at least not according to
orthodox PCT. Sustained and increasing error may do so, but it's primarily
sustained and increasing error in the genetically determined intrinsic
variables that affects the rate of reorganization. Disturbances for which
the existing control hierarchy cannot compensate may induce an increased
rate of reorganization because that implies that somewhere there is a
sustained error. What normally induces reorganization, however, is
conflict, because conflict almost assures that somewhere there is a
sustained error. The conflict may be within one body or between two
(or more) bodies. Either can cause sustained error and influence the rate
of reorganization.

---------------------

<Or is this just a statement of fact, like the fact that
President Clinton is immoral and the House and Senate Republicans
are moral;-)>

It is a statement about perception. I don't determine the morality of Mr.
Bill or Republicans, they do.

I thought only "Mr. Bill" could judge his own morality, according to you.

Was I wrong? If I understood you correctly, from before (and in this
message, too) only you can judge your morality, and you alone could say
that if you did what you perceive him to have done, you would perceive
yourself to be immoral. If that's the case, how can "they" (whoever
"they" may be) determine his morality? They can find out if they have
the power to have him removed from office, but that's a very different
thing, isn't it? Or is it, in your view?

Martin

KENNETH_J_KITZKE1 · February 1, 1999, 6:46pm

[From Kenny Kitzke 990201.1300 EST]

When the higer level system is not sensing error, is it producing output to
set lower level systems reference signal levels? Again, if it is
contributing to the value of the lower level reference signals, why would
it end up setting anything that is conflicting internally?

This gets so confusing. I think PCT explorers throw their hands up and say
PCT, especially HPTC, is so complicated it will never be more useful for
solving conflict internally or with other external people than the claimed
faulty approaches now being used by S-R type psychologists.

Just when I seem to be gaining understanding, I fall back. If my reference
signal at the principle level is eat zero pork, I would program myself to
not buy any bacon. It still seems these reference signals are indeed
aligned so that both can be achieved by a particular program of behavior.
But, I am beginning to see that the use of the term align is what our
behavior is trying to achieve relative to our perceptions and our
references.

If I had a principle of eat no pork. And I also had a principle of eat
what tastes good. If I unknowingly taste bacon fat and it tastes good,
what happens next when I find out its pork I'm tasting? I must reorganize
at a higher level? This kind of conflict at the principle level does seem
clear.

I would say the more they are aligned (non conflictive), the easier it
would be to achieve successful control.

Perhaps this case is the one I think is normal, while you call it unusual.
<Usually this is not the case, and there are many different possible
patterns of low-level perceptions that result in the desired high-level
one. One of the
aphorisms of PCT is "many means to the same end, many ends by the same
means."

You cannot satisy your reference belief "I am a good Jew" simply by not
eating pork. You have to perceive other things to have certain values
as well. But part of the output of a control system maintaining the
perception of "your goodness of Jewishness" at "high" will be trying to
set the reference value for the perception of eating pork at "zero."
However, it is not necessary that it also set a reference value for the
perception of buying pork at "zero." As someone (perhaps you) brought up
as an example, you could be a good Jew and nevertheless buy pork for a
Gentile friend. All the same, "many ends by the same means." Both
perceptions, "eating pork" and "buying pork" are more easily kept at
a level near zero by the action of "walking right past the bacon display.">

This is helpful. I may get it eventually. Or, I may go insane.

I don't want to confuse others. I want them to make PCT clear to me. A
well organized hierarchy would seem to be one that is nicely aligned and
can be perceived as meeting references in the most ways.

That may be what is really confusing me -- I don't properly understand
reorganization. I'll try to study that a little more. That might resolve
some of these moral belief concerns I have, because I do not sense that my
moral beliefs change very often. So, I might hold my eat no pork reference
for a lifetime, but might vary my buy no pork reference so that my wife's
Gentile father can eat a big juicy ham for his Easter meal a my house.

This too seems helpful to understanding reorganization, that it can result
from internal or interactional sustained error. I like that.

<<I don't determine the morality of Mr. Bill or Republicans, they do.>>

Yes, Mr. Bill determines his own moral beliefs and the Republicans each
determine their own and you an I determine our own. Our morality is
defined by how well our perceived behavior reflects them.

Bill Clinton is not being impeached for being immoral. He is being tried
for violating the very laws of the land that he swore before God to uphold.
If so, the constitution allows the Congress to remove him from office. It
seems just and fair to me.

It is immoral to commit adultery for me, but it is not a crime. But,
perjury about any material fact is both immoral to me and illegal in
society. Being popular as a president does not enter into these
definitions. It only determines how many will perceive being happy or sad
after the final outcome is announced. Allowing popular opinion to
determine moral beliefs for others would foster corruption unfanthomable
and much chaos and coercion. Or, so I perceive.

Best regards,

kenny

R_Marken · February 1, 1999, 7:53pm

[From Rick Marken (9901201.1250)]

Martin Taylor (990128 0:16) --

What Powers is saying is that reference perceptions (i.e. the
values of the reference signals) vary _because a higher level
system is bringing its own perception to its reference level_.

Kenny Kitzke (990201.1300 EST) --

This gets so confusing. I think PCT explorers throw their
hands up and say PCT, especially HPTC, is so complicated it
will never be more useful for solving conflict internally or
with other external people than the claimed faulty approaches
now being used by S-R type psychologists.

HPCT is really very simple once you understand how a simple
control system works. Think about the basic tracking task.
In that case a fixed reference (for the position of the cursor)
results in a fixed perception (of cursor on target) because
a lower level output (position of the mouse) varies as necessary
to prevent disturbances from affecting the perception (of cursor
on target).

To think of this in HPCT terms just think of the varying output
as a varying _reference_ for a lower level perception (of mouse
position, say). Now you can see how a higher level perception
(cursor position) is protected from disturbance by variation
of a lower level perception (mouse position). The lower level
reference (for mouse position) is not "aligned" with the higher
level reference (for cursor position). If you keep typing to
think of the operation of a control system this way you will
just remain confused. Your confusion will disappear as soon
as you see that "alignment of references" has nothing to
do with the operation of a control system (or a hierarchy
thereof).

Control systems vary lower level references (or outputs,
if they connect to the environmnet) _as necessary_ to keep
higher level perceptions matching higher level references.
Just study the basic control diagram; that should do it. Once
you know how the basic control system works, you can see why
it is often necessary to vary lower level references to
protect higher level perceptions from disturbance.

For example, suppose that you are controlling for a higher
level perception like "being kind to others". One way to
control this percpetion is by being "honest" to other people.
So the perception "being kind to others" is equivalent to
"cursor on target" in the basic tracking task; the reference
for "honesty" is equivalent to the reference for mouse position.

In most cases it may be true that you can keep the "being kind
to others" perception in its reference state by setting the
reference for honesty at a high level. But suppose that a
person asks questions (disturbances) which, if given an honest
answer, would make your percpetion of "being kind to others"
deviate considerably from its reference. For example, the
person might ask "how long do I have to live, Dr." What you do
(automatically, just as in the tracking task) is give a less
than completely honest answer (you say, "I don't know" instead
of "less than a month").

The complex part comes from the fact that you might have other
control systems that are _also_ setting the reference for
"honesty". In that case, you get might _conflict_. For example,
the system controlling for "being kind to others" tries to lower
the reference for honesty in order to keep the "being kind to
others" perception at its reference; the other system, controlling
for "following the letter of the law", say, tries to set a high
reference for honesty to keep the "following the letter of the law"
perception at its reference). The result is _conflict_. The
system controlling for "following the letter of the law" doesn't
get the perception it wants; the system controlling for "being
kind to others" doesn't get the perception it wants. What you
will see (if the person speaks at all) is what you heard
Clinton saying in his depositions; something in-between being
nice to your others and following the letter of the law.

Does this help?

Best

Rick

···

---
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

Tim_Carey1 · February 1, 1999, 9:08pm

[From Tim Carey (990202.0710)]

[From Rick Marken (9901201.1250)]

Does this help?

Sure does. Thanks for a wonderully clear description.

Great post.

Tim

KENNETH_J_KITZKE1 · February 2, 1999, 4:40pm

[From Kenny Kitzke (9901202.1200 EST)]

Yes. Thanks.

I never tried too hard to understand HPCT becuase 1) at a part time pace, I
had a hard enough time understanding PCT and 2) HPCT had a lot of admitted
wiggle room about how many levels, types, etc.

I think I am free of the alignment of references idea. BTW, what also
helped is the point in MSOB p38 where Bill Powers distinguishes between a
control hierarchy versus a command hierarchy. The perception I have of a
hierachy structure is the top level sets what is supposed to happen below.
Of course, in business, what is wanted to happen changes far too
frequently.

I still feel that even though once you get HPCT, it might be easy to
understand, it is still difficult to understand until you play with it
enough to get comfortable with it. Like juggling.

Respectfully,

Kenny