Taylor's diagram

[Dag_Forssell]

[From Dag Forssell (920703)]

Martin Taylor 920629 1515

It just occurs to me that we take Martin's 23 level chart and fold it on
the mirror line, then interconnect the control systems across so we
control all the perceptions up and down. We are back to the diagram as
we know it, but with an expanded understanding of it.

This is true, but I'd rather not do that. One of the points of making the
diagram is to show a relation between a "Boss Reality" and a controlled
percept. Now if that Boss Reality exists, it is accessible to another
control system (e.g. an experimenter). The other control system can focus
on (perceive) the same complex environmental variable (CEV) and perhaps
attempt to control its perception of the CEV, disturbing the first control
system's perception of it. We can diagram the interaction his way, taking
the top half of this diagram as representing the whole of my earlier one,
and the bottom half its mirror image:

               ECS
Person 1 / \
             /| |\
        ====^=^===V=V====
             \| |/
              \ /
  World CEV
              / \
             /| |\
       =====V=V===^=^======
             \| |/
Person 2 \ /
               ECS

If you fold the original diagram about the mirror line where "the rubber
meets to road", you lose this view. Much more crucial, you lose the view
of somewhat off-focus disturbance, which was the core of the discussion
about VOT.

Bill Powers (920629.1600)

The 11 levels below the line through level 0 are, of course, in the observer.

Bruce Nevin (Wed 920601 08:59:24)

The mirroring around level 0 reminds us to what extent apparent
structure in the environment is a reflection of structure in the control
hierarchy, projected there by the observer. We assume vice versa, but
that can only be an assumption. (Right, Wayne?)

So it is worse than perhaps Martin has said. Not only may the observer
(the investigator) identify the wrong environmental variable V as that
controlled by the observed control system, the two parties may also have
differently structured control hierarchies, and so may parse the
environment differently into environmental variables.

Martin, I was gone two days, or I would have commented on your post sooner.
I think your presentation is both exciting and dangerous.

Exciting because of the relationships you want to portray. Dangerous
because it will invite misinterpretation.

                                      One of the points of making the
diagram is to show a relation between a "Boss Reality" and a controlled
percept. Now if that Boss Reality exists, it is accessible to another
control system (e.g. an experimenter).

Your diagram creates the illusion that the "Boss Reality" is structured in
a hierarchy outside the control system, and is accessible as the same "Boss
Reality" to another observer. This is patently false. That is the dangerous
part. Your diagram may be very useful as an illustration or teaching tool,
however. That is the exciting part.

Your description fails to make the distinction forcefully, if at all. (If
you do, it escapes me). Bill makes the distinction - and Bruce (I think).

About a year ago, I portrayed in a diagram to Bill (as a teaching tool) the
process of perception up all 11 levels. Separately, I tried to portray the
process of control down all 11 levels. In a third step, I meant to marry
the two. Bill did not let me get away with it. (I have not given up on the
teaching tool.)

It took me time, but I now think I understand the "behavior of perception"
reasonably well. I think it is very hard to grasp the close integration
that HPCT suggests of both perception and control up and down the hierarchy
and sideways in between each level. I believe it is much easier to say: I
understand; I agree, than to explain and demonstrate understanding.

In one of my posts over a year ago, I related the perception of room
temperature as the difference between the rrrrrrrrr song of the sensing
neuron and the rrrrrrrr song of related memory reference. You may remember
the song part. As perceptions are passed on up the hierarchy, imagination
enters into the process as well, to flesh out the picture as needed.

While I agree that there is a Boss Reality phenomenon we call temperature
(and model as molecular motion - I have never seen a molecule or its
movement, but have adopted the systems concept), it seems clear that it is
"accessible" to two different observers ONLY by comparison with the
subjective reference in each observer. Therefore the same Boss Reality can
never be accessible in the same way to two different observers.

The uninitiated reader of your diagram may be led to believe that you
advocate something that is true of the "Boss Reality" and that two
observers can agree on with certainty. A careful note will have to
accompany the diagram each time it is used, to convey what it portrays and
what it does not portray. As I understand you, the diagram may be intended
only as a simplistic teaching tool for those who have not yet understood
the concept of behavior of perception.

Dag

[_Martin_Taylor]

[Martin Taylor 920703 15:15]
(Dag Forssell 920703)

                                      One of the points of making the
diagram is to show a relation between a "Boss Reality" and a controlled
percept. Now if that Boss Reality exists, it is accessible to another
control system (e.g. an experimenter).

Your diagram creates the illusion that the "Boss Reality" is structured in
a hierarchy outside the control system, and is accessible as the same "Boss
Reality" to another observer. This is patently false. That is the dangerous
part. Your diagram may be very useful as an illustration or teaching tool,
however. That is the exciting part.

I don't think it is patently false. What IS patently false is that we can
ever know whether it is true. All we have access to is our own perceptions.
That we can control them suggests the existence and character of the Boss
Reality (if there were nothing out there that we could perceive as a
Complex Environmental Variable, we couldn't control the perception of it).
I think all this goes without saying. It applies to all attempts to employ
The Test. The experimenter always has to assume that there is a Boss Reality,
and that the CEV being disturbed as part of The Test is one that has a
corresponding percept in the subject (and that the subject exists, as part
of the Boss Reality).

I don't see anything new, difficult, or dangerous in any of this. Either there
is some reality out there or there isn't. If there isn't, we are (sorry, I am)
just having fun in our own head(s?). If there is, we manipulate it and see
others manipulating it. The only structure of the "out there" that we can deal
with is the structure we give it through our perceptual/cognitive machinery.
And there is neither more nor less reason to deal with the "truth" of the world
as structured than to deal with it as a set of mathematical equations describing
quantum chromodynamic systems, or whatever today's favoured physical substrate
might be.

Any instance of The Test depends for its success on the subject controlling the
perception of the same CEV that the experimenter perceives. The Test is there
to see whether this identity exists. If it does, reality has, in part, the
same structure to the subject as to the experimenter. A lot of my "statistics"
argument with Bill and Rick hinges on the unlikelihood of the experimenter
hitting on the same CEV as the subject, or even having the same structure
of reality as the subject. It becomes even harder when there is a cultural
or species gap between the experimenter and the subject.

Of course it's all in the head. That doesn't mean it's any the less in the
world. And that is not dangerous. What is dangerous is to assume that your
way of seeing the world is the only way, and must be the right way, and that
anyone who sees it differently is lying, bad, dangerous, and to be suppressed.
That's what's dangerous.

···

---------------

About a year ago, I portrayed in a diagram to Bill (as a teaching tool) the
process of perception up all 11 levels. Separately, I tried to portray the
process of control down all 11 levels. In a third step, I meant to marry
the two. Bill did not let me get away with it. (I have not given up on the
teaching tool.)

What did Bill not allow you to get away with? If outputs derived from error
signals go as references to the same lower level ECSs that the perceptual
inputs come from, the upward flow of perception is intimately married with
the downward flow of action. Indeed, it is this presupposition that we here
are banking on, in our belief that we can train a control net much faster and
more precisely than we can train an ordinary S-R neural net such as a
pattern recognizer. I thought that the reciprocal connection was an aspect
of HPCT that was pretty much taken for granted. Is it not?

Martin