Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!ucsd!sdcsvax!beowulf!demers
From: demers@beowulf.ucsd.edu (David E Demers)
Newsgroups: comp.ai
Subject: Re: evolution of intelligence
Message-ID: <6922@sdcsvax.UCSD.Edu>
Date: 5 Aug 89 19:30:12 GMT
References: <2153@hub.UUCP> <19229@vax5.CIT.CORNELL.EDU>
Sender: nobody@sdcsvax.UCSD.Edu
Reply-To: demers@beowulf.UCSD.EDU (David E Demers)
Organization: EE/CS Dept. U.C. San Diego
Lines: 68

In article <19229@vax5.CIT.CORNELL.EDU> b27y@vax5.cit.cornell.edu (Michael Gray) writes:
>	Isn't it just a little far fetched to think that genetic 
>"advancements" like intellegence happened using only the mechinisms of
>random selection of genes and survival of the fittest.  

No.  While the odds of ONE useful adaptation occurring in any
particular reproductive event may be small, there are literally
billions of these events.  Once anything develops through
genetic recombination which provides enhanced fitness, that
individual is more likely to reproduce and pass the characteristic(s)
on.  It doesn't take many generations for the trait to sweep
through the population, even if it only provides a very slight
survival/fitness edge to the holder.   Think of an analogy with
compound interest...

>Before anyone worries: I am not arguing for a higher power (God etc.)
>I just think that those two mechinisms don't seem to be enough..  I realize
>that the evidence doesn't stack up much in my favor.

Well, it's mostly trying to understand the statistics.  I don't
think humans generally are able to grasp the effect of many 
"experiments" over a long period of time.  Remember, it has
taken a LONG time to get to where we are today.

>questions:
>	What about "reverse" evolution?  Some researchers (I do not remember
>all the details) I recall seem to have found evidence of speices which
>evolved attributes which made survival HARDER.  More specifically: A type of
>pike, whose teeth where evovling so that they stick outwards, which made
>cating and eating prey MORE difficult?  This deosn't seem possible under the
>orignal Dawrininan discriptions.

I suspect that there is some explanation based on increasing fitness.
Survival is not the only thing involved.  Species need to 
optimize reproductive ability.  For example, pheasants' tails do
not seem to assist in survival, but DO assist in attracting a
mate.  But I don't know anything about the fish you mention...


>	This of course implies to me that there may be something more to
>evolution, some other factors influencing how we reproduce.  The only
>preminition that I have had is that the mechinism of mutation may not be as
>random as it seems.

Evidence seems to be that most mutations are harmful, and that
mutation rates are fairly low.  I don't think that there is
much evidence for non-random mutations.  Anyone know?

>Is there any evidence which supports this? or am I just opening myself up
>to net.flame as heretic of the month. I am also interested in results of
>Genetic algorithms which may contadict me.  
>--misha--

While dabbling in Genetic Algorithms, I have been impressed by the
robustness.  I've used GA to search very large discrete space
and almost always can find a global optimum.  There is a good
book by David (oops, last name is Goldman or Goldberg...) on
genetic algorithms, including lots of code fragments.  

As to evolution in general, you might read "The Blind Watchmaker"
by Richard Dawkins.  It strongly advocates his ideas and
shows how the seemingly improbable "random combinations &
mutations" are not so unlikely after all.  The key idea is
that most benefits come from combining already existing
"subsystems", rather than from mutations.

Dave DeMers
demers@cs.ucsd.edu