Re:implementation

From: Marchal <marchal.domain.name.hidden>
Date: Thu Jul 22 12:06:03 1999

Jacques M Mallah wrote:

>> I don't understand. Do you believe or not that consciousness
>> is maintained in a well-implemented computation running on a UTM, where
>> all inactive registers (we know them because we have run that computation
>> before) have been eliminated ? Yes or No.
>
> I'm not quite sure what you mean. By 'register' what exactly do
>you mean? My answer may depend on it.

I was using 'register' in a generic sense for a "physical" part of a
concrete computer.

>> [BM] I guess you mean NO, if only because you tell us "inactivity does not
>> disqualify that piece". But this means that the presence of
>> consciousness
>> during an execution of a computation depends on the presence of an
>> inactive
>> piece, i.e. inactive during that computation.
>>
>> [JMM] >As
>> >pointed out above the piece would likely not be inactive since it would
>> >have to perform the computation to verify the run.
>>
>> [BM] I believe you try to say something here. I don't see exactly what you
>> are talking about. What do you mean by "verify the run" ?
>
> [JMM] In the case where the sequence is driven by the random inputs,
>which was the one being discussed, if it happens that the correct sequence
>is not followed then some action was to be taken. To do this some device
>must test to see if the correct sequence is being followed. In Maudlin's
>case that was easy because his implementation was clock style (false).

Do you mean any clock style implementation is necessarily false ?
Consciousness would not 'supervene' on a 'physically active'
Babbage Universal Machine ?
Could you explain me in what sense Maudlin's implementation is false ?

BTW, do you think the 'reading book' implementation of "the neuronal
(or chemical) activity of Einstein brain" (in 'The conversation with
Einstein's
brain' by Hofstadter in Mind's I) is necessarily incorrect, making
Achille talking with a zombie (Einstein) ?

>> Note that the argument shows also that any computation can be
>> executed in a
>> counterfactual correct way with an arbitrary small amount of
>> arbitrary physical activities.
>
> I have not 'noted' that, and you have not made any case for that
>at all. If you are using Maudlin's example, as I've already said, it's a
>false implementation much like Chalmers' clock and dial.

Would you agree with the following definition of "correct implementation"
(CI)
of a machine (relatively to a set of neighborhoods) ? A CI is an
implementation such that, for a set of variations of that neighborhood,
it will give the correct executions, i.e. the "intended" outputs and
absence of outputs.
The neighborhood is just what gives the inputs to the machine.
 
The possible variations of the neighborhood gives the counterfactuals.

Do you believe in zombie ? What do you think about the Turing Test ? Your
opinion on these matters could help me to understand what you mean by
false
implementation.

>> I agree so much with you !
>> but if you belief
>> 1) than consciousness is invariant for functional substitution which
>> preserves the counterfactual structure ;
>> 2) than consciousness supervenes on the physical activity implementing
>> the computation, then you are in trouble because consciousness can be
>> made
>> to supervene on some spurious physical activities.
>
> Then give an example.

Olympia.

I know you already told us that she is a false implementation, but I don't
see why.

The implementation is just bizare, but not in a relevant way, I think.

Maybe Chalmers-Putnam's implementations are "false" (for me too), because
it is not clear for me how
to use them effectively (like a concrete computer or like a brain, etc.).
These are more like description of computations, these are not
relatively well 'connected' or 'interfaced' to us. (We made the
neighborhood here).

But Olympia is, isn't ? Despite her story she recuperates her ability to
manage the counterfactuals. Despite her body handicap she recuperates her
ability to manifest herself. She can even dream.


[So what ?
So, I think the computationalist shouldn't associate consciousness with
the
'physical' activity of the computer, but with the (relative) computation
itself.
So the computationalist must explain how the belief in 'physical
activities and laws'
emerges in the possible Turing-tropic discourse along possible
computations.]


Bruno
Received on Thu Jul 22 1999 - 12:06:03 PDT