> On 3 Feb 2019, at 04:35, Russell Standish <li...@hpcoders.com.au> wrote:
> 
> On Sun, Jan 20, 2019 at 04:06:49PM +0100, Bruno Marchal wrote:
>> 
>> But when used in physics, this type of inductive inference assume not only a
>> reality, but a “brain-mind” identity, which is not consistent with the
>> mechanist hypothesis.
>> 
> 
> How so? All it assumes is that there is a relationship (correlation if
> you will) between elements of phenomenology. If something is
> phenomenally true (observed) , then some other thing is likely to be
> phenomenally true.
> 
> It is up to the learning algorithm to figure out what the relationships are.
> 
> None of this assumes reality, nor any sort of mind-brain identity.


Imagine that you let a ball falling. To compute the probability that it will 
hit the ground, people will usually assume that there is a ground, that there 
is ball, and that the ball obeys some law, so as to make the 
computation/prediction.
They will also assume that their mind remains attached to the body doing those 
computation/prediction.
Yet, with mechanism, this does not work, as there is no ball, no ground, and no 
identification possible between your first person possible experience and some 
world in which you would be there. The only way is to take into account all 
computations going through your mental state, as this one is attached to the 
infinitely many computations doing this in arithmetic, and the statistics will 
be given roughly by the number of computation which realise the experience of 
seing the ball falling on the grounds, “divided” by all computation leading to 
the initial state and where the ball does or not fall on the ground. 
Mathematically, it is more subtle, because the accessible states of the machine 
is structured by the logic of self-reference (probability one is given by the 
mode with “<>t” in the provability variant). That includes the computations 
involving white rabbits, speed quicker than light, etc. (a bit like the virtual 
particles on quantum filed theory, which can also “violate” the physical laws, 
like physical laws can violate in dreams, and thus in the apparence related to 
some computations. 
A learning algorithm also supposed some stable stream of inputs, which with 
mechanism have to be justified from that statistics on *all* computations.  It 
is not a problem in applied AI, but it is the problem we have to solve (and 
that the Löbian machine do solve in arithmetic) when trying to get a coherent 
theory of the relation between mind and matter appearance.
Just to involve a material universe, or a god, does not work, as it would make 
such God or Matter into a magical thing capable of making some computations 
more “real” than other, without changing anything in the computation, and that 
violates the “yes doctor”, as it call for something not Turing emulable (be it 
a substantial Matter or a God).

Bruno





> 
> Cheers
> -- 
> 
> ----------------------------------------------------------------------------
> Dr Russell Standish                    Phone 0425 253119 (mobile)
> Principal, High Performance Coders
> Visiting Senior Research Fellow        hpco...@hpcoders.com.au
> Economics, Kingston University         http://www.hpcoders.com.au
> ----------------------------------------------------------------------------
> 
> -- 
> You received this message because you are subscribed to the Google Groups 
> "Everything List" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to everything-list+unsubscr...@googlegroups.com.
> To post to this group, send email to everything-list@googlegroups.com.
> Visit this group at https://groups.google.com/group/everything-list.
> For more options, visit https://groups.google.com/d/optout.

-- 
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at https://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.

Reply via email to