Amicus Brief

[Bruce_Abbott]

[From Bruce Abbott (950609.1145 EST)]

Bill Leach 950608.20:14 U.S. Eastern Time Zone --

Your Honor, we have examined the amicus brief filed by your "friend of the
court." We find in his statements great insight and wisdom concerning all
those cases wherein he agrees with our position. As to the rest, well,
nobody's perfect. However, we believe he will come around to our side after
we have had the opportunity to offer our response (and perhaps a small
$$token$$ of our appreciation) in those areas where we appear to disagree.

The initial "No real control system can keep the controlled perception
_exactly_ at reference at all times regardless of disturbance." IS true
in the absolute sense... and in that sense is completely irrelevant to
the discussion at hand.

Your Honor, this is DIRECTLY relevant to the discussion at hand, in that we
are arguing that stress may arise even when there is control. Control is
never perfect, and error can lead to stress.

Moreover, the objection suggests a serious misunderstanding of basic
closed loop negative feedback control system operation on the part of
Counsel. To wit: The statement that "In the basic ... there is no
output to counter ... when the error is zero." is again clearly true but
wholely irrelevant.

The error signal value will be at a non-zero value pretty much 100% of
the time when control exists. However, the important consideration (both
from a discussion/analysis point of view and, it appears from ample
evidence, the control systems view) is that the amount of error will be
very small as long as control exists and rise to near full possible value

Your Honor, my client has logged time on DEMO1 and DEMO2 and understands
full well how the basic control system operates. The Friend-of-Court
assumes in his argument that the system in question is high gain and lacks
significant lags. Neither of these assumptions is necessarily true of a
given real control system, in which case the error may at times be
significant even though reasonable control is maintained.

The error signal value will be at a non-zero value pretty much 100% of
the time when control exists. However, the important consideration (both
from a discussion/analysis point of view and, it appears from ample
evidence, the control systems view) is that the amount of error will be
very small as long as control exists and rise to near full possible value
when "good control" is not obtained.

Your Honor, this has not been our experience on compensatory tracking tasks.
The error has certainly been less than the "near full possible value" and
yet not "very small" either, large enough to arouse some degree of stress.

That is in the basic control loop, error signal value immediately jumps
to maximum output as soon as the controller is unable to make perception
match reference to within the deadband of the controller.

Your Honor, a control system that "slams the needle" on the output as soon
as there is dectable error by definition is a high-gain system and does not
represent the general case.

OTOH, Counsel's description of motive in his objection is a "nice
sounding PCT like description" but probably not at all accurate.

The "friend of the court" suggests that first a "MOTIVE" is a perception.
Secondly a "MOTIVE" is specifically a conscious perception. Thirdly
these is associated with this "MOTIVE" the perception that it is a
"reason" why certain goals are desired.

It is further suggested that "MOTIVE" includes conscious perception by
others concerning why the subject might or might not be controlling a
particular EV.

Objection, Your Honor. Motives may become objects of perception (one may be
aware of one's motives) but often are not ("unconscious motivation").
Motives-as-perceptions are inferences as to what the motives are; they are
not the motives themselves. We may be wrong about these inferences, both
about our own motives and those of others, but that does not change the
motive itself. To state the motive is to state what goal-state is being
sought or defended.

Regards,

Bruce

[Bill_Leach1]

<[Bill Leach 950609.19:08 U.S. Eastern Time Zone]

[From Bruce Abbott (950609.1145 EST)]

We find in his statements great insight and wisdom concerning all those
cases wherein he agrees with our position. As to the rest, well,
nobody's perfect.

Interesting statement. Are you saying that IF I agree with you then my
statements contain "great insight and wisdom" but IF I DO NOT so agree
then they fail to contain "insight and wisdom"?

Your Honor, this is DIRECTLY relevant to the discussion at hand, in that we
are arguing that stress may arise even when there is control. Control
is never perfect, and error can lead to stress.

There are a number of concepts here in which we need to "come to terms"
over just to find out if we are even talking about the same things or
not.

To start with, I absolutely agree that NO control system can control its'
perception to exactly the reference value at all times. In an absolute
and exacting sense, perfect control probably exists much much less than
1% of the time. This fact is irrelevant to the sort of discussion that
we are having.

Control is "good" basically when the perception is maintained close
enough to the reference that the control system is "satisfied". Not a
terribly useful definition but it is a terribly important concept. It
does not matter how effective the control appears to us observers, the
only factor of import is whether the subject system considers control to
be good or bad based upon its own criteria.

"Bad" control is the situation wherein the perception is NOT being
maintained to within acceptable limits about the reference value.

Stress is the existance of unacceptable error. Much discussion has taken
place on CSG-L concerning stress in the past. Some of the generally
accepted conclusions:

    Stress itself probably is not a "real" phenomenon but rather a lable
    that we apply to a property, that we "notice", that must exist in a
    self-organizing control hiearchy. (ie: There is probably no "Stress
    System").

    Further, the term Stress generally applies to a perception. That is
    when we "feel frustrated" or the like, we consider ourselves to be
    "under stress". The perception itself is undoubtedly related to the
    actual metabolic effects sensed that are themselves controlled
    perceptions.

    Increasing error generally means increasing stress value.

    Individual error signals probably have a threshold for onset of
    stress and grouped error likely do as well.

    Multiple small errors, each of which _might_ not individually be
    resulting in any stress probably combine to create a "general" stress
    level perception.

    There may be a "priority" associated with controlled perceptions
    (actually there undoubtedly is such). It appears that an inability
    to bring some perceptions under control at all is not "stressfull"
    while even mildly poor control of others is very Stressfull".

    The PIF for the reorganization system probably includes not only a
    priority sensitive scheme but probably also an integrator.

    Reorganization probably begins at some level of error above
    unacceptable.

    The rate at which reorganization proceeds is probably a function of
    the extent to which error(s) exceed their acceptable limits.

    If Stress is viewed as the activation of reorganization (ie: error
    amount is perceived to exceed threshold limit) then most "Stress" is
    not consciously perceived at all.

If a control system is controlling a perception to within acceptable
limits then there is no resulting stress (with respect to that control
system). Stress could result from other perceptions such a perceiving
that one is "exhausted" or even that one "will" fail to control in the
future. Even irrational perceptions can "cause" stress. For some
people, the perception of stress itself causes stress (cardiac and ulcer
patients might easily fall in this catagory).

So yes, control is never "perfect" and error _can_ lead to (a perception
of) stress. Or excessive error IS stress depending upon just what you
would like the term stress to really mean.

Your Honor, my client has logged time on DEMO1 and DEMO2 and understands
full well how the basic control system operates. The Friend-of-Court
assumes in his argument that the system in question is high gain and
lacks significant lags. Neither of these assumptions is necessarily
true of a given real control system, in which case the error may at
times be significant even though reasonable control is maintained.

Bruce, I am only going to say that in my over 30 years of experience with
closed loop negative feedback control systems that HIGH OPEN LOOP GAIN is
a basic fundamental requirement for successful control in all but very
limited and very well designed environments. In my experience a control
system displaying error values above a few percent of full scale is
"barely hanging on" and will likely soon fail altogether. Momentary
spikes in the error signal to even 100% of output are usually not of
concern for most systems.

The internal signal values in the models that we create to run on digital
hardware may in reality have very little resemblance to the equivalent
signals in the "real thing".

Your Honor, this has not been our experience on compensatory tracking
tasks. The error has certainly been less than the "near full possible
value" and yet not "very small" either, large enough to arouse some
degree of stress.

Unless you have been making some measurements that no one else know how
to even begin to approach making the connections, the "errors" that you
are talking about are only incidentally related to the errors within the
organism. In addition, knowledge of the magnitude of an error signal
does not, at this point in time, give us a clue as to that signal's
relationship to reorganization (or like system).

Motive

Sorry, I don't buy the idea at all that there is something called a
"motive" that actually exists as an entity within a living system other
than as a conscious perception.

One of the very serious problems that exist in all such discussions today
is that we so readily equate "described and labeled" phenomenon to actual
functional requirements within the organism.

If the motive is the goal, then lets drop the term motive as it is far
more ambigious than the term goal.

"The Motives are the reasons why we do or want things" is a common
definition for the term. It is also devoid of any objective meaning.

-bill