It was good to hear from you again. I think some of the questions you have
raised are standard questions. So here are the answers.
1) On the question of who the controller is, if there is one, in any system
which has interacting subsystems:
I will try to explain this in the context of the TV example, which is a
widely used example. I quote here from another response that more clearly
states your argument.
"You state that although a person is influenced by what is on TV, they are
still the "controller" of the ON/OFF signal. It may be proper to say this,
but not necessarily correct. Based on your definitions, a more consistent
analogy is that the master controlling system (the cathode ray tube in the
TV-identified as a controller as the TV will work in its absence) generates
a signal (A) that is transmitted to a subsystem (in this case the viewer).
If this signal contains a particular structure the subsystem will send a
signal back (feedback-B) to the control system to terminate the transmitted
signal (A). The viewer may hold the ON/OFF switch but by definition is not
the "controller". As for the hunger example where it was blood sugar level
(the signal) that controlled the hunger response, so in the TV case it was
the image/sound (the signal) that controlled the viewer response. The point
being that because of the more general implications, this analysis would
probably be more acceptable if the human viewer was replaced with an
artificial subsystem."
The feedback in these systems is the source of all confusion. In many
engineering systems, the controlling subsystem is easily identified and its
function tested out in a non-feedback mode. You might have tried this out
yourself if you ever had installed a thermostat for a heating or cooling
unit. After installation, one simply rotates the dial or presses some button
a few times to check out if the ON/OFF function of the thermostat really
works. So another test of a controller in any system, natural or artificial,
is that it can also be operated in a non-feedback mode. And this
non-feedback mode of operation clearly establishes which subsystem controls
whom; it should clearly show that the controlling subsystem is not dependent
on the feedback of information from the subservient system, whereas the
subservient system clearly depends on the inputs from the controlling
subsystem for its operation.
So perhaps a simplified, non-feedback mode of operation of the TV system
will clarify all the confusion. Suppose I set up the TV on an auto remote
control. Let's say a person, who is not watching the TV, simply switches the
TV ON and OFF every 10 minutes or some other fixed time interval. There is
no question of the content of the TV influencing the auto remote control
(the person with the remote control) here; so the TV no longer controls the
remote controller (the person) with its "content," in the sense argued in
the above quotation. There is only one-way communication here, from the
controller to its subsystem. I hope this helps to clarify the notions of a
controller in a feedback system and how to identify such a controller.
2) You said: "If the brain controls the body, it is also true that the body
controls the brain."
Apply the simple test explained above and it should be clear who controls
whom. You would agree that walking or running or crawling is one of the
functions where the body and the brain operate as an interacting feedback
system, with lots of visual feedback (information) to the brain about
position and environment. Will the brain be able to "operate" the body in
these modes (walking, running, crawling) without the visual feedback? Of
course it can - you can still walk or run or crawl with your eyes
blind-folded or shut, can't you? So the brain can clearly operate in a
"unidirectional" mode, a term that you had mentioned as a possible
characteristic of a controller. Again, as stated above, this
"unidirectional" mode of operation, without the feedback information, is all
that needs to be shown to establish a subsystem as the controller in any
system.
3) You said: "It is an oversimplification no matter how its stated."
I don't know what you mean by "oversimplication" - the notion of the brain
as a controller? I hear scientists talk about the "motor control" function
of the brain all the time, which I suppose is the same thing. Is that
"oversimplification" too?
4)You said: "We do not benefit from such broad generalizations."
It may not be obvious to you, but you are also using certain generalizations
(distributed, interacting, recurrent systems) to argue against mine. Read
carefully your arguments and you will understand. By the way, generalization
is the essence of science.
To understand the benefits of this particular "control theoretic"
generalization, you would have to get real deep into the nature of some of
the algorithms that we call brain-like. For example, all of the
connectionist algorithms are based on these control theoretic notions,
although never acknowledged explicitly. I am quite sure that if you look
carefully at your own algorithms, you will find that they also use control
theoretic notions. There is no escape from these notions. You cannot build
these brain-like systems without acknowledging control theoretic notions.
The main benefit is that we can build much better brain-like systems once we
acknowledge these notions.
5) You said: "Your relentless quest for a controller of the brain is a
search for a humunculus."
I don't know exactly what in the brain would qualify as the "humunculus."
Searching for sources of control in the brain is not the same as trying to
find "a brain within the brain." These controllers in the brain can have
very limited functionalities. It is obvious from genetics that these natural
systems do come with a lot of "built-in" instructions. So I don't think the
search for sources of control at a level slightly higher than genes is
irrational or absurd. The scientific evidence may already be there.
By the way, in distributed, interacting, recurrent systems, there is also
the assumption of control - a set of "operating rules" (e.g. a learning law
or some adjustment rule) that exercises "control" over some "hardware" (e.g.
a neuron or a cell). So the notion of "control" is already being used. The
basic argument is over how much "hardware" can these controllers control.
Can they exercise control over much more "hardware" than we think? I think
it is as simple as that.
6) You said: "Your thinking would be much better clarified by examining
specific issues than by vague, loosely defined notions."
Control theoretic notions are well-established in engineering. There is
nothing "vague" or "loose" about it. Design and construction of brain-like
learning algorithms are a specific issue in this context.
7) You said: "For example, what things control you? What things do you
control? Can you decide? Can you draw a firm distinction? Can you be
certain those things that you control do not control you, and vice-versa?"
The TV and other examples might provide the answers to these questions (who
controls whom).
8) You said: "What makes this an interesting or productive question?"
The need to design brain-like systems.
9) You said: "Consider the possibility that while things control (or
influence) one another, the relationship is rarely unidirectional."
As explained above, the TV viewer (and other examples) can clearly operate
in a non-feedback, "unidirectional" mode. Here you have clearly stated one
of the the real tests for a controlling subsystem - that whether it can
provide operating instructions to the subsystem it's controlling without
getting feedback from it. I can give you plenty of examples, from both
natural and artificial systems, where you can isolate the controlling
subsystem and operate it in this "unidirectional" mode and the system will
still work, although not very well. But this "unidirectional" test is used
all the time in engineering system and that's one of the real tests for a
controller.
10) You said: "In fact, the most interesting, complex and relevant examples
are of reciprocal influence, recurrent networks, feedback, modulation,
interaction."
I have no disputes here.
11) You said: "The rest of it is all plebian, pedestrian and simple. Fodder
for the dim-witted."
I hope none of us on this or any other mailing list would need to respond to
this kind of statement.
Tony, I hope I have answered all of your questions.
Asim Roy
Arizona State University