# Control BY perception?

[From Rick Marken (930824.2130)]

I just couldn't resist commenting on Martin Taylor's (930824 15:00)
comment on Tom Bourbon's (930824.1028) comment on Michael Fehling's
(930823 18:40:21) comment.

Fehling said:

Similarly to PCT (I think), our model requires that
the organization and its decision making units accomplishes this control in
terms of its interpretations of environmental conditions.

And Tom said:

... an organism
accomplishes control *of* " its interpretations (TB: perceptions) of
environmental conditions," not "in terms of" that interpretation. This is
another significant topic for us to explore.

And Martin says:

I would have read "in terms of its interpretation" as translatable to
"by means of the perceptual input function" rather than "the current
value of the perceptual variable" as you did. If you are right, the
rest of your argument follows, but if I am, he is much closer to PCT
than you allow.

So you are saying that an acceptable way to describe the operation of
a control system is to say that "it is organized to accomplish control
BY MEANS OF the perceptual input function". This sounds completely wrong
to me. Control means maintaining a variable in a fixed or variable
reference state, protected from the effects of disturbance. We know
that the variable controlled by a control system is the perceptual
variable, p, so control means that p = r; the perceptual variable is
kept matching a variable, r, that is a specification of the reference
state of the perceptual variable. The perceptual function, f(), transforms
an environmental variable (or variables), i, into the perceptual variable,
so p = f(i). The statement above implies that f() is the means by which
control is accomplished. But what could f() -- the perceptual function --
have to do with the fact that p=r? The only sense I can make of this
is to assume that you were thinking that the output variable, o, which
is used by the system to keep p matching a potentially varying r and
to resist the effects of disturbances to p, is guided in some way by
the output of the perceptual function -- which is p. So you are saying,
I think, that o = f(i) -- control actions are a function of perception.
This is the way most people (including MOST conventional "manual control"
theorists -- excluding the two that Greg Williams found out about) think
of control -- as an input-output process. This makes sense in terms of
conventional psychological thinking -- which sees perception as the start
of a cause-effect chain that ends with "actions". But PCT shows (both
mathematically AND experimentally) that this model of control is precisely
(and surprisingly) wrong. It is NOT TRUE that o = f(i). In fact, outputs
do not depend on inputs AT ALL; they depend on r and the typically
undetectable disturbances,d, to the controlled perceptual variable:

(1) o = r - 1/g(d)

where g() is the feedback function relating system outputs to inputs.
So the outputs that keep p under control do not depend AT ALL on i or
on a function of i (note that neither i nor f() show up in equation 1).
So it is completely -- exactly -- incorrect to say that control is
accomplished "by means of the perceptual input function". Tom's
statement above was precisely correct -- control systems control an
"interpretation" (the output of f(i), which is p) OF environnmental
events. Control is the control OF perception; there is no control
BY perception. That's what make PCT such a pain for conventional
psychology, which is based on the idea of control BY perception
(even if this control is mediated by complex intervening events).
Unfortunately, control BY perception doesn't happen in control
systems -- not even a little, eensy, weensy bit.

Best

Rick

[Martin Taylor 930825 10:45]
(Rick Marken 930824.2130)

Sorry, Rick. I'm not going to resist your disturbance. You are quite
right, of course, in what you say. You just misinterpreted what I had
intended. Too much shorthand.

The perceptual signal is what is controlled. No question. But the perceptual
signal is determined by the perceptual input function acting on its
sensory inputs. Therefore, control is "by means of the perceptual input
function." It is, of course, equally by means of the comparator, by
means of the output function, by means of the fact that the output affects
the sensory input. I just thought that Tom unnecessarily, but perhaps
correctly, read Fehling as being unaware that the controlled variable
was the perceptual signal.

In this kind of communication it's probably a good idea to be ungenerous.
That way, you can flush out possible misunderstandings rather than let them
rest uncorrected. I just have a tendency sometimes to think that if something
can be read as agreeing with me, then the author does agree with me.
In everyday communication, that's good, because it reduces annoyance.
In developing a technical construction, it's not good, because further
development depends on the agreement being real.

Anyway, sorry not to fight. Maybe next time.

Martin

From Tom Bourbon [930825.0945]

Signing off of the net two or three hours before Rick Marken puts me at a
replied to a post in which Martin Taylor commented on my reply to Michael
Fehling.

[From Rick Marken (930824.2130)]

I just couldn't resist commenting on Martin Taylor's (930824 15:00)
comment on Tom Bourbon's (930824.1028) comment on Michael Fehling's
(930823 18:40:21) comment.

Fehling said:

Similarly to PCT (I think), our model requires that
the organization and its decision making units accomplishes this control in
terms of its interpretations of environmental conditions.

And Tom said:

... an organism
accomplishes control *of* " its interpretations (TB: perceptions) of
environmental conditions," not "in terms of" that interpretation. This is
another significant topic for us to explore.

And Martin says:

I would have read "in terms of its interpretation" as translatable to
"by means of the perceptual input function" rather than "the current
value of the perceptual variable" as you did. If you are right, the
rest of your argument follows, but if I am, he is much closer to PCT
than you allow.

Then Rick said:

So you are saying that an acceptable way to describe the operation of
a control system is to say that "it is organized to accomplish control
BY MEANS OF the perceptual input function". This sounds completely wrong
to me. Control means maintaining a variable in a fixed or variable
reference state, protected from the effects of disturbance.

Rick went on to give a detailed account of why he (and I) thought it was
important to draw attention to the *possibility* that Michael's phrasing
implied a different interpretation of that which a control system controls
and of how a control system achieves control. Control *of* perception, not
control *by way of* perception, is the distinctive, and the most
bedeviling, feature of PCT. I concur with the math and the logic of the
man from the City of Angels and I will not repeat all of his reply.

though I agree with much of it, there are a couple of points in which I
would be more generous to him than you were."

I hope Michael did not think I was less than generous in my coments to him.
As I tried to indicate in my post, I was trying to identify topics in his
posts on which there *might* be differences in our interpretations of
control and of control systems. I did not presuppose that there *were* such
differences and I tried to refrain from imputing to Michael any particular
understanding, or lack of understanding, of PCT. I was, and I remain,
interested in learning more about what he thinks concerning the topics I
identified.

Tom to Michael:

One apparent difference is in our use of ideas about information flow and
transfer. To date, none of the published material on PCT modeling,
including modeling of social phenomena, has included measures of
information. They have not been needed. Martin Taylor and his associates
are at work on projects in which that circumstance might change.

Martin to Tom:

Again, there is a difference of reading. I interpreted Fehling's
"information" to be in the sense of "meaning," which is to say "the change
of state of some high level perceptual variable." What I am aiming at,
and what I see no signs of in Fehling's notes, is the quantitative
analysis of how much information (reduction in uncertainty) occurs
at various places in the hierarchy as seen from various viewpoints.
I think that his usage would have to be taken into account in any PCT-based
approach to social interaction, whereas mine will (I hope) provide a
deeper understanding of the processes that occur in any control system,
hierarchic or not, social or not.

Again, Martin, rather than impose my interpretations and readings on
Michael's remarks about "information" and its role in control by
organizations, I tried to identify a possible difference in interpretations,
and to suggest that this is a topic for Michael and PCTers to explore. I
remain interested in learning his further thoughts on that topic.

Martin to Tom:

The question of organizational purposes is, as you say, likely to lead
to some debate. The purposes are clearly in the individuals, but the
interactions among the individuals, with the consequent reorganizations,
are likely to align them sufficiently that it is a useful shorthand
to say "the purpose of the organization." Of course, if "the organization"
is identified with "the dictator" or "the CEO" the shorthand becomes
longhand.

Agreed.

Until later,

Tom