james at teoth.fsnet.co.uk (James Teo) wrote in message news:<3c4575bd.7088104 at news.freeserve.net>...
> On 15 Jan 2002 15:05:37 -0800, mats_trash at hotmail.com (mat) wrote:
> >what are people's views on the darwinian and computational approaches
> >to explaining the mind? Evolutionary psychology is very popular these
> >days, with the way our minds work explained in terms of how they
> >provided survival advantage over the eons. Computational theories
> >suggest that there is a syntax or language of thought using which our
> >minds can extract meaning an construct ideas about the world. Usually
> >these two approaches are viewed as approximately the same. Evolution
> >has honed the computational 'modules' of our mind. I just wonder what
> >other people's views are on this.
>> Well, my 2 pence worth.
> I wouldn't call them "approximately the same" since the methodology
> and level of investigation is very different.
> - Evolutionary psychology tends to ask "Why do human beings think or
> act like that?"
> - Computational theories tends to ask "What are the modules which make
> up the mind and how do they do what they do?"
> They seem to very different in that context. Not contradicting but not
> really overlapping either since the questions being asked are
> qualitatively different. I think a lot of evolutionary psychology is
> hand-waving and often of dubious merit though.
The reason I suggest that they are similar is that many of the most
prominent theorist currently regard them as such - the mind as Turing
machine etc. For example Steven Pinker, Plotkin, Dennett, Fodor (with
reservations) and almost any AI people. The way the theories are
brought together is by the darwinian selection of the computational
modules that makeup brain/mind. Those 'syntax parsers' that served us
well were propagated. Similarly in evolutionary psychology the main
claim is that our actions can be explained by the way they must have
been advantageous (in the past or now). Since the evoultionary
psychology must suppose that cognitive function is highly divisible
(otherwise the whole lot was selected for at once, and evolution works
on defined and distinct attributes) it has to postulate distinct
modules. And since these modules must operate a well defined function
then it *must* be able to be modelled computationally. I don't think
the modularity thesis is all that attractive, which would negate both!