rsn_ at _comcast.net
Thu Dec 2 08:24:00 EST 2004
On Thu, 02 Dec 2004 07:57:32 GMT, "AngleWyrm"
<no_spam_anglewyrm at hotmail.com> wrote:
>"Matthew Kirkcaldie" <m.kirkcaldie at removethis.unsw.edu.au> wrote in message
>news:m.kirkcaldie-7151BA.18171102122004 at tomahawk.comms.unsw.edu.au...
>> In article <f%xrd.429771$wV.213471 at attbi_s54>,
>> "AngleWyrm" <no_spam_anglewyrm at hotmail.com> wrote:
>> Yes, nearly all neurons have a part which receives signals (dendrites),
>> which then decide whether they will fire or not. If the cell does fire
>> it sends a signal down its axon, a long tube which may be super-short
>> and contact the neighbouring neuron, or may extend all the way down the
>> spinal cord and make contact with a motor neuron in the lower back.
>> Some neurons have sensors down in the feet, and generate signals which
>> travel all the way to the neck. All of this transmission is one-way.
>> There's not really a map of what goes to where: in the brain, everything
>> connects to everything using one set of cells to go forward, and another
>> to come back. In the body, sensory fibres send signals in toward the
>> spinal cord or the brainstem, and motor fibres come from those same
>> places to make contact with muscles or glands.
>> There's a lifetime to be spent figuring it all out! Anything in
>> particular you're interested in?
>Yes, with respect to neural nets and artificial intelligence. I have seen
>many such networks, and they involve a set of input neurons, which pass
>signals to one or more layers, and then on to an output set. What I'm
>wondering is if this directedness between neuron layers is an accurate
>model. I have seen pictures of neurons, which look like a complex root
>structure of inputs, and it gets me wondering.
>If axons can be long (as in longer than just adjacent neurons), do some of
>these axons feed signals to neurons that--either directly or indirectly
>through other neurons--supply inputs to that very neuron? More technically,
>is it correct to consider the brain as containing cyclic directed graphs?
Ken Collins has already referred you to neuroanatomy text which
provide a large scale "wiring diagram" of many aspects of the brain.
Note, however, that the harder you look the more connections you see.
The patterns of connectivity shown by the neuroanatomy are just the
starting point. Matthew Kincaide has also indicated that virtually
everything is connected to everything else provided you allow multiple
An excellent resource is "Synaptic Organization of the Brain" by G.
Shepherd. It is particularly valuable in emphasizing different levels
of circuits -- microcircuits involving interacting portions of cells
(within a fraction of a mm), local circuits in a small brain region
(within 1 or 2 mm), and larger pathways and networks involving
multiple brain regions (many mm or cm). It is only the latter
circuits that are covered by the subject of neuroanatomy. If you want
to compare real brains with artificial neural nets, it is the first
two that you need to look at.
Artificial neural networks are really very cartoonish simulations of
real neural networks. Real neurons function in a much more complex
way, often using purely graded interactions without action potentials
and threshold logic. Real neurons show enormous plasticity. Real
neurons are connected in enormously complex ways. One cubic mm of
brain can contain about ten thousand cells, each one receiving input
from thousands of its neighbors and sending outputs to thousands. As
I mentioned just before, these local interactions can well occur
without action potentials. The cyclic graphs you mention (closed
loops of connectivity)) are everywhere in local and long-distance
That is not to say that artificial intelligence and neural networks
aren't valuable systems to study . It is just that they represent
only a tiny sliver of the true complexity of nervous systems.
More information about the Neur-sci