Hierarchy

[Richard_Marken2]

[From Rick Marken (2003.11.29.1100)]

Marc Abrams (2003.11.29.1305) --

Rick, my questions _never_ concerned your spreadsheet model.

Oh really? Here's what you said:

Besides being incomplete, to say that
Rick's model is a working _HPCT_ model I believe is intellectually
dishonest. Sure it's a 'working' model. The question becomes, of what?
Which
levels in the hierarchy does Rick's model purport to represent?

These sure looked to me like questions concerned with my spreadsheet
model.

My questions
_always_ centered around the current hierarchy. I only introduced your
model because it was used as a claim that the current hierarchy _was_
in
fact modeled.

Well, I certainly never used it as such a claim and I don't know of any
other competent PCT researcher who has done this either.

I said I thought that was intellectually dishonest.

I suppose it would be somewhat dishonest if anyone did it. Who made
this claim? Please show me the quote.

Both you and Bill perceived things differently. You both thought I was
saying that your spreadsheet model was invalid.

I actually had no idea what you were claiming. All I know is that you
asked what my model was a model of and what levels in the hierarchy my
model purported to represent (see above). I answered that question:
it's a tutorial model of a general three level hierarchy controlling
intensity, sensation and relationship type perceptions.

Actually,
I was very impressed by your model. So impressed that I initially
thought
the current hierarchy was in fact true.

That's another reason I answered your question: so that others would
not make this mistake. The spreadsheet model was built to show that
hierarchies of control systems work and how they work. The "truth" of
the hierarchical model has to be determined by experimental research.

I have come to think otherwise.

That's good. Now I would suggest that you start studying my spreadsheet
model to see how working models of control systems work.

You really should look at Martin Taylor's web page. I am not
suggesting he
has the answers.

I certainly will.

Best

Rick

···

---
Richard S. Marken
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

[Marc_Abrams1]

from [Marc Abrams (2003.11.29.1407)

[From Rick Marken (2003.11.29.1100)]

Oh really? Here's what you said:

> Besides being incomplete, to say that
> Rick's model is a working _HPCT_ model I believe is intellectually
> dishonest. Sure it's a 'working' model. The question becomes, of what?
> Which
> levels in the hierarchy does Rick's model purport to represent?

These sure looked to me like questions concerned with my spreadsheet
model.

Really? I read it differently. See my last sentence and two questions? Am I
questioning the validity of your model or the hierarchy? Think about it. I
acknowledged that it was model (of a hierarchal control system) I questioned
what the model represented (i.e. _what_ hierarchical control system)

Well, I certainly never used it as such a claim and I don't know of any
other competent PCT researcher who has done this either.

Rick, I'm not going to waste my time going through the archives to make my
point but I disagree with your statement here.

> Actually,
> I was very impressed by your model. So impressed that I initially
> thought
> the current hierarchy was in fact true.

That's another reason I answered your question: so that others would
not make this mistake. The spreadsheet model was built to show that
hierarchies of control systems work and how they work. The "truth" of
the hierarchical model has to be determined by experimental research.

So why do you think I made this mistake? Because when I asked to see a
working model of the current hierarchy I was told that your spreasdsheet
model was it. I _WAS NOT_ given the caveat you just gave, but I'm glad your
mentioning it now. I would like to see a similar statement at your site with
the model.

> I have come to think otherwise.

That's good. Now I would suggest that you start studying my spreadsheet
model to see how working models of control systems work.

Yes, understanding the mathematics behind your model is a goal of mine. I wo
uld suggest you study Martin Taylor's page and look at an alternative to
your working control systems. You might find it illuminating.

I certainly will.

Excellent. I'm interested in hearing your impressions.

Marc

[Richard_Marken2]

[From Rick Marken (2003.11.29.1530)]

Marc Abrams (2003.11.29.1407)

Rick Marken (2003.11.29.1100)]

These sure looked to me like questions concerned with my spreadsheet
model.

Really? I read it differently... I questioned
what the model represented (i.e. _what_ hierarchical control system)

Right. This question is concerned with my spreadsheet and I answered it.

Best

Rick

Richard S. Marken
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

[Bill_Powers]

[From Bill Powers (2003.11.27.1543 MST)]
For general interest:.
The 11 levels I have proposed have the following properties:

1. A perception of one level is composed of perceptions of the next lower
   level (and sometimes from still lower levels when there are
   any).
   I think I got onto this from reading the old Gestalt psychologists. When
   you examine an object, you can perceptually analyze it into smaller
   objects, in the way that a chair is seen to be made of arms, legs, rungs,
   decorations, and whatever other parts there are. It can also be seen as
   part of a display of furniture in a store, and the store can be seen as
   part of the shape of a city, and so forth. Going to the smallest possible
   parts, like atomic particles, or the largest, like galaxies, requires
   imagination. This is a way of analysis that retains the main theme,
   objects, and considers different size scales. This is a hierarchy of
   size, such as J.G. Miller proposed in his Living Systems. However, a chair can be analyzed in a different way. When I look at a
   chair in the corner of this room, I see dark and light brown shadings,
   greyness in shadows on the floor, I see lines and curves. None of these
   things is an object. They are the sorts of things you see in any object,
   in different combinations. My generic name for them is sensations.
   Question: is it possible to perceive an object when no sensations are
   being perceived? I invite my listmates to try this out. Try to find an
   object which would still be there if none of the sensations you see in it
   were there. And you can try this in different sensory modalities. A sound
   “object” might be a chord, with individual.pitches being the
   sensations that comprise it. Can you hear a chord when no pitches are
   being perceived?
   I call the “object” level “configurations,” to apply
   to all the senses. I think that in every example, we find that the
   elements of a configuration which are not just other configurations are
   in fact sensations, and that there can be no configurations unless some
   sensations are present. Sensations can be present without objects, but
   not vice versa.
   That is an interesting fact about perception, if true. Even the old
   Gestalt psychologists never put it that way – in fact, they seemed to
   think that you perceive the higher level before the lower, which my
   experiences contradict. You may attend to the higher before the lower,
   but I claim that the higher can’t exist without the lower. Try to imagine
   a sensation made of zero intensities.
   This suggests that one perception can be a function of a set of other
   perceptions. We can show this in a model by showing a perceptual signal
   coming out of a box, and a set of other perceptual signals going into the
   box.
   Another property:
2. In order to control the state of a configuration, such as by
   distorting a circle, it is necessary to alter the states of the
   sensations that make it up.
   
   Clearly, there has to be more than one sensation to create either shape
   above. In the Paint program, I can left-click on a white box in the
   palette, and draw the circle or ellipse white on white. Of course I see
   nothing on the screen, even though the program is dutifully turning all
   the pixels within a circular or elliptical region white (same for all
   black or all any color). The background is exactly the same
   color-sensation, so there is no configuration.
   In this picture, what makes the difference between a circle and an
   ellipse is that some white pixels have turned to black. That is a change
   in a set of sensations. In the Paint program I used to make these
   objects, I could stretch the circle into various degrees of ellipticity
   by dragging the mouse. This motor action changes the color of pixels from
   white to black on the screen in an organized way, turning one
   configuration into the other. So it seems that in order to control
   the state of a configuration, I must vary the states of
   sensations.
   These two properties of perceptions one level apart can be seen, I hope,
   in any example of perception from lowest to highest in my list. A
   perception of any given level, if analyzed into elements that are not
   simply more examples of the same perception, comes apart into perceptions
   of the next lower level (property 1). And to control a perception of any
   given level, it is necessary to vary perceptions of the next lower
   level (property 2). These properties were always being kept in mind as I
   tried to identify more levels.
   At some point while building up the whole hierarchy of 11 levels, a
   process that has taken about 45 years, I realized that a perception of a
   given level can, in some cases, be composed of perceptions from
   any lower level, not just the next lowest. So there can be a
   configuration of intensities, or a sequence of configurations. It’s hard,
   on the other hand, to imagine a program composed of principles, or a
   sensation made up of cubes.
   This is how the hierarchy began. The next step was to try to see how the
   basic control-system model, applicable at any one level, could be used to
   represent these properties of the whole perceptual system. It would
   require that a set of perceptions of one order be controlled as a means
   of controlling a perception of a higher order. The higher order system
   had to tell the lower ones whether their individual perceptions were too
   faint, just right, or too intense. This implied that the reference
   signals of the lower systems are adjusted by the higher ones.
   After that, I was able to set up some simple models that had these two
   properties (Rick expanded on them to six systems at three levels in his
   spreadsheet model). We’re still handicapped in that we don’t know how the
   perceptual functions of higher-level systems really work, so for the
   present we’re stuck with using the human perceptual systems, giving
   people tasks to perform at various levels and applying the one-level
   basic control model. We can set up a model for any one level, and run it
   and match it to real behavior, but we have to cheat on the perceptual
   input function: it simply creaates a perceptual variable that corresponds
   to what we can observe in the environment. We can’t show how that
   perception is derived from lower-level aspects of the environment. To do
   that, we would have to know a lot more than anyone now knows,
   anywhere.

All this is, of course, very time consuming, and often calls for
resources or skills that are unavailable. All of us, I think, have many
experiments in mind that we would like to do, but most of them languish
on the shelf. Two people or half a dozen people can only do so
much.

If anyone wants to propose a different kind of organization they are
perfectly free to do so. But random experimentation isn’t likely to lead
anywhere, and there has to be some sort of data, even if only personal
observations, to give substance to any proposal. I think that the facts
about perception described above are fairly solid, although the actual
levels proposed are probably much shakier than the basic principle itself
is. These relationships must either be shown not to exist, or be
incorporated into any model. I can’t think of any other but the
hierarchical model, but maybe someone else can make a different
organization work.

Best,

Bill P.

(Attachment 238187d.jpg is missing)