Hi Rob,
--- Rob D <8rjd1@qlink.queensu.ca> wrote:
> I think some pretty solid evidence against that model is aging and
> brain
> deterioration. If one cell recognized each thing people would start
> not
> being able to recognize individual things as those brain cells aged
> and
> died. Instead, when people age it is a systematic deterioration as
> many
> things get fuzzy, rather than at age 40 not being able to recognize
> cats and
> only cats.
> To me this is evidence that one cell isn't solely responsible for
> high
> level concepts such as grandmothers.
> Rob
Yes, you're absolutely right. I don't believe in the grandmotherneuron
theory either (although the Schamujuk-DiCarlo model I mentioned, which
is a fairly recent model intended to explain findings in the animal
laboratory research, still uses these kind of representations!). But I
don't believe in the distributed theory as well, at least not in the way
currently formulated. My understanding of parallel distributed neural
networks, is that many neurons represent one thing. If one neuron dies
because of old age, than the other neurons are still able to represent
something like a grandmother, although slightly less perfect. But
mathematically, it's the same kind of representation. After the right
input, the representation of your grandmother will 'light' or 'fire',
saying it's 'on'. One advantage of the distributed model, indeed is that
the representations are not as brittle as in the grandmotherneuron
theory. Another advantage, is that other old people besides your
grandmother can be presented by at least a portion of the same neurons
representing your grandmother. That is, a network can be trained to
respond in the same way to 'old people' as opposed to responding in
another way to 'young people'.
That's my current understanding of distributed networks. But I hope
there's someone out there reading this who has a better understanding of
these things, and can better explain these matters...
With friendly greetings, Patrick.
__________________________________________________
Do You Yahoo!?
Try FREE Yahoo! Mail - the world's greatest free email!
http://mail.yahoo.com/
MOQ.ORG - http://www.moq.org
Mail Archive - http://alt.venus.co.uk/hypermail/moq_discuss/
MD Queries - horse@darkstar.uk.net
To unsubscribe from moq_discuss follow the instructions at:
http://www.moq.org/md/subscribe.html
This archive was generated by hypermail 2b30 : Sat Aug 17 2002 - 16:01:57 BST