[ExI] Do digital computers feel?

Stathis Papaioannou stathisp at gmail.com
Mon Feb 6 07:12:34 UTC 2017


Brent Allsop wrote:

<When you talk about "*only observable behaviour*" you are assuming a
definition of "observe" that is completely qualia blind.  There isn't
something special about qualia, but there is something qualitative, which
can't be observed by simple "*only observable behavior*".  You can detect
and represent qualia with any physical behavior you want, but you can't
know what an abstracted representation of what you have detected
qualitatively represents unless you know how to interpret that behavior
back to the real thing.  In order to include the qualities of conscious
experience into a definition of observation, you must provide a definition
of observe that properly qualitative interpretation of any abstracted
representations into whatever it is that has a redness quality being
observed.>

But I'm only asking at this point about *observable behaviour*, ignoring
qualia completely. It is my contention that if we do this the qualia will
emerge automatically and it is your contention that they won't. But in
order to figure out who is right you have to consider the experiment as I
have proposed it; you can't assume your conclusion in the premises.

<I imagine a simple-minded engineer working to design a perfect glutamate
detection system that can't be fooled into giving a false response by any
other not glutamate substance or system.  It is certainly possible that
some complex set of functions or physical behavior, like glutamate, is one
and the same as something we can experience as a complex redness, right and
that nothing else will have the same physical function or quality?>

I don't understand this paragraph. Do you accept that it is possible to
make a reliable glutamate detector, a device that tells us only if the
substance in question is glutamate or not?

<Once your simple minded engineered glutamate detector is working, it will
never find anything that is glutamate in the rods and wires engineered to
simulate glutamate in an abstracted way.  Also, without having the correct
translation hardware, you will not be able to interpret any abstracted
representations of glutamate (redness), as if it was glutamate (redness),
let alone, think it is real glutamate (real redness).>

That's all OK - the glutamate detector just detects glutamate, real
glutamate, and nothing but glutamate. So if there is glutamate in the
synaptic cleft, the detector in the postsynaptic neuron will detect it. In
this example the detector is not replacing glutamate but the glutamate
detector in the neurons, which is the glutamate receptor protein. To
replace the glutamate you would have to find another molecule or
nanostructure that is released when glutamate would be released and that
stimulates the glutamate receptors in the same way as glutamate does.

<And of course, when you neural substitute the glutamate and its detector,
out for some simulation of the same, the neural substitutuion fallacy
should be obvious.  It will only work when you completely swap out the
entire detection system with something else that knows how to properly
interpret that which is not glutamate, as if it was representing it.  Only
then will it be *observably the same behavior*.  But nobody will claim that
your simulation has any real glutamate being used for representations of
knowledge - glutamate being something that physically functions identically
with glutamate (or redness) without any hardware interpretation mechanism
being required.>

So are you agreeing that if you replace the glutamate with a substance
which is released when glutamate would be released and which stimulates the
glutamate receptors when glutamate would stimulate the receptors, the
neurons would fire in the same order and for the same duration as the
unmodified neurons would have? Remember this is just a question about the
*observable behaviour* of the system. Once you answer this question (yes or
no) you can then answer the additional question of whether the red qualia
would be preserved in the modified system.


--Stathis Papaioannou

On 6 February 2017 at 15:14, Brent Allsop <brent.allsop at gmail.com> wrote:

> When you talk about "*only observable behaviour*" you are assuming a
> definition of "observe" that is completely qualia blind.  There isn't
> something special about qualia, but there is something qualitative, which
> can't be observed by simple "*only observable behavior*".  You can detect
> and represent qualia with any physical behavior you want, but you can't
> know what an abstracted representation of what you have detected
> qualitatively represents unless you know how to interpret that behavior
> back to the real thing.  In order to include the qualities of conscious
> experience into a definition of observation, you must provide a definition
> of observe that properly qualitative interpretation of any abstracted
> representations into whatever it is that has a redness quality being
> observed.
>
> I imagine a simple-minded engineer working to design a perfect glutamate
> detection system that can't be fooled into giving a false response by any
> other not glutamate substance or system.  It is certainly possible that
> some complex set of functions or physical behavior, like glutamate, is one
> and the same as something we can experience as a complex redness, right and
> that nothing else will have the same physical function or quality?
>
> Once your simple minded engineered glutamate detector is working, it will
> never find anything that is glutamate in the rods and wires engineered to
> simulate glutamate in an abstracted way.  Also, without having the correct
> translation hardware, you will not be able to interpret any abstracted
> representations of glutamate (redness), as if it was glutamate (redness),
> let alone, think it is real glutamate (real redness).
>
>
> And of course, when you neural substitute the glutamate and its detector,
> out for some simulation of the same, the neural substitutuion fallacy
> should be obvious.  It will only work when you completely swap out the
> entire detection system with something else that knows how to properly
> interpret that which is not glutamate, as if it was representing it.  Only
> then will it be *observably the same behavior*.  But nobody will claim that
> your simulation has any real glutamate being used for representations of
> knowledge - glutamate being something that physically functions identically
> with glutamate (or redness) without any hardware interpretation mechanism
> being required.
>




-- 
Stathis Papaioannou
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20170206/2161e046/attachment.html>


More information about the extropy-chat mailing list