But this is trivially solved. If you have two branches then the probability is 50%. If you 17 branches seeing spin up and 16 branches see spin down, then the probability is 48%, and so on. I realize that the details are not trivial to work out in practice (QM itself is not trivial), but the general “problem in the philosophy” you refer to is a non-existent problem.
The equivalence of MWI and “QM” makes any argument here obsolete.
It would have to fully determine consciousness—the simple meaning of which is: fully fix all the facts about consciousness—in order for physicalism to hold. Since the same instance of consciousness.exe running can, according to you, lead to two different conscious experiences (consciousness of each branch separately vs. consciousness of the superposition), evidently the physical facts do not determine the mental ones, and physicalism does not hold in your scenario.
Simply to be empirically adequate: my experience is such that from one conscious state, I evolve into another. This is explained by the evolution of the physical state if the mental supervenes on the physical; since that is not given in your interpretation, this explanation is lacking and needs to be supplied by an extra-physical rule.
Because again, otherwise it does not account for our experience. I experience the wave function to collapse into one distinct state; the anthropic selection does not account for this, unless it could differentiate between distinct states within the superposition.
Collapse interpretations do; modal interpretations do; the many worlds interpretation as it is usually framed does. In fact, explaining this is the reason for any interpretation in the first place!
This presupposes that you have as many branches as are needed for the Born rule to hold; in other words, in order to get the Born rule probabilities, you must assume (and it is a genuine assumption) that there are just the right number of branches to get the Born rule. This is tantamount to just flatly assuming it. And the more general problem—how to get probabilities at all, as opposed to how to get the right numbers out—is far from a non-problem; it goes directly to the meaning of the concept of probability.
Take again the example with the outcome of some probabilistic experiment—say the tossing of a coin—undertaken in two different rooms. Within each room, it is trivial to assign probabilities: if each coin is fair, you get heads and tails half the time. But now assume the coins are correlated such that the coin in room 1 always shows the opposite of room 2: what is now meant by the probability to get tails? Certainly, here you can move to conditional probability: the probability to get tails for the coin tossed in room 1 is perfectly well defined. But in the MWI, the outcome of the experiment defines the branch: that is, each coin is transferred to a room depending on the face it shows. So the conditional probabilities are trivial: the probability that the coin in room 1 shows tails, if showing tails is what caused it to be transferred to this room in the first place, is of course 100%.
The equivalence of MWI and QM is exactly what’s at stake here: if the MWI cannot reproduce the probabilities of QM, then it is in fact not equivalent to it. And the fact that it is perfectly well possible to assign probability measures to worlds such that the observed statistics disagree with QM means that a non-trivial argumentation is needed (this is recognized even by proponents of the MWI, as is evidenced by the frequent attempts—by Deutsch, Wallace, Saunders etc.—to provide just such an argumentation). If you just flatly assume this equivalence, then of course you get the right statistics; but also, your interpretation becomes explanatorily sterile, as you assume what you intend to explain.
I think you are using a very liberal use of the word “physicalism”. From wiki:
*Physicalism is a philosophical theory holding that everything which exists is no more extensive than its physical properties
*
This is exactly true in the current example: the physical facts uniquely determine the mental ones; it just happens that they determine there to be multiple mental ones. This should not be surprising, as it is true even in classical mechanics. You can have dual descriptions of the same physical phenomena, and there is certainly no reason why you cannot have multiple consciousnesses that simultaneously result from (and therefore be completely determined by) the same underlying physical phenomena.
The “extra-physical rule” you speak of is no different from in the Copenhagen interpretation (or any other interpretation). In Copenhagen, the “extra-physical rule” is to roll dice. In the MWI, the “extra-physical rule” is to rule dice. The only difference is that the reason the dice are necessary is explained by the MWI, whereas in Copenhagen they are not.
Regarding C1, each version of you has a history associated with some evolution, but that does not imply that there is any meaning in assigning some kind of coherent existence to a single consciousness that you can follow along a history. Positivistically, the point is that after a measurement takes place, there is a collection of versions of you each associated with a different set of memories associated with different histories. It is completely meaningless to try to put a tag or “soul” on each consciousness and say that it is that consciousness that is evolving. There is no “primitive this-ness” to individual consciousnesses.
It does account for our experience; it shows that each branch must experience collapse. Collapse is explained. In each branch. Of course assigning the probabilities that arise due to the branch densities is a technical hurdle, which you abuse me with later on in your post 
however this is not relevant to what I thought the discussion was. The fact is that such probabilities will obviously exist. I think it is reasonable to table for the moment the discussion of how to get the particular Born rule, as opposed to some other probability rule, out of the MWI structure. The fact is that there will obviously be some rule that is similar to the Born rule here (which for the sake of the present argument I find sufficient to prove the point I thought we were at odds over).
No they do not. You said “Anthropic selection only says that you should find yourself in any of these states; it does not select one world from all possible worlds”.
And neither does the Copenhagen Interpretation select where the wave function will collapse to. The only difference is that in the MWI (as I have put it forward) there is a physical explanation for the apparent randomness.
I’m sorry but I just don’t see the point you are trying to make. It seems obvious to me that there is a version of an observer who sees tails, and a version of an observer who sees heads, and that the probability of finding yourself to be an observer who sees either heads or tails is therefore 50%, not 100%.
Well, that’s the same as saying ‘there can be multiple pictures emerging from the same set of pixels’. Does that really seem right to you? But as I said, I’m willing to accept this; but then still (cf. again Albert and Loewer) you’d have to account for the evolution of minds. And if you’re willing to just sacrifice this, as you seem to be later in your post, then you could just as well do that at the level of the physical (since we know the physical only via the mental), and get rid of any evolution there instead. The problem is that then your theory isn’t a theory anymore.
Regarding the dice, I think most people would hold exactly the converse statement to be true, that the Copenhagen ontology is intrinsically probabilistic, while it is difficult to even give a coherent meaning to the term ‘probability’ in the many worlds context. And of course, in Copenhagen (if endowed with a realistic ontology), the process of reduction is very much physical; it’s simply a break with the usual unitary evolution.
I don’t understand the problem here. I experience collapse to a certain state; anthropic selection only explains collapse to some state. If that were all that is needed to solve the measurement problem, then there’s be no problem, for in some state I am after every evolution! So if that’s all the work your anthropic selection does, then it doesn’t do any work at all.
The problem is just that I (and from my reading of the literature, basically everybody else) don’t think that it is at all obvious that probabilities follow from the MWI. The problem is not a technical one; it’s a crucial conceptual hurdle to making the MWI consistent with quantum mechanics and observation. If you can’t get the Born rule, the many worlds program simply is a failure.
For the Copenhagen interpretation, collapse is a genuinely probabilistic process: the branch of the wave function to collapse to is selected at random. This solves the problem of why we observe a certain outcome completely. But such a thing simply does not exist in the deterministic many worlds interpretation; and if it did, you would not have gained anything over Copenhagen.
I’m sorry (likewise), but ‘it seems obvious to me’ just isn’t a very good argument here. Have you had a chance to take a look at the Hemmo and Pitowsky paper? I think it lays out the basic problems much better than I could.
From my point of view your argument is much the same. You keep asserting that something doesn’t make sense, but I still can’t make heads or tails (pun intended) of what exactly it is. I have a similar reaction reading the Hemmo and Pitowsky paper. They have multiple paragraph asserting things like “Our point is that there is no
sense at all in which any probability measure is naturally picked out by the
many worlds theory”, without, as far as I can tell, justifying such statements. I really wish we can just work out this very simple example, because I just don’t get how on earth you don’t have a clear probability measure in the example you gave. Could you please elaborate… I must be missing something.
That is a simplification that amounts to a straw man. However, yes, the fact that there are dual descriptions of physical phenomena is a fact, not a hypothesis, and provides a rather convincing proof that yes, it does seem right to me.
I don’t see at all what you are saying here (and yes I have read the Albert and Loewer ref).
Then I don’t understand how you are defending the statement that MWI requires some “extra-physical” rules which Copenhagen does not. Copenhagen maps a wave function to a collapsed state using pure randomness (something I think doesn’t make sense, but let’s leave that for another discussion). This is a rule. The MWI maps a wave function to a collapsed state by anthropic randomness (a given consciousness having some probability of finding themselves in a given branch). This is also a rule (positivistically the same rule). The difference is that in the MWI the rule is explained as a natural consequence of nothing other than Schrodinger evolution. Again, as far as I understand the core of your argument here, I think it is irrelevant to go into a technical argument about there not being a proof yet that the Born rule is the unique probability measure resulting from the MWI picture, because you are arguing something more fundamental: that extra-physical rules are required. I still cannot make sense of this.
I seriously think there is something very big here that you are not getting. I just can’t understand the rational behind these assertions. You provide a “because”-like statement:
If that were all that is needed to solve the measurement problem, then there’s be no problem, for in some state I am after every evolution!
But I don’t follow why you think there is no problem because “in some state I am after every evolution”. You really need to flesh this out somewhat. Not because I am unconvinced of your argument, but because I just have no idea anymore what you are arguing.
From my reading the consensus among Everettians (which seems to make up a very large proportion of notable physicists) is that it is very obvious that “probabilities follow from the MWI”. The problem has just been deriving the Born rule specifically. I won’t deny that if ultimately you can’t get the Born rule, then the MWI will be a failure. But on the other hand saying that it is not obvious that probabilities follow from the MWI is just completely wrong.
Well, physical collapse doesn’t really solve the problem, because it is logically inconsistent, and ill-defined. But even if it was, the anthropic selection I have described also “solves the problem of why we observe a certain outcome completely”. You keep repeating that it doesn’t, but I don’t follow why. Repeating that there are multiple worlds and therefore no rule for selecting out of them without one that is “extra-physical” betrays somehow not understand what anthropic selection is. From the point of view of an observer the rules are exactly the same; the MWI provides an explanation for the seeming randomness, just as statistical mechanics provides an explanation for thermodynamics.
I think the most pedagogically clear way of making an accurate analogy between your coin toss idea and the MWI is something like the following:
You are taking part in an experiment. Scientists have perfected a cloning machine, kind of like a star-trek transporter, except it does not destroy the original copy. You walk into the cloning booth, they clone you, and then you are given a card. The original copy receives card “A”, and the new copy receives card “B”. From the point of view of each copy, he is “you”, and his evolution has been continuous. However from the point of view of one copy, he received card “A”, and from the point of view of another copy, he received card “B”. From “your” point of view, as you enter the experiment, you will have a 50% chance of finding yourself with card “A” or “B”. You can repeat the experiment as many times as you like, and if you keep track of your results, you can populate a statistical distribution of "A"s and "B"s that is consistent with p=0.5. Personally I don’t find anything at all controversial about this.
What the hell? Somehow, my response to this thread vanished… Let me try again (luckily, I had most of it saved):
Let’s take the example you provide in your third post:
The difference with the many-worlds interpretation is that there’s an objective fact here as to which one is me, and who is the clone. In the many worlds interpretation, there are just person 1 and person 2, on equal ontological footing. And moreover, who is person 1 and who is person 2 is only decided by who gets card A and who gets card B. So there’s just no meaning to talking about the probability that person one gets card A—since person one is defined by getting card A, that probability must be 100%.
But, you’re saying, this is different from the probability I should assign to finding myself with card A, say. First of all, let’s note that for this to make sense, you need to postulate the transtemporal identifiability of minds—otherwise, there’s just no ‘me’ in the sense that I exist both before and after the split; but more about that point further down. The problem with this is that for me, there is no objective difference regarding the number of clones that are produced: it could be one, two, or a hundred thousand. So I would have to have some way of knowing these ratios beforehand to assign probabilities; but there does not appear to be a way to do this (again, there is no reason beyond brute postulation that branch densities should correspond to Born frequencies).
And even if there is some way to attach appropriate probabilities to some split, what makes it such that these are the probabilities that I will observe? Branching histories exist in which I get the correct frequency, certainly—but so do those in which I won’t. And it’s not obvious that the former somehow dominate the latter: certainly, for a very large number of repetitions, the probabilities of getting any particular history—of taking any particular string of A and B out of a hat, so to speak—will equally approach 0. But then, since the relative frequencies of A and B in the total ensemble is 1:1, one should expect that, no matter the actual branching probabilities, one will end up with as much A- as B-cards in the long run.
I don’t follow. There are dual descriptions of the same set of pixels; that doesn’t change the picture. Do you think the mental facts depend on the description you choose for the physical ones?
Well, if you deny the transtemporal identity of minds, then all that really exists is just a random jumble of observer moments. How could I, in such a situation, develop a rational belief in quantum mechanics, for instance? All my memories I must regard as faulty, and thus, I can’t have built up any experiences, and empirical evidence for the validity of QM (or any given scientific theory). Furthermore, the falsifiability criterion becomes reduced to incoherence: any prediction I make becomes null and void, because I won’t be there to check it, and because any outcome is realized somewhere somehow. So if you deny the identity of minds over time, you deny the possibility of doing science, because that is predicated on the coherence of experience—which is then not given. Before accepting this, I think it would be preferable to accept outright dualism, or a modification of the unitary dynamics.
I actually agree with you there, which is my chief reason for rejecting Copenhagen.
No. In the Copenhagen interpretation, if you buy the possibility of a probabilistic selection rule, then the outcome is very different from what happens in the MWI. The collapse essentially throws away part of the wavefunction, while another part is made actual; in the MWI, all parts of the wavefunction are kept, and stand on the same ontological footing. Thus, Copenhagen here explains more, and that same explanatory power must be demanded of the MWI to make it a viable alternative.
The extra-physical rules are required only in your version, if you want to have multiple different minds supervening on the same physical state, and have there be a nontrivial evolution of minds (of course, if you are willing to sacrifice the latter, you won’t need such a rule; but I think what you sacrifice is greater than what you gain). It doesn’t arise in the ordinary MWI, because there is only one way to be aware of a superposed state, and that is the separate awareness of its components.
Well, just take the ordinary unitary dynamics, after which I am in a state like |"0"0> + |"1"1> both in Copenhagen and many worlds. Copenhagen now introduces the collapse in order to explain why I have in fact a definite experience: half the state is thrown away, the other half is given the Mark of Reality™. You’re effectively saying that it doesn’t need that; but then, I also don’t need anthropic selection, since it can’t distinguish between the two states anyway.
I’m not sure what to do here other than point you to the references I’m familiar with. Take for instance the many-worlds article on the Stanford Encyclopedia of Philosophy, where the section on probability starts with:
[
](Many-Worlds Interpretation of Quantum Mechanics (Stanford Encyclopedia of Philosophy))
Note that this is written by Lev Vaidman, certainly a notable physicist and Everettian.
As I said, the Copenhagen interpretation provides an additional rule according to which a definite state is selected out of the superposition: the collapse dynamics. One part of the superposition is discarded, while another is made actual. This is what makes Copenhagen consistent with experience. You assert now that the rule is not needed, that anthropic selection can do the same work. But that, to me—and I’m sorry—amounts essentially to using the concept of anthropic selection in some quasi-magical way, as an explanatory McGuffin: after measurement, I find myself in a superposed state; recognizing that this is a problem regarding our actual experience, Copenhagen then postulates the collapse to actually change the state to some definite one. You, on the other hand, claim that your anthropic selection can do the same work; but it can’t even differentiate between the states in the superposition. It just cannot do the work done by the collapse, since in the MWI, all parts of the superposition are on the same ontological footing. But then, there is just no reason for me to find myself in one state rather than the other, or in any distinct state at all.
The closest you’ve come, as far as I can tell, to provide a justification for the supposed power of anthropic selection is asserting that maybe our consciousness is just the type of thing that only exists in a part of the superposition, but not in the whole state. But this is tantamount to assuming your conclusion; and also, note that if it is possible to have a consciousness existing that supervenes on the whole state, again anthropic selection (which only selects worlds in which conscious observers are possible against worlds in which they are not) cannot do the work needed to select one kind of consciousness over the other.
But then, maybe I don’t understand what anthropic selection is. Perhaps you can just try and clear up for me exactly what you take it to be, and how it is supposed to do the work you need it to? Maybe work with the clone example: how is it anthropically selected that I am one of these clones, and not another? Because that is what needs explaining—it just doesn’t suffice to explain that I am one of the clones; that much is trivial.
This is what’s provided by all other interpretations, and the reason for there being interpretations in the first place. Copenhagen here invokes the collapse, effectively killing all clones save one. Many worlds—as usually conceived—claims that the awareness of all clones together is the same as the awareness of each of the clones separately (which you don’t, since you assert that it’d be possible to have a ‘group awareness’ of all the clones that is different from each one’s singular awareness). Many minds asserts a brute-force probabilistic rule. Hidden variables (and to some extent, modal theories) just claim that only one clone is real, the others relegated to mere possibility. Consistent histories constructs a set of mutually equivalent evolutions throughout all possible cloning events that lead to you being the clone you perceive yourself to be right now. The transactional interpretation uses the clone you start out being, and the one you end up being, to single out one chain of clones leading from one to the other. Objective collapse theories assert that there is a random process independent of measurement that leads to one of the clones being the right one. And so on.
The point being, all of these interpretations give a reason why we see a definite state instead of an indefinite one; I just fail to see how yours does the same work.
HMHW, I am going to snip out these quotes because they get to the core of the issue, and if we don’t settle it then only incoherence will proliferate:
I agree. We should focus on this example. If we cannot agree about that, then the rest is hopeless. To that end there are a few points you make that are relevant to that example:
I think it is reasonable to design the thought experiment in such a way that it is the same as in the many-worlds; both the original and the clone are on equal ontological footing, because the clone is exact, and once the cloning process has been performed, there is no experiment which can distinguish between the two. They are completely on the same ontological footing and there is no meaningful objective sense in which the two copies should be differentiated. If you point out the fact that, perhaps, the original stays in the same place during the cloning process, and the copy appears in a new place, then you are missing the point. It could just as well been that the original was transported to the new location and the copy was put in the original’s place. We have no way of knowing or distinguishing the two possibilities because the clones are exact. So yes, there is meaning to talking about the probability that “you”, in such an experiment, get card A or B. As a participant in such an experiment you will empirically find that p=0.5.
As far as I can tell, if you agree with this, then you agree with what I have been saying: probability arises naturally from the MWI in a way that should, barring some continued misunderstanding, not be considered controversial. The question of the specific Born rule is another matter. What we are discussing now is whether or not a probability measure associated with the branching universes makes sense.
If you don’t agree with the above, please consider first the actual process of participating in such an experiment. Do you disagree because you believe in some kind of “soul-like” trans-temporal identify that is unique that cannot be copied by the clone machine?
See, this is interesting, because I think that in order for your position to make sense, it is you who must postulate transtemporal identifiability of minds (see above). I do not postulate it.
Yes!
Because at any given moment “you” are one out of many observer moments, and the grand collection of observer moments has a density spectrum that is given by the wave function. The vast majority of observer moments (just as in the clone example) see a random sampling from the wave functions they interact with (just as the vast majority of clones repeating the experiment will see p=0.5, while an infinitesimal number will always see “A” or “B”), and therefore see behavior that is consistent with the rules of QM.
Even if you are somehow existentially averse to the non-trans-temporal identity of your mind, it is still true in the MWI that there is a wave function evolving that corresponds to your physical state, and that that physical state has memories (“A”, “A”, “B”, “A”, etc), and will call itself HMHW and has a rational belief in QM. There may be a “random jumble of observer moments”, but of course consciousnesses are non a random jumble of thoughts – they are (in a sense by definition) those linear narratives plucked out anthropically from the jumble.
No. That is confusingly trying to both assert that your mind has trans-temporal identity and yet that that trans-temporal identify is maintained across a self-contradictory history. It is like saying that at time t1 you are HMHW, and at time t2 you are Frylock, and that therefore at time t2 Frylock’s memories are faulty, because Frylock should really be HMHW and have HMHW’s memories. Frylock thinks he is Frylock, but really he is HMHW with Frylock’s memories and Frylock’s body. This kind of garbled thinking is just wrong wrong wrong. At any given moment there is just Frylock, with Frylock’s memories, and HMHW with HMHW’s memories. They each have their separate experiences encoded in their memories.
No. There will be future observer moments to check it, who will call themselves HMHW. They will think they are you (and I think they would be pretty frustrated if you told them they weren’t you) and they will check it and verify the prediction. In general if there is some previous observer moment that make a prediction, later on there is a collection of branches with a memory of that prediction. If that prediction was p=0.5, then again just as in the clone example, the branch densities are such that unless an observer moment is very unlucky it will find itself to end up in a history (ie have memories) consistent with that prediction.
As I have explained above, I think you do not appreciate the argument. I do deny the identity of minds over time, and yet come to a very different conclusion: from the point of view of any given observer moment at any given time, science works just fine. This again can be seen from the clone example. The experiment participant can make the p=0.5 prediction, and if you just follow what happens as the experiment progresses, you find that any given observer moment will be very likely to find memories consistent with p=0.5.
No, because the question of who is me in this experiment is determined by getting the card A or B. It’s not the case that either of the two copies after the cloning process is me, and then gets the card A, or the card B; both copies are me (absent the card-giving). If only the copying had taken place, there would be no operational difference in the world, provided both copies are indeed on equal ontological footing. It is the card-giving that individuates both copies, makes them different from one another; since thus their only difference is predicated on the card they receive, attempting to formulate a conditional probability based on which card they receive (i.e. which ‘clone’ they end up being) is flatly circular.
No.
Oh yes, of course I do! I’m a physicalist after all: as such, I believe that the mind fully supervenes on the physical. Thus, since there is a clear matter of fact regarding the evolution of the physical state, there is just as clear a matter of fact regarding the mental one—they’re not different, after all. The question simply does not arise for the physicalist, which is part of what makes the position so attractive for me.
I’d still very much appreciate it if you could give me some narrative as to how anthropic selection does what you need it to do, perhaps along the lines I gave in my last post for the other interpretations.
No. I exist only at some point in time, if there is no transtemporal identity—I am just a single observer moment. I have not been before, I will not be in the future. How can I then have experiences? How can I have memories? The very concepts simply don’t apply in your ontology.
This simply does not follow. If there is not connection between minds at point t[sub]1[/sub] and minds at t[sub]2[/sub], then you are not entitled to conclude that there will be any observers with memories of my prediction. Retrospection may convince you that this has always been the case, but of course, in your model, retrospection tells you nothing about what is actual.
Take the clone example. At any given branching, the clones are outfitted with minds; the minds they get have no relation to the minds ‘they’ had, or rather, to the minds that existed at the previous instance in time. Thus, you cannot draw any inference from the minds that were to the minds that will be; science does not apply to explain their experience because there is no experience there to be explained. Experience entails transtemporal identifiability, or rather, needs it as a prerequisite. Otherwise, for any experience, one can ask: whose experience was it? And the answer can never be ‘mine’, for the concept of my past self is just not defined.
Let card “A” be given to the copy who emerges from the machine that was walked into. Let “B” be given to the copy who emerges from the machine that the transported copy walks aware from. The card-giving does not individuate both copies; the card “A” is not given to the “original” because no one knows which one is the original. They are both exactly the same, and there is no meaning in trying to assign one to be the “original.”
I am a physicalist. I die hard eliminative materialist. And yet we seem to disagree about whether the mind has trans-temporal identity, and whether the mind fully supervenes on the physical.
Slightly modifying your MWI example (modifications in bold):
The awareness of all clones together is the same as the awareness of each of the clones separately and the awareness of each aware combination of clones.
Although I don’t think my bolded addition is really relevant to most of what we have been recently discussing. The current state of our disagreement about the clone example indicates to me that you would not agree with your own description of the MWI before I had modified it above.
This to me is the physicalist mystery of consciousness. In the physicalist description, consciousness is constructed out of some arrangement of matter at time t1, t2, t3, etc. Of course there is no transtemporal identity – that would be to posit an extra-physical rule. Arrangements of matter are not unique. Due to the translational invariance of spacetime, for example, the arrangement of matter that gives rise to your consciousness at time t1 can be erased at (x1,y1,z1) and put at (x2,y2,z2) without affecting your consciousness (since it does not affect physics). Therefore two copies can swap position, and from the point of view of physics, their “identify” has not interchanged. Similarly with translations in time, and so forth. This clearly demonstrates that the transtemporal identity you seem to be positing here (as I interpret you to have explained it) is extra-physical.
From my own physicalist position, consciousness is nothing other than the fact that observer moments tend to like to use this word “consciousness” and attribute it to themselves. This is not particularly surprising, since each observer moment is derived from a deterministic wave equation and has memories that are consistent with some history, and each observer is a complicated mechanical being that processes information and has labels that it attributes to itself.
Wrong. The Schrodinger equation tells us that there will be minds at t2 with memories of your prediction.
How are you defining “actual”? I posit that your definition is not coherent.
They have the exact same relation to their minds at the previous instance in time that your current mind has to the mind that preceded it. In every case there are minds at different time slices which are perturbations of minds at previous time slices. What is the difference? The current evolution of your mind is a succession instances in time in which the body that is sitting in front of your computer is outfitted with a different brain that is a perturbation of its last brain. This is just newtonian mechanics. The relation between clone minds is exactly the same as the relation between your mind at time t1 and your mind at time t2. In both cases you must equally come to grips with the existential worry that somehow “you” at t1 is not the same as the “you” at t2, and in both cases the question about contiguousness and consciousness is exactly the same.
Science is positivist. It works perfectly well, and this is why I keep pushing you to consider the clone example. Positivistically, follow the experience of a clone repeating the experiment over and over again. The clone need not be overcome by existential dread because it does not have transtemporal identifyability. Each clone, at each observer moment, can ask itself whether it’s current set of memories are consistent with the prediction that is stored in its memory.
This is an assertion of an extra-physical rule.
But the question you want to ask is meaningless. All we have access to is the current moment. The concept of your past is defined positivistically in science, and is defined by the memories of any given observer moment.
Now you are assuming that who’s the original and who isn’t is a determinate fact, i.e. that both aren’t on the same ontological footing.
If you say that the mind doesn’t supervene on the physical, then you’re not a physicalist, whether you say so or not.
We’re going in circles here: then your account is not empirically adequate, because I never experience anything else but being in a single, definite state (nor does anybody else seem to).
Then you’re just misunderstanding what I’m saying. The addition of the bolded part makes it such that there are now two inequivalent ways of being aware of the same physical state, with no way to differentiate between them; in many worlds, there is just one way, and hence, no problem.
Only if you’re saying the physical evolution is also underdetermined. Otherwise, since the physical and the mental are in one-to-one correspondence, there’s simply no problem; one could stop talking about the mental completely, since if fully reduces to the physical.
Wait, you’re saying the physics is identical, and thus, the mental is identical—and thus, I need an extra-physical rule??
It is not, in your description—that’s the whole problem. The wave function does not determine the observer moment, but at best some class of observer moments.
It does no such thing. The wavefunction has no memory: the state it collapses to (whether actually or apparently) does not depend on its history.
In order for a history to be actual, it must have, in some meaningful way, happened—it is not enough to simply (and in general, falsely) believe to have been in states a,b,c,…; there must have been something with which I share some measure of identity that has been in those states in order for my memory to be more than a mere mirage.
Which, as you’re saying, is none.
This is meaningless without a way to compare minds across time slices.
I don’t think anybody has really thought that since the 60s, when even the positivists were saying that that whole thing was just a really silly idea, and yeah, everybody should probably just move on.
There is no such thing without the possibilities of identifying minds across different time slices!
Oh dear, no! Think about an account where a single mind supervenes on a physical state: the evolution of that physical state fully determines that of the mental state. No extra-physical rules there. And a completely coherent account of experience: I can identify my past selves, and what happened to them causally affects me. I am in a very real and simple way just the future version of a mind at some previous moment; other minds are not. This is needed for any kind of experience; otherwise, everything is just a brain-in-a-vat sham.
Typically, the past is defined causally: what is in my past is everything that could have causally effected me. But this simply fails in your account, because there is no causality in the mental realm, and the physical is insufficient to fix it. And without such causality, there’s just no science, I’m sorry.
So basically, what I’m saying is, that in your model, there are minds at t[sub]1[/sub], minds at t[sub]2[/sub], minds at t[sub]3[/sub], each of which is a lone observer moment flickering into existence, only to again be extinguished. No mind at t[sub]3[/sub] can claim to be the continuation of a mind at t[sub]2[/sub] any more than any other; no mind at t[sub]3[/sub] has been in existence at t[sub]2[/sub], t[sub]1[/sub], or any other point in time. Thus, no mind has a history: it is, and then it is not. The concept of the passage of time does not even apply to it. It might have memories (though how the act of remembering, or any kind of cognitive activity, works if you just exist for one moment is not trivial either), but those memories don’t refer: there is nothing that stands in the relevant causal relation to such a mind. There is no causality at all, because no minds can influence one another, because there is no change in minds. Nothing these minds believe has content, nothing they think has meaning, because no term they use refers. There is no experience to such a mind, because there is no change; but since science is concerned with experience, there can be no science. These minds cannot meaningfully form beliefs at all, because they never can have sufficient reason to believe anything.
I don’t know. It might be that this works as a solution to your problem; but personally, it would just amount to an admission of defeat. I don’t think this is a rabbit hole worth jumping in.
Or put even more simply, there’s just no meaning to saying: “I have observed that ____, hence I believe that _____.” No such observation has actually occurred. Even if one allowed that each observer moment produces a single observation, then, without anything tying those together, there would just be all possible combinations of observations floating in perfect isolation from one another. You could construct strings of observations out of them that support quantum mechanics, strings that don’t, and strings that are altogether unreasonable, and there is no rule telling you to prefer any string over any other. Every position would be equally valid—or rather, equally invalid and meaningless.
Frankly, if I found that some position forced me to such a conclusion, I would consider the position reduced to absurdity.
It looks like there is an argument being put forth that identity across time is needed to avoid BIV problems.
What is needed are the right casual connections between a mind and the world. And such connections can obtain between things where identity over time fails. Indeed, the very things we want to refer to, to avoid BIV, are not identical (or share any interesting identity) with a mind.*
However, you go on to deny causality between minds (which is a nastier result than even being forced into a BIV result, to my mind). My question is, do you (think you) need the (quantum) physics, in addition to lack of identity-over-time, to get failure of causality between minds? (I’m guessing you do, but I find this conversation rather hard to follow, so I may as well just ask.)
*Actually more caveats are needed; I guess just, “some of the things we want to refer to” is good enough.
Explicitly I am not saying that.
If there is one thing that I am sure of in this world it is that I am a physicalist.
How do you know? Entities that do experience it (if such an entity exists) are not you. Your logic is like “I never experience being a fish”, therefore fish don’t exist.
No, the mapping isn’t between two consciousnesses and the same physical state (although I still do not have the slightest clue as to how this is inconsistent with physicalism). One consciousness is mapped to one physical state, another to another physical state, and then, perhaps, there could be a consciousness that is mapped to both of the physical states together. But even if there are two inequivalent ways of being aware of the same physical state, so what? Physicalism still holds: a physical state is mapped to X. X happens to be a set of awarenesses. Who are you to say what a valid X is that the physical state must be mapped to? Consciousness is, after all, a set of properties, so even in the case that X represents a single consciousness, you are still mapping to a set of something.
If you assume the physical and the mental are in one-to-one correspondence, you are adding a rule: that the physical and the mental are in one-to-one correspondence. There is no reason having to do with physicalism that this must be the case. Physicalism just implies that there is a correspondence, and that nothing outside of the physical is responsible for that correspondence. In the case of anthropic selection nothing outside of the physical is responsible for the selection. No extra rules are needed.
You seem to be saying that if there is some magic “you-ness” that surrounds and binds the molecules in your body and brain, such that if you were completely destroyed and then replicated, “you” would cease to exist, and “someone else” would start to exist. Even though this process of being destroyed and recreated is physically equivalent to the regular transportation of your body and brain through spacetime. You seem to be wanting to add an extra-physical rule in order to allay some kind of existential worry that “you” will cease to exist each time your wave function evolves a tiny bit.
But that class is determined from the wave function. What I said was correct: “each observer moment is derived from a deterministic wave equation.”
Of course it does. The statistical distribution of collapses at t2 entirely depends on the wave function at time t1. If I predict at time t1 that p=0.5, then at time t2 there will be a distribution of me, almost all versions of which have a memory of making that prediction. The entire point of QM being predictive is that while collapse is random, it is random within a statistical distribution determined by the Schrodinger equation, and that the distribution of collapses at t2 depend on the distribution at t1.
This is an assertion without any logical support. I can keep asking “why”?
No! For the love of all that is holy, please, please, just walk through the experiment. You walk into the clone booth at time t1. Now there are two physical states that are identical to the state at t1. Both have the same subjective continuous experience that results from whatever comparison across time slices that is your preference, since, if you are a physicalist as you claim, the fact that one of the clones was transported from (x1,y1,z1) to (x2,y2,z2) should not be physical by translational invariance. Therefore 50% of the time “you” will find yourself getting card “A”, and 50% of the time card B.
I don’t see how this is so. It sounds like you are again advocating for an extra-physical rule – one that somehow identifies minds across time slices.
How is your experience of consciousness in any way contingent upon your ability to “identify my past selves”? Your ability to create an intuitive abstraction of a physical process that happens to please your subjective sense of aesthetics, a sense of “continuity”, should have nothing to do with the relationship between that physical process and the consciousness that results. You insistence on such a requirement amount to an extra-physical rule.
You also are in my description, in the same way that you are a descendant from a common ancestor..
This is just an assertion. Why is it needed for any kind of experience?
Yes. There. Is. Just follow the clone example. In the clone universe, science works just fine. For each clone history, the experience is exactly as what is seen in a collapse model. It’s past determines its future up to a statistical distribution. There is not the slightest difference.
What’s being proposed (by iamnotbatman, whom I again ask to correct me if I am mis-representing his position) is that a set of minds is associated with every physical state. Furthermore, there is no relation between minds at different points in time—no mind has any greater claim to be the future version of a mind in the present moment than any other.
The brain-in-a-vatness comes about due to the unrelatedness of each mind to both past and future: since it exists only in a given moment, the world, as far as it is concerned, would be the same if there were no past, and no future, just the physical state in that moment (something which one can postulate to be the case in the real world—see last Thursdayism/omphalos arguments—, but which necessarily, or so it seems to me, is the case here). Since this is true for any of these minds, there also cannot be any causality (not one mediated physically, at any rate).
In a wider sense, I believe that there is no sufficient reason for belief in any scientific hypothesis in this scenario. Let’s take the clone example: at any point in time, there is a set of clones, and a mind associated with each clone (and certain sets of clones, as well). These clones are differentiated by making distinct observations—or receiving different cards. So let’s concretely say there’s two clones, one who has card A, and one who has card B. Then advance time by some amount, and let a splitting event take place. There are now four clones, each of whom has received a card, and each of whom has a mind. But neither of these minds are identifiable with any of those from before.
Now, say a clone at time 2 has card A, and a mind that is aware of that fact. This mind is not the successor of any of the minds at time 1. Thus, there is no matter of fact to the assertion that the clone had card A (or B) at time 1. The clone may remember having had card A, but his memory is not grounded in anything. One might suppose that maybe somehow the wavefunction ‘remembers’, thus giving him the ‘correct’ memory, but not only is the ‘correct’ memory ill-defined, also the only relevant differences between the wavefunctions at any point are exhausted by the allocation of cards, if the analogy to the many worlds picture is to hold. So there is no extra information that the wavefunction could somehow bestow on the clone.
But then, if we look at a long sequence of such splittings, there is no matter of fact of whether some clone had sequences ABABBABAABA… or AAAAAAAA… or ABABABA…, etc. Hence, there is no experience associated with the clone—which is as one would expect, since he only meaningfully exists in a single moment. But then, he cannot form, on the basis of his experience, any scientific hypothesis, and be justified in doing so.
You say “in my model”, but the logic you are putting forward applies just as well in classical mechanics.
There is causality in the sense that the minds are a result of the deterministic evolution of a wave equation.
These are just assertions. I think they are good ones, but again, they are true just as well in classical mechanics. You are just beginning to essentially move this into a discussion about how consciousness in a classical world is possible or has any coherent definition beyond the trivial (“consciousness is a label that certain entities apply to themselves”).
That depends on your definition of “I”. If you want to be finicky about the wording, try:
"This current information processor has a memory of a slightly different information processor having observed X, hence this current information processor believes Y.
Anthropic selection is what binds things together! You may be starting to get it.
Even in the Copenhagen Interpretation, you can get very unlucky and have a long string of events that do not support QM. It is no different in MWI.
Then perhaps you are not a physicalist because of the existential dread that results from the conclusion of your line of reasoning. That’s not a very good way to do science, but hey, it’s a very common attitude.