How Do We Persist Through Time?

Almost all human persons have the intuition that we continue existing through time. Though we may colloquially say “I was a different person back then,” or “I will be a new woman moving forward,” we do not mean to say in any serious, substantial, or metaphysically meaningful sense that the person we were in the past or will be in the future is a qualitatively different human being than the person we are now. Without the belief that personal identity persists through time, such things as punishment for past crimes, property law, and reconnecting with an old friend, would not make sense. Because this is such a deeply held intuition, it has received a substantial amount of philosophical attention spanning back to the ancient period.

Though there are several proposals for what constitutes one's continued existence through time (a.k.a. the persistence conditions of transtemporal identity), the most popular are (a) psychological continuity views and (b) brute physical relation views (Olsen 2019). The former holds that an individual's identity, and therefore their continued existence, depends on some form of psychological continuity. This may be continuity of first‐person memory, of consciousness, or of some other psychological characteristic. Psychological continuity views preserve the intuition that we would survive if our consciousness were uploaded to a computer or if we woke up in a different body, but seem incompatible with the claim that I was a fetus or could ever enter into a persistent vegetative state, as neither of these involve consciousness in any recognizable sense.

Brute physical relation views propose physical persistence conditions, that is, an individual continues to exist if their body exists, and their existence can be traced back to each stage of their development (including prenatal). Generally, these views center around the claims that we are biological organisms, namely human animals, and our selfhood is inextricably bound up with our physical bodies. Whilst physical (sometimes, “animalist”) views can comfortably accommodate the intuition that I was once a fetus, they are less compatible with the idea that a person could continue to exist in some futuristic setting in which their consciousness was separated from their body and their body destroyed.

One way that personal identity and persistence conditions are explored is through thought experiments that ask whether identity would be preserved in a range of scenarios involving altering or removing various characteristics. One such set of cases in the literature is known as “fission cases,” which considers whether a person's identity could be preserved after splitting into two (or more) persons. Without MWI, fission cases are interesting but currently unrealizable philosophical exercises. MWI, however, implies that fission (albeit in a slightly different sense) may have been part of reality since long before philosophers existed to contemplate it.

Fission Cases

We now know that a person can survive the removal of one brain hemisphere with their psychological characteristics largely intact. Historically, philosophical discourse on fission cases has played on this idea, imagining what might happen if each hemisphere of an individual's brain were transplanted into two host bodies, resulting in two present persons (often referred to as lefty and righty) who are each psychologically continuous with one past person (Parfit 1984, chapter 12). I will begin by considering fission as simply the splitting of an individual into two persons, each of whom is psychologically continuous with the prefission individual.

The problem is as follows: let a be a being who will at some future time t undergo fission, and let l and r be those two persons who exist post fission. After t, is a identical to lefty, righty, both, or neither? The physicalist must conclude that a has ceased to exist, as a’s original body is no more. However, both lefty and righty are psychologically continuous with a. If only one of them existed, it would seem like a straightforward case of continued existence according to the psychological continuity theorist. But how can we make sense of identity when there are multiple persons who have an equal claim to being a?

According to Leibniz’ law, if I am numerically identical with some object, I must share all the same properties with that object. Thus, if a is numerically identical with lefty and numerically identical with righty, meaning a shares all the same properties as lefty and righty, then, following the transitivity of numerical identity, lefty and righty must be numerically identical. Quite clearly, though, lefty and righty are not numerically identical, as they occupy distinct spatial locations, presumably have different streams of consciousness, and can operate entirely independently of one another. Thus, they do not share all the same properties, and so they are different persons. Another way of putting it is that, in the words of Derek Parfit, ‘identity is a one‐one relation,’ and so collapses into incoherence when branching occurs (Parfit 1971, 10). Fission, then, is sufficient for a to cease existing.

Bodily continuity theorists are generally happy to accept this conclusion—the destruction of the original body constitutes a’s death. Perhaps matters are a little more complicated if the new body is an exact copy of the old body, but strictly speaking the destruction of the original body is a sufficient condition for a’s death. Things are more challenging for psychological continuity theorists, however, as they seem forced to conclude that both lefty and righty are identical to a by virtue of having psychological continuity with a, violating Leibniz’ law. Two of the solutions to this problem that have been offered are the so‐called multiple occupancy view, and the no‐branching view (n.b. the “no branching” view refers to the branching of persons within a world in standard fission cases, and is not to be confused with the branching of worlds that we will come to shortly) (Olsen 2019).

The multiple occupancy view, in brief, claims that if there is fission in your future then there are multiple occupants in your body up until the fission occurs. Before t, a was actually lefty + righty. Both lefty and righty are psychologically continuous with a, as before t both lefty and righty were multiple occupants of a. Therefore, their identity is preserved post‐fission. The no‐branching view, on the other hand, holds a's continued existence depends on any future persons’ psychological continuity with a, and the added caveat that no other being is psychologically continuous with a. Therefore, the no‐branching view claims that before t, a exists, and after t, a has ceased to exist. Instead, lefty and righty have come into existence. If just righty had come into existence, sharing psychological continuity with a, then righty would constitute the continued existence of a. Once lefty is introduced, however, a has ceased to exist, as an individual cannot survive a branching event. In short, fission is death.

Who Am I in Many Worlds?

Whether any of the above solutions are feasible in MWI is a highly important question. A brief defense of the multiple occupancy view, developed from ideas first presented by David Lewis (1976), is offered by Simon Saunders and David Wallace. Lewis claimed that the problem of continued identity post‐fission can be solved by claiming that each person is made up of an aggregate of temporal parts or stages. On this view, if an individual will at some future time undergo branching into two copies, then there are timelessly two people, and the pre‐branching individual's person stages are shared by both. One can attribute two sets of thoughts to these person stages, as there are two persons who later branch. So, at any one time, each “person” actually contains a multitude of overlapping person‐stages whose futures will later diverge.

Although Saunders and Wallace acknowledge that Lewis’ metaphysical view ‘has won few supporters’, they nonetheless propose applying it to MWI in the form of a semantic argument (Saunders and Wallace 2008, 295). Their aim is to address a common critique that MWI cannot accommodate probability and uncertainty particularly regarding what an individual should expect their future to hold (as everything physically possible happens, and these events are experienced by many versions of us). They propose that person stages are unaware of which future they will end up in as they cannot know which future branch is theirs, and that this solves the problem with accommodating uncertainty and probability within the Everettian picture. They argue that this use of indexical uncertainty, namely uncertainty about one's contextual location in future branches, recovers the notion of probability necessary for MWI to reflect quantum observations in which uncertainty and probability are incontrovertible.

Importantly, their primary concern is to develop a working semantics, not a robust metaphysic of transtemporal identity (Wallace and Saunders 2008, 303). This is because Saunders and Wallace deny that there are metaphysical truths about personhood and identity other than those fixed by our ordinary usage. Whilst they are not making robust metaphysical claims about identity, their argument is an example of how the multiple occupancy view might play out when talking about identity in an Everettian context. For the robust metaphysical argument, one should look to Lewis.

Parfit (rightly, in my view) notes that the multiple occupancy view “greatly distorts the concept of a person,” and should be rejected (Parfit 1984, 256). As I see it, whether the multiple occupancy/person stage view works for bifurcation is somewhat by‐the‐by—it collapses into absurdity in the context of MWI due to the sheer scales involved. In a recent popular book on MWI, physicist Sean Carroll writes:

We don't know how often branching happens, or even whether that's a sensible question to ask … but we do know that there is a lot of branching going on; it happens every time a quantum system in a superposition becomes entangled with the environment. In a typical human body, about 5,000 atoms undergo radioactive decay every second. If decay branches the wavefunction in two, that's 2⁵⁰⁰⁰ new branches every second. (Carroll 2021, 119–20)

If branching occurs as frequently as this, then at any one time I am not one person. Instead, I contain a non‐denumerable infinity of persons/distinct person stages and am woefully mistaken in believing that I am one person whose experiences and perspectives are unique. This is both extremely difficult to square with our intuitive understanding of our own identity and hard to make metaphysical sense of—we would need extremely compelling reasons to accept such a conclusion.

Not only would we contain incomprehensible multitudes, but we would not even know who we are at any one time. Which of the multiple occupants or overlapping stages is me, if only one can be? There is no good answer to this question. Paul Tappenden agrees, writing: “an utterance of ‘I’ in Lewisian contexts of multiple utterance cannot straightforwardly be assumed to refer to the utterer. The semantic rule [‘the word “I” refers to the speaker in any sentence in which it occurs’] cannot be adopted without further ado” (Tappenden 2008). There is no means by which individuals can reliably refer to themselves, and not some other person‐stage that will later branch off. Thus, the word “I” cannot have self‐referential certainty. This is a significant problem for the multiple occupancy view.

Moreover, claiming that at each moment of our lives we contain perhaps an infinite number of overlapping persons who share our thoughts and experiences stretches our reasonable, common sense, understanding of identity beyond breaking point. Wallace and Saunders write that “intuition, if we are talking of physical discoveries, doesn't come into it” (Wallace and Saunders 2008, 302), and “our intuitions about what is ‘reasonable’ or ‘imaginable’ were designed to aid our ancestors on the savannahs of Africa, and the universe is not obliged to conform to them” (Wallace 2010, 69). Yet I argue that (at least for the theist) our intuitive understanding of personal identity as a one‐one relation between individuals and themselves (and not a one‐many relation between an apparent individual and a nondenumerable infinity of person stages) must be in some sense meaningful. We should not need to abandon the idea that “I” is a word by which we can, with certainty, refer to ourselves, unless provided with overwhelming evidence for doing so. Such evidence is not, in my view, available at present.

Happily, the no‐branching view avoids these problems of overpopulation and self‐referential uncertainty. Rather more unhappily, however, it implies that I cease to exist each time branching occurs. The postsplit persons are not me; they are new persons who have stepped fresh into existence as I was destroyed by the splitting of worlds. Yet this too feels totally alien to our intuitive understanding of our own identity as it leads to the conclusion that the six‐year‐old child I remember being is (a) not me and (b) is dead. Though I am psychologically continuous with the child, she has not survived the many branching events that have occurred in the twenty years that divides us. Moreover, the no‐branching view holds that each person with whom I am psychologically continuous only exists for a very short time (if the above Carroll quotation is anything to go by, it may be as little as a fraction of a second). On this view, I am not even the same philosopher who started writing this article, even though I currently hold the ideas in my head, remember doing the research, and feel as though I have been writing it the whole time.

Even the bodily continuity class of persistence conditions cannot help here, for presumably individuals do not have the exact same body before and after the split. Before the split, there was one body; after the split, there are at least two. Much in the same way as the post‐split persons have psychological continuity with the pre‐split person without having the exact same mind, so too will the post‐split persons feel as though they have bodily continuity although (following Leibniz’ law) it cannot be the exact same body. Clearly, whatever one's views on persistence conditions are, Everettian branching presents significant problems for personal identity.

One possible solution is offered by Parfit, who argues that philosophers are mistaken in thinking that personal identity is what is most important in fission cases. We should not ask “am I identical with some past or future person,” but “have I survived?” Whilst identity is an all‐or‐nothing phenomenon, most of the relations important to survival admit of degrees (Parfit 1971, 11). Parfit argues that after t, a is identical neither with lefty nor righty, but that this does not matter. What we should care about is survival, even if the being that survives is not numerically identical with us. Furthermore, “we ought to regard division as being about as good as ordinary survival” (Parfit 1984, 261).

Carroll echoes this, writing that the idea that each of us is the exact same person from birth to death was never more than a useful approximation, so it should not be too concerning that MWI encourages us to abandon it. On MWI, we must accept that “the lifespan of a person should be thought of as a branching tree, with multiple individuals at any one time, rather than a single trajectory – much like a splitting amoeba” (Carroll 2021, 139–40). There are many versions of us, each of whom can be traced back to an original person with whom all subsequent persons are psychologically continuous. So, on this view we survive, even if we need to reconceptualize exactly what this survival looks like.

Tappenden's solution is to adopt the “stage view” of personal identity, developed by Theodore Sider, whereby persons are not an aggregate of person stages many of whom can be within a single individual at a given time (à la the multiple occupancy view) (Tappenden 2021, 2.1). Instead, persons are stages, and one individual contains maximally one person stage at any one time (Sider 1996, 441). Therefore, any utterance at a time has a token which is associated with the unique body, itself a stage, which is the body of that person at that time (Tappenden 2008).

This view is similar to perdurantism whereby persons are comprised of temporal parts and are thus never wholly present at any moment in their lives (contra eternalism in which persons persist in their totality through time). It differs, however, in its more radical claim that persons are not comprised of stages; they are stages (Sider 1996, 433). In applying this to MWI, Tappenden argues that an individual survives branching by being continuous with pre‐branching temporal parts (viz. past temporal counterparts), and individuals who branch off each other share past parts but are not identical post‐branching (viz. they do not share future temporal counterparts). Although detailed consideration of this view falls outside the scope of this article, I encourage the interested reader to turn to Tappenden for further discussion. My own feeling is that this is the most plausible option currently available.

Much has been covered and very little solved in the preceding paragraphs. If we take a no‐branching or bodily continuity view, then the fission of many worlds is death. If we accept the multiple occupancy view, then we can survive branching but at a high price. We must dispose of our intuitions about the uniqueness of our identity, and we can no longer use the term I with self‐referential certainty. If we take Parfit's view, however, then we need not worry about fission cases of the kind incurred by MWI, as “the relation of the original person to each of the resulting people contains all that interests us – all that matters – in any ordinary case of survival” (Parfit 1971, 10). Tappenden's view also allows persistence, whilst preserving our intuition that identity is both singular and self‐referentially nonarbitrary.

Alleviating these worries is required to preserve the internal coherence of classical theism in the context of MWI. Theism holds that individuals have personal relationships with their God that are developed over their course of a lifetime. It also holds that individuals are held morally and eschatologically responsible for their actions in this life and the next. For this to remain coherent, it must be the case that individuals can continue to exist throughout the various moments of their lives regardless of branching they may or may not undergo. If not, it seems deeply unjust to hold an individual responsible for sins committed by (a numerically nonidentical) earlier version of themselves. It also becomes hard to understand how an individual can cultivate and develop their ongoing relationship with God throughout their lifetime. In addition, whatever issues are raised for personal identity correspond with concerns regarding the soul. Can souls survive branching? Do they split? Do we contain multiple souls at any one time? These are highly important open questions.

Moreover, solving the personal identity problem in MWI is also important christologically. All the aforementioned issues relating to personal identity may be applied mutatis mutandis to the personal identity of Christ. If Christ cannot survive branching events, meaning he is not the exact same person at various points in his life, then this is problematic for the Christian. If the Christ crucified is not the Christ risen (because identity cannot be preserved after world‐branching) then lots of theological problems arise regarding the eschatological promise of resurrection typified in the resurrection of Christ. Moreover, according to the Chalcedonian definition, Christ is the divine Logos incarnate. Either (1) Christ only exists in one branch (the branch in which the human body of Christ in that branch contains the unique divine Logos) meaning all other branches contain “zombie” Christs and the people in those branches are, in fact, worshiping idols, or (2) Christ's Logos fragments into an infinite number of pieces as it undergoes the multitude of splitting events that occur during Christ's lifetime. Each of these possibilities is theologically troubling. However pressing these problems are, it is not within this article's scope to examine it in addition to the above discussion of personal identity. The salient point is this: if numerical identity cannot be squared with MWI, then theism faces serious problems. It is, therefore, of the utmost importance that the theist can make sense of the aforementioned issues surrounding personal identity in an Everettian context.

Horrendous Evils and Maximal Suffering

The problem of evil holds that the God of classical theism possesses the attributes of omniscience, omnipotence, and perfect goodness, and therefore should be aware of evil, able to prevent or eliminate evil, and motivated to prevent or eliminate evil. Yet, evil exists. Therefore, the existence of a creator God with the aforementioned attributes is problematized. In this section, I argue that MWI raises both new and an enlarged versions of this problem to which fresh responses are required. I will develop both the new and enlarged problems of evil in dialogue with Marilyn McCord Adams and Stuart Sutherland's article “Horrendous Evils and the Goodness of God.” Adams and Sutherland identify a particular class of evils, the existence of which is especially problematic for the existence of the God of Christian theism, which they name horrendous evils. These are: “evils the participation in (the doing or suffering of) which gives one reason prima facie to doubt whether one's life could (given their inclusion in it) be a great good to one on the whole” (Adams and Sutherland 1989, 299). In other words, horrendous evils are evils so intensely devastating that experiencing them leads to doubt about whether one's life was worth living. From such evils, individuals may never recover.

On MWI, there must be versions of me living out the worst possible iteration of my life, including participating in horrendous evils. Everything that is physically possible (i.e., compatible with the initial conditions and the evolution of the universal wavefunction in accordance with the Schrödinger equation) happens, meaning that there will be versions of me suffering acutely from genetic diseases brought on by quantum effects on genetic mutation (Al Khalili and McFadden 2014) and versions of me in branches where many other forms of suffering abound. This problem has a relative dimension, insofar as those individuals are living a life that is (by comparison) far worse than other possible versions of their life. It also has an absolute dimension—those bad versions of an individual's life emerge by necessity out of the structure of the Everettian universe. If it is physically possible for versions of me to participate in evils, horrendous or otherwise, then there are versions of me in some branches who are participating in these. If Adams and Sutherland are right, then these versions of me are suffering to such a significant degree that the positive value of their life is utterly engulfed. There is reason to worry about this from an individual's perspective insofar as horrendous evils will be in at least one of their futures; there is also reason to worry about this from a God's eye perspective as there will be far more (beloved) individuals experiencing acute suffering if Everettian QM accurately describes reality. This is a state of affairs, I argue, that a loving God should want to prevent.

Whilst an optimistic theist may wish to invert the problem and claim many benefits to Everett's view (i.e., there are many versions of you living good, and some even living the best, versions of your life possible), the benefit to some is, I argue, vastly outweighed by the price others must pay. Morally, joy and suffering are not equal opposites—it is not justifiable to cause someone intense pain for no reason other than so you can feel an equally intense pleasure. Morality simply does not work like that.

Adams and Sutherland draw an important distinction between two descriptions of God that is helpful here. On the one hand, God is a provider of global goods, and on the other hand, God is a loving parent of each individual creature (Adams and Sutherland 1989, 302). Generic, global solutions to evil (e.g., a free will defense) cannot solve individual problems because they ignore the latter description of God as a being who loves and cares for each individual person. Any such argument regarding a global distribution of goods and evils ignores the plight of the vastly increased number of individuals who suffer greatly, recognizing only the global provider aspect of the divine nature and ignoring that God is supposed to love each individual person. Though suffering may be balanced by the abundance of good on a global scale, for those individuals who suffer most, MWI is unimaginably cruel. Echoing the oft‐referenced sentiment of Ivan Karamazov, a world that allows an innocent child to suffer in order that some other good is obtained is an abhorrent world in which we should not want to live. ⁵ We ought not to take lightly the fact that so many more people exist to suffer acutely on MWI than in a singular universe, even if MWI contains a correspondingly greater amount of good. This is the enlarged version of the problem of evil.

The Problem of Evil as a Cosmological Problem

It is the case, however, that even if we inhabit a singular universe there exists an inordinate amount of suffering. Therefore, one might reasonably claim that whilst the problem of evil may be enlarged on MWI, we already have a rather large problem to contend with. Perhaps the Everettian world‐view's contribution does little to reshape the contours of this ancient and persistent debate. For this reason, I wish also to set out how MWI raises the problem of evil in a new way, building on the arguments of the previous section. As aforementioned, suffering follows from the very structure of the Everettian world/s. It is no longer a contingent matter whether one's future will contain horrendous evils, extreme suffering, or profound loss. Somewhere in the Everettian world/s, there will be versions of you experiencing all these things. This gives a new string to the problem of evil's bow—if God exists then God chose to instantiate a particularly cruel set of physical laws that makes suffering emerge, by necessity, for a great many persons.

One can reasonably postulate that an omnipotent God could have created an entirely different set of physical laws to avoid this problem altogether. It seems that there are conceivable universes that obey quantum laws which an omnipotent God could have created that do not branch off ad infinitum creating a multiplicity of individuals many of whom are necessarily suffering greatly. Robert Russell calls arguments from, and responses to, evil that concern the very structure and laws of the universe “cosmic theodicies,” as they elevate the problem of evil to one of cosmology (Russell 2007, 124–25). Cosmic theodicies claim that we cannot blame God for certain evils as they are necessary consequences of some structural component of the universe. Russell rightly rejects the claim that God did not have a choice about certain fundamentals of nature (e.g., whether the universe branches or not) when creating the world. Surely, Russell asks, a truly omnipotent God could have created another type of universe altogether, in which the cruelties of our own could have been avoided?

Cosmic theodicies only push the problem back one step. Instead of God being directly responsible for individual instances of suffering, God is indirectly responsible for suffering as God created the features of the universe which cause that suffering. Though God may not be directly responsible for the suffering of the unlucky Emilys, God is responsible for creating a universe in which the suffering of some Emilys (though it is initially arbitrary which ones) is inevitable. Indeed, if we can survive branching, then maximal suffering is necessarily in (at least one of) the future(s) of every single person. Why a loving God would choose such a design is an open question.

Though I have argued that this is a new version of the problem of evil, it bears important similarities to another theodical problem that has received recent attention in the science‐and‐religion literature: the problem of evolutionary theodicy. Both problems are grounded upon the claim that God chose morally reprehensible (or, at least, morally questionable) laws of nature from which evil and suffering follow by necessity. Just as the theory of evolution by natural selection contains intense suffering and vast waste for speciation to occur, so too do the laws of Everettian QM entail extreme suffering for a number of unfortunate individuals. Perhaps what we need in light of the previous discussion is an Everettian theodicy, which may resemble an evolutionary theodicy of the kind developed by those working in the biologically focused areas of science‐and‐religion and the philosophy of religion (Southgate 2008; Sollereder 2019). We are now able to turn to the final worry, which relates to both Christ and Salvation.