Re: Invariant Recognition, Grandmother Cell, and Memory Hierarchy

"dan michaels" <feedbackdroids@xxxxxxxxx> wrote:
> Curt Welch wrote:

> Regards how and why these hierarchicall trees form over time, that's
> clear. Genetics. You can't get this level of complexity in the brain
> without it being specified by the genome for this to happen. That
> Linsker stuff that you and the behaviorists were so enamored of a year
> of so ago is useless to create
> this level of hierarchical complexity. Not a chance. So, then you want
> to ask why it is that the genome specifies that this should be so. I've
> addressed his in prevous posts to this thread - but I may be the only
> person in the
> world who sees it this way. Ce la vie.

Even though I do lean more to the idea of nuture over nature than you do, I
don't doubt for a second that huge amounts of the structure is directly
specified geneticaly.

What I constantly say however is that the genetic constructed parts aren't
very interesting or useful to the AI problem.

The AI problem requires us to understand how we learn all the things which
can't be genetical specified - like all the parts of speaking English which
can't be genetically specified. We might have langauge hardware in the
genes, but we don't have English in the genes. How do we get English into
our heads?

How do we learn to drive a car? Car diving is not genetically coded, so
what type of hardware is produced by the genes that can transform itself
into a car controller?

If the visual cortex is all hard wired geneticaly and is not a learning
machine, then it's of almost no use in understanding AI.

--
Curt Welch http://CurtWelch.Com/
curt@xxxxxxxx http://NewsReader.Com/
.

Relevant Pages

  • The Future wont look like the present
    ... with genetics. ... billion years ago the highly complicated molecule DNA had emerged. ... increase in complexity. ... At the moment computers have an advantage of speed, ...
    (sci.bio.evolution)
  • Re: Predictions "by evolutionary theory"?
    ... indistinguishable from Behe's "irreducible complexity". ... Genetics 3:422-499. ... breathe a word about ALL the characters (in what? ... including an ignorance of what "irreducible complexity" means. ...
    (talk.origins)
  • Re: Characterizing complexity
    ... >>Richard Dawkins mentioned an idea to characterize complexity in his ... relative complexities of those organisms. ... > * Organism complexity as specified by genetics. ...
    (sci.bio.evolution)
  • Re: Characterizing complexity
    ... > Richard Dawkins mentioned an idea to characterize complexity in his ... > phrase 'repeat N times.' ... Organism complexity as specified by genetics. ...
    (sci.bio.evolution)
  • Re: Stephen Wolfram vs. Charles Darwin on natural selection
    ... via a developmental program that is mostly specified by genetic factors. ... So Kolmogorov complexity would seem to be a natural measure. ... within population genetics synthetic models. ... reversible nested sets of genetic factors specify complexity and not ...
    (sci.bio.evolution)