Re: >H soft incremental uploads

From: Eugene Leitl (Eugene.Leitl@lrz.uni-muenchen.de)
Date: Mon Aug 12 1996 - 09:44:53 MDT


On Fri, 9 Aug 1996, Eric Watt Forste wrote:

> Transhuman Mailing List
>
> At 2:28 AM 8/9/96, Eugene Leitl wrote:
> >As low-artefact brain cryopreservation is an absolute prerequisite for
> >uploading (soft incremental uploads are a myth, imo) I am perfectly happy
> >with the Prometheus project.
>
> What about the near-future use of cognitive prostheses? I mean neural-net
> chips that are trained to mimic functions of very small pieces of human
> brain tissue that might be used to cure the kind of aphasias and other

If the lesion area is very small and the function itself has been mapped
extensively or can be easily learned afterwards I see no problem with
this approach.

Note however: current (and mid-future) semiconductor integration
density is insufficient for the substitution of but the most puniest
tasks the brain routinely accomplishes, heat load on tissue will be
considerable and the microelectrodes are certainly not bioinert enough nor
their long-term impacts known.

This might be mostly useful in repairing spinal cords lesions, imo. There
seems to be virtually no personal differences in the spinal cord, just
basic reflexes.

> cognitive deficits that small ischemic strokes and other brain injuries can
> cause. I think economic demand within the health-care industry will lead to

I think economic pressure will merely tend to pronounce the patients to
be incurable... Already hitech medicine inflicts a heavy, soon likely to
grow intolerably high financial drain upon the populace/medical insurance
of leading industrial countries. As the age histo distribution grows
increasingly skewed & skeweder, medical costs are bound to explode (and
they do).

> heavy funding of this technology path, and once the engineering research to
> develop silicon cognitive prostheses is well underway, the same
> technologies can be used to enhance and expand "normal" function.

Silicon photolitho 2d circuitry is totally insuitable for brain
augmentation emplants, both in terms of integration density (number of
neurons substituted) and computational power (since you can't go totally
maspar because of low physical connectivity).

This asks for 3d quantum dot arrays/molecular circuitry + advanced active
cooling which ain't yet exactly there.

> Once people start using silicon neural-nets to expand and enhance normal
> brain function, one can imagine a smooth path to a point in which one's

1) I doubt it will be neural nets, more likely we will see coupling of
   brittle computer algebra packages (Mathematica, MatLab, etc.) first.
   bioNN computational prowess is pretty awesome -- we'll need some time
   to approach this order of magnitude. Complementary skills.

2) It is _not_ smooth, as additional circuitry installation is grainy --
   you plug in new modules

I once thought one might design a protein-based molecular circuitry,
which induces a soft upload by "infecting" the brain, autoreplicating,
scanning neurons and substituting them for crystalline protein molecular
circuitry to emulate the orginal neuron functionality.

Alas, apart from more trivial reasons, the integration density
achievable with soft nanotech does not permit scanning/substituting at this
level. If _at all_, it might be only feasible with strong nanotech
(Drexlerian nanoassemblers), which' feasibility has not been shown
sufficiently yet. (Though speculative technology can be fun, we ought not
to cross over in the realm of the improbable or even mythical).

If this fine-grain incremental upload is infeasible, merely beckoning to
the surviving set of strange attractors constituting your personality
with virgin circuitry will depend on their evolution-shaped propensity to
cross over into new territory. Since this never happens in real life (though
lesion tolerance & recovery is based on semi-dormant, redundant circuitry
taking over the part of vanquished one), it is not likely to work.
(However, the faint possibility that it might be possible should be
sufficient to spawn off an research task force to shed some light on this
difficult subject).

> "identity" has gradually moved "into" the silicon prostheses, because

I doubt identity has the properties of a liquid. (If it had, it would be
extremely nice of the evolution to provide us with it without
evolutionary fitness requiring such a useful trait).

> eventually there would simply be more and faster computational power on the
> prosthesis side. The sorts of computational structures that neural networks
> support (minds, in this case) can be extremely protean and flexible.

Alas, I have grown to distrust this Proteus guy (... you never know who he'll
be next ;). Most neuroscientists I know are highly sceptical about
unbound malleability of cerebral circuitry. If incremental soft uploads are
possible at all, they won't come before Singularity -- and that will be too
late. Contrary to Vinge's headbands, I doubt we'll see much of it prior to
Singularity, when every single rule is likely to be broken -- including
the necessity to rely on headbands (or heads).
 
> The logical end result of this technology path would be something similar
> to Greg Egan's "jewel" or Ndoli device. I don't think the possibility of
> soft incremental uploads can be so easily dismissed until after we learn

I am sorry if I should have given the impression of my dismissal as easy,
I should have given arguments.

> more about the way neural networks (both artificial and natural) work.
>
> Eric Watt Forste <arkuat@pobox.com> http://www.c2.org/~arkuat/

'gene
_______________________________________________________________________
| ui22204@sunmail.lrz-muenchen.de | cryonics, nanotechnology, |
| Eugene.Leitl@uni-muenchen.de | >H transhumanism, [...] |
| c438@org.chemie.uni-muenchen.de | "deus ex machina, v.0.0.alpha"|
| http://www.lrz-muenchen.de/~ui22204 | >H: "alpha-->omega" |



This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:35:43 MST