On 31 January 2014 18:30, Bruno Marchal <[email protected]> wrote: > > <snip> >
> OK. But you could also start by saying something like the POPJ assumes by > default a primitively-physical basis). > > Especially that it is certainly arguable that comp does not solve it to > our *entire* satisfaction yet. > OK. Actually, I'm trying to persuade Craig that it still applies on a primitively-sensory basis. But not in comp. Hopefully. > BTW, although you say that Craig can perhaps avoid the POPJ by appealing > to a non-comp theory, ISTM that the problem of reference is still there so > long as his "fundamental-sense" theory relies on causally-closed extrinsic > *appearances". > > I think Craig does not believe that his fundamental sense relies on > causally-closed extrinsic *appearance*. he would say that sense makes those > causally-closed extrinsic appearance, which makes sense in comp, actually > (to bad he believes only non comp guaranties that). > > Of course his theory does not explain mind, consciousness or sense, as it > assumes it. > And I fail to see how it relates to the *appearances*, except by making a > sort of naive identification of sense with some matter (up to some > "convolution" which he does not describe in any precise way). > But if he makes that naive identification (modulo any convolution, which I've offered him the opportunity to explain) the POPJ can still bite him. > > However, under questioning he's so far been rather unclear about this > aspect of his theory. > > > That's clear. He assumes sense, and try to make it into a form of matter, > sometimes. May be the last reference to tTegmark might help him. It seems > to be a form of panpsychism. > It would seem so. But POPJ can still bite panpsychism, I think, although this doesn't seem to be widely recognised. My post to Craig elaborates on this. > Are such appearances causally closed? Do we not rely on such "physical" > consistency? Maybe, sometimes, who knows, whatever. I might go so far as to > say that he's been dodging the question. > > > By assuming sense, he dodges the mind. And by being unclear of matter, > well he might dodge the issue of matter too. > > It is still better than the person elimination of the materialists. > I agree. > <snip> > > The plausible resolution of the paradox, if I've understood you, lies in > the capability of the machine to refer to non-shareable but incorrigible > truths beyond formal proof and demonstration. > > Yes. > In more than one sense, and those sense are related. > > One sense can be attributed to Gunderson, and is very simple. Once you > have build some numbers of robots, having enough cognitive abilities to > recognize themselves and name the other robots, it will recognize some > basic difference between itself and the other, just by the virtue of being > itself. > Like "not seeing his own neck". > > UDA does not need more than that simple assymetry. It provides the comp > solution of the problem why am I the W person and not the M person. A > negative solution, as it says "nobody could have predicted that". > Here appears already a stock of 1-truth, or 1-1 truth, which are non > logically justifiable and sometimes unexpressible (having non definite name > or description). > > But formally, we get more senses for this, all deriving directly or > indirectly from incompleteness. > > If you want the usual boolean logic of any extrinsic 3p, enough rich to > describe itself (like we could ask for an explanatively close physics) > extends into a modal logic, naturally, when that 3p self is taken into > account. That's the modal logic G. G is the logic of the 3p self in a 3p > reality. > OK. > > But by incompleteness, some truth about that 3p self cannot be logically > justifiable by that 3p self, and Solovay theorems gives the precious gift > of a modal logic of the whole self-referential truth (whole at the > propositional modal logic level: it is not the whole truth!). That is the > logic G*. > OK. > > To give the simple but important example, the consistency, that it is the > non provability of the false (<>t = ~[]f) is an example of true statement > (trivially if we limit ourself to sound machines), which is not provable by > the machine (by the 3p self about its 3p self, at the right level of > descriotion: here by construction). > > G* is decidable, and so a correct machine can "produce" a lot of truth > about itself that she cannot justify logically. > > G and G* still operate purely in the extrinsic (to use your term for the > 3p, if you don't mind). > OK. > > If the machine can grasp G and in a different way G*, she might decide to > pick up some truth in G*, and bring them back on earth, by adding them to > G. (and later, comp will be something like that, somehow). > > This can be done in many ways. Adding the truth "<>t" at the bottom level > makes the machine inconsistent. Adding it slightly higher, makes the > machine more competent, more relatively speedy. > Is the acceleration achieved by using this truth as a "bet" on a reality? If so this, reminds me of Daniel Kahneman's "system 1" (intuitive decision making), described in his book Thinking Fast and Slow. System 1 is faster than rational evaluation (system 2), but at the risk of error if used outside its domain of application. In any case, the whole notion makes sense, ISTM, in the context of the requirement to survive in an unknowable "extrinsic" environment that can only be approximated "intrinsically". We are forced to bet in order to survive, but at the risk of annihilation if we bet wrongly. No guarantees. > > But instead of adding them, the difference between rational belief, or > proof, or justification, with truth, entails that we can apply the > definition of the knower, that is the first person, by Theaetetus. We add > "<>t" to the definition of a new modal box, like []p & <>t. > > We can define, in arithmetical terms, a knowledge notion, by defining to > "know p", by "to believe p" together with the truth of p, and the weaker > notion above ([]p & <>t) > > It works, and this define, in arithmetical terms, a notion of knower, (and > observer) and the knower obeys a temporal logic of subjective state of > knowledge close to Brouwer theory of the subject (an intuitionist logic). > It is S4Grz. > That these different logics already exist is amazing. > > Then - if we are machines - our own incontrovertible faith in, and ability > to refer to, such indexical "facts" may serve as the warrant that also > delivers our fellow machines from zombie-hood. > > > Yes. > Not only that. It makes also the rational machines open to the idea that > humans are non zombie *too*. > That's a relief! David > > Bruno > > http://iridia.ulb.ac.be/~marchal/ > > > > -- > You received this message because you are subscribed to the Google Groups > "Everything List" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/everything-list. > For more options, visit https://groups.google.com/groups/opt_out. > -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/groups/opt_out.
