IUBio

Wavelet answer: Re: question: job of a neuron

Matt Jones jonesmat at physiology.wisc.edu
Tue Sep 11 11:16:15 EST 2001


"John H" <John at faraway.com.au> wrote in message news:<Iqhm7.1668$iH4.125149 at ozemail.com.au>...
> Thanks Richard,
> 
> I'm not in a position to judge the wavelet idea, and one of my internal
> rules is to try (!) to avoid making judgements about things I don't
> understand. Hence my final question about testing it against a simple
> nervous system. That seems to me to be the next logical step in the exercise
> and for the purposes of proving\disproving wavelets the best possible
> strategy.
> 


Yes. This is exactly the appropriate response. It's a nice sounding
collection of fancy words all strung together:

In article <005901c13565$03aca660$6d02030a at RBlue> Ron Blue,
rcb5 at msn.com writes:

>The backaction propagation wave 

>effects the wavelet modulator responses that the neuron will
emit
 

>interference patterns

> Neurons are oscillators

>The oscillation is not constant, therefore it is wavelet.


Each word in isolation means something specific, but when they're
crammed together like that, their meanings cancel each other out.
Plus, there's a lot of imprecise use of words that makes me
suspicious. For example:

"backaction propagation wave" - this appears to be a reference to
what's known as "backpropagation" in the context of neural network
training algorithms, or alternatively, "backpropagating action
potential" in the context of cellular electrophysiology. The two
meanings -might- actually be related to each other, but neither are
clear here from the context, and further, "backaction propagation
wave" is simply not a phrase used in any of the literature I'm
familiar with.

"wavelet modulator responses" - What's this supposed to mean? A
wavelet is a convolutional kernel used to expand a signal into a set
of orthonormal bases, exactly analogous to a sine wave in Fourier
analysis. This leads into signal processing language, where words like
"modulation" and "response" are common. But what's a "wavelet
modulator response", and how would a neuron "emit" one?

"interference patterns" - Interference patterns are properties of
excitable media, of which the brain is an example. But there is
probably not one single reliable experimental report of -that- sort of
"interference" being observed  in the brain. If I'm wrong, someone
please provide the citation to a peer reviewed paper.

"Neurons are oscillators." - No, most of them are not. There are some
relatively special cases of pacemaker neurons, but even in common
pacemaking circuits, most of the neurons -are not- spontaneously
oscillatory. People who work on rhythmic circuits are constantly
trying to identify pacemaker units, and often cannot find them.

 "The oscillation is not constant, therefore it is wavelet." - An
oscillation that is not constant is not a wavelet. A wavelet is a
static series of numbers. It is not an oscillation, or a standing wave
like in a guitar string.


Having said all that, I do believe that wavelets (the real kind, as
described at www.wavelet.org, and at
www.amara.com/current/wavelet.html (I love this site!)) have a place
in the study of neural function, just as Fourier analysis does. It
would not surprise me -at all- if there were parts of the brain whose
main job it is to perform wavelet transforms on incoming signals, like
the auditory system performs Fourier transforms on sounds. However,
being an elegant and cool sounding idea does not make it a scientific
fact, or even a proper theory. No amount of fancy words will do that.

Only EXPERIMENTAL DATA and/or CLEAR REASONING will do that. 

Cheers,

Matt




More information about the Neur-sci mailing list

Send comments to us at biosci-help [At] net.bio.net