The many worlds anthropic principle is explored here from the a priori perspective of rationalist metaphysics, within the framework of modal logic. It is shown how the apparent contradictions of quantum superposition can be thought of in terms of different levels of world models. The framework of modal logic is used, but given the rationalist assumption that all possible worlds exist. There is thus no absolute distinction between possibility and necessity. To take the point of view of a conscious being in a world, however, is to adopt some such distinction--something we must do in order to do physics. This paper is intended to lay the groundwork for future attempts to develop theories of what is necessarily true in a world with conscious entities. It also contains some tentative speculations on the difficult issue of death and quantum mechanics.
Of course, this is an immense task, and this essay will only take a few faltering steps towards this goal. It will help if the reader has some understanding of modal logic, the lambda-calculus[2,3] and rationalist/idealist metaphysics. My metaphysics draws primarily from Parmenides, Descartes, Leibniz, Bradley,[8,9] Everett[10,11] and Carter.[12,13] I will provide only a brief overview of these ideas. See the references for more details.
I will lay our groundwork by describing, in terms of possible worlds, one of the primary conceptual problems with idealism. We will see how modality allows us to think more clearly about it. I will also show that modern physics gives us good empirical reasons to believe that this is actually how the world works.
The other fundamental principle of rationalism was introduced by Parmenides: "Whatever can be spoken or thought of necessarily is, since it is possible for it to be, but it is not possible for nothing to be." Parmenides sees two ways of viewing the world, one objective and one subjective. From the objective stance, all possibilities are equally real, and so the everyday world of our experience has no a priori claim to be more real than any other logical possibility. From the subjective point of view, however, it is only natural for us to see our own world as special and to seek the best explanation for it we can find.
In Leibnizian terminology, this means that all possible worlds exist objectively. We, as inhabitants of one of these worlds in particular, would like to explain the nature of our conscious experience. In the past couple of decades, the Anthropic Principle has become a popular way of doing this.[11,12,13] Since all possible worlds are equally valid from the most objective stance, we need a selection principle--some a priori reason to place our world in a characterizable subset of all possible worlds. If we can build a model describing this subset, that model should provide us with an explanation for much of the order we see around us. The version of the Anthropic Principle I will use starts with Descartes' Cogito: "I think, therefore I am." The a priori selection principle we are looking for selects out only those worlds that contain conscious life. This is the only feature of our world that we already know must exist, for if it did not, we would not be here to ask the question in the first place. This principle--that the universe must have features that allow consciousness or we would not be here to be conscious of it-- is called the "Anthropic Principle"; the version of it described here in terms of a plurality of worlds is the "Many Worlds Anthropic Principle", or MWAP.
V is the valuation function that defines which symbols appear in which worlds. R is the accessibility relation, which defines which worlds are possibilities from the perspective of any particular world. Recall that from Parmenides' objective stance, all possible worlds are real. However, within a particular model, only the worlds in W are considered possible. So to work within a model is automatically to take a subjective stance and intentionally exclude some worlds from the universe (for instance, figure 1 arbitrarily excludes the world in which both A and B are true).
But not only is W limited to a subset of all the worlds we could possibly build from W, but R limits each world as being possible only from the point of view of certain other worlds. In figure 1, for instance, w1 is a possibility from the perspective of world w3, but not vice versa. So, since A is true in w1, we say that A is possible in w3. But from w3's perspective, there are only two possible worlds: w1 and w3. Since A is true in both, we can also say A is necessary in w3, since it is true in all possible worlds (from w3's point of view). C, however, is true in w3 but not w1, so it is possible but not necessary in w3. In more formal notation, we say:
A narrower subjective stance would be that class of models containing any conscious life that also happens to fit my own personal criterion for self-identity. I might call this "my universe": the set of all possible worlds that contain an individual I identify enough with to call myself. So the world in which I just scratched my elbow, even though I did not scratch my elbow in this world, contains a conscious life form that is not literally me here now, but is close enough that I consider that person to be an alternative version of myself. So I say it is "possible" that I "could have" scratched my elbow just now. If this other individual was in my precise situation except that he did not have the appropriate conscious state and memories necessary for me to identify with him as being an "other" me, I would not call it a "possible world" (not, that is, from the perspective of "my universe"). I will call this "my universe" perspective the "personal identity class of models".
Many other perspectives are possible. A still narrower perspective, but a metaphysically very important one, is the view that sees only things that happen to my actual self as possible. Only one stream of consciousness is allowed to define the world. This view simply eliminates other selves as possibilities (by this, I mean other selves that I identity as "me", not the existence of other conscious beings, which is still allowed). The individual who scratched his elbow just now in that other world is now considered to be no more me than any other consciousness. This is similar to the personal-identity view, but with a narrower view of personal identity, restricted to only one stream of consciousness. I will thus call this the "strong personal-identity class of models" or the "stream of consciousness models". The earlier example, which allowed more than one stream of consciousness into my definition of self, I will sometimes call the "weak personal-identity class of models". It is clear that many different definitions of self are possible, depending on how weak we make the definition. I will generally take the stream of consciousness models as the strongest possible, but in fact there is an even stronger, narrower way to define oneself, one which seeks to go no further in its viewpoint than the Cogito of Descartes. This is the view that associates the self only with one's current state of consciousness. While this maximally sceptical view of self is philosophically important as a starting point, few people ever really adopt this as a serious notion of personal identity, so I will consider it a personal identity model only in the degenerate sense, choosing to more generally call it simply the "conscious-state-identity class of models".
Stream of consciousness models and conscious-state models are closely related. The latter provide us with our philosophical starting point, but we will almost always want to extend our view to the weaker stream of consciousness model, allowing that we are the same person today as yesterday. It is the stream of consciousness model, then, that is the most metaphysically important, as this is the one that selects out the world that we generally consider to be our everyday "world", the observable universe, past, present and future. One might think that this leaves us with just one world in our model, accessible to itself, but unable to "see" any other worlds. But this is only one kind of model in this class. Stream of consciousness models include only worlds that can be identified with a particular stream of consciousness, in which that stream is "true". This gives us at least one "world", but a model in this class could break things down into further "worlds", based on some additional criteria, giving us a sort of hierarchy of models, with only the root or top model containing a solitary world. For instance, a common metaphysical position to take is that each time slice of our universe is its own "world". There is no absolute justification for connecting up the world as it existed yesterday and the world of today and calling them a single world. From this viewpoint, the world at time t is considered a separate world (but in the same universe) as the world at time t+1 . The fact that they follow each other in time is a consequence of the model. So, in a stream of consciousness model, world w' is accessible to world w if either world contains a conscious being that can remember being in the other. Perhaps this definition needs tightening, but we will leave it at that for now. The point is that the world of yesterday is "accessible" to me only because I can call it to conscious memory. Note that if we "time-slice" according to conscious states, then the stronger conscious-state model can be subsumed lower in the hierarchy, underneath the higher level stream of consciousness model. We will see more explicitly how this works later on.
But what of the world of tomorrow? The stream of consciousness view seems to require that tomorrow be accessible, which may seem counter-intuitive at first. But recall that accessibility merely defines relations between worlds, and does not necessarily refer to cognitive or perceptual accessibility. You cannot remember tomorrow, but in this class of models we are saying that it is accessible to you because you-in-the-future remembers being you-now. So future time slices can be accessible to past slices, in this model. For any given time slice, tk, there are many possible worlds at tk+1, but only one at tk-1. In other words, I can only remember one past, but there are many futures with persons that can remember being me. This means that which time-slices get included in the stream of consciousness model depends on one's current conscious state, even though the stream of consciousness model was supposed to transcend that particular state! In other words, it turns out to be impossible to form a concept of oneself based on a single stream of consciousness that includes past, present and future, unless one is dead and can categorically say that one has no future, in any world. Otherwise, we are compelled to allow future time slices to be accessible. So we will clarify our definition of stream of consciousness models, so that they are always defined relative to a conscious-state model, or are given relative to an entire lifetime from birth to death, if any (since, in a way, a "true" stream of consciousness model is of the latter variety and cannot be built until one has lived one's life to the end and there are no more possible selves that one could yet turn into).
I will use the term "idealist" to mean all of these various philosophers from the most objective nonmentalist to the most subjective mentalist. What these views all have in common is a rejection of "material realism", the notion that the everyday "matter" of the world has some kind of objective reality of its own, as matter, and apart from its status as a possible idea or its status as an actual idea in a mind. Even those idealists who believe that ultimate reality is nonmental will nonetheless usually agree with the subjective idealists that everyday material objects like chairs and rabbits are somehow mental in character, since they are highly dependent on the viewpoint of the observer, and are only "partial aspects" of a larger nonmental reality. It is this kind of nonmentalist idealism that I will argue for. The "larger" reality is simply the totality of all possible ideas, including in it all possible worlds, of which ours is but one. Thus, the "mental" nature of material objects comes in simply as the fact that they cannot be defined or picked out without bringing in oneself. They are the objects in "my world", as opposed to all other worlds. While this is certainly not "material realism", it is still a brand of "realism", since there is still an external reality--it just is not equivalent to our everyday notion of material objects in space and time. However, almost all forms of idealism have in some sense been realist. Even Berkeley believed in an external reality, but he equated it with God rather than matter. Arguably, the only idealists that are true anti-realists are solipsists, who believe only in their own mind, but it is very difficult to find any genuine solipsists, living or dead. It is thus more correct to contrast idealism with materialism rather than realism, although in the twentieth century, "realist" usually means "material realist".
By historical accident, material realism is generally associated with "analytic" philosophy. But in fact, the whole notion of some possible worlds existing and some others not existing is thoroughly unanalyzable--I would claim even mystical. A wholly "analytic" philosophy would consider only that which can be rationally conceived of, and thus logically analyzed--anything else would be passed over in silence. It is ironic that the term "analytic philosophy" can thus be taken as virtually synonymous with "objective idealism", a brand of philosophy that real-world analytic philosophers considered it part of their founding mandate to eradicate. The reason for this is that modern idealism began with Berkeley, firmly entrenched in the self-contradictory subjective camp. It became, over the years, much more objective and rational. Bradley, who represents perhaps more than anyone the brand of idealism the analytics were rejecting, had brought things almost, but not quite, back to the objectivism of ancient idealists like Parmenides and Plato, but he was still ultimately a mentalist. The analytics, like Bertrand Russell and G.E. Moore, were largely rejecting this mentalism--I think justifiably. However, they threw the baby out with the bath-water, going whole-hog for a thoroughly mystical naive realism, instead of continuing the evolution of idealism, a la Bradley, into a nonmentalist and fully analytic variety.
Instead of answering this right away, I would like to look first at an example. I will pretend that my conscious-identity model contains only two worlds. This model is defined from the perspective of time t0. In world wH, I have a hat. In world w~H, there is no hat at all. At time t0, I am approaching a table. I believe that I have the hat on my head. The hat has no perceivable effects on my conscious state, even when it is on my head. I am utterly oblivious to it. But I know this. I am perfectly aware that this particular hat can sit on my head without being felt at all (perhaps it is a very light hat, or I have such a thick head of hair, I cannot tell it's there). So it happens that at time t0, in both worlds, I am under the belief that the hat really is on my head. In wH, I am correct, while in w~H, my belief is false. When I get to the table at time t1, I reach to take off the hat. In world wH I find the hat, and place it on the table. But in world w~H I find to my surprise that I am not wearing a hat after all.
Now assume, for simplicity, that my conscious state consists only in knowing what time it is, and whether I believe there is a hat in the world or not. Assume for now that I have only two possible beliefs: either I believe the hat is there or I believe it is not. I am not capable of reserving judgment or being uncertain about it. We will abbreviate these states of affairs as follows:
All of this remains to be worked out in detail, but for those familiar with computational systems such as the lambda-calculus, here is a hint as to how to think about it: H and ~H can be thought of as self-contained computer programs, with no inputs and no outputs. As such, each is a complete "universe" in its own right: they are metaphysical things-in-themselves. But B and ~B refer to qualities of wH and w~H. As such, they are like test-functions that can be applied to H or ~H to determine if the relevant partial aspect is there. Equivalently, they are facts derivable from H or ~H, but not facts from which H or ~H can be derived. I will use the Leibnizian terminology, and call H and ~H "monads". But to even view H and ~H as corresponding to computer programs, we need to develop the role that inference rules play in our system, which will not be addressed in this essay. So, for current purposes, consider all four symbols as atoms. The negation sign cannot at this time be considered a symbol in its own right. ~H is simply an atom, contained in some worlds and not others. There is as yet no concept of negation in our system.
Because H is a monad, we want to say that it is the world, not merely an aspect of the world, like B. H has this characteristic for some levels of analysis, but not when we consider a "world" to be a time slice. For then, the "world" is not a metaphysical thing-in-itself, but is itself only a partial aspect. It is the whole universe of worlds--the model--that is then considered analogous to a monad. So I will apply the word "monad" to the model-level, rather than the world-level. An atom that is in all worlds of a model could be said to be monadic with respect to that model. "Monadic" can be read either as "a consciousness-containing program in the lambda-calculus with no inputs or outputs", or "true in all worlds in a model". Unless otherwise specified, I will use the term in the former sense, since that is the more metaphysically important (and Leibnizian) sense of the word. The latter I will use only if it is specified what model the "monadicity" is with respect to.
You might object that these atoms cannot even be said to exist in more than one model, since they are defined by their model, with no structure of their own. However, if on a higher level of analysis, we can find a one-to-one mapping between the role of an atom in one model and an atom in another model, then they could actually be called the same atom.
Figure 2 shows the two worlds starting at t0: one in which I have on a hat, and one in which I don't. So far, we have been thinking of these two selves as two different versions of me (like the me that scratches his elbow and the me who doesn't). But since my consciousness "B + t0" is identical in both worlds, one could equally well say that at t0, the same person is occupying two different worlds simultaneously! When I reach the table at time t1, I find the hat in world wH, and now perceive it to be on the table. But in w~H, I find that, much to my chagrin, the hat doesn't exist at all! The result is that my belief changes from B to ~B and now our model contains two different people with different conscious states: consciousness "B + t1" occupies wH and consciousness "~B + t1" occupies world w~H (see figure 2).
This is a consciousness-based analysis, since we are defining two worlds with individual streams of consciousness. But other models are possible. Since wH(t0) and w~H(t0) are identical, if we subscribe to the Principle of the Identity of Indiscernibles, we can equate the two and construct a model with a single world at t0: wH~H = w(t0) in figure 3.
Here, time t0 has a peculiar status. H and ~H are both to be found here. Let's call this state of affairs H~H. But does that mean (H v ~H) or (H & ~H)? The answer is either or neither. Recall that the logical connectives do not yet have any meaning in our system. In figure 2, we would likely think of it as a disjunction across worlds (H v ~H), so it might make sense to think of figure 3 as containing a conjunction in one world (H & ~H). The former is a tautology, the latter a contradiction. Yet the actual state of affairs described is supposed to be the same!
But there is not as yet any notion of contradiction or tautology in our system. Certainly, it will be possible to define figure 3 as contradictory, but that depends on how we develop our logic. For now, consider wH~H a single world with both "a hat on my head" and "no hat on my head" somehow "superimposed" on each other. There is nothing contradictory about this in itself. It is only when developed in a particular logic that this might be considered contradictory. For now, just consider that we have a new atomic symbol: H~H.
Is H~H a new monad, to replace both H and ~H? Not in the important sense of conscious-identity, since it only holds at t0 and not t1. So H~H is only a partial aspect of the universe; H and ~H are still the monads. But we could take H~H as a monad with respect to a model that considered some partial aspect of both the w(t1) worlds to be important enough to justify unifying the two worlds into one. In figure 3, we could focus on the fact that both worlds at t1 contain some version of me. This is a more objective stance, that of "personal identity" rather than consciousness. Now things are switched around; H~H is the monad, and it is H and ~H that are partial aspects! So the extent to which something is considered a thing-in-itself depends on the model in which we are working. To make something a monad, eliminate all worlds in which it is not true from the model. If the resulting worlds all contain a single stream of consciousness, then this is a monad in the important metaphysical sense.
Figure 4 shows the three levels of analysis we have talked about. Individual worlds contain more and more of reality as we go up the figure from the bottom to the top. A traditional idealist goal is to show how it can all be unified into one Absolute world of which all other worlds, including ours, are partial aspects. 
Figure 5 expands the analysis from figure 4 into the larger picture. Below the temporal level, the universe can be further analyzed into individual subatomic particles (the symbols shown in the diagram stand for photons, electrons and quarks). At the very bottom of the diagram is the level of individual particles, where we have taken the extreme step of considering, not just each time slice, but each individual particle, as its own "world".
Fig.5: The big picture.
Earlier, we used the Principle of the Identity of Indiscernibles to unify two worlds into one, based on the fact that they contained the same consciousness. Since it is a well established fact of quantum physics that any two electrons are indiscernible, we can do the same sort of thing here. This gives another level of analysis where all individual electrons in the universe are considered as the same electron (the level of particle types). The universe now contains a single electron, zipping backward and forward in time. In fact, at this level of analysis, the time slices have no meaning. Movement "backwards in time" is no different than movement to the left or right. Richard Feynman  gives a very accessible description of this slightly twisted way of looking at electrons. It may be twisted, but it is in complete accord with the quantum equations.
Higher up in the diagram, we find the level of our observable universe--the level of conscious-identity (or strong personal-identity). Higher still, we find levels of analysis that, one might say, "transcend" our own particular universe. This transcendence is not some mystical notion of a rationally inaccessible higher plane of reality, like that imagined by the subjective mentalists or the material realists. The "higher reality" here is nothing more than Turing Computability or Universal Language--"whatever can be spoken or thought of". Ultimately, everything underneath this in the diagram is simply an enumeration of all possible computer programs. So the Absolute Monad would be a computer program that produces and runs every possible computer program, including itself. Such a program would, of course, never stop running. 
Now let's go back to our primitive stick-man example, and look at the model we get when we flatten out figure 4 into a single model (figure 6). We could (fancifully) add in our Absolute Monad, and it would be true in all worlds in figure 6. If our system were the extent of all possible analysis of anything, then this would be the final goal of idealist metaphysics achieved. But of course, our system falls far short of that! So I have not included the hypothetical Absolute Monad, but the model is nonetheless highly objective compared to the more subjective models that came before. The absence of an absolute monad means that nothing is true in all worlds--this "universe" is just a collection of partial aspects. This makes it metaphysically incomplete, but that is as it should be--a simplistic little model of a stick-man with a top-hat ought to be inadequate, obviously. In fact, the model we have constructed does not really include a stick-man or a hat. These were just external interpretations, introduced as aids to the imagination--fanciful ways of talking about the very simple mathematical system shown in figure 6. In any case, complete and total objectivity is rarely desired. Usually, we want to analyze something in terms of some particular restricted model, which is inevitably incomplete from the point of view of some more objective stance.
Fig.6: A relatively objective view (Leibnizian monads are checked).
The method of rationalism has always been to accept reason and question experience. Descartes' Cogito asks us to take only the most basic given (consciousness) as empirical fact. Although the universe of the stick-man and his hat is strange, modern physics has revealed something the nineteenth century idealists would not dared have hoped for: empirical evidence that the world really does work this way! In fact, quantum theory's infamous Schrödinger's Cat thought experiment  is identical to the hat example in all the essentials. Quantum theory allows any possibility to occur (within certain arbitrary boundary conditions--for us, the conscious-identity version of the Anthropic Principle). In the hat example, only two possibilities exist at t1, so at t0 we would say the "experiment" had a 50% chance of one result and a 50% chance of the other. At t1, we would experience a "collapse" of this "superposition of states" into one state. But from a more objective stance, this would still be one state where both results have occurred. I do not experience this more objective stance, since it is above the level of individual consciousness and includes two differently conscious versions of me.
In real life, of course, a hat's description is made up of many sub-programs; it is a complex beast, inter-related with my brain and the rest of the world in a complex way. So it takes only a tiny change in the hat to affect my consciousness, and the superposition of states would be retained for only a very short time (such systems are called "chaotic"). In our thought experiment, the hat behaves in a weird quantum manner because we pretended that a hat could exist at the level of the base resolution, or granularity, of the universe. The hat is a "quantum" thing. It has no inner complexity. It is connected up with the rest of the world in a very simple, nonchaotic way. So a great disturbance can be made to the hat (such as its disappearing) without affecting consciousness (except, of course, that it was just a fantasy that consciousness could exist in such a simplified universe in the first place!).
In the real world, however, there are objects close to the fine resolution of the universe, which can be modified in significant ways without affecting a nearby human's state of mind. Electrons, for instance, have a common habit of being in multiple places at the same time (or, according to the Many-Worlds interpretation, occupying more than one world at the same time), as well as disappearing from one location and reappearing at another, without travelling the distance in between. The quantum world is very much like the idealist's universe. Even the table's turning into a pink bunny-rabbit is allowed in quantum theory. In fact, such a possibility is automatically part of the quantum wave equation simply because it is a possibility. Tables turn into rabbits in only a vanishingly small percentage of worlds, however, so (although it happens in some worlds) we never observe it in practise.
Perhaps the kind of modal analysis sketched out in this essay could eventually show that all the strangeness of quantum mechanics really does follow directly from a rationalist metaphysics. Much work remains to be done, however.
But what if we could a priori eliminate worlds in which we die as improbable (or worlds in which we prematurely die, if some kind of death is at some point a logical necessity)? One might argue for this on the basis that such short-life-span worlds will eventually be weeded out at the moment of "death" anyway. In this theory, the only worlds that have much probability in the first place are those in which we avoid death "naturally", without the need for fluky, improbable "miracles". This would be to take the "true" stream of consciousness view, only considering streams that are complete from the beginning of life to the end (if any). If that could somehow work, we could posit some kind of "maximum life-span" principle--that longer lives are more probable, in the space of all possible lives, than shorter lives. If so, we can expect medical science to provide us with indefinite life extension before we reach the age where miraculous flukes would be required to keep us alive. This might mean that the most probable time to be born in the history of the universe would be right around the technological cusp where science becomes advanced enough to extend life indefinitely. Someone who was born much before the cusp would have a life too short to be very probable. Someone born much later would experience a shorter life solely by virtue of having been born later (at least, if eventual death is a logical necessity--if instead life were literally eternal, the person born later would always and forever be younger than the person born earlier, but the total life-spans of both would be infinite, making it unclear, to me at least, what the probabilities would be). If the universe eventually ends, it would mean that eternal life is a logical impossibility, but that we can expect to survive to the bitter end (billions of years instead of mere decades). The life-span of the universe would thus be roughly the maximum possible life-span of a conscious being. The maximum life-span principle would explain why we seem to have been born at such a fantastically pivotal time in human history. This theory would also allow a proof for immortality (or more precisely, maximum logically possible life-span), one that does not require thermodynamics to suddenly work differently when we get old, and would allow us to live indefinitely without being in a decrepit state for perhaps vast periods of time.
On the other hand, I'm not sure how such a theory could really work. If we base our philosophy on the Cogito, we can only take our present consciousness as our starting point to reason about such things. To take the more objective, but not fully objective, view of true stream of consciousness seems arbitrary. It simply provides us with the most comforting, easy-to-swallow view of reality (which, of course, has nothing to do with whether it is true). The probabilities, being subjective, must surely be based on one's current conscious state. That leads us back to either the "eons of barely-alive decrepitude" theory or the "it's logically impossible, given our current condition, to live much past one hundred" theory. The first seems suspicious because it would violate the laws of thermodynamics (at least under their usual scientific interpretation). The second seems implausible considering the truly enormous variety of logical possibilities under consideration. Yet the true stream of consciousness model seems to be arbitrary wishful thinking.
On the other hand, it may be completely fallacious to extropolate our current consciousness into the future like this. So far, my reasoning has gone something like this: "the only futures that are relevant to me-now are those for which me-now is a valid past state. Therefore, those are the ones I will use to talk about my future probabilities, even if they are all wildly improbable." However, there is no metaphysical principle tying you to these futures as opposed to other futures of selves that are close to being future yous but not quite, or for that matter selves that are nothing like you. If we are as strict as we possibly can be about applying the Cogito, all we can really do is presume that our current conscious state is a reasonably probable one out of the set of all conscious states (including ours and all other possible creatures). So it is not really legitimate to declare quantum miracles highly probable, just because they are required to maintain a future self that is literally our future. All we can really say is that certain consciousnesses are more probable than others, and one in which quantum miracles had to occur is highly improbable, and so one should not expect to ever find oneself in such a state. So a more realistic view of the future is simply to realize that it involves many possible consciousnesses, and that some of these will satisfy our arbitrary concept of "me" and some won't. But whether we focus on those that do, or not, it is not sensible to view our future as consisting of quantum-miraculous selves. If we choose to focus on future consciousnesses that are very much like "me-now", we are free to do that, of course, so long as we choose relatively probable ones. It is thus more reasonable to extropolate oneself into the future by imagining a self that is almost but not quite consistent with one's current self, but that has high probability, than by imagining--as we were earlier--a more strictly consistent future self that has extremely low probability.
This gives us reason not to worry over being faced with an incredibly improbable eternal state of decrepitude as opposed to being saved by medical science, but still does not tell us how to view the future if there are no probable futures containing us, no futures where medical science saves us without any miracles. Am I then to conclude that the most probable future either contains a future me or a future somebody else? There seems to be something wrong with that. While it may seem reasonable to extropolate into the future towards a self that is almost a future me-now, extropolating into someone who is just a different person altogether is plain silly. So are we not then stuck with extrapolating into the highly improbable quantum-miraculous future-me, if we are to extrapolate at all? Perhaps we shouldn't extrapolate at all, if that means a future that is miraculous. A miraculous future is also quite apt to be very unstable, and unpredictable. Such a future might be more like a half-conscious dream-state with no clear linear direction of time, than anything like a solidly envisionable future in something we would call a "world" , and what sense does it make to extrapolate into such a state and call it the "future"? In the end, of course, one must realize that it is an arbitrary decision what one chooses to extrapolate into in the first place, and all that science really need provide us is a reasonable explanation for who we are right now. In any case, it seems likely to me that a miraculous future self would lack many of the features required for a stable self identity. I also currently see no way to really prefer longer life-spans as more probable than shorter ones, an assumption that it seems to me would be simply wishful thinking, since it would require calculating probabilities from the viewpoint of an entire life-span, and that seems to violate the spirit of the Cogito. Nonetheless, the idea may still have some merit, and is an intriquing one. Still, I see no reason to currently make any bold claims as to the mortality or immortality of the soul based on quantum physics. We may be immortal--in which case it will probably be medical science that saves us, not a mystical spiritual soul or quantum miracles--or we may be quite mortal and destined to die. A more complete understanding of quantum mechanics as a rational theory, which we have only just begun to explore in this paper, may help in all this; it may well be that the tools to properly answer the death question are just not at hand in our current scientific arsenal.