Bob Mottram wrote:
2008/11/5 Richard Loosemore <[EMAIL PROTECTED]>:
At the end of the
day, if you end up with some problems in the code because you transcribed it
wrong, how would you even begin to debug it?
Brains and digital computers are very different kinds of machinery.
If I were to copy the circuits of a computer by scanning successive
slices of it optically or with a laser there would indeed by mistakes,
including registration errors due to imprecise reconstruction of the
deformed material as it was sliced. These errors, if not corrected,
would lead to catastrophic failure (or at least some quite severe
malfunction) of my copy.
Brains however are not nearly so sensitive to small errors, and in
some cases fairly extensive damage can be sustained without causing
the entire system to fail.
I don't agree that WBE (or connectomics) is a "wild goose chase", and
quite apart from the scenarios dreamed of by Moravec and Kurzweil I
think it will lead to medical and prosthetics advances. Having
chatted with some of the people doing photography of brain tissue I
think we may actually be closer to being able to reconstruct detailed
models than many people believe.
In general, it is not correct to say that "Brains ... are not nearly so
sensitive to small errors", because this kind of robustness has only
been demonstrated for such problems as gross physical damage.
You get a very different picture if you consider the (strong)
possibility that a slight change in the balance of certain
neurotransmitters can lead to such problems as epilepsy, migraine and
delusionals disorders.
Or, consider the possible role played by spike timing in dendritic
trees, which looks like it could be a very subtle, distributed effect.
For all we know, the critical layout of the afferent neurons is a result
of a developmental process that depends on subtle molecular mechanisms
going on inside the branches. In other words, the exact wiring really
matters, and that exact wiring is *maintained* by detailed adaptive
processes going on at the molecular level.
Unless the researchers doing the WBE decide to emulate right down to the
molecular level (and I do not believe they are suggesting this at any
time in the near future), we may well find that when the system is
switched on, the adaptivity of the system is compromised and the
placement of the synapses gradually starts to diverge from the layout
that makes the system work. So the system may be coherent at first and
then just disintegrate.
Now, these kinds of problems would simply never show up in a brain that
was damaged in the usual way (by gross physical damage, for example),
because that kind of damage does not go in and make strange adjustments
to the way that the mechanisms balance one another.
By contrast, if you stick some LSD chemicals into the system, it starts
to go crazy pretty quickly.
The general point I am trying to make is that we are not in a position
to make the kind of strong, optimistic statement about the robustness of
brains that you are claiming, because we do not know enough about the
kinds of "bugs" that can occur in the system, and we do know that some
subtle changes have large effects.
That lack of knowledge about the kinds of bugs we will encounter must
then be combined with the fact that, in this extraordinary case, we
would have no way to address those bugs because we would be deliberately
avoiding the problem of understanding how the system works.
Unknown bugs + no debugging techniques = disastrous project plan.
Richard Loosemore
-------------------------------------------
agi
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/
Modify Your Subscription:
https://www.listbox.com/member/?member_id=8660244&id_secret=117534816-b15a34
Powered by Listbox: http://www.listbox.com
