Modelling the human brain by modelling its evolutionary emergence

mat mats_trash at hotmail.com
Mon Feb 25 07:46:17 EST 2002


It is a little clearer now and many of the problems I raised are
dissipated becuase in fact your project is clearly NOT MODELLING THE
EVOUTION OF THE BRAIN.  If you start out with 'neuronal assemblies'
then what you are in fact doing is more akin to modelling the
embryonic and post-natal development of the brain.  Nothing to do with
evolution.  This is not to denegrate the merits of the project, but to
just be explicit about what you are doing. What your project may do is
highlight some of the developmental mechanisms by which the immature
brain responds to its environment.  But again this is not evolution.

A few points concerning your 'new' project though - Similarly with the
problems I raised to the 'evolutionary project', the insight you will
gain from this experiment is very much more limited than I think you
appreciate.  All that you will be able to do is perhaps abstract some
general rules as to morphogenesis of the CNS, but because you do not
start the modelling from a system that is as richly detailed as the
'real' developing brain, you have noe way to ensure that the
development you observe is in any way a reflection of the real brain. 
Though it may perform in elementary functional testing in a similar
manner to the brain, that is no guarantee that 'whats inside the box'
is the same.  Electronic cars and those with internal combustion
engines can both travel at 30mph, but that doesn't mean they are the
same underneath the bonnet.

Also, from where do you know that cortical columns are 'simple
hopfield networks'?  I think it very unlikely that all cells within
the column are connected to all others.  Many will be connected more
thna once, some not at all.  Again this bears out what I say above -
where you start from has a huge influence on where you end up, and the
conclusions you can draw from the project.

Lastly, what are your 'analyses'?  What data have you looked at in
coming up with your project?  Be explicit - not just the thoughts of
some people - actual data about cortical architechtonics.




More information about the Neur-sci mailing list