Closed Loop -- Models and Their Worlds

CSGnet Archive CSG2001
21

[From Bill Curry (2001.03.26.2300 EST)]

Rick Marken (01.03.26.1700)
>Bruce Nevin (2001.03.26 18:17 EST)--

> When you say "persistent disturber" (or when you previously
> said "the disturber is not inclined to find a different set of
> actions to control his perceptions") I think you mean that the
> disturber is controlling a perception of disturbing the other's
> control.

Not at all. There are many reasons why the kid might persist in
being a disturbance. Maybe he just wants to keep hitting on Mary
Lou in the first row. Or maybe she wants to practice her role in
_Grease_. The disturber might not even care that he or she is being
a disturbance to the teacher.

This interesting thread at times (like many others) seems like a game of
"follow the bouncing CV". Is a disturbance evidence of a controlled
reference to disturb, a side effect variable, or both? Since each case
is possible it leads me to question whether the TEST is capable of
reliably teasing out THE variable being controlled by a real person, or
is it better positioned as a useful tool for thought experiments and
suitably constrained modeling experiments?

In a hierarchical living control system, a high level perceptual variable
(like "I am a bad-assed dude") purportedly can set a host of lower level
references that similarly control a multitude of other variables. So it
seems we are never just controlling a single variable at any point in
time. Any given behavior can be expressing the organism's efforts to
control a complex melange of principles, programs, sequences, etc.
Controlling for catching a fly ball can be affected by the sun angle, the
fielder's fight with his girl friend the previous, night, a stretched
achilles tendon, and a fan mooning him from the bleachers. Moreover, in
a dynamic system, control is shifting from moment to moment based on the
changing perceptual stream and the shifting gains occurring in the
engaged systems.

Is it appropriate to consider CVs as isolated entities?

Best regards,

Bill C.

···

--
William J. Curry, III 941-395.0088
Capticom, Inc. capticom@olsusa.com

22

[From Rick Marken (01.03.28.1920)]

Bill Curry (2001.03.26.2300 EST)--

This interesting thread at times (like many others) seems like a
game of "follow the bouncing CV".

Then you must not be following it;-)

Is a disturbance evidence of a controlled reference to disturb,
a side effect variable, or both?

In PCT, a disturbance is not evidence of anything. A disturbance
is the name of the effect of an environmental variable on a
controlled variable.

Since each case is possible it leads me to question whether the
TEST is capable of reliably teasing out THE variable being
controlled by a real person, or is it better positioned as a
useful tool for thought experiments and suitably constrained
modeling experiments?

Since people control many variables simultaneously it makes no
sense to talk about "THE variable being controlled by a real person".
The test is used to determine whether a real (or simulated) person
(or other living system) is controlling a particular variable at a
particular time. It can do this accurately and reliably no matter
how many other variables the system may ne controllng at the same
time.

In a hierarchical living control system, a high level perceptual
variable (like "I am a bad-assed dude") purportedly can set a host
of lower level references that similarly control a multitude of
other variables. So it seems we are never just controlling a single
variable at any point in time.

Correct. To see how it works, try running my spreadsheet hierarchy
model at http://home.earthlink.net/~rmarken/demos.html. This
model controls a total of 18 different perceptions; 6 different
perceptions of three different types (levels). You can use this
spreadsheet to prove to yourself that you can test for control
of any one of the eighteen perceptions in the hierarchy. The
perceptions are defined by the equations in the perceptual cells.
For example, one perception controlled (at level 2 in the hierarchy)
might be p2.3 = 1*i.1+1*i.2-1*i.3-1*i.4*1*i.5-1*i.6. By applying
disturbances to this variable (disturbances to each of the inputs,
i.n) you can see that this variable varies far less than expected.

If you actually compute a stability ratio -- expected over observed
variance -- for this variable you will see that the stability value
will be quite high. If you did the same test for a variable that
is not controlled -- say p2.x = -1*i.1-1*i.2-1*i.3-1*i.4*1*i.5-1*i.6
-- you will find that the stability ratio for this perception is quite
low; nearly 1.0. The test, then, reveals that this variable is
under control. The test (in this case using the stability ratio)
reveals a controlled variable despite the fact that seventeen other
variables are being controlled at the same time, many of them in
order to control p2.3 and others using p2.3 as the means of control.

Is it appropriate to consider CVs as isolated entities?

No. And that's why they are not treated as such in PCT.

Best regards

Rick

···

--

Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: marken@mindreadings.com
mindreadings.com

23

[From Bill Curry (2001.03.30.1500 EST)]

Rick Marken (01.03.28.1920)

Bill Curry (2001.03.26.2300 EST)--

> This interesting thread at times (like many others) seems like a
> game of "follow the bouncing CV".

Then you must not be following it;-)

Thanks for the clarification. I will follow the bouncing ball through
your spreadsheet!

Best regards,

Bill C.

···

--
William J. Curry, III 941-395.0088
Capticom, Inc. capticom@olsusa.com