Le 09-juin-05, à 01:19, Jonathan Colvin a écrit :
I don't believe in observers, if by "observer" one means to assign
special
ontological status to mental states over any other arrangement of
matter.
I don't believe in matters, if by "matters" one means to assign special
ontological status to some substance, by which it is mean (Aristotle)
anything entirely determined by its parts.
This is similar to the objection to the classic interpretation of QM,
whereby an "observation" is required to collapse the WF (how do you
define
"observer"?..a rock?..a chicken?..a person?).
Yes, but Everett did succeed his explanation of the apparent collapse
by defining an observer by "just" classical memory machine.
But this was in response to a comment that "it was time to get serious
about
observer-moments". An observer is such a poorly defined and nebulous
thing
that I don't think one can get serious about it.
My definition is that an observer is a universal (Turing) machine. With
Church's thesis we can drop the "Turing" qualification.
Actually an observer is a little more. It is a sufficiently "rich"
universal machine.
To be utterly precise (like in my thesis) an observer is a lobian
machine, by which I mean any machine which is able to prove "ExP(x) ->
Provable("ExP(x))" for any decidable predicate P(x). ExP(x) means
there is a natural number x such that P(x), and "provable" is the
provability predicate studied by Godel, Lob and many others.
But then I need to explain more on the provability logic to explain the
nuances between the scientist machine, the knowing machine, the
observing machine, etc. You can look at my sane paper for an overview.
I'd note that your
definition is close to being circular.."an observer is something
sufficiently similar to me that I might think I could have been it".
But how
do we decide what is "sufficient"? The qualities you list
(consciousness,
perception etc) are themselves poorly defined or undefinable.
Consciousness can be considered as a first person view of the result of
an automatic bet on the existence of a model (in the logician sense) of
oneself. From this we can explain why "consciousness" is not
representable in the language of a machine. And consciousness get a
role: self-speeding up oneself relatively to our most probable
computational histories.
It should develop in all self-moving mechanical entity.
I define variant of "first person view" by applying Theaetetus'
definition of knowledge (and "popperian" variants) on the Godel
self-referential provability predicate.
Perhaps you could try to tell me what do you mean by "matter?"
Bruno
http://iridia.ulb.ac.be/~marchal/
