[Comp-neuro] Re: Attractors, variability and noise

Walter Read read at csufresno.edu
Thu Aug 21 20:51:59 CEST 2008

   Just a couple notes in response to Jim's comment.

> Obviously, the question of what we can and can't understand as  
> individuals, or groups of individuals is somewhere near the core of  
> this discussion.  Many would claim, as Brad does here, that of  
> necessity our descriptions of things (brains) can not exceed the  
> ability of humans to communicate what they understand.

      First, the issue of "communicability" isn't the same as the issue of "complexity". We 
constantly deal with systems that are too complex to allow any micro description to reliably 
predict global behavior - the Roman Empire, crop forecasts, choosing American Idol. For 
perhaps a more familiar example than Mirage jets, Windows XP ran to about 40 million lines 
of code and Vista goes closer to 65 million. (I mentioned these numbers in a class once and 
afterwards a student came up, apparently genuinely disturbed. "Really?" "Yes." "Gee, then I 
guess Bill Gates *doesn't* read every line before software is released!") But this doesn't mean 
we can't generally make policy or predictions based on our collective, imperfect 
understandings. And any particular failure can usually be traced back to a particular place. 

   Second, communicability is cultural and historical. At one time, notions like gravity or 
unbounded space or limitations on the speed of light would have been considered 
"paradoxical". We still refer metaphorically to "the fixed stars" and "sunrise" but most people 
have a basic idea of the later ideas and can imagine and think about a heliocentric world. Our 
ability to form new ways of thinking grows much more slowly than our ability to generate 
data but it does grow, in response to the efforts of those who explain the ideas or show their 
everyday usefulness, those who might be called storytellers. (Cf Galileo's popular writings 
and John Donne's early poetry.) 

   The information systems people distinguish "data" and "information". The data is 
fundamental but it has to get translated into news we can use.


Walt Read
Computer Science, MS ST 109
CSU, Fresno
Fresno, CA 93740
Email: read at csufresno.edu
Tel: 559 278 4307
        559 278 4373 (dept)
Fax: 559 278 4197


More information about the Comp-neuro mailing list