Re: Dreams and Machines
Rex Allen wrote: > On Sat, Jul 18, 2009 at 11:55 AM, Bruno Marchal wrote: > >> I am OK with all this. It has to be like this if we take the comp hyp >> > > So what are your thoughts on my question as to whether abstract > concepts other than numbers also exist in a platonic sense? For > example, the idea of "red"? > > So obviously we can cast everything as numbers and say, "In this > program, 0xff00 represents red". But RED is what we're really > talking about here, and 0xff00 is just a place holder...a symbol > for what actually exists. > > In your view, Bruno (or David, or anyone else who has an opinion), > what kinds of things actually "exist"? What does it mean to say that > something "exists"? > Well, since you asked, I think "exist" is always relative to some domain; so we should use "exist" in different senses. First of all I think epistemology precedes ontology. We first get knowledge of some facts and then we create an ontology as part of a theory to explain these facts. Facts are obtained in different ways. Chairs and tables and people exist at the most basic level of epistemology, i.e. we directly perceive them. Sometimes it is argued that we don't really see tables and chairs, we see 2-D patches of color and infer tables and chairs. This is the error of the misplaced concrete. Perhaps as infants we saw patches of color, but as adults our brain processes information differently and we directly perceive 3D objects. That we have theories of vision that tells us we're "really" experience certain excitations of the visual cortex or that tables and chairs are "really" quarks and electrons with lots of empty space are beside the point. Those are ontologies built on other theories that were inferred from perception of macroscopic 3D objects. Something similar happens with mathematical objects. We learn language intuitively and built into language are certain logical and mathematical structures so that we come to perceive conjunction and disjunction and the natural numbers and some other concepts directly. Do these mathematical objects "really" exist? I'd say they have logico-mathematical existence, not the same existence as tables and chairs, or quarks and electrons. Similarly we may, in another domain, say that Sherlock Holmes violin exists but Sherlock Holmes tuba does not, based on the reading of Conan Doyle. Brent Meeker > It seems to me that maybe consciousness is actually very simple. It > is just a group of platonic ideals, like red, that are related to each > other by a point of view: "I like red", or "I see a red sphere." > > Maybe what is complicated is constructing or identifying a causal > structure (e.g., a machine, a brain, a program, etc) whose evolving > state can be interpreted as representing a series of "connected" or > "related" instances of consciousness. But the machine (physical or > otherwise) is NOT that consciousness, the machine just represents that > consciousness. > > In this view, consciousness itself consists directly of the abstract > platonic ideals that form the contents of a given moment of > consciousness. > > > >> It remains to explain the relative stability of that illusion. How and >> why some dreams glue, in a way sufficiently precise for making >> predictions about them. >> > > Maybe unstable illusions exist, but, being unstable, don't ponder such > questions? > > Obviously we have such conscious beings here in this world, with > schizophrenics and the like. > > So your questions about "why are my perceptions so orderly", would NOT > be universally valid questions, because there are conscious entities > whose perceptions are NOT orderly. > > And I would say that even my perceptions are not consistently orderly, > as when I dream I often experience strange scenarios. > > To say that dreaming and hallucinating are special cases I think is to > make an unfounded assumption. It would seem to me that orderly > perceptions are the special case, and dream-logic realities would be > the norm. > > If consciousness is in some way a result of computation, then a > program that generates all possible mind-simulations will surely > result in the vast majority of resulting minds experiencing > dream-logic realities, not "law-and-order" realities like ours. > > I think Sean Carroll (who I'm reasonably sure would disagree with > everything I've proposed above, but still) had a pretty good point on > such "counter-intuitive" predictions: > > "The same logic applies, for example, to the highly contentious case > of the multiverse. The multiverse isn’t, by itself, a theory; it’s a > prediction of a certain class of theories. If the idea were simply > “Hey, we don’t know what happens outside our observable universe, so > maybe all sorts of crazy things happen,” it would be laughably > uninteresting. By scientific standards, it would fall woefully short. > But the point is that various theoretical attempts to
Re: Dreams and Machines
On 19 Jul 2009, at 04:43, Rex Allen wrote: > > On Sat, Jul 18, 2009 at 11:55 AM, Bruno Marchal > wrote: >> >> I am OK with all this. It has to be like this if we take the comp hyp > > So what are your thoughts on my question as to whether abstract > concepts other than numbers also exist in a platonic sense? For > example, the idea of "red"? Numbers are not enough. Even assuming first order logic. Then assuming "we" are digitalizable machine, this can be proved: Numbers are not enough. Numbers together with addition and multiplication are enough, and it is "absolutely" undecidable (for us, and us = any universal machine/ number) if there is any richer ontology. Numbers and addition + multiplication is a structure already "Turing universal". With addition and multiplication (and logic) you can already define the computational states and the pieces of histories going through them. You can understand that if you assume comp, all the computations going through the state of self-introspecting agent imagining "red" already exists as much as numbers. All the proposition of the shape "the machine i goes through states S" are, when true, elementary theorem of arithmetic, and they are accompagnying by "dense sets of proofs or relative realisations"). In the arithmetical Platonia, you already have all universal machines, and all their computations, which makes already place for big amount of "abstract concept" existing "platonically" (= like the numbers). And then you can define the modalities or point of view of those machines, by realizing that they will be aware (they have access too) the gap between platonist truth and what they can prove, and ... You may read the paper on Plotinus here, i.e. click on "pdf" on the right of "A purely arithmetical, yet empirically falsifiable, intepretation of Plotinus" on my url http://iridia.ulb.ac.be/~marchal/ You can see, well, not my thought on the subject, but the thought of the universal platonist machine. A machine is platonist when she believes, proves, asserts, the instanciations of the principle of excluded middle principle. > > > So obviously we can cast everything as numbers and say, "In this > program, 0xff00 represents red". But RED is what we're really > talking about here, and 0xff00 is just a place holder...a symbol > for what actually exists. Probably so. > > > In your view, Bruno (or David, or anyone else who has an opinion), > what kinds of things actually "exist"? What does it mean to say that > something "exists"? Assuming comp, something S exists ontologically when you can prove that S exists in Robinson Arithmetic (a very weak, yet universal, theory), And something S exists epistemologically when, let us say, you can prove in Robinson Arithmetic that there is a universal machine mentioning S. Technically it is far more elegant and sophisticate. See the eight hypostases (points of view) in the plotinus paper (or look for Plotinus or hypostases in the archive of the list). Instead of Robinson Arithmetic, you can take any first order specification of any universal system, machine or lnaguage (be it Conway's Game of Life, FORTRAN, LISP, prolog, Basic, c++, ... up to modular functor from quantum topology or knot theory, or number theory itself. > > > It seems to me that maybe consciousness is actually very simple. It > is just a group of platonic ideals, like red, that are related to each > other by a point of view: "I like red", or "I see a red sphere." Yes. > > > Maybe what is complicated is constructing or identifying a causal > structure (e.g., a machine, a brain, a program, etc) whose evolving > state can be interpreted as representing a series of "connected" or > "related" instances of consciousness. Yes. The difficulty is that consciousness, from its internal view, can only be related to an infinity of states belonging to high infinities of infinite computations. Third person consciousness, like the consciousness of my friend, is locally attachable (by guess) to a brain. "My consciousness" is not "attachable to a brain, only to an enumerable infinity of brains/machines/numbers weighted by non enumerable infinite histories. > But the machine (physical or > otherwise) is NOT that consciousness, the machine just represents that > consciousness. Indeed. The machine can represent 3-consciousness, like my identity cart can represent myself. 1-consciousness is related to a continuum of machines. This follows form the UDA. 1-consciousness is ignorant which "places" it occupies among continuum of histories. > > > In this view, consciousness itself consists directly of the abstract > platonic ideals that form the contents of a given moment of > consciousness. Not directly. It needs a self-reference, that is no more than two diagonalisations. Computer science suggests, and arguably forces entities to relate to themselves relatively to
