On Feb 4, 1:05 am, Colin Hales <c.ha...@pgrad.unimelb.edu.au> wrote:
> Stathis (....Down below...)
>
>
>
> Stathis Papaioannou wrote:
> > On Thu, Feb 3, 2011 at 9:35 AM, Colin Hales
> > <c.ha...@pgrad.unimelb.edu.au> wrote:
>
> >> This means we are hooked into the external world in ways that are not
> >> present in the peripheral nerves. Looking at the (nerves pulses) signals, 
> >> it
> >> is impossible to tell if they are vision, smell, touch or anything else.
> >> Those that think that a computer can add this extra bit of connectivity to
> >> the external world, believe in comp/COMP. When you replace the brain with a
> >> model of a brain using a computer, that "extra" bit, the connection with 
> >> the
> >> outside world we get from our qualia,...the qualia created by the brain
> >> matter itself, is replaced by the qualia you get by 'being' the computer.
>
> >> If you believe comp/COMP, then you believe that the computer's model -or -
> >> the computer hardware itself -  somehow replaces the function of the 
> >> qualia,
> >> by analysing the sensory signalling, which is fundamentally degenerately
> >> related to the external world. Only a human with qualia can, from sensory
> >> signals, provide any sort of model for our 'computer-in-a-vat' that might
> >> stand-in for an external world. Having done that, the world being explored
> >> by our computer-in-a-vat is the world of the human model generated from the
> >> sensory signals, not the world itself. When an encounter with the unknown
> >> happens, then the unknown will be chacterized by a human model's response 
> >> to
> >> the unknown, not the (unknown) actual world. The extent to which these
> >> things are different is the key.
>
> >> Neuroscience is beginning to progress from NCC (Neural correlates of
> >> consciousness) to EMCC (electromagnetic correlates of consciousness).
> >> Researchers are slowly discovering that certain aspects of cognition and
> >> behaviour correlate better with the LFP (local field 
> >> potential/extracellular
> >> field) than mere action potentials.
>
> >> If the EM fields are the difference, then in replacing the fields of the
> >> brain with the fields of the computer running a model...and your
> >> qualia/cognition go with it.
>
> >> So when you think of the 'input/output' relations for a computer, the
> >> sensory signalling is only part of it. There is another complete set of
> >> 'input' relations, qualia, that together with the sensory signals, form our
> >> real connection to the outside world. So the old black-box replacement idea
> >> is right - but only if the black box has a whole other set of 'input'
> >> signals, from the qualia. The only way you can computationally replace 
> >> these
> >> signals is to already know everything about the external world already. 
> >> Your
> >> alternative? Keep the qualia in your 'black box'. To me that means
> >> generating the fields as well.
>
> >> Don't get me wrong. Lots of really nifty AI can result from the
> >> 'computer-in-a-vat'. However, that's not what I am aiming at. I want AGI. G
> >> for General.
>
> > Can the behaviour of the neurons including the electric fields be
> > simulated? For example, is it possible to model what will happen in
> > the brain (and what output will ultimately go to the muscles via
> > peripheral nerves) if a particular sequence of photons hits the
> > retina? If that is a theoretical impossibility then where exactly is
> > the non-computable physics, and what evidence do you have that it is
> > non-computable?
>
> Lots of aspects to your questions.... and I'll try and answer Bruno at
> the same time.
>
> 1) I am in the process of upgrading neural modelling to include the
> fields in the traditional sense of simulation of the fields. The way to
> think of it is that the little capacitor in the Hodgkin-Huxley
> equilvalent circuit is about to get a whole new role.
>
> 2) Having done that, one can do simulations of single unit,  multiple
> unit, populations etc etc...You may be able to extract something
> verifiable in the wet-lab.
>
> 3) However, I would hold that no matter how comprehensive the models, no
> matter how many neurons ... even the whole brain and the peripheral
> nerves...they will NOT behave like the real thing in the sense that such
> a brain model cannot ever 'be' a mind. The reason is that we 'BE' the
> fields. We do not 'BE' a description of the fields. The information
> delivered by 'BE'ing the field acts in addition to that described by the
> 3rd-person-validated system of classical partial differential equations
> that are Maxwell's equations.
>
> 4) A given set of photons,  can result from an infinity of different
> configurations of the distal world. A single red photon can come across
> the room from your xmas decorations or across the galaxy from a
> supernova. It is a fundamentally degenerate relationship. Yet the brain
> inherits enough information to converge on a visual scene that captures
> the difference. HOW?

a) It guesses. It isn;t always right.
b) the eyes receive millions of photons. It doesn;t build a scene
from them individually, it comes up with a hypothesis
that explains them jointly, a gestalt. That greatly
constrains the problem space (although not always uniquely,
as the Necker cube shows).

> I think I know, but that explanation is too long
> and doesn't matter. The fact is that the EM fields deliver _extra_
> information inherited from their relationship with space itself. It has
> to. There's no place else or it to come from!

That is, again, a magical solution to a non existent
problem.

> 5) Regardless of my wacky ideas about space, I'd like to reinforce the
> implications of the particular case of the scientist, who is trying to
> find out about the distal natural world from position of fundamental
> ignorance. If you claim that we have enough information to overcome the
> degeneracy, then you already have what the scientist wants...knowledge
> of the unknown external distal world....so you are not actually doing
> science. You already know.

You are conflating the idea that we have enough
information to get started with the idea that we know
everything

>This is the killer logical position. If you
> say a computer can do it, you are saying, in effect, that science does
> nothing/already knows everything.
> ============
>
> In the end, then, I am not saying that there is something uncomputable
> in the sense that it is impossible to 'simulate' it. You can simulate
> anything! What I am saying is that if you could _you wouldn't bother_
> because you'd already know everything. To accurately simulate a
> scientist you have to simulate (a) the scientist and (b) the entire
> environment of the scientist, when the scientist is trying to uncover
> the unknown and you can't simulate it because you don;t know it. I use
> the scientist as a model for the generally intelligent behaviour.


And that conflates not being able to simulate intelligence with
not being able to simulate an environment

-- 
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To post to this group, send email to everything-list@googlegroups.com.
To unsubscribe from this group, send email to 
everything-list+unsubscr...@googlegroups.com.
For more options, visit this group at 
http://groups.google.com/group/everything-list?hl=en.

Reply via email to