.nr LL 6i
.nr PO 1i
.nr VS 24
.PP
I am adequately convinced by Putnam's argument and proof that there need not
be a unique reference-relationship between words in a language and objects
in the world; that there can be different interpretations of the language
such that the same sentence under both interpretations would hold the same
truth value.  But I am not convinced that the truth of Putnam's claim has a
necessary or significant implications for Cognitive Science, although I will
show where it might.  There are three arguments with which I will counter
Putnam's claim; the first two I think he already addresses, but I will
present them just the same.  The last argument he does not address, and the
answer to it would indicate whether his claim is relevant to Cognitive
Science.
.PP
Presumably the significance that Putnam's claim would have for Cog Sci is
that, supposing that we have invented a machine that seems to have human
intelligence, there would be a possibility that this machine's
.I "notional"
world is completely different from ours.  It would be possible that the
machine's idea of a cat is our idea of a cherry, yet when we talk to it
about cats it seems to be thinking about our cats.  So 
.I "functionally"
(to us) the machine seems to have human intelligence, but yet the concepts
that are going around in its `head' are radically different.
.PP
But isn't that only true if we are dealing with purely linguistic evidence?
Perhaps through
.I "talking"
about cats and cherries we think that the machine is working as we are, but
talking doesn't necessarily make direct reference to the real world, only to
itself.  So let's ask the machine to draw, or to point to a cat.  Then we
could see what it's really thinking about.  However, I would assume here
that Putnam would say that the machine's `point' and `draw' are different
from ours, and they are different such that when combined with it's `cat'
(our cherry) it produces the `right' behavior (it points to or draws a cat).
.PP
Okay, time for the second argument.  Let's think about how this machine that
we've invented comes to know what it does.  Let's assume (and I think it is
a good assumption) that this machine has learned in a similar way that
children learn.  It learns words by their reference to real objects, just
like children, so how could it go wrong?  I assume here Putnam would argue
in a similar fashion as above, namely that this machine, for whatever
reason, builds a different intensional world inside its head as it learns;
that its apparent connection with the word `ball' and the object ball has
nothing to do with the fact that the ball is a ball, rather there are other
things going on (in which the ball probably does play some part).  This
would probably mean that the way it learns is different from ours; the
learning mechanisms it uses are different.
.PP
I am not so much convinced by the above argument, and my reason for this
leads to my third argument: given that we've carefully modeled the learning
after what we (will) know about human learning, why should we have any
reason to believe that the machine's notional world is different from ours
when ours are the same among each other?  Why do we have any reason to
believe that person X's notional world is the same as Y's (and everyone
else's), yet the machine's is different?  If the answer is that there is no
reason, or that X's could be different from Y's, then it should not bother
us at all that the machine's be different; the only thing that would matter
is functional equivalence.  But if the answer is Yes, we do have reason to
believe that there is a difference between the machine and human but not
interhuman, then where does the difference come from?  It seems to me that
the only response to that would be that it comes out of the physical
differences between machine and brain.  I do not say that that is wrong, but
I would need to be convinced.  I think the physical differences could lead
to important differences in other realms, but it does not at all seems
obvious that physical differences (that are taken into account when
developing the `software') need lead to such drastic, across the board
differences in notional worlds and reference.