Newsgroups: comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news3.near.net!noc.near.net!paperboy.wellfleet.com!news-feed-1.peachnet.edu!usenet.eel.ufl.edu!news.mathworks.com!udel!gatech!howland.reston.ans.net!vixen.cso.uiuc.edu!sdd.hp.com!hplabs!hplntx!curry
From: curry@hpl.hp.com (Bo Curry)
Subject: Re: What's innate? (Was Re: Artificial Neural Networks and Cognition
Sender: news@hpl.hp.com (HPLabs Usenet Login)
Message-ID: <D3r01o.7AA@hpl.hp.com>
Date: Thu, 9 Feb 1995 19:41:48 GMT
References: <3gtu3i$rf3@mp.cs.niu.edu> <D3GBKw.F5D@spss.com> <3guoku$bci@mp.cs.niu.edu> <D3L79A.IJ4@cogsci.ed.ac.uk> <3h66fb$c85@mp.cs.niu.edu> <D3n9IJ.GBz@hpl.hp.com> <3h8qjn$ovj@mp.cs.niu.edu> <D3p5K4.F8v@hpl.hp.com> <3hbh7i$f2u@mp.cs.niu.edu>
Nntp-Posting-Host: saiph.hpl.hp.com
Organization: Hewlett-Packard Laboratories, Palo Alto, CA
X-Newsreader: TIN [version 1.2 PL2]
Lines: 49

: In <D3p5K4.F8v@hpl.hp.com> curry@hpl.hp.com (Bo Curry) writes:
: >This is true. But to complete a functionalist argument such
: >as it seems you're trying to make, you have to postulate
: >some mechanism or physical constraint. This has not been
: >forthcoming.

Neil Rickert (rickert@cs.niu.edu) wrote:
: I have been saying, in several postings in this series, that I
: favored a learning process.  I take this to be something like
: induction which is based on statistical evidence.

But that's too vague to be of much use.

: Chomsky's case for UG would be stronger if he had produced a design
: for a UG, and demonstrated with computers that this design would
: work.  He has not done this.  Part of my skepticism about UG is that
: nobody has presented a persuasive mechanism whereby there could be an
: innate UG.

Of course his case would be stronger. But the program you've outlined
above is the work of an entire discipline, not of one man,
however brilliant.
One great virtue of Chomsky's formulation is that it makes testable
predictions. There is a limit to how much we can learn about
the organization of the brain by observing behavioural
regularities, without disassembling the hardware. I certainly
hope and believe that that will be done, and the computer programs
will be written, etc - but I don't expect Chomsky to do it.
Darwin had a very strong theory, based on observations -
but it took Mendel and Crick to present a persuasive mechanism.

If someone had written a program which could learn to talk
grammatically *without* having any preimposed structure to
facilitate this (say a giant, amorphous neural network),
then I would immediately concede the point.

: In the circumstances, I think skepticism is the best stance.  I would
: rather assume that there is no UG, with the risk of having to change
: my mind if sufficient empirical evidence to the contrary is
: developed.  If, instead, I were to assume UG as the default
: assumption, I might be misled and miss some of the important
: evidence.

I'm not arguing in favor of dogmatism.

The question is: does assuming a UG lead one into fruitful
lines of research?

Bo
