On Thu, Jul 07, 2005 at 04:51:23PM +1000, Stathis Papaioannou wrote:

> I have no problem with the idea that everything about a person's 
> personality, memories etc. is physically encoded in his brain, and that in 
> principle, sufficiently detailed knowledge about his brain should allow an 
> emulation on a computer which would be just like the original person. The 

Fair enough. You seem to suddenly deviate from this position at some point
below, though.

> problems are:
> (1) what is the level of detail of neuronal information required;

It doesn't matter, if that information is present in the vitrified brain.
Preliminary results look good http://leitl.org/docs/cryo/
More results will be forthcoming in the next 1-2 years (this is something I
know, not guess).

> (2) can this requisite information be preserved in a post-mortem specimen;

See above. No showstoppers, under optimal conditions.

> (3) can the information be scanned or read in a way that can be used in a 
> computer model;

Yes, though TEM is probably not sufficient. Scaled up cryo AFM has more than 
resolution, and allows individual sampling of ablated molecules. In times
where CO molecules are individually sorted by the isotopes by numerical 
control, and assembled into elaborate circuits, this shouldn't require much 

> (4) can each subsystem of neuronal function relevant to cognition be 
> modelled closely enough to allow emulation;

This is the most difficult point: you have to build a system which can
abstract models, building at least 2-3 hierarchies, until you arrive at an
isofunctional model well mapped to the hardware used.

I have ideas in that direction, but nothing has been tested yet.

> (5) given adequate information and adequate models, is the computer power 
> available up to the task of emulation in anything like real time?

Near future will give us systems built from moles of bits (by self-assembly
of individual molecular circuits). Pretty speedy
systems, enough for a speedup of 10^6, if not more. The difficulty lies in
obtaining a model which is isofunctional to the original. By the time you
have that model, hardware will not be a bottleneck. 

> I believe the level of detail required and the complexity of the required 
> models is grossly underestimated. Simply getting a 3D image of a brain down 

No offense, but given the level of your ignorance, how do you know who 
has estimated what? 

> to electron microscopic detail, including all the synaptic connections, 
> would be an enormous task, and it probabaly wouldn't tell us any more about 

Yes. You need more resolution than TEM, btw. That's what automation is for.

> the mind of the brain's owner than a picture of the books on a library 
> shelf would tell us about the book contents. I would bet more on mediaeval 

I told you we have results that there's probably enough preserved for the
tissue to be retransplantable(!). Once it's in the dewar, time stops. I told
you we have current methods allowing you to resolve submolecular structures
in cryopreserved tissue. There's no fundamental reason why you can't image
kg-sized vitrified objects at atomic resolution, at least transiently where
it matters (what is this transmembrane protein, and how is it been modified, 
the membrane itself is rather boring). This is easily verified even with only 
online information.

The information is preserved in the structure. Are you claiming that there is
something missing? What, precisely, then?

> monks decoding the data on a DVD sent back in time than I would bet on 
> scientists decoding the contents of a human mind from cryopreserved brain 
> sections.

You'll see individually accurate numerical models of simple critters +
virtual models within the 20-30 year time frame. We could do this now with C.
elegans, given 5-10 years, and a considerable budget.

> If mind uploads were to become a reality, I think the best strategy would 
> be research into brain-computer interfacing.

What is your estimated time frame for advent of technology like

I, personally, am not holding my breath. YMMV.

Eugen* Leitl <a href="http://leitl.org";>leitl</a>
ICBM: 48.07100, 11.36820            http://www.leitl.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE

Attachment: signature.asc
Description: Digital signature

Reply via email to