Re:[MURG] Artificail Hippocampus - some REAL research

["Martin O. Baldan" <martinob@arrakis.es>]

Hi, Randal and all



Wow, I was astonished by that piece of news!  I read it in the newspaper and
thought it was an exaggeration. I didn't know that computational
neuroscience was so advanced.



If I got it right, they didn't know the inner structure of the hippocampus,
they just recorded the outputs for different inputs and designed the chip
accordingly. I have the impression that this kind of approach is widely
used. Well, I have a small question about that strategy: is it universally
usable all over the brain, or are there some conditions the wiring pattern
of neurons must meet? I mean, in digital electronics, you can easily model a
combinational circuit by recording its output for each input (a truth table)
because each input can only produce one particular output (that is, the
circuit has no memory); but if the circuit is sequential it gets much more
complicated and may well be impossible to obtain a good model by setting
inputs and reading outputs. A trivial example: a cell phone, where the
buttons are the input channel and the screen is the output channel. If you
enter the wrong PIN several times (and then the wrong PUK) then the phone
gets locked and you can't learn anything else by pushing buttons.



Do similar problems arise in neuronal circuits? If not, why? I guess this
analogy is probably too crude, but I'd really like someone to explain me the
relevant differences.



Yours,



                  Martin





>Message: 4
>Date: Thu, 13 Mar 2003 09:06:39 -0500 (EST)
>From: "Randal A. Koene" <rak@minduploading.org>
>To: murg@minduploading.org
>Subject: [MURG] Artificail Hippocampus - some REAL research
>Reply-To: murg@minduploading.org
>
>Hi there,
>
>All this talk about how to approach getting some real research done...
>
>Well, in an encore to the link I posted about the challenge to
>the Canadian Neuroscience Foundation, now a group in California
>working with Theodore Berger have started testing an
>artificial hippocampus - a topic very dear to me.
>
>http://www.newscientist.com/news/news.jsp?id=ns99993488
>
>Specialized as I am in all things hippocampal, I know of course that
>knowledge of hippocampal function is far from complete, so this is
>clearly an early test. It isn't really about the hippocampus as much
>as it is about the ability to link to signal pathways within the brain
>and to process the information flow in specialized CPUs. That is
>the start of a very wide and very promising field of neuronal
>prosthetics.
>
>You'll notice that they didn't start by coming to us, the experts in
>mind uploading to ask where they should start. ;)
>
>Quite to the contrary, we should be going to them, to work with them.
>That is where you can do the most good. Whatever your skills are, try
>to work IN the field, then make a difference when it comes to
>directing progress toward WBE.
>
>Cheers,
>       Randal