On Fri, May 10, 2019 at 12:26 AM Bruce Kellett <bhkell...@optusnet.com.au> wrote:
> From: Jason Resch <jasonre...@gmail.com> > > On Thu, May 9, 2019 at 7:47 PM Bruce Kellett <bhkellet...@gmail.com> > wrote: > >> On Fri, May 10, 2019 at 10:18 AM Jason Resch <jasonre...@gmail.com> >> wrote: >> >>> On Thu, May 9, 2019 at 7:02 PM Bruce Kellett <bhkellet...@gmail.com> >>> wrote: >>> >>>> On Fri, May 10, 2019 at 9:36 AM Jason Resch <jasonre...@gmail.com> >>>> wrote: >>>> >>>>> >>>>> On Thu, May 9, 2019 at 3:09 PM 'Brent Meeker' via Everything List < >>>>> everything-list@googlegroups.com> wrote: >>>>> >>>>>> >>>>>> Would it make a difference if they compute the same function? >>>>>> >>>>> >>>>> Not from the perspective of the function. If the computation is truly >>>>> the same, there is no way the software can determine it's hardware. >>>>> >>>>> >>>>>> If so then you might as well say it would make a difference if they >>>>>> were run on different hardware. >>>>>> >>>>> >>>>> From the outside it might seem different. E.g. instead of silicon >>>>> some other element, foreign to the chemistry of this universe, might make >>>>> for a more appropriate substrate. >>>>> >>>> >>>> But the computations that comprise a conscious mind also, ipso facto, >>>> comprise the whole universe. >>>> >>> >>> I don't see how this follows. Is the computer on your desk the whole >>> universe? Is it not able to run an isolated computation which is not >>> affected by what other parts of the universe are doing? >>> >> >> The computer on my desk is not conscious! >> > > Maybe. I'm not sure we can conclude anything so easily. But in any case > it can illustrate the point that a computation need not be identical with > the whole of the universe that contains it. > > Not all computations are conscious. But those that are must include a > large part, if not all, of the physical universe. > Such a computation sounds more like a Boltzmann Brain than the type of universe I think we inhabit. I think our brains, weighing only a kilogram or two make up a small part of the observable universe. > > So if the computations are the same, the conscious, AND THE UNIVERSE in >>>> which it resides, are the same. There can, therefore, be no "outside" from >>>> which the consciousnesses and universes are different. >>>> >>> >>> Couldn't what we take to be the physical universe be a simulation run in >>> computer within a very different universe? Clearly then the outside and >>> inside view would be very different. >>> >> >> But the theory is that the physical universe is a statistical construct >> over all computations running through your conscious self. >> > > You're jumping ahead to the final result of the computation, and continue > to jump back and forth between different levels/definitions of universe. > > If the theory is to make any sense, any reference to "the universe" must > be consistent with every other reference to "the universe". > > > Einstein's paper on special relativity begins with one concept of time, and by the end of the paper refines it to a more accurate conception of time. I don't think this is any different. Here we start with a conventional meaning of "physical universe", and by the end we show how the reasoning leads to a more refined concept of universe. > To clarify, let me enumerate stages of the argument such that we can be > clear which one we are speaking of: > > 1. Your brain can be replaced with a functionally equivalent physical > component which implements its functions digitally (here we change nothing > about our assumption of what the physical universe is) > 2. Following from #1, your consciousness can supervene on an appropriately > programmed digital computer > > That implies that my consciousness is just a single computation, > potentially implemented on a linear Turing machine. > > > That I said supervene implies that a many-to-one relationship between computations and mental states. This is no different from any other functionalist or even physicalist theories of mind. That there are neutrinos going through your brain means there are different physical states, but these neutrinos (supposedly) don't alter your conscious state. Therefore, that many different different patterns of neutrinos in your brain result in the same mental/conscious state suggests a many-to-one relationship between physical states and mental states. > 3. Due to Church Turing and #2, the underlying implementation of the > computer (the programming language, the physical material, the laws of > physics, the universe it happens to run in) are irrelevant, only the > functional equivalence at the low-enough level (substitution level) is > important to preserve consciousness (note that nothing to this point has > changed anything about our assumption of reality, the ontology, etc.) > 4. Assuming arithmetical realism (which implies the existence of all > computations) and #3, this implies all conscious states exist in > arithmetic. (this makes redundant the assumption of physical universes that > are distinct from physical universes, here we modify our ontological > assumptions about what a physical universe is) > > Why should we assume arithmetical realism? > You don't need to, but if you do this implies the infinite computations, and therefore the existence of all possible mental states. It leads to the "many histories", and can explain many physical facts. For example, this paper outlines several: https://arxiv.org/pdf/1712.01826.pdf - Simple physical laws that are probabilistic - Persistent regularities - An external world that contains the observer - Inter-subjective agreement on physical laws - Subjective immortality - Simple initial conditions - Observation of a universe that evolves in time - Observation of a universe with an absolute beginning in time All of these conclusions are reached constructively from the very basic assumption that all observer states exist (as detailed in the paper). This is a motivation for assuming arithmetical realism. (It's the simplest theory, which assumes the least, and is consistent with all the facts at hand). > > 5. Given #4, and the fact that an infinite number of indistinguishable > programs implement your conscious state (e.g. different below your > substitution level), and given that these programs may diverge in the > future, then making predictions about future experiences (the focus of > physics) now becomes a statistical question regarding the distribution of > unique programs existing below your substitution level. We have now > reached the "reversal" (the laws of physics can be derived from the > arithmetic concerning conscious programs which exist arithmetically, here > we acknowledge that no observer exists in any single universe). > > If the programs are indistinguishable, how can they 'diverge in the > future'. > The programs are indistinguishable from the first-person point of view, in terms of the consciousness it supports, not from its third-person description (e.g., my neutrino example). Another example is my operating system and browser example. The browser may start of identically on two different operating systems, but then diverge as new information from the local environment enters the browser (a sub-program of the operating system). > It has not been specified what part of the execution stream of any program > gives rise to some conscious moment -- is it short or long; one machine > state or many; in sequence or not? > If it is a complex state, it may involve many machine states. For example, consider any arithmetical function, which takes an integer and returns an integer. Let's say the function takes any integer N and returns 1 if it is prime and 0 if it is composite. This function, viewed atomically, may require billions of independent machine states and micro-instructions before it completes the computation. At no single time is the state of machine identical with the function being computed, rather the computation must be viewed in its entirety. Also, there are many ways the function could be computed, as you hinted, it could be serial, or parallel, it might look for factors from N-1 down to 2, or from 2 up to N-1, it might only search from 2 to sqrt(N), it might use some number field sieve method. All of these are "functionally equivalent", because at the highest level, looking at only the inputs and the outputs, it is impossible to distinguish between any of these possible implementations. I think this analogy carries forward when considering conscious states: there can be many implementations, and the machine states should not be confused with the conscious states, which instead are more like high-level functions. > I have a real problem understanding what might be meant by the statistical > question that determines your actual future. Statistics applies to the > behaviour of ensembles of individuals. The individuals are not affected by > the statistics. So if there are an infinite number of computations passing > through my current conscious state, diverging in the future (as in the > past), then that corresponds to many individuals that might share some > experience -- there is no single individual that encapsulates the > statistics of the ensemble. > The problem arises because a consequence of computationalism is that teleportation is in theory possible. If your brain state is destroyed at one location and re-instantiated elsewhere you will survive. Now what happens when your brain state is re-instantiated in a million different locations? You have a non-zero probability of subjectively perceiving yourself to end up in any of those million locations. While your threads of computation do not interact, you can end up as any of them, despite any location you might start from. > > So from the evolution of the view of what is meant by physical universe, > we see there are at least 3 connotations: > A) The first view where a universe is a causally isolated physical > structure which may or may not contain observers > B) The second view where a universe is a relatively stable (perhaps > shared) observation in the mind of some observer(s) > C) The third view where there exists a unified set of metaphysical laws, > applicable to all observers, and in principle these laws can be derived > from the arithmetic of self-reference, there is no longer the notion of an > observer which belongs to a universe as each observer is supported by an > infinity of similar, but distinct computations > > > These concepts of 'the universe' are not consistent. If you are claiming > that notion C) can be obtained by an argument starting with A), then your > argument is incoherent. > Getting to B and C does follow automatically from A, but from assumptions of computationalism and arithmetical realism, as I detailed in steps #1 - #5. Over the course of those steps, I detail when the notion of "physical universe" is amended by the reasoning or the assumptions. > > So when you say "the theory is that the physical universe is a statistical > construct over all the computations running through your conscious self", > you are correct that this is the logical end and conclusion of the theory > of computationalism. But when I said you could implement any consciousness > in any universe where it is possible to build a Turing machine, I am > talking about the "level A" type universe. (which I acknowledge to be > redundant and eventually eliminated in the theory, but use this example for > pedagogical purposes). > > Any concept introduced for pedagogical purposes must be consistent with > the endpoint of the argument. That is not the case here. You start with an > independent physical universe and claim to derive a universe based on an > infinity of conscious computations. This is a reductio. > What is the logical problem, can you point it out to me? > > So any external universe is part of that construct through your >> consciousness. So appealing to an external universe running a simulation >> does not help at all. >> >> Remember, consciousness is the sum over all computation that pass through >>>> that particular conscious state, so in this theory your AI, be it in >>>> silicon or the Game of Life, cannot be conscious, because it is a single >>>> computation. >>>> >>> >>> That all subjectively indistinguishable computations going through that >>> state are a possibility means the consciousness cannot identify itself with >>> any one particular thread of computation. In this sense that consciousness >>> is not the same as one of the programs passing through that state. But to >>> say the consciousness is not identical with one of the computations is >>> different from saying that computation is not conscious. >>> >> >> The trouble here is that that is an unproven assumption. >> > > It follows from there being a level of digital substitution (the > computationalist assumption). Let's say we can substitute your > consciousness with a computer that emulates each atom in your neuron to a > precision of 20 decimal places. This means we could also substitute your > consciousness with a computer that emulates each atom using a precision of > 50 decimal places, or 100. Your consciousness would be no different. > However, given enough time these simulations would eventually diverge. > > That relates to substrate independence: it says nothing about different > threads of computation. You take one thing to imply something completely > different -- which is illogical. > Just as for substrate independence there is a notion in multiple realizability (in the physicalist sense) there is in functionalism an analogous notion of multiple realizability with functions (as in my example above with different implementations of functions that determine if a number is prime). > > If the future of any conscious moment depends on the statistics over the >> infinite number of computations running through that state, then a single >> computation gives a conscious moment that does not have a coherent future. >> > > I think we are in agreement here. The experiences of a single computation > would/could eventually discover it has deviated from the normal expected > statistics. > > Taking statistics of individual conscious moments is not a meaningful > exercise. > I don't think so. It might mean that given enough time a simulated AI could eventually determine it is being simulated. I need to think about this more. > > Neither does a single computation exist in a coherent world, since >> physics, and the appearance of matter, is also the result of the statistics >> over the infinite number of computations. >> > > Once we replace our "level A" view of physics, with the realization in > "level C" that there is no physical world, then it becomes another question > altogether where we can really perform computations or not, or just create > the appearance in our own minds of running computations. > > So answering "Yes" to the doctor is merely a mental aberration? > > > However I believe that physical universes exist as programs which > themselves contain sub-programs which are observers. Otherwise I am not > sure how we would explain our apparently shared reality. > > The existence of an independent third person world is certainly a problem > for computationalism. You have not solved that problem here. > > > I think Markus Muller (https://arxiv.org/pdf/1712.01826.pdf) has gone a long way towards doing so. > Who am I e-mailing if not some other conscious observer with which is > connected somehow as part of a larger structure? This also implies that if > I, as an actor in this larger program, fashion a computer, it can implement > another sub-program which may also be an observer. True our infinite > diverging paths of consciousness will diverge and split and recombine, > etc., but in each case we can expect to find ourselves to be a computation > that is consistent with our existence. > > That is probably the biggest leap of faith of them all -- how can you > possible ensure this? Most computations do not correspond to anything > useful for our existence. > > > What are you referring to when you say "most computations"? I don't disagree with your point, but you can say the same about everything in physics. Most space doesn't have anything to do with life, etc. > To be conscious as a human generally would imply a sub program that has > existed in a much larger program which has been on going for long enough > and with the right conditions for human beings to evolve and arise. For > this reason, I believe the local physical laws (not global) are strongly > dependent/correlated with the mind in question. > > Why do local physical laws differ from the global laws? > For clarity, by local laws I mean things like the mass of the electron (or even the existence of something that could be called the electron). Global laws are things like the points I enumerated in Markus Muller's paper, such as "physical laws are simple and probabilistic". Local laws differ for different observers because they are different possibilities compatible with conscious observers. > If the laws are mind independent at the start of the argument, they cannot > become dependent on your mind, or anyone else's, through the course of the > argument. > They are mind dependent for those universes that contain minds. > > And consciousness must be embedded in a coherent "world" in order to >> exist. >> > > Does it? I am not so sure. > > What are you conscious of? Consciousness itself? Or of the qualia of > experience of an external world? > > > The latter, I guess I wasn't sure what you meant by coherent world. I think I get it now and agree (at least for the majority of minds/conscious states, some might inhabit delusional or incoherent realms, be Boltzmann brains, etc.). Jason > > This makes one suspect that YD + CT, leading to computationalism, is not a >> coherent theory. >> > > Based on your questions and objections I can tell you have put a lot of > good thought into these matters. I look forward to our future discussions > on the topic. > > Jason > > -- > 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 everything-list+unsubscr...@googlegroups.com. > To post to this group, send email to everything-list@googlegroups.com. > Visit this group at https://groups.google.com/group/everything-list. > To view this discussion on the web visit > https://groups.google.com/d/msgid/everything-list/146da7b0-950d-c620-9dd1-462766179669%40optusnet.com.au > <https://groups.google.com/d/msgid/everything-list/146da7b0-950d-c620-9dd1-462766179669%40optusnet.com.au?utm_medium=email&utm_source=footer> > . > For more options, visit https://groups.google.com/d/optout. > -- 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 everything-list+unsubscr...@googlegroups.com. To post to this group, send email to everything-list@googlegroups.com. Visit this group at https://groups.google.com/group/everything-list. To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CA%2BBCJUh1j0yNCFYoaGzaYL%3Dr9KuXdvB6T3Hh3MVdrGt9OdG9Wg%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
