bill at darwin.bns.pitt.edu
Mon Mar 10 16:04:37 EST 1997
kspencer at s.psych.uiuc.edu (Kevin Spencer) writes:
> Whatever this number turns out to be, I wouldn't put any faith in it.
> The brain isn't a computer based on binary logic. It certainly doesn't
> store information as bits.
Information theory doesn't apply only to binary logic, and the
information capacity of the brain is an important measure regardless
of whether information is stored in "bits". It is particularly
important if you are interested in how much detail is present in
memory. Suppose, for instance, somebody claims that people are
capable, when correctly tested, of remembering everything they ever
saw, just as though they were looking at it all over again. Well, the
information content of a bitmap representation of the visual world, at
the level of detail of the human visual system, is easily higher than
1 megabyte per view. Suppose you get one view per second, for ten
hours per day (both underestimates). In one year, the accumulated
information comes to about 10^13 bytes. This value, curiously, is my
own best estimate of the total information capacity of the human
brain. We see then, if the estimate is correct, that human memory
cannot possibly store experience in the form of a sequence of raw
visual images. If moment-to-moment visual experience is stored at
all, it must be in a highly compressed form.
More information about the Neur-sci