What the Neocortex Does

Kevin K. KK at _._
Fri Aug 4 17:25:49 EST 2000



Harry Erwin wrote:

<snip>

My main problem with your view is the mismatch between the incredible
power and complexity you claim for nerves/brains, versus the simplicity
of the overt behavior of the organisms which have them. I don't see
anything particularly complicated about the behavior of bats (or most
people, for that matter!). Eating, defecation, regulation of the
heartbeat, flying... Behaviors like these can be easily copied by simple
programs. Even the ultimate bat achievement -- capturing prey in flight
through echoranging -- is not that difficult. Military radar for
identifying and tracking multiple targets performs essentially the same
function, using digital computers.

The situation is even worse with more primitive organisms. If the brain
is such an incredibly complex (indeed, unsimulatable) computing device,
then why is it wasted on dumb finite-automata style tasks like opening a
gill flap? It may not be possible to exactly simulate the brain of a
rabbit with a digital computer (just as it is impossible to exactly
simulate the behavior or any real system), but it is very easy to
simulate the *behavior* of a rabbit with a digital computer. Rabbits
have an extremely limited repertoire (running, standing, digging,
grooming, mating, eating, fleeing, fighting etc.) which they repeat over
and over. So where is all the brain power going? According to your view,
these animals are endowed with an infinite state computer, so why can't
it outperform (say) a 486 with an 8MB hard disk? Why don't we use rabbit
brains as archival storage devices?

It strikes me as very bombastic to claim that an animal which has a hard
time learning "yes/no" is somehow going to be the model for an abstract
machine with greater power than a Turing machine. Bats are mysterious
animals, but not that mysterious. You might just as well say that a rat
brain is superior in processing power to all the linked computers of the
Internet (which, after all, is a mere finite state machine).

Church's thesis is very strong and time-tested, and your comment about
it is what led me to post. To disprove it, you must produce a machine
which outperforms a TM like a TM outperforms a finite automata or a
push-down stack. So far, I haven't seen anything from you in that area
except obfuscation.

Kevin K.






More information about the Neur-sci mailing list