On 12 Aug 2009, at 04:32, David Nyman wrote:

> 2009/8/11 Bruno Marchal <marc...@ulb.ac.be>:
> Bruno, thanks for your detailed responses which I will peruse closely.
> Meanwhile, I finally managed to locate on FOR an apparently coherent
> summary of the MGA (which I understand to be the essence of UDA-8).
> Here is my understanding of it:
> The MGA presents the case of a TM with a prepared tape that specifies
> a computation deemed - per comp - to instantiate a specific phenomenal
> conscious sequence.  Since a TM is invariant to the details of
> physical implementation, we can set out to produce a maximally
> reduced, or 'lazy', version of this particular machine, called Olympia
> (perhaps evoked as reclining languorously on her chaise longue).
> Given the finite length and specific execution paths of the
> computation, we can determine the precise sequence of tape location
> states that it entails.  Knowing this, we can arrange a simple
> mechanism to produce precisely these states by turning the
> corresponding tape locations on or off in a sequential left-to-right
> process.  We can further arrange for the machine to be able to produce
> the states required for each non-invoked path by means of similar
> additional pre-arranged mechanisms, even though these would not in
> fact be activated by the computation as specified.  In this way we can
> ensure that both actual and counterfactual conditions could be dealt
> with by the simplest possible combination of minimised and idle
> mechanisms.
> If I've got this more or less right, the burden of the MGA seems to be
> that a physical TM instantiating a specific computation deemed to
> evoke a conscious state could have its activity reduced to such a
> minimal level that it is no longer plausible that consciousness could
> supervene on it.  Whilst this does indeed strike one as intuitively
> very strong, as long as there continues to be activity at *any* level
> a diehard material-computationalist could still claim that it wasn't a
> final knockdown blow.  I see now that - as you remarked - my own
> argument includes a first approximation of this, since my reference to
> the arbitrarily many implementations of the computation would of
> course include Olympia.
> But the burden of my argument is in fact different, since - setting
> Olympia aside - it questions the very plausibility of what is assumed
> by comp - i.e. that a conscious state fixed in the *same* state by
> computational specification could be capable of invariance in the face
> of entirely *different* physical activities, if it is deemed to be
> *caused* by these activities.  This idea in fact appeals
> simultaneously to two separate causal principles - i.e. computational
> and physical.  Hence it strikes one as inherently a sort of
> non-interactive dualism, as implied by the computation's - and hence
> the conscious state's - insensitivity to changes in physical activity.
> The solution then seems obviously to be to throw one or other of these
> supposed causal principles out, i.e.:
> 1) either it is the case that consciousness simply supervenes on
> particular physical activities whose computational status is
> irrelevant;

... but then comp is false. OK? And thus comp implies "2".

> 2) or it is the case that consciousness supervenes on computation
> itself independent of physical activity (the conclusion that you in
> fact draw from the MGA).
> In the second case - i.e. the reversal of number and matter - I agree
> that you can save any role for primitive matter only at the cost of
> rendering it dualistically epiphenomenal in the sense sometimes
> attributed (IMO incoherently) to consciousness in materialist
> accounts.

But then by UDA1-7, not only "stuffy matter" would be epiphenomenal,  
but it would have absolutely no relationship with any observation,  
making it entirely spurious.

> As to UDA1-7, I think I see now that of course you assume mechanism,
> but that you reserve the argument for number-matter reversal until
> UDA-8.  Perhaps this order might better be reversed?

Like in my theses? May be. MGA is far more subtle than UDA1-7, that is  
why I have eventually decide to put it at the end.


Note that what you describe here as MGA is Maudlin's later and  
different argument. MGA is also immune to an objection made by  
Russell, which is that QM does "realize" the couterfactuals. Maudlin's  
argument can be saved from this with a version where Olympia simulates  
classicaly the quantum evolution of the brain. So MGA is more simple  
and direct. Anyway, the conclusions are the same, comp forces the  
abandon of the physical supervenience thesis. So comp forces to  
restrict the supervenience thesis on the mathematical computations  
(computationalist supervenience).


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 
For more options, visit this group at 

Reply via email to