What is the problem

[Martin Taylor 940906 18:00]

Rick Marken (940906.0900)

what I was trying to get at was the idea that the
only "problem" faced by a hierarchical control system is getting it's
perceptual signals to match its reference signals. There is no really other
problem - - real or imagined.

Two or three weeks ago I would have agreed with this, but now, after the
discussion with Avery Andrews and Bill P. on ESVs etc, I'm not so sure.
In one sense, what you say is obviously true--the control system has no
access to the world except through its perceptual signals. But in another
sense, the world can "blindside" you; and the problem that presents to the
organism is to survive. That, surely is THE problem, from which the problem
of controlling perceptions derives. One must survive in order to control
perceptions, and one must control perceptions in order to survive.

But attempts to control perceptions are not enough, because you can be killed
by things you don't perceive at all (the thrown knife in the back, intense
gamma radiation, strange bacteria, the sniper bullet when you go shopping
so that you don't starve in Sarajevo). The problem is how to control
perceptions in such a way that you are unlikely to be killed or damaged by
something you don't or can't perceive. That seems to me to be the "only"
problem faced by a hierarchic control system that is a living system.
(Being killed by inadequate control of something you CAN perceive is
part of the generic sub-problem that derives from the "only" problem--
"getting its perceptual signals to match its reference signals.)

Martin

Tom Bourbon [940907.0826]

[Martin Taylor 940906 18:00]

Rick Marken (940906.0900)

Rick:

what I was trying to get at was the idea that the
only "problem" faced by a hierarchical control system is getting it's
perceptual signals to match its reference signals. There is no really other
problem - - real or imagined.

Martin:
. . .

In one sense, what you say is obviously true--the control system has no
access to the world except through its perceptual signals. But in another
sense, the world can "blindside" you; and the problem that presents to the
organism is to survive. That, surely is THE problem, from which the problem
of controlling perceptions derives. One must survive in order to control
perceptions, and one must control perceptions in order to survive.

Yes, but . . .. All an organism, as a perceptual control system, can do is
control its perceptions. Survival in the world-as-such is a side effect --
an important one, mind you, but nonetheless a side effect of perceptual
control. Notice the low likelihood of the events you present below as
examples of "blindsiding."

But attempts to control perceptions are not enough, because you can be killed
by things you don't perceive at all (the thrown knife in the back, intense
gamma radiation, strange bacteria, the sniper bullet when you go shopping
so that you don't starve in Sarajevo).

Everyday occurrences for every person? The probabilities may be non-zero,
but for the species as a whole, they are effectively zero. As for those of
us who end up, "in the wrong place at the wrong time," as the media love to
put it, "them's the breaks."

The problem is how to control
perceptions in such a way that you are unlikely to be killed or damaged by
something you don't or can't perceive. That seems to me to be the "only"
problem faced by a hierarchic control system that is a living system.

How is this a problem for the individual system? I am severely "differently
abled" at comprehending, for example, we might be controlling perceptions so
as to avoid, for example, being squashed by falling objects from space?
Perhaps some further development of this theme by you will help me, Martin,
or I may simply need a few cups of coffee to jog the little grey cells.

Later,

Tom

[Martin Taylor 940907 11:30]

Bill Leach 940907.02:13 EST

Marin write:

... But in another sense, the world can "blindside" you; and the
problem that presents to the organism is to survive. That, surely is
THE problem, from which the problem of controlling perceptions derives.
...

I posit that this is a case of shifting view points which Martin tacitly
admits with:

(What do you mean "tacitly." My intention was to FORCE a shift of
viewpoint!)

In one sense, what you say is obviously true--the control system has no
access to the world except through its perceptual signals.

This I believe is precisely what Rick is intending.

I know, and I quite agree(d) with Rick. The point of my posting was to
argue that when one looks at any individual organism, if can be seen as
solving the problems of which it can be aware--and that's ALL it can do,
making Rick correct. But when one looks from outside the organism, with
different perceptual functions operating on different data, one can see
that the perceptions the organism controls are "its" means of solving a
different problem--survival. The organism cannot be aware of all of the
side effects of its perceptual control, but all now living organisms are
descended from ancestors who did act in ways that allowed them to survive
long enough to propagate their genes. Those ancestors solved, unknowingly,
the survival problem. Lizards don't stay too long in the desert sun, but
move to shade before they dry up and shrivel away. They don't control
perceptions of shrivelling up (at least I would suppose not), but they
probably do control a perception of something related to body temperature.
Not evaporating away is a side effect, but it does solve (partially) an
important problem.

Yes, it is definitely a shift of viewpoint, as we so often must do, between
the organism itself and the observer of the organism and its environment.

Somehow I have trouble considering the death of the control system due to
external causes over which it can not have accurate perception in advance
of mortal danger (if any) to be rightfully considered as a control system
failure.

In some sense, of course, it is, in that the control system can no longer
control after it dies! But I don't think I suggested that failure to
observe the unobservable was a failure of any of the Elementary Control
Systems within the hierarchy of the organism. If anything, it could be
construed as a failure of evolution to provide the organism with the
appropriate perceptual possibilities. (Tom Bourbon said [940902.1451]
in response to Gary Cziko:

It's kind of neat to consider that our brains evolved via blind variation
and selection, they function using blind variation and selection (as in
control-system reoganization), and we are now using computers and test
tubes [TB: and E. coli] to harness the power of blind variation and
selection to evolve
useful programs, organisms, and products. It looks like blind variation
and selection "all the way down" (and up) to me.

And that looks like control, from top to bottom and back again.

And it is a failure of THAT control system, if one goes along with Tom
(which I don't, there being no reference signal or perceptual signal
in evolution).

Tom Bourbon [940907.0826]

But attempts to control perceptions are not enough, because you can be killed
by things you don't perceive at all (the thrown knife in the back, intense
gamma radiation, strange bacteria, the sniper bullet when you go shopping
so that you don't starve in Sarajevo).

Everyday occurrences for every person? The probabilities may be non-zero,
but for the species as a whole, they are effectively zero. As for those of
us who end up, "in the wrong place at the wrong time," as the media love to
put it, "them's the breaks."

Exactly. That's how evolution works, to make those probabilities low, at
least in the period of life before gene propagation. Some species solve it
by producing huge numbers of offspring in an environment within which those
"breaks" occur to almost all members of the species, but a few survive. We
don't. We produce few offspring, and act so that many of them do survive
to reproduce. But we don't control perceptions of "survival." Survival
is a side effect of what we do control--a problem we solve without knowing
that we do so.

The problem is how to control
perceptions in such a way that you are unlikely to be killed or damaged by
something you don't or can't perceive. That seems to me to be the "only"
problem faced by a hierarchic control system that is a living system.

How is this a problem for the individual system?

Perhaps some further development of this theme by you will help me, Martin,

Perhaps it might help if I reword your question by changing "for" into
"solved by." Looked at from within the individual system, it is not
a problem. Looked at from without, the individual system solves it, but
only probabilistically--it cannot be otherwise, since it does not involve
perception OR control.

I trust the shift of viewpoint now seems more coherent?

Martin