NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/bit/listserv/csgl/864

Text File | 1992-08-31 | 10.3 KB | 196 lines

Comments: Gated by [e-mail]
Path: sparky!uunet!paladin.american.edu!auvm!BEN.DCIEM.DND.CA!MMT
Message-ID: <[e-mail]>
Newsgroups: bit.listserv.csg-l
Date:         Mon, 31 Aug 1992 14:35:33 EDT
Sender:       "Control Systems Group Network (CSGnet)" <[e-mail]>
From:         [e-mail]
Subject: Re: Manipulation
Lines: 185

[Martin Taylor 920831 1330]
(Bill Powers 920829.2130) and (Greg Williams 920830)

I hope this is short.  Time presses.

Bill.  I think this is another occasion in which the language we use points
in opposite directions while the thoughts are close.  But I'm not sure.

----------------------------------------
>Martin Taylor (920829.1315) --
>
>>One should never say that one "controls" the location of the car
>within >its lane on the road.  One should say that one controls the
>percept of >the location of the car....
>
>The stages of Satori. Before I understood, I said that I controlled
>the car within its lane. While I was understanding, I said that I
>controlled the percept of the car within its lane. After I understood,
>I said that I controlled the car within its lane (for I know only the
>percept in any case).

Well, yes.  That's the normal way of looking at it.  But when we are trying
to be precise, and communicate exactly, then we have to go back to separating
the controlled percept from the CEV, which is what I was trying to do.  We do,
easily but sloppily, use third-stage language most of the time.

>>We need a different word for what happens to the CEV corresponding to
>a >controlled percept.  For lack of a better word, I propose here to
>use >X-control (externalized control). One can X-control passive CEV's
>in >the real outer world.  One cannot control (P-control?) them.
>
>I resist this terminology. If the (hypothetical) external CEV (complex
>environmental variable) does not behave as the percept does, then the
>CEV has been misdefined; it must entail aspects that are not matched
>in the perceptual function of the controlling system.

As you put it here, ALL CEVs are and always will remain misdefined, inasmuch
as there probably exists no construct whatever in the outer world that behaves
precisely as the perceptual input function that defines the CEV does.  All we
can know is those aspects of the outer world that correspond to some perceptual
function, but our actions affect the real world as it is, not as we perceive it.
We perceive the effects our actions have on the real world as filtered through
our perceptual input functions, and that filtering is both noisy (intrinsically)
and subject to error (contextual effects).  The CEV we think we are controlling
corresponds only by fluke with any CEV controlled by anyone else, and even
then we cannot know that this correspondence has occurred.

What we can know is that we can perceive someone else acting on the CEV we
are trying to X-xontrol--i.e. we perceive that our percept corresponding to the
CEV is changing in coordination with some perception of action by someone else.
We might even, applying the Test, perceive that the other is X-controlling a
CEV highly correlated with one we are controlling (or disturbing).  We cannot
perceive or determine by any finite procedure what the other is P-controlling.

>I want to minimize
>reification of that hypothetical reality. To speak of a CEV apart from
>the perceptual system that defines it for a controlling system is to
>give the CEV an independent existence of its own, to objectify it, and
>to imply a third-party observer who can know the true form of the CEV
>which may be partly misrepresented in the controlling system. There
>may be many variables in the external world. But they are not
>organized into CEVs.

It is this paragraph that makes me think we are trying to say the same thing.
It is what I was trying to get across in making the P-control versus X-control
distinction.  X-control is X both for X-ternal and for "hypothesized variable."
X-control is based on the notion that what you perceive has some identity in
the outer world.  Maybe it does, and maybe it doesn't.  You can't know.

>There is only one form of control: control of the percept.

Yes, that's my main point.  And using that as the sense of "control", we CAN
control other people, as you demonstrated in a posting last week.
==========

>A disturbance that has a transient effect only will not materially
>affect the behavior (output) of the control system. If the control
>system could react within the interval we classify as "transient," the
>disturbance would be successfully resisted and there would be no
>transient effect on the perception. Remember that the main means of
>producing predictable outputs from a control system is to apply a
>disturbance THAT IS CANCELED BY THE ENSUING ACTION.

You can't have it both ways.  Either there is an error signal that results
from the disturbance and generates an output signal that resists the
disturbance, or there isn't.  If there is, it is an indication that the
disturbance did affect the percept.  If there isn't, we don't know whether
any percept was affected, but if one was, it remains altered.

>Control systems evolve and reorganize until they can react adequately
>on the time-scale typical of normal disturbances. Don't think of
>disturbances as transient "events." They are simply independent
>variables that influence the controlled variable. They may be brief,
>or like gravity they may be present permanently.

All feedback loops have transport delay.  All linear feedback systems can
be characterized by their bandwidth and phase response.  Non-linear ones
are harder to characterize, but among their characteristics are information
capacity measures.  All of these apply to the dynamics.  I think the word
"adequately" in your first line is key.  It means that the organism will
survive, behaving this well, and that no excess resources will be applied
to ensure better behaviour than that.

>I repeat that an influence (or disturbance) is neither control nor
>determination. It is simply an influence, which may or may not have
>any effect, depending on other influences that may be present at the
>same time (such as the output of a control system).

Well, I refer again to your clarifying posting, which I had better look up...
Yep.  Here it is (920828.0900)

>People are not generally content merely to apply an influence to
>someone else's behavior and accept the result. To do that would be to
>accept the fact that the other's behavior is most likely to proceed as
>before with no effect from the supposed "influence." What happens in
>reality is that if the first attempt at influence fails, as it is most
>likely to do, there will be continued attempts involving varieties of
>influences and increasing force behind the influences. The influencer
>reveals the fact that this is an attempt at control, not merely at
>influence. The object is to have a particular effect on the other's
>behavior that matches the influencer's goal for the effect.
>
>The transition from mild and innocuous attempts at influence to
>concerted attempts to control is inevitable, for the simple reason
>that mere influence has almost no effect.

The same applies when we try to "influence" a ball to go up an incline by
blowing at it.  If we just blow, it may go up and come down again, or it may
go sideways, or something else.  If we really want it to go up, we modify the
direction and strength of blow according to our perception of what the ball
is doing.  No difference when we "control" another person.  We keep trying
different "influences" until the person does what we want or until we give up.

The original (confusing) claim against controlling another person was that
we could not, from outside, reliably set a reference level for any ECS within
the other person.  This view is the one I was seeing as irrelevant, since
we control only our own percepts.  Even internally, no ECS controls another.
An ECS controls only its own percept.  It can not set the reference level for
another ECS, though it can influence that reference level in conjunction with
influences from possibly many other ECSs at its own level.  And even if an
ECS was the only influence on the reference level of some lower-level ECS, it
would still not be controlling that ECS.  If it could perceive directly the
reference level that it was setting, it could (X-)control that, but the lower
ECS might react in any number of ways, depending on the correctness of its
organization, its gain, the environment within which it was trying to control,
and so forth.

No, control of another has nothing whatever to do with setting reference levels
for another.  Control of another may be achieved easily if it so happens that
the other's reference levels come to lead to actions that are perceived as
appropriate to the reference levels in the "controller."  But the other's
reference levels are a red herring in the discussion of whether one person
can control another.

===============
Greg,

>>What is meant is one person controlling another person's actions
>>surreptitiously, leading to that other person doing something one could not
>>get the person to do simply by asking.
>
>I thought we'd already reached an agreement that one person cannot control (in
>the PCT sense) another.

Well, I haven't reached that agreement (see above).

>In sum, I am exploring (with your aid) the PCT foundations of (c). There is
>never any PCT-control "from the outside," is there? Success in manipulation
>depends on proper functioning of the influencee's control system (note that
>con men complain about the occasional mark "too stupid" to understand the
>game); it depends most crucially on the influencer having a reasonably good
>model of the influencee's control structure -- and this is one of the main
>topics for investigation in a GTM based on PCT. More precision in our
>definitions has great value, I think, for minimizing confusions due to this
>or that word meaning one thing to one person and another to someone else, so
>let's keep at it, OK? I'm learning a great deal from this.

I quite agree, and that is why I introduced X-control to contrast with real
(P-)control.  And with P=control, one person most certainly CAN control
another, whether it be by force (i.e. in a conflict situation) or by taking
advantage of presumptions about the other's reference levels to set up
disturbances that lead to the desired actions.  Each method fits exactly
within the normal PCT definition of control.

Martin