Re: Conscious of the parts

From: Mike Dougherty (msd001@gmail.com)
Date: Thu Aug 31 2006 - 19:51:51 MDT


On 8/31/06, Rik van Riel <riel@surriel.com> wrote:
>
> Tennessee Leeuwenburg wrote:
>
> > Assumption of scaling.
>
> Speaking of which, researchers are looking at how crayfish (and
> other invertebrates) can get reasonable sensory input and reactions
> with *way* fewer neurons than required by mammalian brains.

To me this suggests that maybe we will be able to produce artificial
> intelligence that is more efficient than human brains, resulting
> in a smarter intelligence even with computers having less raw
> bandwidth.

"Invertebrates such as insects and crustaceans achieve similar movement
> and sensory outcomes to humans. For example, finding food and selection
> of appropriate mates and nesting sites. Where humans use millions of
> neurons to achieve such outcomes, invertebrates do it with thousands.
> Where humans use hundreds, invertebrates may use as few as six," he says.
>

Insects or crustaceans have a different kind of scalability which affords
that kind of efficiency per unit - the redundancy of many individuals
provides for a number of them to fail without jeopardizing the continuity of
the species. I imagine that the genetic signature of millions of gnats
nearly identical, so that if even a significant number (let's just say 80%)
of them 'failed' to avoid death before reproduction, that there is still a
viable repopulation possibility without a huge loss to the evolutionary
program represented by "gnats" - If the same 80% kill rate were applied to
the human population, it would likely have a much larger impact on the
overall survival of humanity. The redundancy of a large number of low-cost
units is a different type of scaling than an expensive unit with redundant
system failsafes.

* of course I am not an evolutionary biologist, anthropologist, or qualified
scientific authority. I also do not intend to kill 80% of the human
population. That would just be rude.



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:57 MDT