[Martin Taylor 931013 18:30]
(Bill Powers 931010.0640)
A long message, Bill, but I hope this will be a short reply. [Later:
it turned out not to be. Sorry.]
As I read your posting, I agree with most of it, and don't see what I said
that leads you to think I wouldn't. I think that the problem may hinge on
Again, don't mix "perception" with "signal." The argument is
that the "whole word event" may be (badly) controlled before
the signals that the analyst sees as its components can be
controlled and turned into perceptions.Here, however, you seem to be admitting at least that the
components must be represented as neural signals before any
"event" can be controlled.
That word "admitting" is the point, I think. I was asserting it throughout
my posting, in agreement with your tutorial corrections. Here's the key:
In the above, you said "don't mix 'perception' with 'signal.'"
Not in general, but I define all signals generated by input
functions at any level as perceptual signals.
I didn't make it plain, perhaps, that I was defining as "perception" only
those signals that are the subject of control, not all signals that
occur within the input functions of ECSs. I think that is a useful
distinction, but it might have been better to use a term such as
"C-perceptions," meaning any signal that is a perceptual signal in an
ECS, whether at that instant being controlled or not.
It was clear to me then and still is now that the hierarchy of
perception and control in the brain is carried out by physically
distinct volumes of the nervous system, spatially separated from
one another and sharing no neurons. Within one hierarchical
level, it is still possible that global networks of distributed
functions exist rather than discrete specialized functions. But
the networks of one level do not interpenetrate those of another
level. The levels are connected -- and separated -- by upgoing
and downgoing signals. The neurons comprising a lower level are
not just a subset of the neurons that comprise a higher level.
These are different sets of neurons in different places.
I have no problem with this, or with any of the similar passages
in your posting. I suspect there may turn out to be an issue on which
I queried you in Durango, as to whether there might be microstructure
within a conceptual level. I don't remember your exact answer, but I
am pretty sure you were not dogmatically opposed to the idea. I believe
there is quite a bit of microstructure within most, if not all, levels,
and that this microstructure develops over time.
One byproduct of my hierarchy of physically-separated levels is
that we no longer need expect that a single organizing principle
generates each successive level. The behavior of one level is not
reducible to the behaviors of the lower-level systems. If you
eliminate a higher-level system, the character of behavior that
goes with that system disappears and can't be reconstructed from
the behaviors of the lower level systems. There is no reason why
a world made only of shapes and configurations should suddenly be
perceived in terms of motion. There is no reason why a world of
relationships should suddenly develop a characteristic known as
categories. Each new level is, in terms of the existing levels,
gratuitous and surprising.
Once we get past the first sentence, I agree wholeheartedly, but that
first sentence gives me problems.
I don't see what "physically separated" has to do with it. Neither do
I see that differences in the characteristic operations of the different
levels is incompatible with there being a "single organizing principle,"
at least at the neurochemical level of analysis, and perhaps higher.
The only way to establish the existence of event control by an
infant is to test for it and observe it. This is the ONLY way to
avoid attributing to the infant levels of control that the infant
does not actually possess yet. If you make such judgements on the
basis of "some similarity of overall pattern," you are most
likely to be projecting patterns onto the data that you can
perceive but that the infant cannot. The passive observation of
patterns is an exercise in introspection.
Good point. I plead guilty, but I think Bruce has probably bailed me out.
Most likely, to my mind, control is being learned at both levels
at once. That is, there are certain vocal components that are
under good control and others that are not. At the event level,
event control first becomes good for the components that are
under the best control at a lower level, but control of other
events must await mastery of the control of other components.
Also, I would expect that after event control progresses for a
while, it creates situations in which reorganization must return
to the component level in order to allow the events to match
better what is being heard. Old components must be modified and
new ones must be added.
I think this is correct, and not incompatible at all with what I have
been saying. If it is, I plead lack of writing skill in expressing
what I think.
Bruce was proposing that syllables and phonemes are created
_after_ the perception of a "whole word event", by breaking down
the whole event into constituent parts. Thus the syllables and
phonemes are the result of analyzing (taking apart) a whole
event.
I didn't read him that way. The way I read it was that reorganization
to permit control of syllables and phonemes occurs _after_ some control
has been achieved at word level. The syllables an phonemes are not
the results of analyzing _A_ whole word event.
Look at it another way. The baby, as soon as it is born and perhaps
before, must be able to control some perceptions or it dies. One means
of control (action, that is) is to cry. The controlled perception may
not be one we adults would recognize, but we can label it perhaps as
"non-hunger, non irritation, ..." Each occurrence of "cry" results
eventually in some degree of improvement in the state of the perception,
but often the change happens only after a long time during which other
perceptions are disturbed (food is offered when wetness should be reduced,
and so forth). When the cry starts to become differentiated into different
kinds of noise, mother begins to act so that baby's perceptions are controlled
more quickly and accurately. Mother need not be perceived as such (though
there seems to be some evidence she can very early be distinguished from
other people). She is just a piece of the environment, and reorganization
leads to differences in crying that affect different erroneous perceptions.
Now _this_ cry affects the "hunger" perception, differently from the way
it affects the "wet" perception, whereas _that_ cry seems to affect the
"wet" perception instead. So the perceptions of those cries become controlled.
Extend this to a period 15-20 months later. Vocalizations are becoming
much more controllable. As you say, baby can now control the perception
of different kinds of sounds. Lots of different babbling practice helps
that to happen. Now baby finds that some strings of perceptions enable it
to control other perceptions, such as the joint presence of Daddy and a ball.
But this doen't mean that baby has learned ANY phonemes or words. Or
morphemes. Baby can control its own perception of its acoustic productions
to an extent sufficient to enable it to control other interesting perceptions
by (as we observers know) affecting the actions of other people. What
more need it do? It doesn't need any of the analytic apparatus. It doesn't
get any benefit from controlling syllables or other smaller structures.
That comes later.
···
============
On mutuality, see my posting to Rick today. I don't think I was repeating
anything from the "helping" literature, but if I was, the viewpoint was
sufficiently different that I didn't see it. That might make the posting
worth soemthing. I'll look over your analytic comments later. I have to
rush now. Concert day.
Martin