Humpty Dumpty (was: reflection)

Derek L. VerLee
Wed, 20 Sep 2000 01:13:02 -0400 (EDT)

My two cents, (and S. Pinker's two cents as well).  I am something of a
newbie.  To let you know where I am coming from, I nearly have a BS in CS
from Michigan Tech.  Its not [insert revered CS program here], but its not
a bad program, and I do a bit of reading on my own.  I am on the list
because just about everything about TUNES fasinates me deeply.  

I dont intend this as a flame to anyone, but the whole conversation about
computers not being sentient was confusing the hell out of me, because I
couldn't understand why we were having it, and I think other people were
confused as well.  Well, finally I get it.

However, while  are on the topic (hopefully for not much longer ;), Steven
Pinker, in his book How the Mind Works, talks about this issue.  I'm not
going to quote directly, because I dont have the book in front of me, but
he is an advocate of the computational theory of the mind, IE, that the
brain is a machine (a feed forward, back-prognation neuro network, or
something like it).  He says that some people think that computer
scientists are being inacurate or overly loose with their language when
they say that a computer "knows" or "understands" something, however he
looks at it differently.  He thinks that computer science is "finally
demystifying" these terms.  The computer does know, in the sence that if
information is available in the current context, it is known (which is
really exactly the same way I would define "know" for humans).
In my opinion Sentience need not, and should not, enter the conversation.

The term reflection, when speaking about the mind, the relevent part of
the definition is "the returning of the mind to that which has allready
occupied it," is simular to what we are using the word reflection for in
computer science.  From the previously stated example, a process gets an
object representing information about itself from the operating system,
and then manipulates it in some way.  The operating system reflects that
change by some functioncall on the shared object containing the process
information.  Here, the system is returning to the shared info, or "that
which has allready occupied it", and almost by (but not quite) 
coincidence, reflection in the way we are using it fits the dictionary
definition.  After all, the term reflection was undoubtably choosen
because it "felt" like human activity of the same name.


                            \  /
                        -----><-----  (do you really believe this?)
                            /  \

On Tue, 19 Sep 2000, Jecel Assumpcao Jr wrote:

> While I can see what Lynn is trying to point out, the fact is that the
> computer science community has stolen many words and given them new
> meanings: bit, protocol, hardware and so on.
> My use of "reflection" might confuse newbies, but it would be best if
> they simply get used to it. This is the meaning that they will find in
> a huge number of conferences, thesis and papers, books and web pages.
> The book I wrote a chapter for is a good example: "Advances in
> Object-Oriented Metalevel Architectures and Reflection" edited by Chris
> Zimmermann.
> Since this list provides a very specific context, I feel free to say
> things like "the program knows about distances but not about areas"
> instead of "the code reifies the concept of linear measurements but not
> the concept of two dimensional measurements". Tunes is not about
> artificial intelligence nor cognitive science, so I don't think there
> is any harm in being loose with the meaning of some words.
> I am trying to be helpful, here, not argumentative. Please let me know
> if I failed...
> -- Jecel