specting the ignorance of others?

CSGnet Archive CSG1998
1

[From Oded Maler (980714)]

Rick Marken (980713.1630):

Bruce Gregory (980713.1015 EDT) --

> I infer the "giving respect to the wishes of others" involves
> control at the level of principles. This control is exercised
> by adopting Plans (at the level of programs)that involve such
> actions as asking others what they would like to do. Spreadsheets
> do not appear to be well-suited to modeling programs.

It's not modeling programs that's the problem (for spreadsheets
or any other programming tool); it's developing a perceptual
system that can _perceive_ that a particular program is happening.
A program _perceiving_ system would have to act like a function
whose input is the time sequence of states of what might be a
certain type of program and whose output is a signal indicating
whether the input is a certain program type or not. I don't know
how to build such a program perceiving function offhand;

                                                  ^^^^^^^^
And "onhand" you do?

but I know
it's possible because it's been done -- by the human brain. Once
you have such program perceiving functions, you can have control
systems that have a reference for perceiving particular types of
programs (that is, references for particular outputs of the program
perceiving functions). These control systems would vary their
outputs (logic, loops, etc) as necessary to make the intended
programs keep happening.

Best

Rick

It's really amazing. You have no clue at all about perceptual
functions at all the zillion levels between the sensations and the
highest levels. Yet you allow yourself to extrapolate authoritatively
from your tinkering with few scalar variable models into questions of
programs, principles, world views and what not. As long as you don't
know how to build any of these functions (not to mention a function of
a "simple" object like a chair), all your high level contemplations
are as valid as any other cognitive psychology model or
hypothetical architecture.

2

[From Bruce Gregory (980714.1300 EDT)]

Rick Marken (980714.0920)

Well, maybe not "authoritatively". I have done experiments to
show that people can control programs; to control a program
they have to be able to control a perception of the program.
Any model of control of programs (or principles or world views,
if they can be shown to be controlled) will have to have some
means of producing perceptions of input variables that can be
compared to references.

A very simple form of program involves a sequence of perceptions. I gave an
example of a proposing program today. It seems that the perception being
controlled is the perception that you are experiencing the perceptions in
the sequence. In the sequence I described the perceptions seem relatively
easy to produce and to model. How would I go about implementing this on a
spreadsheet? (This may be too complex to answer in any detail, but I'd
appreciate some suggestions.)

Bruce Gregory

3

[From Rick Marken (980714.0920)]

Oded Maler (980714)]

It's really amazing. You have no clue at all about perceptual
functions at all the zillion levels between the sensations and
the highest levels.

Actually, we do have some clues. Bill describes them in
Chs 8 - 13 of B:CP. I've made my own modest contributions
in my "Hierarchical behavior of perception" paper at

http://home.earthlink.net/~rmarken/papers.html

There is also an experiment (that actually works on some
computers;-)) that can be used to study some higher level
perceptual functions:

http://home.earthlink.net/~rmarken/ControlDemo/HP.html

Yet you allow yourself to extrapolate authoritatively from
your tinkering with few scalar variable models into questions
of programs, principles, world views and what not.

Well, maybe not "authoritatively". I have done experiments to
show that people can control programs; to control a program
they have to be able to control a perception of the program.
Any model of control of programs (or principles or world views,
if they can be shown to be controlled) will have to have some
means of producing perceptions of input variables that can be
compared to references.

If you can think of some way of producing controllable perceptions
other than by computing a function of inputs and producing a
controllable output then that's fine with me; this is just a
natural way to think of it for me. And frankly, I can't think
of any model of perception that doesn't basically work as
a mathematical function; whether the output of the function
is scaler or not doesn't matter to me, as long as it can
be controlled.

As long as you don't know how to build any of these functions
(not to mention a function of a "simple" object like a chair),
all your high level contemplations are as valid as any other
cognitive psychology model or hypothetical architecture.

How the functions are built is irrelevant to my questions
about the validity of S-R and cognitive models. Now wipe the
froth from your mouth, enjoy France's World Cup victory and
re-read B:CP. You'll feel much better in the morning;-)

Best

Rick

···

---

Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/

4

[From Rick Marken (980714.1620)]

Bruce Gregory (980714.1300 EDT) --

A very simple form of program involves a sequence of
perceptions.

Actually, a _contingent_ sequence. Many different sequences
can be generated by the same program. For example, consider
the program

STATE = STARTSTATE
FOR i = 1 to 8
PRINT STATE
if STATE = A then STATE = B else STATE = C
if STATE = C then STATE = B
if STATE = B then STATE = RANDOM(A,B,C)
NEXT i

The following sequences can be generated by this program
(I think; I did this hastily).

ABCBBABA
BCBBCBAB
CBCBCBBA

A progam perceiving function would have to produce the
same output (perception) for each of these sequences --
indicating that they were generated by the same program --
and a different output for similar sequences (like ACBAABAC)
that are generated by a different program.

In the sequence I described the perceptions seem relatively
easy to produce and to model. How would I go about implementing
this on a spreadsheet?

It's hard to model things that occur in temporal sequence in
a spreadsheet -- but it could be done. Let's say the sequence
is composed of four perceptual elements (1,2, 3, and 4). I would
represent the the temporal elements of the sequence as a sequence
of cells. These cells represent the lower level perceptual inputs
to a sequence perceiving function. This sequence perceiving
function would be some kind of nifty logical function that is
"true" only as long as the temporal elements of the sequence
are being filled with the appropriate input values _in a particular
sequence_. Let's say the function is designed to detect the
sequence 1,3,2,4 (and assume 0 means "no input yet"). Then the
function would have to be true for the following temporal
sequences (a maximum of 8 cells for teh temporal rangle):

1,1,0,3,0,2,4,0
0,1,3,3,0,0,2,4
1,3,0,2,0,0,2,4
1,3,2,4,0,0,0,0

and false for sequences like

1,1,0,3,0,0,4,2
0,1,3,3,0,4,2,4
1,3,0,2,0,0,1,4
1,3,2,3,4,0,0,0

Me:

A program _perceiving_ system would have to act like a function
whose input is the time sequence of states of what might be a
certain type of program and whose output is a signal indicating
whether the input is a certain program type or not. I don't know
how to build such a program perceiving function offhand; but I know
it's possible because it's been done -- by the human brain. Once
you have such program perceiving functions, you can have control
systems that have a reference for perceiving particular types of
programs (that is, references for particular outputs of the program
perceiving functions)...

Bruce Abbott (980714.1255 EST)

It would seem from this description that one would have to have
a complete specification of a program (plan) in memory prior to
the attempted execution of the program

Then you didn't understand my description. There is no need for
a program to be stored in memory; the program perception (the
output of the program perceiving function) is a perceptual signal
that remains at a particular value as long as the inputs to the
function (such as the A,B, C sequences above) are the outputs
of a particular program. If the outputs of the behaving system
are actually producing the program, then the program is presumably
being perceived because the output of the program control system is
specifying perceptions to be produced by lower level control
systems (logic, relationship, configuration). The system I imagine
for controlling program perceptions is analogous to the level 3
logical relationship perception control systems in my spreadsheet
hierarchy. In that case we have systems controlling the "truth
value" of some logical relationship perceptions (like "A>(B-C))
by varying the references for lower level perceptions (of A, B and
C) which are themselves controlled by varying the still lower
level perceptions of which they are composed.

Best

Rick

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken/