the liver and the brain

dan michaels feedbackdroids at yahoo.com
Mon Sep 6 11:49:36 EST 2004


Wolf Kirchmeir <wwolfkir at sympatico.ca> wrote in message news:<b0O_c.3090$lP4.238270 at news20.bellglobal.com>...
> Lester Zick wrote:
> 
> > On 5 Sep 2004 10:11:48 -0700, feedbackdroids at yahoo.com (dan michaels)
> > in comp.ai.philosophy wrote:
>  [...]
> >>In case you didn't understand quite what I was getting at ... how does
> >>the new-born baby zebra know what its mother is?
> > 
> > 
> > I don't think it knows what its mother is. But don't forget the
> > problem is bidirectional. The mother is also figuring out who its
> > offspring is and if bound to its offspring through maternal instinct
> > provides offspring with considerable help knowing who its mother is.
> > I think an offspring knows where its food comes from, who licks it,
> > cleans it up, who's looking after it. Maybe there's something more to
> > mother-offspring bonding, possibly genetic or an instinctive bond of
> > some kind that isn't obvious. But I doubt there's comparable
> > instinctive knowledge of what lions are. If a precocial offspring can
> > figure out who its mother is and vice versa, chances are it can avoid
> > considerably more danger following its mother than trying to know the
> > myriad potential predators and dangers in objective terms and trying
> > to avoid them individually and collectively. It's a question of
> > numbers in addition to subjective knowledge of mother-offspring bonds.
> > 
> > Regards - Lester
> 
> You've obviously never lived on a farm.
> 
> "Subjective knowledge of mother-offspring bonds" indeed! ROTFL.


Hmmm, would a baby zebra separated from its mother not know at all
what to do? Once the M-O bond is broken by distance, the baby is a
goner.

Well, I can't tell you what's in the mind of a baby zebra, but I can
tell you that, when you train NN's, that it pays to train them using
*both* correct patterns [true positives] and also using incorrect
patterns [true negatives], otherwise they never learn to discriminate
well.



More information about the Neur-sci mailing list