Newsgroups: comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!satisfied.elf.com!news.mathworks.com!news.duke.edu!news-feed-1.peachnet.edu!insosf1.infonet.net!internet.spss.com!markrose
From: markrose@spss.com (Mark Rosenfelder)
Subject: Re: What's innate? (Was Re: Artificial Neural Networks and Cognition
Message-ID: <D31HFJ.6oH@spss.com>
Sender: news@spss.com
Organization: SPSS Inc
References: <3fosrd$2if@mp.cs.niu.edu> <3g4br6$dga@mp.cs.niu.edu> <D2z22I.7FK@spss.com> <3g673d$7pl@mp.cs.niu.edu>
Date: Fri, 27 Jan 1995 00:59:43 GMT
Lines: 84

In article <3g673d$7pl@mp.cs.niu.edu>, Neil Rickert <rickert@cs.niu.edu> wrote:
>In <D2z22I.7FK@spss.com> markrose@spss.com (Mark Rosenfelder) writes:
>>Neil Rickert <rickert@cs.niu.edu> wrote:
>>>Once again, there would have to be existing structures capable of
>>>learning syntax.
>
>>Not really; the ability to learn syntax may have been added by the new
>>or modified structures, even if 90% of the modules used were already there.  
>
>But then we are back where we started.  If the changes required to
>learn syntax are that simple, then there is something already wrong
>with the "poverty of stimulus" argument.

Only if you assume that the changes to the brain to prewire syntax, and
the child's process of learning syntax, are comparable.  I don't see that
they are.  An analogy might help here: evolution's job would have been like
a computer language designer: how do you indicate argument-passing by value 
rather than by reference?  should "not equal" be <> or != or NE?
how strong is type checking?  The child's task is like that of a programmer
trying to read and write programs in the language given plenty of examples,
but no manual.  He faces the same questions, but he has a harder task
(especially if, as in the worst case, the designer made decisions arbitrarily
and inconsistently).

>>>These grammatical properties seem well suited to the use of a
>>>language in human-to-human communication, dealing with the kinds of
>>>activities humans do.  Alternate grammars that have been suggested to
>>>not seem as well suited.
>
>>This is IMHO the most promising line of attack against UG.  If given
>>features of UG can be shown to derive from some general constraint(s)
>>on cognition or communication, Chomsky's case is severely weakened.
>
>But Chomsky would never admit that.  He distinguishes between an
>Internal language and an External language.  The E-language is for
>communication.  But Chomsky seems to argue that it is the I-language
>which is the subject matter of linguistics, and it is with respect to
>the I-language that UG is important.  This is a brilliant tactical
>move by Chomsky.  It allows him to make his theory completely
>impervious to the empirical evidence, while still claiming that
>investigation of the the I-language is an empirical problem.

I don't know enough about how Chomsky defines I-language to evaluate
this criticism.  If it really sets it up to insulate the theory from the
facts, so much the worse for it.  But if there is some principled, observer-
independent method, even if indirect, for getting at the I-language from the
E-language, that's fine.  (A model here might be linguists' methodologies
for determining the phonemes of a language.)

>>Perhaps a better way of saying this would have been: If there is such a
>>thing as a general learning facility, why does it become almost impossible
>>to learn new languages fluently after puberty, when it's no harder (and
>>often easier) to learn other things?
>
>The "other things" that are easier to learn are mostly those that
>depend on prior knowledge, which simply was not available to the
>child.  Where prior knowledge was not important it is harder for
>adults.  Children learn how to use computers with great ease.  Adults
>who have not previously dealt with computers find it very difficult.
>Adults who have not previously learned to bicycle have great
>difficulty learning.  I imagine adults have more trouble than
>children in learning rap music.  I could probably find many other
>example.  I think your claim that adults learn faster is simply wrong.

Adults don't learn *everything* faster.  Perhaps we should just say that
age is not the only factor in ease of learning.

>>>>                               Consider the class of grammars as powerful
>>>>as that of a natural language.  To oversimplify drastically, the child 
>>>>has to determine which of these possible grammars is actually in use;
>>>>evolution need only pick one randomly.
>
>>>Ah, yes.  The "... and then a miracle happened" argument.  This seems
>>>to be consistent my earlier rhetoric about creationism.
>
>>What's the miracle?  
>
>A rather complex set of genetic mutations all happened at about the
>same time.  Although most mutations are harmful, by great luck these
>in combination just happened to be beneficial and to give us a UG.
>You don't find that miraculous?

Now *this* is a creationist argument!  It's basically the argument from
design: such and such an organism is too complex to have developed by chance.
