VS: Controling observables (was Re:... long live William T. Powers)

[Dag Forssell 2017.06.15.13.00 PDT]
Martin, Rick,
I would be most interested in your evaluation of Rupert’s discussion of
the engineering perspective on control, spelled out in Chapter 3, section
3.4 The Conventional Error, in his Ph.D. dissertation
(2000).

To me, this reads as a very complete, easy to understand explanation of
what has been and still is wrong with the engineering perspective on
control.

Off the top of my head, I can’t think of any other so complete and clear
discussion of this.

Rupert’s dissertation can be downloaded here: 1.8 MB.

[
https://www.researchgate.net/publication/317381281_Visual_Control_in_Natural_and_Artificial_Systems

](https://www.researchgate.net/publication/317381281_Visual_Control_in_Natural_and_Artificial_Systems)I am thinking about posting it at pctresources.com, with a front page
that points to the PCT sections of this dissertation. Rupert said
OK.

I think this dissertation will speak to the engineers I want to address
with my PCT story.

Best, Dag

···

At 11:53 AM 6/15/2017, you wrote:

[Martin Taylor
2017.06.15.14.22]

For once, I am commenting on a message from Rick with which I largely
agree. I like that, when it happens.

[From Rick Marken
(2017.06.15.0945)]

Martin Taylor (2017.06.14.08.52)–
MT: I have had to rethink my earlier “quibble” about the
use of the word “stabilize” [Martin Taylor 2017.05.12.13.14],
because I realize that stabilization is indeed the core of any kind of
control process. The question is the stabilization of what?

MT: Here are the two canonical diagrams, one for engineering control
and one for perceptual control. Functionally, they are the same except
for the omission of an analogue to the PCT “Perceptual
Function” in the engineering diagram.

RM: Also missing from the engineering diagram is a representation of
independent disturbances.

True.

The Load is not a disturbance
because it is in the path of the feedback loop; it’s analogous to the
feedback function in PCT (which is not shown in this diagram).

The load certainly is not a disturbance, but it is where a disturbance
arrow would enter the loop. For example, if the Load is a circular saw
and the reference/input is its rotational velocity, disturbances would
caused by variations in the material being cut. The Load is more
analogous to the CEV than to any other single part of the PCT loop, but I
think it probably should be distributed over the PCT diagram, in some
cases partly joining the “Plant” as part of the output
function, in other cases standing alone.

MT: Even thought they are functionally identical, they are usually
interpreted differently, besides which they confusingly use the same
labels for different parts of the loop.

RM: � Right. The variable that is called the Reference in the PCT diagram
is called “input” in the engineering diagram. And the variable
that is called � “input” in the PCT diagram is called
“Feedback” in the engineering diagram. This confusion is one of
the reasons why engineering psychologists were unable to apply control
theory correctly to the behavior of living systems, the other reason
being that engineering psychologists (like all psychologists) thought
behavior was a process of output generation rather than control.

I suspect this is correct, and can be used when trying to get control
engineers to understand PCT. The use of the same terminology for
different things and different terminology for the same things does make
the transition unnecessarily difficult. But that’s not something that can
be changed, especially because in the two different environments
“input” in particular enters as a clearly different place. The
Engineer does “input” a reference value, whereas the organism
does “input” data through its senses.

[]

�

MT: We don’t need to explore the similarities and differences between
these diagrams. All we need to do is ask just what is stabilized. In the
Engineering diagram it is the difference between the “input”
and the “feedback”, whereas in the PCT diagram it is the
difference between the “reference” and the
“perception”.

RM: Good point. Only the error is kept constant. As you note, the
controlled input will vary to the extent that the specification for this
input varies.�

MT: … it may be extremely difficult or even impossible for an
external observer to determine whether a control system is acting if the
reference value is changing as unpredictably as is the disturbance or
more so (PCT language) and the observer can see only the PCT
“input”.

RM: I don’t believe that that’s true. In my Mind Reading demo, for
example, the computer is able to “read the mind” – that is,
determine the controlled variable – of a person who is varying the
reference for the state of the controlled variable at least as
unpredictably as the disturbances to that variable (which are sine waves
and, thus, quite predictable).

Yes, the computer can do this, but the computer “knows”, and
more importantly (if I remember correctly) uses the exact values of the
disturbances at any moment. An outside observer does not. However, an
external observer might be able to guess the controlled avatar if the
subject uses a reference value that has a statistical distribution
different from that of the disturbances. It could be difficult if the
subject is good at mimicing the computer’s statistics for its disturbance
production.

�
MT: Of course, if the reference value changes only slowly or is
temporarily fixed, an external observer will see the value of the
externally observable “input” variable to be stabilized, which
I think is why there is confusion over the use of the term
“stabilize” in PCT.

RM: I don’t think “stabilize” is a “term” in PCT.
It’s not to be found in the index to B:CP, for example. I think it’s just
a word that can be used to mean what it means: “keep from
changing”.

Exactly. It’s one word that can be imported from everyday language
without having to be modified for use in PCT. That’s what makes any word
more widely understood and less technical jargon. You know what
“stabilize” means without knowing PCT, and when you come to
know PCT your understanding of it hasn’t changed.

So when you want to talk about
keeping some variable from changing you could use the word
“stabilize”. But that could apply to a variable that is
stabilized by opposing force (like the stability of the walls of my
house) or by control (like the stability of the temperature of the water
in my swimming pool).

True. Is that a problem? Lots of words mean different things in different
contexts, and there’s no confusion. If you want to relax in the sun
listening to the rippling water of a Highland brook, you don’t go to the
roof of a financial institution to sit on the “bank”.

Martin

[From Dag Forssell (2017.06.17 11.40 PDT)]

[Rupert Young (2017.06.17
13.00)]

DF: I would be most interested in your evaluation of Rupert’s
discussion of the engineering perspective on control, spelled out in
Chapter 3, section 3.4 The Conventional Error, in his Ph.D. dissertation
(2000).Â
I’ve just been updating versions of these based on Wiener’s
original, with additional PCT labels. Do these look correct in terms of
labels and separation between organism and the world?

Rupert,

I just checked the archive and see that you came to CSGnet in September
1997.

You discussed many items through September 2001, then took a break and
came back in 2012.

Your dissertation was never mentioned on CSGnet until I did just
recently.

Your images were embedded and hard to see in Eudora, so I picked them up
in the Embedded folder and pasted them into InDesign. I’ll attach a pdf
of that.

Seems to me that the first diagram is the valid one. That one shows the
reference, comparator and Error inside the organism.

Was the second one Wiener’s version?

Best, Dag

update diagrams.pdf (164 KB)

[From Dag Forssell (2017.06.17.1420 PDT)]

Fred has a good point here. I missed that end in my pdf reply.

Perhaps if you move the vertical next to Sensory input out, move Sensory
input to the left side of it, and include all of the effector (all
muscle?) the diagram could feature a clean dotted line alignment.

Best, Dag

16f0a861.png

16f0a951.png

···

[Fred Nickols (2017.06.17.1435
ET)]

Well, my test message got through so I’ll try this one again.

I think if it were up to me I’d place “Sensory Inputâ€? in the second
diagram inside the organism and over by the comparator.

Fred Nickols

From: Rupert Young
[
mailto:rupert@perceptualrobots.com]
Sent: Saturday, June 17, 2017 7:51 AM
To: csgnet@lists.illinois.edu
Subject: Re: VS: Controling observables (was Re:… long live
William T. Powers)

[From Rupert Young (2017.06.17 13.00)]
DF: I would be most interested in your evaluation of Rupert’s
discussion of the engineering perspective on control, spelled out in
Chapter 3, section 3.4 The Conventional Error, in his Ph.D.
dissertation (2000).

I’ve just been updating versions of these based on Wiener’s
original, with additional PCT labels. Do these look correct in terms of
labels and separation between organism and the world?

[]

[]

Regards,

Rupert