From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Tue Oct 09 2001 - 02:52:55 MDT
Samantha Atkins wrote:
>
> Really? Across the board? Tell me, what scientific methodology
> to you use to determine you are in love? that you are happy?
> that a painting is really good (or bad)? that a person is or is
> not to be trusted and how much?
Science is a formalized tool that reflects a greater whole - for example,
the experimental method reflects (among other things) the Bayesian
Probability Theorem. I don't need magic to know whether or not I'm
happy. There's a certain subjective sensation I call happiness, and it's
not exactly hard to tell whether it's present, and at what intensity, and
why. The question I ask myself ultimately reflects a physical fact about
my brain-state, and an external observer who wanted a definite answer to
the question could find one. If I were confused about the definition, an
external observer could discover everything there was to know about the
confusion.
Not everything involves controlled experimental studies, but that's
because controlled experimental studies aren't always necessary to find
simple answers to routine questions. It doesn't mean that there's
something outside of science. I don't use a scientific methodology to
determine whether I'm happy for the same reason I don't use a scientific
methodology to determine where I left my wristwatch; it would be
overkill. It doesn't mean that the location of my wristwatch is a deep
truth that science can never comprehend.
Asking an ambiguous or observer-dependent question does not mean that the
underlying reality is uncertain or subjective. It doesn't even mean that
the question is outside of science. If science can precisely describe the
ambiguity or the observer-dependency, science can combine that description
with the description of reality to yield the best available answer to the
question.
-- -- -- -- --
Eliezer S. Yudkowsky http://singinst.org/
Research Fellow, Singularity Institute for Artificial Intelligence
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:11:16 MST