Zipf's law and cortical areas
Thu Aug 3 22:51:53 EST 1995
It is commonly thought that the cerebral cortex is composed of
columns, or modules. Each module processes some aspect of a receptive
field, such as orientation or movement direction. Groups of modules
form functional areas that display the entire represented field. The
features represented can become abstract as one moves away from the
primary field, and the receptive fields can become larger so that the
functional areas become smaller.
My question is: what is the relationship between the size of the cortex
as a whole and the distribution of size of functional areas? Can it be
expected to follow Zipf's law*?
This question is prompted by the following. I study the mustached bat
auditory cortex, which is quite small (about 13 sq mm). Despite its
size, it contains more thandiscrete areas, some of them specialized
to process biosonar signals. I have evidence from corticocortical
connections for a columnar organization in this cortex (i.e., labeled
patches about 0.5 mm wide). However, some of the areas are so small
that they may be composed of as few as 1 column.
I therefore wonder, do the larger cortices of more commonly studied
animals contain extremely small areas that process some specific signal
attribute? If so, the potential for a very large number of areas would
Yet, it is clear that in larger brains there are some areas that are
large, much larger than any in the mustached bat. If the size of the
largest areas scaled with brain and body size, while the size of the
smallest areas stayed constant at one or a few columns, could
predictions be made about how many areas are likely to exist as a
function of brain size? Has this been done?
I would appreciate an education on these issues, particularly as to how
or if Zipfs law could apply here.
E-mail or post is fine
* As I understand it, Zipfs law is a purely empirical observation that
many things in biological systems scale according to the rule that as
larger maximum values occur then more and more intermediate and small
values also arise, to fill in the distribution. Is this description
close? Are there equations with parameters?
More information about the Neur-sci