The law, he said briefly; getting somewhere with conventions

[From Bill Powers (931030.1230 MDT)]

Lewis Henry LaRue (931029.1510 EDT) --

Hugh Gibbons has written more extensively about "the law itself"

Gibbons, H.; Justifying Law: an explanation of the deep structure
of American law; _Law and Philosophy_ _3_(1984) pp. 165-279.

He will probably send you a reprint on request.

This paper was written before Hugh's more extensive explorations
of PCT but after he had read my book. He was trying to deduce the
highest level of principle he could find in the way the law is
actually administered. Here is a quotation from an early part (p.

"Ordered coercion is the central aspect of law. Law may be more
more. It may be the vehicle by which the dominant class exerts
control. It may be a socializing force, teaching people how to
behave in complex situations. But it _must be_ a pattern for the
use of coercion in society.

"... It requires the idea of will, that one is a subject,
proceeding under the control of one's own plan. Coercion is the
overwhelming of the _will_ of one by another. That is the subject
of law. Law determines when and under what conditions the will of
one person will control another."

"... At its root, law is a scheme for answering a single
question: Upon what basis is coercion justified? Or, when is it
right for the will of one to control another?"

And finally, on p. 170, after rejecting several other possible

"I suggest that the following answer generates a structure of
propositions that fit our law quite well (and are hence an
approximation of its deep structure): Coercion is justified when
it is part of a general scheme for expanding the range and power
of the will of each person. One may control [against] the will of
another only if that is justified somehow by an expansion in

"In what follows I will explain American law as a scheme for
expanding the ambit of individual will."

( [against] added above by me).

On the very next page he introduces a control-system diagram as a
way of illustrating what "will" and "purpose" mean, taking his
argument off the level of metaphysics and into the real world of
human organization (as we see it under PCT).

While Hugh sometimes plays fast and loose with the sacred texts
of PCT, he is a sufficiently clear and original thinker that I
prefer to raise only the most major objections, and then only
when doing so seems productive. His departures may contain the
seeds of future improvements in the theory. So don't take Hugh's
renderings of PCT amiss if they depart from mine, because at
bottom he gets the point quite well enough to have earned a
license for experimentation with embellishments and alternatives.

You comment:

I first became unhappy on p. 5, where he states: "Law exists to
guide and limit the choices that we make." As a historical
statement about the genesis of law (back in the mists of time),
this seems false. As a social science statement about the role
that law plays in contemporary life, it also seems false. I
think I can defend my two assertions about past genesis and
present role, but I don't want to go into detail unless others
feel the need. For the moment, let me say that Hugh's comment
is far too "functionalist" and too "rationalistic" for me.

I don't think you would have much trouble convincing Hugh that
this "purpose" of law is a loose attribution; I assume he meant
that this was the effect that law had, rather than its intended

I join you in disagreeing with the statement that the law is a
control system, but again I would expect Hugh to give up that
wording rather easily, in favor of saying that it is used by
human beings as an instrument of control. The human beings are
the only control systems involved, the law being a means to
enable them to achieve their purposes. But the purposes of early
laws may not have been what we now read them to be.

As to what the law "is", that depends on the level from which you
view it. As a system concept, it is a vast collection of
institutions, principles, customs, and procedures managed by
individuals in a large array of professions from professor to
bailiff. It is conceived of as an entity despite constant changes
in form and personnel, like a baseball team.

At the level of "laws," the law is a set of contingencies, like
the if-then clauses in computer programs, which impose a cause-
effect structure on the world in which people conduct their
affairs. If it were not for laws, people could park anywhere they
like as long as they please, with no unwanted consequences. Laws,
at least as they read on the books, establish, for performing
certain actions, side-effects that have negative or positive
value to the actors: a fine for overparking, or a tax exemption
for donating to a charity. So laws add social properties to the
physical world, properties that have to be taken into account as
people act to control what is happening to them.

Laws in no way prevent people from performing any action, or
cause them to perform any action. All that laws do is state
consequences. If you're willing to pay a $10 fine for every hour
over the limit, you can park as long as you like -- which, if
parking garages are charging $15 per hour, may be a bargain.

So what Hugh says is right: the essence of law is in the
coercion, because coercion makes certain that you will not be
able to avoid the consequence. You may choose to pay the $10
fine, but you can't choose to overpark and NOT pay the fine -- at
least in principle. The law contains coercive mechanisms of
sufficient power to overwhelm anyone's direct attempt to change
the contingencies. Again, of course, in principle. People have
many ways of avoiding consequences, and the law is imperfectly
implemented, but we're talking design, not application.

There's a level between cause-effect programs and system
concepts: principles. This, I glean from what Hugh says, is where
judges work. Specific laws are applied or not applied according
to the principle at issue. This is a fuzzy area where "the law"
ceases to be a simple list of acts and punishments, and something
more general is considered. Hugh has expressed his First
Principle as if it exists at this level: respect for the will of
others is a principle. But I believe this principle is only a
reflection of a system concept, a concept under which a person is
viewed as an autonomous entity pursuing internal purposes, with
freedom to do so being synonymous with the human condition.

From this conception of human nature, it follows that "expanding

the ambit of individual will" is a principle required to support
the system concept. It isn't the only such principle, but it is a
good expression of what is required for a person to live free --
or for as many as possible to live as free as possible. Hugh uses
the concept of a human being as an autonomous control system as
the highest justification for his principle: the principle simply
expresses what is needed for control by individuals of their own
lives to work as well as possible. Hugh's claim is that
principles of this kind actually do determine the kinds of
decisions that are made in courts of law (good ones, I presume).

Hugh claims that the law is a "control system", and
furthermore, he seems to state that it is a control system is
the very same sense that an individual is a control system.
*Dubitante* on that claim.

Yes, I agree. The law is a "control" in the same sense that the
volume knob on a radio is a "control." Left to itself, the volume
knob controls nothing: it is a _means_ of control when used by a
human being for a certain purpose. The law has no purposes of its
own; it is simply a means by which people who make and apply laws
can act toward achieving their purposes.

As usual, I strike out boldly into fields of which I know next to
nothing. It might be more informative for you to hold forth on
the relationships you see between PCT and the law.

Maybe you can persuade Hugh to get on the network. He started to
do so once, but gave up because he was "too busy." I find that
excuse irrelevant and immaterial. We need him here, don't we?
What could be more important than that?


Bruce Nevin (931028.1403) --

Well, it looks as if this jumbled mess is settling down into some
sort of order. Everything you say about social conventions now
makes sense, with no false notes that I can hear. In your spare
time (ho ho) maybe you could work all of this up into an essay
that provides the correct translations between those who think in
terms of individual models and those who deal with the phenomena
directly. That might even be worth publication in a linguistics
journal (as well as here).
RE: diphthongs

I've been concentrating on the vowels, assuming that if I (or
someone) can figure out a way to produce them in a realistic
manner, diphthongs can be taken care of by a transition-level
system that creates a smooth change from one vowel-sound to
another by varying the appropriate reference signals. You need
transition systems for consonants, too.

I suspect that in making diphthongs, the vowel sounds execute a
smooth transition from a starting vowel to an ending one, as in
"ah-oo" for "ow" and "oo-ah" for "wah." The formants on my crude
spectrograph swoop smoothly from one state to the other, passing
through all intermediate states. If we already had a level of
control that could produce any given vowel to match a reference
signal, then all the higher level would have to do to produce a
diphthong would be to generate a changing reference signal.
Avery Andrews (931030.1130) --

Emphasizing the role of reorganization on the output side might
help in making the distinctive nature of pct clearer. Suppose
you have a three level control system like Rick's spreadsheet
model, with fixed structure. Then you *can say* that there are
rules for producing outcomes (if the hand is X-much too far to
the left, move it to the right at rate Y, ...). It's just that
these rules are differential in formulation, so you have to
take dynamics into account to understand how they work (and, if
you don't know what the high-level reference settings are,
you'll never really understand what is going on).

This came up in a conversation with Dag Forsell this morning. I
don't think it's so much reorganization that will help as in the
idea that output functions map error signals onto _directions of
change_ of lower-level reference signals. Once a control system
is organized properly for its environment, any kind of error will
produce a change of lower reference signals in a direction that
leads toward smaller error at the higher level. While there may
be no literal "map," the effect of the proper output function is
as though there were a map.

What reorganization does is alter the entire map. When there is
no literal map, it alters the coefficients in the equation that
transforms an error signal into an output signal that contributes
to lower-level reference signals. When the right organization is
found, the map stops changing or the coefficients stop changing,
because now any magnitude or direction of error signal will
create lower-level changes that work in the right direction. This
is sort of like a proof of convergence. If from any starting
condition the system always tends to make the error smaller, it
will eventually converge to zero error. When that is true,
reorganization is no longer needed.

[This, incidentally, suggests another plank in the bridge to the
motor-schema folks. An output computation is indeed needed. But
it is only a computation of directions in a space, not
computations of the location of the current goal. If every point
in error-space maps onto decrements or increments of lower
reference signals that tend to move the point toward 0, then the
outcome (set by the reference signal) is assured.]

Even in Rick's three-level spreadsheet there are really no rules
for producing outcomes. The reason is that Rick also has
disturbances acting on the outcomes, so if the same output is
applied twice, the outcome will be different from one occasion to
the next.

If there is anything that can be taken as a "rule" it is of a
subtler nature. It lies in the definitions of all the control
loops involved. For each output of a control system, the effect
of an output signal by all possible lower-level paths is to cause
a _change_ in the net perceptual signal at the higher level
_toward_ the reference signal. What disturbances do is to alter
the error signal. But however the error signal is altered, it
ends up in another state which the control system is already
organized to turn into an action that tends to reduce the error.

The problem with emphasizing reorganization is that this makes it
seem that all action involves searching randomly for something
that will work, as if every new state of the error signal catches
the control system by surprise. This is actually a very small
aspect of behavior; most control actions are fully automatic, and
the controlled variable can be controlled over its entire range
and for all disturbances within its physical capacity to resist,
with no reorganization at all on either the input or the output
side. It's only control systems that are incomplete that require
reorganization to handle new situations. If every situation is so
new that it invalidates the existing organization, then all
control would have to take place through reorganization -- and it
would be both slow and inefficient.

In the example of gain-reversal, at first this may require
reorganization. But after a while, the reversal takes place
automatically and immediately when the sign-reversal happens. In
the experiments Rick did, there was no learning curve; all
subjects already knew how to do such reversals and did them
immediately. A higher-level control system, we might propose,
simply monitored the relationship between cursor movement and
handle movement, and maintained the required relationship by
reversing the gain in the system below. If any reorganization was
involved, it happened so quickly that it couldn't be detected.
Right, Rick?
Got to go stack wood for the winter, and any nuts and berries I
can find.

Bill P.