1Z wrote: > Whatever properties are picked out by a computation > will be relevant to it as a computation.
Yes, of course. But how are these properties supposed to simultaneously produce a state of consciousness stably linked to the 'computation' when this self-same computation could have been instantiated in arbitrarily many physically distinct ways? The computations would be equivalent, but you appear to be claiming that however they are implemented, arbitrarily many distinct physical properties somehow become equally 'relevant' to generating the same state of consciousness. > There is no requirement that > the same connscious state is implemented > by the same physical state, so the multiple > reliasability of computations is not a problem So you say, but just *what* physical properties are supposed to be relevant and *how* do they contrive always to manifest equivalently within totally different implementations of a computation? Is this just supposed to be a mystery? My point is that under materialism, 'computation' is just a metaphor and what is directly relevant is the activity of the physical substrate in producing the results that we interpret in this way. What's critical to computational equivalence is not the internal states of the physical substrate, but the consistency of the externalised results thus produced. But with consciousness, it's precisely the internal states that are relevant. And here your reasoning appears to become circular - a particular set of physical properties can be construed as 'externalising' a particular set of computational results at a given point in time (fair enough) so, whatever these properties happen to be, they're must also be 'relevant' in generating a specific internal conscious state - and so must any arbitrary alternative set of properties that externalise the same computational results. Only because you say so, AFAICS. By making the rationale for supervention of consciousness on physical activity completely arbitrary in this way (it just *somehow* tracks a 'computation' however instantiated), you've effectively abandoned it as a materialist explanation. Didn't Hofstadter use this sleight of intuition to conjure consciousness from anthills and books - or was he perhaps just joking? David > David Nyman wrote: > > On Oct 9, 6:35 pm, "1Z" <[EMAIL PROTECTED]> wrote: > > > > > What is a "computation itself"? A process? And algorithm? > > > > Bruno covers what he means by 'comp' pretty comprehensively in his > > various posts and papers. > > > Almost all my discussions with him are attempts to clarify it. > > > > Using supplementary assumptions -- such as "only activity counts". > > > > Not sure what you're getting at - do you mean that, under materialism, > > the mere existence (not specific activity) of physical properties > > suffices to generate conscious experience? > > I mean the Activity Thesis > > http://tigger.uic.edu/~cvklein/papers/maudlin%20on%20comp.pdf > > > If so, I don't follow. I > > assume (see below) that, under materialism, experience -> psychological > > activity -> physical activity. > > > > Yes, but it is still quite possible that a class of phsyical > > > systems picked out by some computational(but ultimately physical) > > > set of properties are conscious/cognitive in veirtue of those > > > proeprties -- ie computationalism is a sort of convenient > > > shorthand or shortcut to the physically relevant properties. > > > > But this is the very nub. And it may be dead wrong, so would you > > address this directly? > > What is the alternative? Computaitonalism is just dead wrong, as a > thesis > about consciousness ? > That is possible. Computation is an extra factor, > a ghost in the machine? I don't think that is woth entertaining. > > > What is being claimed (in this form, a general > > appeal to the class of arguments referred to by the UDA 8th step) is > > that under materialism, 'computationalism' (i.e. the 1st variety in my > > taxonomy) precisely *can't* 'pick out' a set of 'physically relevant > > properties' in any stable way, because the physical instantiation of > > any given 'computation' is essentially arbitrary, and can extend to any > > number of diverse physical properties, to choice. > > Whatever properties are picked out by a computation > will be relevant to it as a computation. > > > Under materialism, > > specific conscious experiences should presumably map, or reduce, to the > > activity of an equivalently stable set of physical properties (in an > > analogous sense to, say, specific neurological processes reducing > > stably downwards through the physical substrate). And this can't be the > > case if I can change the physical properties of the computational > > substrate at will, from step to step of the program if necessary. > > It can't be done if you can change the relevant properties. > But then it would not be the same computation. > You can do what you like with the irrelevant ones. > > > So > > the claim is that, under materialism, some other schema than > > computationalism must ultimately be deployed to explain any stable > > *general* mapping from consciousness to physics. > > I don't see the problem. Materialism > (specifically, supervenience) only requires that > the same physical state must produce the same > consious state. There is no requirement that > the same connscious state is implemented > by the same physical state, so the multiple > reliasability of computations is not a problem > per se. > > > I agree that this is a > > bold claim, but it does appear to stem from a basic dislocation in the > > supervention scheme consciousness -> computation -> physicalism. Its > > consequence is that if we wish to claim that consciousness does in fact > > supervene stably on computation, as opposed to the physical itself, > > Computation isn't "other" than physical. > It just means there is a 1-N relationship between states > and realisations. > > > then such computation must itself be defined in a manner unconstrained > > to specific *physical* properties. > > It can be and often is. we define a NAND gate as having certain outputs > in relation to its inputs. What they are in non-relational > terms is left open, and so can be (and in fact is) multiply > realised. > > > This is a reductio devised to show > > the consequences of the starting assumptions. You pays your money..... > > > > > The point is that computationalists can continue to believe in matter > > > so long as they don't believe in numbers. > > > > But if I'm right, they can't also believe that 'computation' - which is > > only arbitrarily constrained physically - is an adequate explanatory > > schema for consciousness. > > Or that it is a unexplained-but-true brute fact? > > > It's just a metaphor, and metaphors per se > > (as opposed to their instantiations) aren't 'real in the sense that I > > am real'. > > > > David > > > > > > > David Nyman wrote: > > > > On Oct 8, 6:29 pm, "1Z" <[EMAIL PROTECTED]> wrote: > > > > > > > Yes. But he says he isn't assuming Platonism, although he must be. > > > > > > > Well, if he is, so what? If we allow him this, what then follows - > > > > isn't this more interesting? > > > > > > > He claims that computationalism is incompatible with > > > > > materialism. That is not modest (or correct AFAICS) > > > > > > > I think the 'modesty' part is meant more to relate to provability > > > > vs.believability, per Goedel/Lob - that we must live with doubt (i.e. > > > > empiricism is ineliminable). As to computationalism, there seems to be > > > > some confusion on the list (and elsewhere) between (at least) two > > > > varieties.At least four! > > > > > > > The first might I suppose be characterised as minimalist > > > > comp, dealing with programs as instantiated in (as one might say) real > > > > - i.e. material - computers. Clearly it would make no sense to say that > > > > this kind of computationalism is incompatible with materialism - i.e > > > > that physical processes can 'compute'. > > > > > > > So how does he get "computationalism is incompatible with > > > > > materialism" out of such interviews? > > > > > > > >From the 8th step of the UDA argument. This attempts to show that if > > > > one (but not you, I think?) starts with the much stronger assumption > > > > that *consciousness supervenes on computation itself*,What is a > > > > "computation itself"? A process? And algorithm? > > > > > > > then it can't > > > > also supervene on the physical.Using supplementary assumptions -- such > > > > as "only activity counts". > > > > > > > AFAICS, this stems fundamentally from > > > > the inability to stabilise the instantiation of a computation, given > > > > the lack of constraint on the material substrates that can be construed > > > > as implementing equivalent computations. Given materialism, in other > > > > words, 'computation' is just a metaphor - it's the physics that does > > > > the work.Yes, but it is still quite possible that a class of phsyical > > > systems picked out by some computational(but ultimately physical) > > > set of properties are conscious/cognitive in veirtue of those > > > proeprties -- ie computationalism is a sort of convenient > > > shorthand or shortcut to the physically relevant properties. > > > > > > > I have to say that I think this may really point to a fatal > > > > flaw in any assumption - within materialism - that consciousness can > > > > supervene on the physical *per computation* in the standard AI sense. > > > > However, consciousness may of course still be shown to supervene on > > > > some physically stabilisable material process (e.g. at the neurological > > > > or some other consistently materially-reducible level of explanation). > > > > > > > Bruon's empirical prediction require a UD to exist. That > > > > > is an assumption beyond computationalism. > > > > > > > But not beyond 'comp', which is a horse of a different colour.A Trojan > > > > horse with Plato in its belly... > > > > > > > The UDA > > > > argument attempts to establish, and show the consequences of, a 'comp' > > > > constrained to CT, AR, and the 'modest empiricism' of 'yes doctor'. It > > > > *assumes* that putative stable conscious experiences are associated > > > > with certain types of machine thus defined. From this stems the claim > > > > that the consciousness of such machines can't simultaneously supervene > > > > on an unstabilisable externally-defined 'material' substrate - in fact, > > > > the 'material' also has to be an emergent from the computational in > > > > this view.You are presenting the conclusions, not the argument. > > > > > > > Comp and materialism start from radically different > > > > assumptions, and have diametrically opposed explanatory directions.The > > > > idea that materialism is not compatible with computationalism > > > is a bold and startling claim. > > > > > > If comp is not "standard" computationalism, the fact that it is > > > incompatible with materalism may be a lot less impactive. > > > comp might simply beg the question. > > > > > > > However, I don't think they treat the *observables* in any essential > > > > way as less 'real', but differ radically as to the source - and here > > > > its does get difficult, because one can no longer simply appeal > > > > directly to those observables - as Johnson failed to note in stubbing > > > > his toe on the stone.The Johnsonian argument can be used as a wayof > > > > establishing the meaning > > > of "exist". It answers the question "what definition of existence > > > is there other than the mathematical one". > > > > > > > How can he come to conclusions about the uneality > > > > > of matter without assuming the reality of something > > > > > to take its place? > > > > > > > Well, in the end we can only believe that whatever it is must be 'real > > > > in the sense that I am real', or where are we?The point is that > > > > computationalists can continue to believe in matter > > > so long as they don't believe in numbers. > > > > > > > No, it's really easy. I am real, or I would not > > > > > be writing this. What you mean is to > > > > > establish it by abstract argumentation is difficult. > > > > > Well, it is. That is why empiricists prefer empiricisim. > > > > > > > Well, as you know, I've also had some discomfort with aspects of > > > > platonic or other possibly implicit assumptions in this approach, but I > > > > think now that it's interesting and fruitful enough to suspend > > > > judgement on this pending further (preferably empirically refutable) > > > > results, without fully committing as a believer - but then that is not > > > > what is demanded. However, I acknowledge the robustness of your > > > > Johnsonian approach to refutation! > > > > David > > > > > > > > David Nyman wrote: > > > > > > On Oct 7, 1:16 pm, "1Z" <[EMAIL PROTECTED]> wrote: > > > > > > > > > Numbers that haven't been reified in any sense, > > > > > > > don't exist in any way and therefore don't behave in any > > > > > > > way. > > > > > > > > > Forgive me for butting in again, but is there not some way to stop > > > > > > this > > > > > > particular disagreement from going round in circles interminably, > > > > > > entertaining though it may be? For what it's worth, it seems to me > > > > > > that > > > > > > Bruno has been saying that you get a number of interesting (and > > > > > > unexpected) results when you start from a certain minimum set of > > > > > > assumptions involving numbers and their relations.Yes. But he says > > > > > > he isn't assuming Platonism, although he must be. > > > > > > > > > As he often > > > > > > reiterates, this is a 'modest' view, making no claim to exclusive > > > > > > explanatory truth,He claims that computationalism is incompatible > > > > > > with > > > > > materialism. That is not modest (or correct AFAICS) > > > > > > > > > and - dealing as it does in 'machine psychology' - > > > > > > limiting its claims to the consequences of 'interviewing' such > > > > > > machines > > > > > > and discovering their povs.So how does he get "computationalism is > > > > > > incompatible with > > > > > materialism" out of such interviews? > > > > > > > > > In achieving these results, AFAICS, no > > > > > > claims need be made about the fundamental 'ontic realism' of > > > > > > numbers: > > > > > > rather one is doing logic or mathematics from an axiomatic basis in > > > > > > the > > > > > > normal way.How can he come to conclusions about the uneality > > > > > of matter without assuming the reality of something > > > > > to take its place? > > > > > > > > > The question of which set of 'ontic prejudices' we in fact employ > > > > > > as we > > > > > > go about our daily affairs is of course another issue.And yet > > > > > > antoher issue is whether the conclusions of > > > > > a valid arguiment must be contained in its premises. > > > > > > > > > It may of course > > > > > > eventually turn out that theoretical or, preferably empirically > > > > > > disconfirmable, results derived from comp become so compelling as to > > > > > > force fundamental re-consideration of even such quotidian > > > > > > assumptions - > > > > > > e.g. the notorious 'yes doctor' proposition.Bruon's empirical > > > > > > prediction require a UD to exist. That > > > > > is an assumption beyond computationalism. > > > > > > > > > But as Bruno is again at > > > > > > pains to point out, this won't be based on 'sure knowledge'. It will > > > > > > always entail some 'act of faith'. > > > > > > > > > To establish what is in some ultimate sense 'real' - as opposed to > > > > > > knowable or communicable - is extraordinarily difficult,No, it's > > > > > > really easy. I am real, or I would not > > > > > be writing this. What you mean is to > > > > > establish it by abstract argumentation is difficult. > > > > > Well, it is. That is why empiricists prefer empiricisim. > > > > > > > > > and perhaps at > > > > > > root incoherent. The debate, for example, over whether the > > > > > > computational supervenes on the physical doesn't hinge on the 'ontic > > > > > > reality' of the fundamental assumptions of physicalism or > > > > > > computationalism. Rather, it's about resolving the explanatory > > > > > > commensurability (or otherwise) of the sets of observables and > > > > > > relations characteristic of these theoretical perspectives. Indeed > > > > > > what > > > > > > else could it possibly be for humans (or machines) with only such > > > > > > data > > > > > > at our disposal? > > > > > > > > > David > > > > > > > > > > Bruno Marchal wrote: > > > > > > > > There is no need to reify the numbers.[...] > > > > > > > > > > > I don't think so. Once you accept that the number theoretical > > > > > > > > truth is > > > > > > > > independent of you (which I take as a form of humility), then > > > > > > > > it can be > > > > > > > > explained quite precisely why "numbers" (in a third person > > > > > > > > view-view) > > > > > > > > are bounded to believe in a physical (third person sharable) > > > > > > > > reality > > > > > > > > and in a unnameable first person reality etc.Numbers that > > > > > > > > haven't been reified in any sense, > > > > > > > don't exist in any way and therefore don't behave in any > > > > > > > way. --~--~---------~--~----~------------~-------~--~----~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/everything-list -~----------~----~----~----~------~----~------~--~---
