[From Bill Powers (961023.0745 MDT)
It's my impression that all ocular motor functions are driven by signals in
other pathways, outside the optic nerve. I think those efferents in the
optic nerve must be doing something to the retinal perceptual systems. What,
I can't imagine.
But it probably has something to do with controlling perception. 'Cause
what else is there to control?
Mere perception itself is active.
Let me spin a fantasy, which is they story I invented to talk about those
memory boxes as containing information about the "state" of the units in
the control stack.
I'd been looking at D'Arcy Thompson illustrations. You know, where he'd
have a fish drawn on three types of coordinate system, cartesian, log, and
polar (whatever the appropriate terms are for these things). It's the same
data points in each case, but because the coordinate systems are different,
the fish forms are different. Thompson's point was about the mechanisms of
development, but I was thinking about perception & about all those neurons
which were going the wrong way in the visual system. I got to thinking
about the zero reference level and began to wonder what would happen if I
took these various fish (or skulls, or whatever), described each form in
cartesian coordinates, and then operated on the coordinate systems for each
fish-form in such a way as to make each fish assume a nearly circular
outline. Since the "input" forms are different, I would have to apply
different transforms to the coordinate systems to produce my circular fish
(a zero reference level taken literally in a bizarre way). Those different
transforms would then by my descriptions of the types of fish.
I then imagined this kind of setup: We have a CRT on which a circle is
painted. Some image is input to the system and the inner mechanism
separates foreground from background, and projects the foreground image on
to the screen so that the largest extent just touches the circle, but
doesn't go beyond. An operator sits in front of the screen and turns knobs
which make global and local changes in the coordinate system used to
project the foreground image on the screen. The operator twiddles the
knobs until the area between the circle and the perimeter of the image
drops to an acceptably low value. When that happens the operator notes the
knob settings and enters them into the log as the object's description.
Any image which can be zeroed out with knob settings close to a value
entered into the log will be considered an instance of that type.
So, the process of knob-twiddling is reorganization, and the log of knob
setting vectors is a memory box. And those efferents going down to the
retina might be carrying the coordinate transformation values down to the
retina -- not literally of course, this is a metaphor, but its the
functional relationships which interest me.
Like all metaphors this one is quite limited, with its major limitation
being that operator into whose head we can imagine just about anything.
What interests me about the adaptive optics set-up I mentioned in an
earlier post is that it is a real physical mechanism. And it has
components linked together in the same funtional relationships as my
fantasy recognition system, except that the adaptive optics system doesn't
have anything which corresponds to the log of output settings.
* * * * * *
Now, about reorganization. Imagine a simple creature which moves toward a
circle in pursing some particular goal. One day the envrionment changes and
the circles are replaced by pentagons. The creature has to reorganize to
follow pentagons or forever leave some goal unsatisfied. So, it
reorganizes a particular reference level so that the circle is replaced by
the pentagon. As long as the environmental change is a long-term one (in
relation to the time-scale on which the creature lives), this is OK. But
if the change is temporary, then the creature is going to be in trouble
when the circles return and the pentagons disappear.
Where such change is relatively frequent we need a better system. We need
to make a distinction between the mechanism which applies the reference
value to behavior and the mechanism which contains a selection of
permissible reference values. Once we have this, the creature can store
the various reference patterns, circle, pentagon, square, octagon, etc. in
the memory unit and apply them to behavior as the environment requires.
Now, what is being reorganized in these two situations? In the first the
reorganization happens to the functioning stack. In the second, the
functioning stack remains the same in all situations. What is being
reorganized is the memory unit which contains reference values.
The paper which Hays and I did on the brain treats the whole neocortex as a
bunch of memory boxes organized in three tiers. We're not committed to the
idea that that is the only memory storage in the brain (I think the
cerebellum might be thought of as a memory unit, one operating much lower
in the stack), but we are pretty committed to the notion that nothing else
is going on in the cortex but memory. But, if you think of the cortex as
consisting of some 100 to 1000 memory boxes, each with input and output
connections to various cortical and subcortical regions, then you're going
to get some very sophisticated and complex dynamics as the system juggles a
50 or 100 (not all cortical regions will be functioning at once)
simultaneous reference patterns, continually "swapping" patterns in and out
of the stack to account for current sensory input and direct motor output.
later,
Bill B
PS Nope, haven't forgotten about mirror recognition, just haven't gotten
to it yet.
ยทยทยท
********************************************************
William L. Benzon 518.272.4733
161 2nd Street bbenzon@global2000.net
Troy, NY 12180 Account Suspended
USA
********************************************************
What color would you be if you didn't know what you was?
That's what color I am.
********************************************************