skaggs at bns.pitt.edu
Tue Mar 11 11:49:59 EST 1997
kspencer at s.psych.uiuc.edu (Kevin Spencer) writes:
> Bill and Gene, you've made some very good points. I guess what bothers me
> about statements I've seen about the "information capacity" of the brain
> is that the estimates vary a lot, because they depend upon parameters with
> very large values that vary over several orders of magnitude. The consensus
> seems to be "a lot", which isn't (to me) all that informative.
I agree. Right now we're in something like the position the ancient
Greeks were in when they first figured out that the world is round,
and tried to calculate its diameter. Some of the estimates were
amazingly accurate, but they varied wildly, and couldn't actually be
used for anything anyway. Still, I don't think you would say that the
value was not important.
> Even if you could calculate a good estimate, would that number be useful?
> As I write this I'm not coming up with any ways to use such an estimate, but
> maybe someone else has some (besides questions like "how large a hard drive
> would I need to back up my brain?" :) ).
In and of itself, the information capacity may not be tremendously
meaningful. The value will come when it can be combined with other
information. In a similar way, the capacity of a disk drive is not a
very meaningful number unless you also know how much space is occupied
by a typical text file or data file. Thus, to make sense of the
capacity of the brain, we will need to know what sort of items it
stores and how much space each one occupies.
There is, though, another more general kind of usefulness, to people
who are working on artificial intelligence. If you are trying to
imitate the functionality of a brain, it is useful to know how the
computer you are using compares with a brain in terms of information
processing capability. If it is greatly inferior, the problem may be
hopeless, but if the two are more or less equivalent, you can hope
that the problem is one of software rather than hardware.
-- Bill Skaggs
More information about the Neur-sci