Re: Neuron Computational Requirements?

From: Robert Bradbury (
Date: Thu Apr 20 2000 - 21:03:40 MDT

On Thu, 20 Apr 2000, Robin Hanson wrote:

I'll take a swing at a couple of these:

> 1) Neuron Inventory -- The number of distinct types of neurons, and
> the number of each type in a typical human brain.

I'm not sure how many types there are currently, but you can bet it
is wrong. As they are finding out with cancer, it can look like a
duck and quack like a duck, but there are unrecognizable differences
between individual duck-twins. The only way to do this is
through careful mRNA abundance statistical analysis on a cell by cell
basis. We are just at the start of being able to do this (by using
lasers to destroy surrounding cells, so you can amplify the mRNA
in the single remaining cell up to the detection level). Doing this on
frozen or cryo-preserved samples could be very misleading due to
the stresses that occur prior to the sample being in stasis.
Over the next 10 years or so, I suspect you will see the number of
cell types climb. I have "heard" [no reference] that ~25% of the
genome is expressed in the brain. This may be due to the difficulty
of separating out different cell types (neuronal, glial, endothelial,
etc.) from each other or it may be due to the fact that the brain is
is simply a *very* complex organ.

> 3) Resolution required -- The spatial/chemical resolution of a scanning
> technology that would be required to sufficiently reliably distinguish
> between the types of neurons, and to distinguish which neurons connect
> to which others via which kinds of synapses. (Assume one is repeated
> slicing and scanning a cryogenically frozen brain.)
If the neurons have unique surface receptors, you can do this today
using labeled antibodies. The problem is that you need to know
the information in (1) to know whether surface markers constitute
unique sets or whether cells with different internal operations
appear the same on the outside. If you are slicing through cell
interiors (as you describe), then you might be able to label internal
molecules in a similar fashion and get a better understanding of the
cell type.

I'm not sure if you asked this question, but I think today's current
scans are quite a way from the resolution I think is required.
You could find out what the slice resolution was from the Brain Anatomy
(Visible Brain?) project at NIH and then check cell sizes in Nanomedicine.
I think you will find that the slices are much thicker than multiple cell
layers and way above axon thicknesses (necessary to trace linkages).
I don't envision any technologies improving on that much until
real nanotech comes along unless someone gets really excited about
applying MEMS micro-scalpels to the brain.


This archive was generated by hypermail 2b29 : Thu Jul 27 2000 - 14:09:39 MDT