Re: Question for Bruno

From: Bruno Marchal <marchal.domain.name.hidden>
Date: Tue, 15 Nov 2005 15:39:19 +0100

Le 15-nov.-05, à 06:21, uv a écrit :



>
>>> uv: that this 'strong
>>> mechanistic' approach of Bruno really leaves a lot of
>>> the real world outside its bailiwick.
>>
>> bm: Could you elaborate just a little bit. I am not sure I understand.
>
> uv: Well you state elsewhere
>
>> bm; What I called "strong mechanism" in "mechanism and
>> personal identity" is what I call "computationalism" today,
>> or just comp. I have no idea if comp is true or false, but
>> it is my working hypothesis. the advantage of that
>> hypothesis is that we can reason about it and even be
>> led to testable consequences.
>
> uv: So that suggests, on your own statement, that comp may
> not be an accurate statement of the real world.




Right. But note that this can be said of *any* theory. QM may not be an
accurate statement of the real world (whatever that is), nor General
relativity, nor .... any theory.
But comp adds something here in the sense that comp makes necessary the
possibility of "not comp". Comp share this feature with the consistency
notion as viewed by some consistent machine. A machine can prove that
if she is consistent then she will never been able to prove its own
consistency. Comp is similar: if comp is true then we can only bet on
it; we will never been able to deduce it from a weaker theory.




> Given
> that, it is reasonable to suppose that a lot of the real
> world may be outside its bailiwick. The less charitable
> interpretation would be that a lot within its bailiwick is
> wrong or incorrectly stated.


I don't care at all for two reasons. Once is that my point is just that
once we postulate comp we get an explanation where the physical laws
and the physical sensations come from including the relation in between
(cf UDA). Second, as I said recently, the machine's psychology or
theology remains invariant for all the main weakening of comp, and
remains sound for *all* weakening of comp, and this transforms my
testability argument for comp into a concrete device for measuring our
degree of non-computationality.
(All this is not obvious, and I am not even trying to explain; I just
point on the results I got and on some recent extensions of those
results).

Now please note that the only scientist who explicitly postulated
non-comp is Penrose. His first "godelian" argument against comp is just
wrong. His second one is formally correct but shows no more that "IF we
are machine THEN we cannot know which machine we are", and this is
indeed a key point because it gives the relation between the first
person indeterminacy on the multiple computational histories (from UDA)
and the incompleteness phenomenon (and indeed this is used when I
"translate" UDA in the language of a lobian machine).

Again, when you say "The less charitable interpretation would be that a
lot within its bailiwick is wrong or incorrectly stated", well I agree,
but this is the case for any theory.






>
> Of course there are a third and a fourth interpretation.
> The third would be a combination of the first two and
> the fourth that comp is correct as it is. But you have just
> said that we can't assume that it is,

I am just saying we will never know it to be true. But as I show, we
can perhaps know it to be false. In that sense comp is a "scientific"
theory in the sense of falsifiable theory (Popper).

So I would say that we can assume it to be true. We can "only" assume
it to be true, and then expect a contradiction with the facts, or not.


> and I think that is
> correct. Bearing in mind the fact that the whole subject
> of mind, brain and mechanism is in flux, that situation
> is not surprising. Ultimately we may be led to category
> theory for the mappings.


I still should say I do have argument in favor of comp, in the sense
that I do have argument in favor of each of YD + CT + AR (= comp as I
make it enough precise to be able to reason on it).

1) YD ("yes doctor; see my last paper of search for doctor in the
archive; it is the assumption you can survive with an artificial
digital brain).

2) CT (Church thesis): all digital universal machine are equivalent.
(modern version!)

3) AR (Arithmetical Realism): Arithmetical propositions (see Podnieks
page for a definition) are true or false independently of "myself". For
example that "18 and 25 are relatively primes" is true or false
independently of me, of the belgians, of the europeans, of the earth
creatures, ... of the milky way creature, etc.

But I am NOT interested in those arguments, if only because comp makes
clear that such arguments are quasi personal and concernes more our
private relation with a doctor or a priester than something we need to
really argue for or sell. I find this funny (and it gives also a
political if not religious relief) : comp makes comp proselytes
inconsistent. Comp makes comp (in term of truth, practice, therapy ...)
a personal affair. Of course to reason from or on comp can be done, but
then all what you get is a theory which offer the laws of physics, and
then you can evaluate comp by comparing the comp physics with the
physics you bet from empiry. And until now, comp succeeds its first
step by showing that the logic of physical proposition cannot be
boolean and even that it is apparently quite close to the quantum logic
of propositions we can infer from experiments.

I am aware of many subtleties here, which makes an english rendering
close to a set of contradictory statements. Here the modal logics of
self-reference helps a lot. We should come back on it perhaps. But now
I must go.

Bruno


PS Russell: About Parfit look at "parfit" or "token" in the archive
(the mirror:
http://www.mail-archive.com/everything-list.domain.name.hidden/
Escribe seems down) to see what I think about his book. En gros, I
agree with him until the middle of his book where he argues that we are
token, but comp makes us type. In any case, Parfit "reason and person"
is quite excellent and enjoyable, and surely useful for this list.

http://iridia.ulb.ac.be/~marchal/
Received on Tue Nov 15 2005 - 09:43:57 PST

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:11 PST