[Neuroscience] Re: Wherefore art thou Neuron Code?

r norman via neur-sci%40net.bio.net (by r_s_norman from _comcast.net)
Mon Apr 9 21:43:44 EST 2007


On Mon, 9 Apr 2007 02:21:30 +1000, "John H." <bingblat from goaway.com.au>
wrote:

>
>"r norman" <r_s_norman from _comcast.net> wrote in message
>news:so671317m0t3013efcj2p2hv29ahrosrnn from 4ax.com...
>> On Wed, 4 Apr 2007 07:30:34 -0400, "Glen M. Sizemore"
>> <gmsizemore2 from yahoo.com> wrote:
>>
>> >Damn! I wrote a rather long reply to this, and left it slightly
>unfinished,
>> >and during the night my computer rebooted and the document was not even
>in
>> >backup. Let me reply briefly, and maybe later I'll revisit the issue. The
>> >modern imaging stuff (a perfect example of how neuroscience obtains new
>> >facts because of physics and engineering rather than by new
>> >conceptualizations) is largely a load of crap. I have always criticized
>it
>> >on conceptual grounds - even if the findings are accurate, and mean what
>> >they re purported to mean (i.e., that part of the brain is more active)
>the
>> >importance of such findings is dubious. OK, such-and-such a brain area is
>> >active - how does that explain the physiology of behavioral function.
>> >Obviously, the fact is of potential importance, but the data are simply
>used
>> >to argue a version of what should be called neurophysiological animism.
>> >"Look Martha! There's where the executive lives!" "And all the file
>clerks
>> >have to live in the hippocampus!" Now it appears that there may be little
>> >reliability in the measurements - oh well, another few decades of pissing
>> >into the wind.
>>
>> There is real value in the imaging work, but not as the "explanations"
>> of behavior that are widely touted.  "Aha, when you do X then brain
>> structure Y lights up.  So Y causes X!"  That is nonsense.   However
>> it does strongly indicate that some specific neuronal circuits that
>> are strongly associated with X are located in Y.  My main problem with
>> it is that  if you see a 1 mm speck of brain tissue lighting up (and I
>> am not sure that the resolution is even as good as 1 mm) then there
>> are still some 1000 to 10,000 neurons possibly involved.  You will
>> never even discover the enormous significance of all those other cells
>> whose locations are scattered enough so that there is no significant
>> change in blood patterns and oxygen usage at those locations.
>>
>> I get even angrier when i hear or read about the "fact" that serotonin
>> is the "depression chemical" or dopamine the "addiction chemical" or
>> whatever the current fashion might be.  Nobody ever says that acetyl
>> choline is the "breathing chemical" even though if you block
>> cholinergic junctions, you quickly die from lack of respiration.
>>
>> >Modularity has, no doubt some sort of verity, but what are the modules?
>> >This, of course, is a version of what I mostly argue around here: unless
>we
>> >conceptualize behavior properly, we can never explain it at the
>> >physiological level. We literally do not understand what we are trying to
>> >explain! Needless to say, I think that behavior analysis has identified
>the
>> >core processes, and behaviorists - most notably the Big BF himself - have
>> >gone far in saying how these core processes interact as complex behavior
>is
>> >produced. To say that this view is, ummm, a minority view, is the
>> >understatement of the century.
>>
>> I happen to strongly believe in modularity.  It is essential when I do
>> computer programming.  It is almost certainly as essential when
>> evolution (or learning) does brain programming.  However the main
>> point is, as you say, "we literally do not understand what we are
>> trying to explain"
>
>
>Just the other day I browsed through a book by Ray Kurzweill (surname not
>right) wherein this AI dude argued that in his lifetime it will be possible
>to upload his self into a machine and so achieve immortality. Now I read two
>neuroscientists stating that we don't really have a clue about what we are
>talking about.

Statements like Kurzweil's are the very reason why two neuroscientists
state we don't have a clue.  

>In his text, "The Wisdom Paradox", Elkhonen Goldberg puts forward an
>interesting twist on the cerebral organisation debate. He raises the
>interesting idea that the right frontal neocortex is primarily about
>learning new skills while the left is about executing learned skills and has
>evidence to support it. Perhaps this explains why learning is so often
>associated with struggle because in imaging studies one typically sees an
>association between the right and sadness, the left and happiness. Tough
>luck for you educators, evolution is working against you. Unless of course
>you don't teach your students to learn but rather just tell them how to
>think. Perhaps the best test for teaching skills will be the number of
>suicides in the teacher's classes. The higher the number, the better the
>teacher.

Gee, associating some function with only, say, a region of the  brain
with some 10 billion neurons really does explain everything, doesn't
it.  And, perhaps, those 10 billion neurons might be able to handle
more than one task without confusing them.  The type of wild
speculation you indulge in, even if only to parody what people really
say, is exactly the problem.  Note:  having a portion of the brain
"light up" in imaging studies does NOT mean that portion of the brain
is "responsible" for the function.  There might be some tiny number of
neurons, perhaps only a few hundred thousand or so, scattered all over
the place that are really the critical factors and this would never be
seen by current imaging techniques.
>
>I find it much more easy to think about evolution giving rise to new
>cerebral modes of function by building on the existing set rather than
>creating ones anew. Yes, one can argue that over time the existing set
>evolves a new module here or there but I think the analogy between the
>evolution "programming" the brain and our programming of computers is deeply
>misleading. Computers aren't programmmed by the environment, animals are.
>Without a changing environment there would be precious little selection
>pressure for new modules to arise. For example, in human evolution the
>commencement of the ice ages 2.5 myr ago and the subsequent changing
>environment was the primary driver of cerebral encephalisation in hominids.
>Yet here the region of largest expansion is the region where we have great
>difficulty assigning "modules": the frontal cortex and to a lesser extent
>the temporal lobe. If evolution was about creating new modules as extensions
>to existing cerebral architecture shouldn't we expect these regions to be
>more precisely delineated?
>
>I'd argue for a "connectionist modularity" rather than a "geometric
>modularity". That is, these modules are not to be found in discrete
>geometric regions but in how those regions are connected to other regions.
>Goldberg comes up with an interesting idea here: he mentions spindle cells.
>These types of cells have extensive connections and are higher in the right
>than the left, very high in human beings, though recent research has
>established that whales may also possess some spindle cells. Generally
>though even in the primate line spindle cells are sparse. So if you're
>looking for an evolutionary marker that distinguishes hominid
>encephalisation than spindle cells might be worth thinking about. Then
>again, the cerebellum also is a late evolver and is involved in many things
>... . That's the problem with reaching back into evolution, one can imagine
>too much.
>
>The areas we see blazing up are not the executors of a behavior, the whole
>organism does that. These regions possess skills that are applicable in
>faciltating that behavior but we should not then assume that only that
>region has these necessary skills. For some striking examples of how an old
>dog can learn new tricks look up "constraint induced movement therapy". A
>stunning finding using this clinical method was in a study of young girl
>with cerebral palsy in Melbourne Australia. They placed under this regime in
>the hope of restoring some right side function to her body. What they found
>with fMRI was that it was the right motor cortex that enabled the remarkable
>recovery. Now try and make sense of that terms of modules or even
>neuroanatomy for that matter. The same is true of hemispherotomy,
>the results are quite stunning. It also well established that with practice
>cerebral regions grow, obviously invading other modules. Take my earlier
>example of "action selection" centres in the mammalian brain. Any one or
>number of these regions may be the prime mover of a given behavior, that
>will be contingent upon the physiological, historical, and environmental
>state of the organism. That is, no one region or collection of region is
>necessarily the executive. Geometric modules containing specific skills for
>executing behavior cannot address these issues.
>
>At the bottom of all this is the fundamental problem of just what are
>neurons, and for that matter glia, doing with all this activity. In these
>days people typically think in terms of information processing. Brains
>somehow analyze the environment and decide accordingly. I've often struggled
>with this information processing model and in these stranger days I prefer
>to think of brains as selecting behaviors rather than analyzing the
>environment. This is much more concordant with evolution because that was
>the primary goal of the first nervous systems. To enhance behavior not
>understand the world. One problem I have with the information processing
>model is that at some magical point in evolutionary history nervous systems
>became interested in analyzing the environment and building internal
>representations of the same as a new way to deal with the world. Some people
>are fascinating by all the feedback loops in the brain and cite these as
>evidence of information processing but I can think of way to approach this
>problem without need to think of any information processing. Now I'll just
>have to go away, be very still, and analyze that ... .

You might also want to consider experimental techniques that have even
a chance of producing some evidence to support your ideas.




More information about the Neur-sci mailing list