Newsgroups: comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!satisfied.elf.com!wizard.pn.com!Germany.EU.net!EU.net!CERN.ch!dxal18.cern.ch!hallam
From: hallam@dxal18.cern.ch (Phillip M. Hallam-Baker)
Subject: Re: What's innate? (Was Re: Artificial Neural Networks and Cognition
X-Nntp-Posting-Host: dxal18.cern.ch
Message-ID: <D2EMAJ.Gsp@news.cern.ch>
Sender: news@news.cern.ch (USENET News System)
Reply-To: hallam@dxal18.cern.ch
Organization: Wot!!! Me ????
References: <1995Jan4.203858.27692@galileo.cc.rochester.edu> <3eh8bs$s9f@mp.cs.niu.edu> <D1yB6E.1HA@spss.com> <3eog4a$ne2@percy.cs.bham.ac.uk>
Date: Sat, 14 Jan 1995 16:39:55 GMT
Lines: 110


In article <3eog4a$ne2@percy.cs.bham.ac.uk>, A.Sloman@cs.bham.ac.uk (Aaron Sloman) writes:


|>Not when I last read any Chomsky. Rather it was a general
|>*framework* for human grammars that was thought to be innate, not
|>any particular grammar. The latter would be inconsistent with the
|>diversity of grammars found in natural languages.

When hearing claims of this sort I see the possibility for an infinite
regression of claims. Unless the term `framework' is well defined then
almost anything can be claimed as an instantiation of that framework. To
take the extreemes we might insist that for Chomsky to be `correct' then
each US citizen should arive preprogrammed with the structure of english 
grammar, for the purpose of argument lets call that position the 
strong-Chomsky position. Clearly this is not what Chomsky is claiming 
On the other hand one could assert that the human  itself was a Chomskian 
`framework' for language and that the functional units for speech production 
and analysis provide us with a disposition to communicate. If we hold it 
to be an inate property of communication that it be possible to represent 
it abstractly as interchange of an ordered series of symbols then this view 
does not in fact imply anything about the structure of the brain except that
it has the capacity to communicate which we already know.

The question is at what level of abstraction is the Universal grammar 
hypothesis supposed to apply? The view that language is influenced by the 
structure of the brain can hardly be controvertial. After all a language
too complicated for humans to understand could hardly be expected to last
long (at least until the invention of UNIX).

At some level of abstraction the structure of the brain must be determined
by genetic factors. It seems unlikely that there is a schematic wiring diagram
hidden away in the DNA. Such a scheme would be unlikely to be robust enough for
construction using the construction techniques avaliable. For a model of
thought to be credible it has to pass the tests of being a viable evolutionary
strategy and a viable constructional strategy.


A Viable Evolutionary Strategy
------------------------------

The most persuasive evolutionary strategy would be for linguistic facilities
to be derrived from more general thought processing mechanisms. Although most
animals have some sort of communication there are few that in any way match 
the complexity of speech. One could attempt to interpret courtship rituals as 
a form of highly structured ritual and claim that speech could be derrived
from modification of the patterns controlling these rituals. This seems unlikely
though. Fixed ritual without interaction is not a basis for communication, what
communication there is is more likely to be explained in terms of situational
conditions where the meaning is in direct correspondence with the signal and not
a complicated function of order.

Consider two strategies for communication with flags. The first of these might have
a red flag for stop and a green flag for go. A more complex strategy such as 
semaphore could communicate very much more information but would require very
much more processing to be understood.

For a theory to be a viable evolutionary strategy one would have to show that 
the development of the capability could evolve as a series of incremental 
changes without sudden and dramatic change in structure. On this ground Chomskys 
assertion that speech is integral to our higher thought processing is plausible
at least. It is more plausible to suggest that evolution caused the reuse or
modification of an existing processing strategy first in the primate and then in
the human to enable superior language capabilities than to beleive that a 
separate language processing center was evolved. Such a development would be
akin to the sprouting of an additional pair of limbs or additional eye, such
developments are genetically feasible but imporobable in terms of evolution 
because there would have to be a large number of intermediate species which
devoted energy to creating organs which had marginal function before the full
advantage was achieved.



A Viable Constructional Strategy
--------------------------------

Two things are amazing about biological systems. The first is the miniscule size
of the genetic codes that describe them, the second is the robustness of the
gestation process. 

We know that many aspects of the brain require nvironmental stimuli
to develop properly. Rats which are kept in darkness at a critical stage in 
their development never gain sight. The external stimulus appears to be
essential for the brain to `bootstrap' correctly.

We also know that the human brain continues to develop after birth and undergoes
a major restructuring at about the age of five, that is after speech and in
some cases mathematics (or theology in the jesuits proverb) have been learnt.
It would be no suprise to me if this restructuring entailed some sort of 
optimisation of fixation process of existing abilities. That it would be 
dependent not only on the genetic code but the condition of the brain at that
time. One could well imagine the evolutionary advantage in strengthening 
abilities that had demonstrated their usefullness.

It should also be no suprise if hormone changes played a large part in the
development of the brain. It would be logical for presence of considerable 
adrenalin during a knowledge imprinting episode to cause the impressions 
gained to be given greater importance than those under normal conditions.


It is of course dangerous to make such extrapolations from our limited 
understanding of how the body works. It seems equally dangerous to me to make
assertions about the existence of a Universal language on the basis of an
appealing logical model of linguistics.


--
Phillip M. Hallam-Baker

Not Speaking for anyone else.
