[Neuroscience] Re: Looking for rat trajectories

Mathias via neur-sci%40net.bio.net (by mathiasDOTfranzius from webDELETEME.de)
Tue Feb 27 11:57:25 EST 2007


> =If you don't mind saying=, Matthaias,
> where do you do your work?
Berlin, Germany

> If it's in an academic environment,
> I'd like to come to give an informal
> talk [if I can afford to get there :-]
Thanks for the offer. I assume you want to change our "big picture" of
how the brain works (as far as we have one). I'm afraid I find those
parts of your ideas I found on ths group rather hard to understand and
I'm reluctant to invest a lot of time in understanding them. Please
forgive me my ignorance, but (apart from this discussion) I'd rather
invest my time for finishing my PhD.

> I can help you [and your 'team] with
> everything you're interested in ac-
> complishing.
Well that is one big claim! You see, such big claims raise suspicions ;)
[...]

> I can help you get a handle on the
> 'complexity', for instance.
> 
> Nervous systems are amazingly-
> Simply-organize.
> 
> When folks 'see complexity', that's
> as a function of absence-of-under-
> standing of the True Simplicity of
> nervous-system organization. [Folks
> implement stuff in round-about ways
> that are analogous to the ways in
> which ancient Astronomers calc-
> ulated in ways that included non-
> physically-real "epicycles", which
> rapidly becomes too-'complex' to
> allow further calculation. When-
> ever 'complexity' shows-up, it's
> =always= the case that that is be-
> cause folks just aren't seeing
> physical reality. Physical reality
> is Simply-organized.
In the modelling domain complexity can be rather strictly defined in 
terms of the function space used to map input variables to output 
variables. If a model explains more of the output variables' variance
using a smaller function space than another model, you can call it 
better. This is just an application of Occam's razor.
So if you provide such a model which operates on the same input data as 
mine and has a simpler structure, I should discard mine.

> This is the stuff that I'd address
> in my informal 'talk', if such op-
> portunity is possible.
> 
> I'll give you a lot.
> 
> It'll require you to reconstruct your
> model.
> 
> But your model will work way-
> beyond your existing hopes for it.]
I'm afraid I'm suspicous again against such a bold claim. But the nice 
thing about science is that you can take my theory and improve it, or 
compare it to yours and show its superiority. I'll let you know when the 
manuscript is published on a preprint server within the next weeks.

> [I understand that it's an 'unusual'
> request, but, "Nothing ventured,
> nothing gained"] [Speaking from
> my own perspective, but, perhaps
> you'll find it so, too :-]

> ken [k. p. collins] 
> 
> 


More information about the Neur-sci mailing list