It doesn't take long to fall far, far behind the stream here. I will be
gone all of next month, so I suppose I will just have to tune in as a
newcomer when I get back But first, one note out of many scribbled
while on the train.
(Martin Taylor 920820 19:15) --
I think we have three candidate reorganization systems that are plausible,
Add to this one more, Bill's fallback position and the one I like as
primary, though it is harder to model. In this view, reorganization is
carried out by the constituents (cells) of the control system
experiencing conflict and chronic error. Call the latter a control
system of order n; then its cells or constituents are a population of
control systems of order n-1 (e.g. nerve cells). The hypothesis is that
when a control system of order n experiences error, it produces
something intrasomatically that, if it persists, is the occasion of
error in its cells as control systems of order n-1. They then begin to
change shape, contacts with one another, reactiveness, etc. They do so
until their local error is reduced. But reducing that order-n-1 error
corresponds to reducing the order-n error, in precisely the sense that
reorganization continues until it reduces the error that occasioned it.
This is difficult to model. Would you have soldered joints respond to
overheating by rearranging themselves? The problem, of course, is that
soldered joints are not control systems. (Similarly for other
constituents of our means for modelling). But conceptually,
intuitively, this is very satisfying to me. Seems to fit with Edelman's
ideas too.
I am fairly new on this net so I hope this is not a "could have been
answered if you looked here" question ... but, do any of the models predict the
top layer to change and if not how are the top levels set in the first place?
If the top levels change (and intuitivly they should), do they change
themselves, help change themselves, or are they changed by the lower levers to
create a sort of circle? Also, how are connections between levels set in the
first place, espicially the top level as it has the potential for being the
most complex wrt sensory inputs?
          -John van Loon
          -no fixed address, anymore
     Well I have to delete my desk top now so I won`t get the answers I want
untill I sign on again. Thanks for controlling my thoughts this summer, course
I can`t drive anymore ... Oh well I think that I can still walk.
[Martin Taylor 920829 13:00]
(Bruce Nevin 920828 13:21:43)
(Martin Taylor 920820 19:15) --
I think we have three candidate reorganization systems that are plausible,
Add to this one more, Bill's fallback position and the one I like as
primary, though it is harder to model. In this view, reorganization is
carried out by the constituents (cells) of the control system
experiencing conflict and chronic error.
So far, it sounds identical to my own position, at least if by "control
system" you mean "ECS". If you mean "hierarchy" I don't see how a hierarchy
experiences anything, conflict and error included. An ECS can experience
error, which could be the result of conflict, but it can't experience conflict
directly.
You go on to expand on how reorganization might be achieved in a specific kind
of hardware(mushware) implementation. I have no objection to that, but the
way it is done would have to be different in other (say intracellular) systems.
I am afraid I have never properly understood your "control system of order n"
concept, so maybe I am quite misreading your intention here.
My notion works entirely within a (any) single control hierarchy, Bill's
requires a separate hierarchy monitoring variables that do not appear in the
percepts of the hierarchy being reorganized, and Jeff's requires a set of
symbiotic control systems that may be hierarchies, but that influence the
one being reorganized only by inducing error into it so that it reorganized
according to my (ad your?) scheme. Does your "order n" and "order n-1" fit
into one of these three categories, or is it something else again?
Martin
PS. Personally, I rather like Jeff's version, although Bill thinks that it
can't work. In time, I hope, we shall see.
But first, one note out of many scribbled