Jonathan contrasts descriptions and what the descriptions describe:
> > Stathis: Your post suggests to me a neat way to define what is special
> > about first person experience: it is the gap in information
> > between what can be known from a description of an object and
> > what can be known from being the object itself.
> But how can "being an object" provide any extra information? I don't see
> that information or knowledge has much to do with it. How can "being an
> apple" provide any extra information about the apple?
Let's remember some naive answers here. First, for a fixed physical
object, there exist infinitely many descriptions. It's a common
belief that beyond a certain amount of accuracy, differences don't
really matter. For example, one ought to be quite happy to teleport
even if there is one atomic error for every 10^20 atoms.
Second, a common interpretation of QM asserts that beyond a certain
accuracy, there is *no* additional information to be had whatsoever.
That is, that there exists some finite bit string that contains
*all* an object's information (cf. Bekenstein bound).
Still, the naive answer is that a description (or even a set of
descriptions) of a physical object is different from the physical
object itself: a physical object is a process, and a set of
descriptions is merely a set of bits frozen in time (and here
we are back again, you know where).
However, I hold with these "naive" answers, as do a lot of people.
And so therefore I proceed to answer the above question thusly:
"Being an apple" provides *no* information beyond that which would
be provided by a sufficiently rich description. Even if an
emulation of a person appreciating the sublime, or agonizing to
a truly horrific extent, or whatever----no information obtains
anywhere that is not in principle available to the experimenters,
i.e., available from the third-person.
You could make the experimenter *hurt*, and then say, "now you
know what it feels like", and given today's techniques, that
might very well be true. But this is only a limitation on what
is known and knowable today; it says nothing about what might be
knowable about a human subject of 20th century complexity to
entities living a thousand years from now.
(We ignore the possible effects on the experimenter's value
system, or possible effects on his incentives: we are just
talking about information as bit-strings, here.)
> Obviously there is a difference between *an apple* and *a
> description of an apple*, in the same way there is a difference
> between *a person* and *a description of a person*, but the
> difference is one of physical existence, not information.
Yeah, that's the way it seems to me too.