class vs. instance variables

Joe Strout jstrout at ucsd.edu
Wed Jul 12 14:34:55 EST 1995


In this post, I propose a distinction between class variables and
instance variables in describing the attributes of neurons.  I hope that
you will be so kind as to offer feedback on this approach.

We think of cell classes as categories defined by a set of common
attributes.  Cerebellar Purkinjie cells, for example, are easily
identified by their large somas, fan-shaped dendritic tree, inhibitory
neurotransmitters, position in the cerebellar cortex, and so on.  These
attributes, which are relatively constant for a class of cells, we can
call class variables.  Other important class variables would include the
presence of various ion channels, metabolic reactions (response to NO,
for example), and so on.

Instance variables are those attributes which vary among members of a
class.  The detailed morphology is an obvious instance variable.  Others
may include the density of various ion channels, the efficacy of each
synapse, etc.

As far as I can tell, almost all work in the past has concentrated on
class variables.  A few papers have taken the detailed morphology from a
particular cell (instance), but then used generalizations to fill in
other instance variables.  This is quite understandable, as obtaining
values for all the instance variables of a cell would be quite difficult.

Of what value is this distinction, then?  In the long term, I'm
interested in neuron emulation -- producing functional models of actual,
individual cells and small neural circuits.  This is in contrast to
modeling a particular class of cell (which is quite common nowadays). 
This would be most easily accomplished by first finding values for all
the class variables.  A particular instance, then, would be characterized
merely by measuring the instance variables.  For this reason, it is
important to determine what the instance variables are, and how they may
be most efficiently measured.

Your comments, on the approach in general or on some class or instance
variables in particular, would be much appreciated.

,------------------------------------------------------------------.
|    Joseph J. Strout           Department of Neuroscience, UCSD   |
|    jstrout at ucsd.edu           http://sdcc3.ucsd.edu/~jstrout/    |
`------------------------------------------------------------------'



More information about the Neur-sci mailing list