Newsgroups: comp.ai.alife
Path: cantaloupe.srv.cs.cmu.edu!das-news.harvard.edu!news2.near.net!MathWorks.Com!news.kei.com!ub!galileo.cc.rochester.edu!prodigal.psych.rochester.edu!stevens
From: stevens@prodigal.psych.rochester.edu (Greg Stevens)
Subject: Re: Is this example alive?
Message-ID: <1994Sep16.042830.7666@galileo.cc.rochester.edu>
Sender: news@galileo.cc.rochester.edu
Nntp-Posting-Host: prodigal.psych.rochester.edu
Organization: University of Rochester - Rochester, New York
References: <940909.070611.4285@cheshire.cc.oxy.edu> <350lmq$b5l@harbinger.cc.monash.edu.au> <NICKB.94Sep15121319@abel.harlqn.co.uk> <1994Sep15.193954.17708@news.media.mit.edu>
Date: Fri, 16 Sep 94 04:28:30 GMT
Lines: 26

In <1994Sep15.193954.17708@news.media.mit.edu> minsky@media.mit.edu (Marvin Minsky) writes:

> [...]

>The standard models of genetic evolution do not provide ways for

>1. inheriting acquired characteristics 
>2. evolving "purposeful" mechanisms within the genome.

As far as inheriting acquired characteristics goes, that is something
that many people have played around with.  At Alife III in Santa Fe
I heard a lecture discussing the differences between simulations of
Darwinian vs. Lamarkian evolution, and how they compare as optimization
methods.

It all depends on the parameters of your genome and its relationship to
structure (phenotype).  If you're evolving neural nets, you can pass
the weight matrix as a chromosome, after it is modified by learning
algorithms through the course of its existence, thus instantiating
a Lamarkian effect.

Greg Stevens

greg@santafe.edu
stevens@prodigal.psych.rochester.edu

