X-Message-Number: 25884
From: 
Date: Sat, 26 Mar 2005 01:43:27 EST
Subject: Uploading technology (2.i.0).

Uploading technology (2.i.0).

 In the first message of that set, I have broken the uploading aproach into 3 
steps:
- What to modelize ?
- How to modelize it ?
- On what technology (ies) implement it ?
 
Here I take the first part of step two: 
As a first look, what has to be modelized fall into two domains:
- Conduction of electrochemical potentials in axon (action potential), the 
neuron output, and in the dendrite tree, or neuron input side.
- Very nonlinerar processing in the soma (body cell) and the synapse junction 
domain. This would include: Presynaptic axon terminal, postsynaptic dendrite 
spine head, and synaptic cleft with action of nearby glial cells.
 
Here, I concentrate on the "simple" conduction problem. The first attempt was 
builtto explain electrochemical transmission in the axon. Because it looks as 
an electric cable in a liquid medium, the first model was derived from the 
submarine telegraph line mathematics framework. This is the so-called cable 
model of neural transmission.
 
Next came in 1952 the work of A.L. Hodgkin, A.F. Huxley and B. Katz. This 

remains today a very good representation, unfortunately the "simple" conduction

has turned into a set of four coupled differential equations, hard to solve for
each small branch
of the neuronal tree. It is not perfect neither: For example, it don't take 
into account the activation-inactivation coupling of Na channels. Temperature 

effects are not fully taken into account and membrane channels are assumed obey
Ohm's law, not quite true in the real world.
 
The best present day "simplified" solution seems to be  the Hindmarsh-Rose 

model, first proposed in: Proc. R. Soc. London Ser.B vol.221, p. 87-102 in 1984.
This two equation system has been upgraded to a three coupled differential 
equations in later works: Proc. R. Soc. London Ser.B  vol. 225, p. 161-193 

(1985) and: vol. 237, p. 267-288 followed by a second paper on pages: 289-312 in
1989. X.J. Wang has published too a contribution in Physica D, vol. 62, p. 
263-274 in 1993.
 
It seems that model would have to be chosen for a neural simulator.
 
In most case, the action potential in the axon has no significant processing, 
it is merly data transmission. All the complexities are then "technology 

dependant" and come from the biochemical nature of the neuron. That has not to 
be 
simulated in uploading.
 
The case for the dendrite side is quite different. Here, branching induce 
some important effects on the signal transmission. A dendridic tree is 

definitively not a simple branching conductor system. The physical nature of of 
the 
signal induces some processing with collective effects on connecting branches.
 
Yvan Bozzonetti.


 Content-Type: text/html; charset="US-ASCII"

[ AUTOMATICALLY SKIPPING HTML ENCODING! ] 

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=25884