[From Rick Marken (2005.12.07.1025)]
Marc Abrams (2005.12.05.1902)
First, and I am being very serious here, what else did you think I was
asking and why?
Sorry, your question was just a bit ungrammatical and I wanted to make sure
that I understood you correctly.
Rick Marken (2005.12.06.1530)
The concept of levels of control is based on several different observations
and tests. First is the observation that certain types of perceptions
depend on others but not vice versa.
As far as I understand it these 'observations' were the reflections of and
within a single individual. I have not 'observed' this to be true in me so
how may I observe this in others?
If you don't observe this in your own experience then it certainly cannot be
evidence of levels for you.
The literature that I have read on perceptions does not confirm these
concepts and ideas. Can you cite some other sources besides those of Powers,
that might help me understand this better?
There are some relevant references in my "Hierarchical Behavior of
Perception" paper that is reprinted in _More Mind Readings_. I think the
most relevant are Hubel/Wiesel (1979), Martin (1972), Palmer (1977), and
Second is the observation that the nervous system itself seems to be
I can organize the nervous system(S) in any number of
structural types I desire depending on what I am interested in showing, and
this in fact has lead me to some of my confusion.
I agree. Then this cannot be evidence of levels for you.
So my question still remains, what does the hierarchy represent, structure
or function, and what data do you have to support this view?
The hierarchy is functional but Bill does point out (in B:CP) some possible
correlates of this proposed functional architecture to the structure of the
Third is the observation that control of some types of perceptions can be
done on a faster time scale than can control of other types of perceptions,
Ok, here your 'observations' are those of a model, not an individual and in
that model a 'perception' is represented by a simple 'signal', again have
you been able to validate this through physiological data? If not, what you
have 'data' on is a computer model not a human, so it might be a fine
theoretical piece of work but how can you validate this empirically?
We have confirmed this observation empirically by comparing the behavior of
a real person to that of a model. That's what I mean by empirical testing.
If you only consider empirical testing to be validation through
physiological data then our empirical tests will not be evidence of levels
for you either.
Fourth is the the observation that some controlling can be seen to be
temporally nested within the controlling done by other systems
Again, from a theoretical standpoint I applaud your effort. Where is the
It's in the comparison of human to model behavior. Again, if you don't
consider this empirical evidence then we are just playing in different
sandboxes, I guess.
Finally, we need hierarchical models to account for some phenomena.
No, you don't need a hierarchy, a network would do just fine
I would find that claim more interesting if you showed me the network model
that can do what was done with the hierarchical model.
If you want to discuss my model I will do so only under the rules and
conditions of critical discussion.
What are those rules?
Well, there appear to be some significant differences between the model in
your diagram and the PCT model.
You bet, BIG differences, glad you noticed.
The biggest difference, of course, is the
"world model" box. What is a "world model"? How does it work? What is it
there to explain?
All very good questions indeed.
Thanks. Got answers?
The reference for what I presume is the perceptual signal
(the output of the "sensory input" box) comes from this world model. Why
does the world model specify the value of the perceptual signal?
Again, another excellent question.
That is, why does the world model care what the value of the perceptual
For the same reason a 'comparator' or 'output' function does.
Ok, this is an answer, but I don't quite understand it. I don't think of the
"comparator" and "output" functions as caring about what the value of the
perceptual signal is. What I was getting at was the fact that, in the
hierarchical PCT control model, it's the higher level systems that "care"
about the value of the lower level perceptual signal. The higher level
systems send references for the perceptions that lower level systems are to
produce so that those lower level systems will produce perceptions that make
the higher level perception be what the higher level system wants. It's not
clear in your model why the world model would want the perceptual input to
be a particular value because, as I recall from your diagram, the world
model never even gets the perceptual signal as input.
Finally, is there some test I can do to demonstrate to myself that your
version of of the PCT model is superior to the current version?
Test for what?
To compare the behavior of the model to the behavior of a living system.
What did I say that made you think I was presenting a 'better' model? I said
I was presenting a different model. A different perspective.
OK. I guess the only reason I can imagine for presenting a different model
is if 1) it predicts behavior better than the original 2) it predicts
behavior equally well but is more parsimonious than the original or 3) it is
precisely equivalent, behaviorally, to the original but givens insights into
the reasons for the behavior of the model that were not apparent in other
representations. Are you presenting your different model for one of these
reasons or for some other reason?
Rick, you really just don't get it. I am not in competition with PCT, my
model acknowledges and accepts the work done by many other people over many
years, including yourself and Bill.
You are absolutely right. I don't get it. What is it that you want from me?
You want me to disprove the PCT model and I have no desire to do that. You
will have to decide whether I am 'doing' 'PCT" or something else, and
frankly I could care less what you or Bill thinks.
Kind of makes it a waste of time, then, for me to tell you what I think
then, doesn't it?
I would much prefer to have us working together
Why do you want to work with me when you don't care what I think? What kind
of work did you have in mind?
I need to know how we will go about discussing this stuff before we
I guess I would like to go about having a critical discussion where we
discuss ideas critically but politely. But I don't understand why you would
even want to discuss this stuff with me at all if you don't care what I
think. I don't mind that you don't care what I think. But I do find it
peculiar that you would want to have a discussion with someone whose
thinking you care not about.
I'm not interested in making models I cannot validate with real world data.
I'm not either, though I think we may have a different idea of what it means
to validate a model with real world data.
Richard S. Marken
Home: 310 474 0313
Cell: 310 729 1400
This email message is for the sole use of the intended recipient(s) and
may contain privileged information. Any unauthorized review, use,
disclosure or distribution is prohibited. If you are not the intended
recipient, please contact the sender by reply email and destroy all copies
of the original message.