🚀 go-pugleaf

In <37mmeb$95@news.u.washington.edu> forbis@cac.washington.edu (Gary Forbis ) writes:
>In article <37mgkc$454@mp.cs.niu.edu>, rickert@cs.niu.edu (Neil Rickert) writes:

>|>                       There is a question of fact, as to whether the
>|> brain actually uses analog methods.  I would not be at all surprised
>|> if the answer is yes.  I was discussing the different question, which
>|> is a matter of principle.  Some people are arguing that, in
>|> principle, cognition must be analog, and cannot be digital.  It is
>|> this "in principle" argument which I suggested was bogus.

>I view the question as an emperical question.

>Can we produce a digital device about the size of the human brain which will
>have behaviors consistent with human cognition?

This is probably a bad criterion.  The brain is a device the size of
a brain.  Does the brain have behavior consistent with human
cognition?  The problem here is that we do not examine the behavior
of the brain -- we examine the behavior of the body.  If we separated
the brain from the body, and finished up with a brain-in-a-vat
scenario, how would we determine whether that brain-in-a-vat had the
right behavior?

In any case, whether or not it is a bad criterion, we are not
currently close to meeting it.

>I suspect that if the answer is yes we will accept that the device has 
>cognitive capabilities.

>At increaing levels of percision the digital device may be harder to identify,
>but judicious may continue to identify the digital device that models above
>the level of thermal noise (maybe not, that's the question.)

>Can a digital device model an analog device to the level of thermal noise
>in no more space than the analog device?

The experience of techology favors digital devices.  The compact disk
holds more than a larger analog disk.  The DAT holds more than the
analog cassette.  Telephone companies run 2 channels at 64kbits/sec
over the same line used for a 3 kHz audio telephone signal.  You
mention the problem of noise, but it is often because of the noise
that digital does better.  The digital signal need only be slightly
above the noise level to allow signal processing systems to extract
the correct information.  Analog signals at only slightly above the
noise level lose most of there information.  It is this digital
resistance to noise which allows higher densities for digital
systems.