On Tue, 21 Dec 1999, Gisle Reigstad Tangenes wrote:
> I guess it's time to reveal the terrible secret: Your list is infected by
> a biological naturalist.
Horrors! But then there are already a lot of crackpots infesting
the list, so by comparison, not really a sign that the neighborhood has
slid downhill. But really, why anyone would think that particular
chemicals would be needed to give rise to consciousness is beyond me. I
would even prefer dualism over that.
> My Searlean objection to the above brand of functionalism
> is, How can computation as such be sufficient to generate consciousness,
> when it obviously isn't an intrinsic process of any physical system?
OK, I assume that by "the above" you meant computationalism. Well
there are of course two possible answers to the above. 1) You are
wrong; it is intrinsic; or 2) Computations can exist without needing to be
implemented by a physical system. More on both below.
> To clarify: There is a distinction between intrinsic and
> observer-relative features of reality. The former include all properties
> that are logically independent of the intentional attributions of
> observers, such as the molecular structure of the object I am sitting on.
> The latter are properties that exist only relative to such attributions,
> such as being a chair.
I don't think the molecular structure is as intrinsic as you
obviously think. But let's move on.
> Note that this is not equivalent to the Chinese Room argument, which says
> that syntacs is not sufficient for semantics; it denies instead that
> physics is sufficient for syntax.
Right. I'm glad you didn't use the Chinese Room "argument", which
is just a foolish bit of rhetoric. The problem you _did_ use is indeed a
nontrivial one.
> Computation and all other syntax is
> observer-relative, and in one sense exists only from a 1. person point of
> view. Please release me from the spell of this simple consideration.
I'll try. First a little history: After Searle presented his
argument that 'computation is observer-relative', Putnam picked up on this
theme and showed that a rock appears to implement all computations.
Chalmers then improved on Putnam's argument and proved that with the
standard type of definitions, any system that could be though of as
containing a clock and a dial implements all computations. (For example,
if two particles are moving away from each other, the clock could be the
distance and the dial could be the speed. With the right mapping from
physical to formal states, this system could be seen as implementing any
given computation.)
But being a computationalist himself, Chalmers did not take this
to be a death blow to computationalism. So he decided that the definition
of implementation was missing something, and that adding extra
restrictions on the mappings that were motivated by our ideas of which
systems should implement which computations could save the day. But his
attempts to do this were unsatisfactory.
However, it is my belief that he was more or less on the right
track, and I have also tried to find such additional restrictions. I
admit that I have not been entirely successful yet, but I do have ideas
that I will persue when time allows. My attempts to do this have led me
to consider the shortest algorithms that can produce certain outputs
related to the labeling of the formal states. One problem is that at the
moment, the shortest algorithm that can output a given string is not a
completely objective concept either (see Kolmogorov complexity); but, as I
have mentioned on this list, I have ideas on a possible way to fix that.
On my web page under "interpretation of quantum mechanics" I discuss some
of the above.
I should also mention that there are a few other people (such
as a student (whose name I forget) of the well-known computationalist
Dennett) who have also worked on the implementation problem, but my
approach is a bit different from what others have done.
OK, so one way to be a computationalist is to assume that some
program such as mine will eventually succeed (or has succeeded, as some
think of my competition) in getting the implementation problem under
control, or at least to think that in principle there is a solution.
But many people on this list have another idea: that computations
exist, as such; and that the physical world we see is just the way a
fairly typical computation models its existance. No mapping from
physical to formal states would be needed; the formal states would
already be the existing states. Showing that a typical computation would
see fairly simple physical laws such as ours is part of what is known here
as the "white rabbit problem"; see the posts about that.
Personally I think all possible physical (or mathematical) systems
exist, and that they implement computations.
- - - - - - -
Jacques Mallah (jqm1584.domain.name.hidden)
Graduate Student / Many Worlder / Devil's Advocate
"I know what no one else knows" - 'Runaway Train', Soul Asylum
My URL:
http://pages.nyu.edu/~jqm1584/
Received on Fri Dec 24 1999 - 18:32:52 PST