Re: Cochlear mechanics (Al Bregman )

Subject: Re: Cochlear mechanics
From:    Al Bregman  <BREGMAN(at)HEBB.PSYCH.MCGILL.CA>
Date:    Fri, 14 Jul 2000 15:18:21 -0400

Dear Harry, Stefan and list,

I did not mean to denigrate physics or physicists.

It goes without saying that a *complete* understanding of any level of
description will include its relation to other levels.  What I intended to
convey with my brief statement was that it is not necessary to relate one's
study to more microscopic levels of analysis *before*  pursuing an analysis
of the system at a higher level.  The only reason for my making such a
statement is the belief held by some researchers (not necessarily those of
this list) that theirs is the truly meaningful level of analysis and that
progress at more macroscopic levels must inevitably depend on their favorite
level.  A far-fetched example might be the statement that no real progress
can be made in the study of the human appreciation of, and skill at, music
until the cochlear mechanism is worked out in full.

My belief is that it is productive for work at all levels to go on in
parallel.  If one cannot find a lower-level explanation for a concept that
has good explanatory value at the more macroscopic level -- good in the
sense that it accounts for a lot of data -- this may indicate a deficiency
at the lower level, not necessarily at the higher one.  In this way,
progress at the higher level can actually serve as a catalyst for
better research and theorizing at lower one.  In this example, the influence
on theorizing is from higher to lower levels, rather than the reverse.

Al

-----------------------------------

Harry Erwin wrote:
I think that it can be argued that complex systems (in the sense of Rosen,
1985) exist and show emergent higher-level behavior that cannot be predicted
by simply modeling lower level processes. However, those lower level
processes remain real, and our understanding of the higher-level behavior
has to be consistent with them. If we lack valid models of those lower level
processes, we will have difficulty formulating higher level models
correctly. Or as I kept relearning during my dissertation research: