NetNews Usenet Archive 1992 #30/NN_1992_30.iso/spool/bit/listserv/csgl/1811

Text File | 1992-12-12 | 10.0 KB | 200 lines

Comments: Gated by [e-mail]
Path: sparky!uunet!gatech!usenet.ins.cwru.edu!news.ysu.edu!psuvm!auvm!UTMBEACH.BITNET!TBOURBON
Original_To:  BITNET%"[e-mail]"
Original_cc:  TBOURBON
Message-ID: <[e-mail]>
Newsgroups: bit.listserv.csg-l
Date:         Sat, 12 Dec 1992 02:37:00 CDT
Sender:       "Control Systems Group Network (CSGnet)" <[e-mail]>
From:         Tom Bourbon <[e-mail]>
Subject:      Re: neuropsych and PCT
Lines: 187

From: Tom Bourbon  (921212   02:13  CST)
Concerning Nerves, Brains and PCT

The continuing discussion on "controlling error" spawned a sub-
theme on the idea that PCTers should pay more attention to
neurophysiology, neuroscience and related areas.


Mark Olson (921129  19:06) posted on "neuropsych and PCT"

>TO Bill or anyone,

>... I have noticed a few things which bother me and maybe you can
>iron out or explain.  My basic observation/complaint is that we
>ought to be doing some more bottom-up thinking in terms of what
>the neurophysiological data suggest.  I say this because it seems
>to me that we have Sensation (color) before Configuration
>followed by Transition (motion) when the neuropsych data suggest
>that color and motion are processed before "objects" are
>discerned (in area TE, or IT).

>It seems that some levels are concerned with temporal
>relationships while others are concerned with spatial
>relationships.  Some levels (Sequence) seem devoted to procedural
>tasks while others seem devoted to declarative tasks (Category).

     Through your readings, you have (inevitably?) picked up the
style used most often in the neuropsych-neuroscience literature.
Writers employ language that supports certain impressions and
inferences in readers, but says nothing definite or meaningful.
One result is that innocent readers believe more is known about
brain structure and function than is the case.  I am not attacking
you, Mark, just stating that words and styles in the neuroscience
literature can be tricky.   You have probably seen the phrase "the
neuro... data suggest" countless times.  But data do not suggest.
People look at data and people decide or believe that the data mean
something -- people suggest that data mean certain things.  I am
not trying to be picky, but honest writers in the neurophysiology
literature -- and most of them are -- are telling you that they
don't really know -- the data merely suggest.  If the data are no
better than that, then they can also be seen to "suggest" other
conclusions to other people.  That is usually what happens.
     When applied to brains, words like "processed", "discerned",
"concerned", and "devoted" have no specific referents.  Researchers
see that activity differs across locations or times or some
combination of locations and times, and they read into the data the
language of communication theory and computer science.  Or they
attribute to cells and "regions" in the brain the achievements and
the attributes of the entire person, or of the species or the
culture.  That is common practice, but it is not good science.


> ... We should take what we know about brain processing into
>consideration.  I agree that we do not want to just go from brain
>data to theory--both are needed.  But now it seems that the
>hierarchy does not jive with neuropsych data.  For instance, it
>doesn't seem to me that processing in the parietal lobe (telling
>us WHERE the "object" is) is in any sort of hierarchical
>relationship with processing in the temporal lobe (telling is
>WHAT the "object" is).

>How, for instance, is the Relationship level instantiated in the
>brain?  If that is the level devoted to noting (temporal)
>relationships like causation and (spatial) relationships like
>above/below, left/right, etc, then does this level exist in
>multiple brain regions?  Higher levels seem to reside in frontal
>regions.

Bill Powers (921130.0730)  had some good things to say on those
topics:

>Unfortunately, neurophysiology is too vague about most perceptions
>to do us much good. While the neurological observations are
>probably all right (identifying activity or lesions in various
>parts of the brain), the identification of the CORRELATES of these
>activities depends on informal subjective observation of the
>world. It isn't the "subjective" aspect I question, but the
>"informal" aspect.

I think by "informal" Bill means something like what I mean when I
speak of neuroscientists' careless use of language.  Bill's
discussion a little farther along, about regions that allegedly
process visual information and tell us where or what, goes directly
to the issue of the incredible sloppiness in much of the
neurophysiological literature.

Later, Bill wrote:

>In HPCT there are at least some principles involved in identifying
>correlates of brain activity. For B to be a higher level of
>perception than A, the existence of B must depend on the existence
>of A but not vice versa. When B is controlled, it must be
>controlled by varying A, but not vice versa. And it must be
>possible to control A without necessarily controlling any B, but
>not vice versa.

     An extremely important point.  The word "levels" can be
applied to nervous systems in many ways, not all of them compatible
or equivalent -- it can refer to "altitudes" or distances above or
below an arbitrary reference point; to presumed complexity,
abstraction or sophistication of "processing";   and so on.  Bill
went on with a nice discussion of this problem and of the
categories of your own perceptions that you were imposing on brain
function and architecture.

>But what we know about brain processing is based on how we
>organize our concepts of perceptions. All we know directly about
>brain processes is that different areas of the brain are active
>under difference circumstances. The whole problem is to
>characterize what is different about those different
>circumstances. So we have to organize our understanding of
>subjective perception before we can use that understanding to give
>meaning to activities in various parts of the brain. Without an
>understanding of how perceptions are related to each other, we
>can't understand what it means when different parts of the brain
>are active.

     This topic cannot be emphasized too strongly.  Taken alone,
anatomical and physiological data have no psychological meaning.
Meaning, importance, and significance flow FROM the facts of
subjective experience TO anatomical and physiological data, not the
other way around.  For example, we do not perceive the world more
clearly, or less so, as a consequence of neurophysiological data;
but we "see" neurophysiological data in the light of, or as ways of
explaining, perception.  It has always been so.
     The problem is even deeper.  Most of the neurophysiological
literature derives from research in classic cause-->effect designs:
researchers apply stimuli and observe responses, or if they are up
to date, they administer inputs and observe outputs.  If you
believe organisms, or their parts, are C-->E systems, you will
study them as though they are and you will interpret your data as
though they are.  But what if they aren't?  What if they exercise
control and you don't even consider that possibility, let alone
test for it?  In that case, you will repeat most of the failures in
the neurophysiological literature.
     Even at the level of a putatively simple reflex, where the
model of C-->E seems safest -- where S-->R with certainty and
regularity, or so it seems, appearances can be deceiving.
Virtually all such reflexes are examples of control -- the alleged
reflexive response just happens to eliminate the effect a
disturbance, mistakenly identified as an independent stimulus, has
on a controlled variable -- the existence of which is not dreamed
of by the researcher.  There are exceptions to this portrayal of
neurophysiology-neuroscience, but not many.
     From the apparently simple level of reflexes, on through the
allegedly most sophisticated and complex neurocognitive feats, the
story is the same:  Researchers treat the living systems with which
they work as though they were input-output devices.  The resulting
data have little to do with what organisms do and how they do it,
therefore, when we use them in our attempts to understand brains,
they (the data) cast more darkness than they shed light.  I wish
that were not the case, but it is -- no matter how impressive the
technology for producing lesions, for imaging, or for creating
colorful cartoons of changing patterns of activity, research in
which organisms are treated as though they were what they are not
tells us very little about how brains figure in the bigger picture.
     And that is why I disagree with Mark's sense of urgency when
he posts:

Mark Olson (921211 10:35:57 )

>Concerning Gary's post about the man who couldn't reach for an
>object, I think  this is a perfect example of how PCT'ers should
>give more consideration to  where these various control processes
>occur in the brain.  The superior  parietal region is evidently
>involved in representing external space (body  centered) and
>prefrontal areas seem to be involved in tasks which sound very
>much like tasks at the Sequence and Program levels.
>I think Bill did a great job of constructing these levels in the
>manner that he  did, but the hierarchy is not going to improve
>much anymore by thinking about  it--we should look at the brain.

     Mark, you will need to go beyond Gary's example, and beyond
the literature on "involvement" of brain areas, before I see that
PCT is in danger if we do not "look at the brain."  My reading of
the present scene is exactly the opposite:  Without insights into
the organization of perception and behavior like those offered by
PCT, much of (cognitive) neuroscience is seriously at risk.

Until later,

Tom Bourbon                             e-mail:
Magnetoencephalography Laboratory       [e-mail]
Division of Neurosurgery, E-17          [e-mail]
University of Texas Medical Branch      PHONE  [phone]
Galveston, TX 77550                     FAX    [phone]
USA