The Principle that Orders Everything within the CNS

Jason Eriksen jerikse at bsd.meddean.luc.edu
Wed Oct 16 23:10:14 EST 1996


In article <32657838.45E2 at postoffice.worldnet.att.net>, kenneth paul collins 
<KPCollins at postoffice.worldnet.att.net> wrote:
>A single principle organizes everything that occurs 
>within our CNSs, including their developmental wiring 
>up...
>
>....our nervous systems are physically structured, 
>throughout their entire extents, so that their 
>functioning will automatically "seek" to achieve a single 
>goal... the minimization of the topologically-distributed 
>ratios of excitation to inhibition that are occuring 
>within them...

Kenneth, since you were so kind to put your thoughts into electronic format, 
I'd appreciate it if you could elaborate your thoughts on these subjects. I'm 
not questioning your basic ideas, I just want to know a little more about the 
following:

(1) Could you further explain the principle and experimental evidence (beyond 
the example you have given of pain avoidance) which supports your hypothesis
of this neural "see-saw" kind of wiring? From my understanding, you are
saying that the cortex and other structures (anything which is presumably
not hardwired) is constantly remapping itself in response to environmental
stimuli in order to achieve a sort of threshold. But I'm not exactly clear
on *how* such a structure would work in dealing with non-linearly separable 
problems because it seems to me, in my ignorance of your theory, 
that what you are envisioning is that at each level of cortex where 
reorganization occurs, there is an algorithm that tells each neuron
to remap its responses (via second messenger systems, protein production, ?) 
to the various synapses in order to minimalize the neuron's excitability. 

However, such a model seems to me to have the structure of a one or two 
layered neural network, which Minsky et al. showed in the 1960s (as I'm sure 
you're familiar with) that this class of network is not capable of resolving 
non-linearly separable problems. Only three-layered networks are capable of 
such resolutions, and from my limited understanding of neural networks, middle 
layers usually do not have inputs with excitatory and inhibitory ratios around 
1:1. How does your system solve this problem under your model?

To put it simply, if we were to look at an individual neuron in the CNS, how 
would you model it with respect to this minimalization procedure? What are the 
(hypothesized) mechanisms that are used in this system? 

>....thus, the topography that is correlated with the 
>greatest degree of noxious stimulation receives the least 
>post-inversion excitation... this means that the effector 
>activations that were responsible for moving the body 
>into contact with the environmental source of noxious 
>stimulation will now be most inhibited... thus, an 
>"arrest reaction" will occur with respect to motion in 
>the direction of the environmental source of the noxious 
>stimulation...

(2) I'm not an expert on the spinal cord by any means, but I thought many of 
these reflexes were mediated at the spinal cord through mono-synaptic synapses 
rather than this rather complicated-sounding scheme, which seems like it would 
involve multiple neural networks. 

Your hypothesis reminds me of the neuroscience concept of 
motion-representation "force fields," where complex neural inputs  integrate 
motion with body position to yield the appropriate force vectors for limb 
movement, eye positioning, etc.

I look forward to your comments, Kenneth. Thanks!

Sincerely,
Jason Eriksen



More information about the Neur-sci mailing list